Froth flotation, a century of innovation.pdf

Edited by

Maurice C. Fuerstenau Graeme Jameson and

Roe-Hoan Yoon

Published by

Society for Mining, Metallurgy, and Exploration, Inc.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

BLANK LEFT PAGE

Society for Mining, Metallurgy, and Exploration, Inc. (SME) 8307 Shaffer Parkway Littleton, Colorado, USA 80127 (303) 973-9550 / (800) 763-3132 www.smenet.org SME advances the worldwide mining and minerals community through information exchange and professional development. SME is the world’s largest association of mining and minerals professionals. Copyright © 2007 Society for Mining, Metallurgy, and Exploration, Inc. Electronic edition published 2009. All Rights Reserved. Printed in the United States of America. Information contained in this work has been obtained by SME, Inc., from sources believed to be reliable. However, neither SME nor its authors guarantee the accuracy or completeness of any information published herein, and neither SME nor its authors shall be responsible for any errors, omissions, or damages arising out of use of this information. This work is published with the understanding that SME and its authors are supplying information but are not attempting to render engineering or other professional services. If such services are required, the assistance of an appropriate professional should be sought. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Any statement or views presented here are those of the author and are not necessarily those of SME. The mention of trade names for commercial products does not imply the approval or endorsement of SME. ISBN-13: 978-0-87335-280-2

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page iii Tuesday, January 2, 2007 7:36 PM

Contents CONTRIBUTORS PREFACE PART 1

v

ix

HISTORICAL ASPECTS OF FLOTATION

1

A Century of Developments in the Chemistry of Flotation Processing History of Flotation Technology PART 2

FLOTATION FUNDAMENTALS

FLOTATION CHEMISTRY

95 133 179 227 259 283 339

373

Flotation Reagents—A Critical Overview from an Industry Perspective Sulfide Mineral Flotation Flotation Chemistry and Technology of Nonsulfide Minerals Depressants in Nonsulfide Mineral Flotation Flotation of Precious Metals and Their Minerals Coal Flotation PART 4

FLOTATION CELLS, MODELING, AND SIMULATION

FLOTATION PLANT PRACTICE

869 iii

425 465 555 575 611

637 681 739 757

779

Plant Practice: Sulfide Minerals and Precious Metals Plant Practice: Nonsulfide Minerals INDEX

375

635

Mechanical Froth Flotation Cells Column Flotation Optimal Designs for Homogeneous, Countercurrent Flotation Processing Networks Modeling and Simulation of Industrial Flotation Processes PART 5

65

93

Some Aspects of Flotation Thermodynamics The Nature of Hydrophobic Attraction Forces Adsorption of Surfactants and its Influence on the Hydrodynamics of Flotation Pulp and Solution Chemistry The Physics and Chemistry of Frothers Surface Characterization and New Tools for Research The Flotation of Fine and Coarse Particles PART 3

3

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

781 845

flotation0.book Page iv Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page v Tuesday, January 2, 2007 7:36 PM

Contributors N.A. Abdel-Khalek Central Metallurgical Research and Development Institute (CMRDI) Cairo, Egypt

S. Chryssoulis Advanced Mineral Technology Laboratory London, Ontario, Canada Jan J.I.R. Cilliers Department of Earth Science and Engineering Royal School of Mines, Imperial College London, England, United Kingdom

Hans Allenius Outokumpu Technology Minerals Espoo, Finland Armando C. Araujo Department of Mining Engineering Federal University of Minas Gerais Belo Horizonte, Brazil

William Ducker Particulate Fluids Processing Center Faculty of Engineering The University of Melbourne Victoria, Australia

Barbara J. Arnold PrepTech Inc. Apollo, Pennsylvania

Robert C. Dunne Newmont Australia Ltd. West Perth, Western Australia

Seher Ata Center for Multiphase Processes University of Newcastle Callaghan, New South Wales, Australia

A. El-Midany Central Metallurgical Research and Development Institute (CMRDI) Cairo, Egypt

Cesar I. Basilio Thiele Kaolin Company Sandersville, Georgia

Hassan El-Shall Center for Particle Science and Technology University of Florida Gainesville, Florida

Trevor Bilney Kanowna Belle Gold Mine Boulder, Western Australia W.J. Bruckard CSIRO Minerals Clayton South, Victoria, Australia

Jan Christer Eriksson Department of Chemistry, Surface Chemistry Royal Institute of Technology Stockholm, Sweden

Subhash Chander Department of Energy and Geo-Environmental Engineering Pennsylvania State University University Park, Pennsylvania

K. Fa Department of Metallurgical Engineering University of Utah Salt Lake City, Utah

v

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page vi Tuesday, January 2, 2007 7:36 PM

Mike Fairweather M.J. Fairweather & Associates Rossland, British Columbia, Canada

Stephen Grano Ian Wark Research Institute University of South Australia Mawson Lakes Campus South Australia

James A. Finch Department of Metallurgical Engineering McGill University Montreal, Quebec, Canada

Michael Habner Kalgoorlie Consolidated Gold Mine Kalgoorlie, Western Australia

Daniel Fornasiero Ian Wark Research Institute University of South Australia Mawson Lakes Campus South Australia

Gregory J. Harbort Julius Kruttschnitt Mineral Research Centre Indooroopilly, Queensland, Australia Martin C. Harris Department of Chemical Engineering University of Cape Town South Africa

Eric K.S. Forssberg Division of Mineral Technology Lulea University of Technology Lulea, Sweden

Thomas W. Healy Particulate Fluids Processing Center Faculty of Engineering The University of Melbourne Victoria, Australia

Douglas W. Fuerstenau Department of Materials Science and Engineering University of California Berkeley, California

John A. Herbst Metso Minerals Optimization Services Colorado Springs, Colorado

Maurice C. Fuerstenau Department of Materials Science and Engineering University of Nevada Reno, Nevada

Ronaldo Herrera-Urbina Chemical Engineering and Metallurgy University of Sonora Hermosillo, Sonora, Mexico

A.R. Gerson Ian Wark Research Institute University of South Australia Mawson Lakes Campus South Australia

G.A. Hope Faculty of Science and Technology Griffith University Nathan, Queensland, Australia

Craig Goodall Lonmin Platinum Marikana, South Africa

Graeme J. Jameson Center for Multiphase Processes University of Newcastle Callaghan, New South Wales, Australia

Barun K. Gorain Corporate R&D/Technical Services Barrick Gold Corporation Toronto, Ontario, Canada

N.W. Johnson College of Engineering University of Queensland Brisbane, Queensland, Australia

Brian D. Gotts Potash Corporation of Saskatchewan Allan, Saskatchewan, Canada vi

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page vii Tuesday, January 2, 2007 7:36 PM

Bert Knopjes Lonmin Platinum Marikana, South Africa

D.R. Nagaraj Minerals Processing Chemicals Division Cytec Industries Inc. Stamford, Connecticut

Janusz S. Laskowski Mining and Mineral Processing Engineering University of British Columbia Vancouver, British Columbia, Canada

J. Nalaskowski Department of Metallurgical Engineering University of Utah Salt Lake City, Utah

R. Lastra Mining and Mineral Sciences Laboratories Natural Resources Canada (CANMET) Ottawa, Ontario, Canada

Anh V. Nguyen Center for Multiphase Processes University of Newcastle Callaghan, New South Wales, Australia

Gerald H. Luttrell Mining and Minerals Engineering Virginia Polytechnic Institute and State University Blacksburg, Virginia

Heikke Oravainen Outokumpu Technology Minerals Espoo, Finland Richard Peaker Metso Minerals York, Pennsylvania

Alban J. Lynch Julius Kruttschnitt Mineral Research Centre University of Queensland Indooroopilly, Queensland, Australia

Antonio E.C. Peres Federal University of Minas Gerais Belo Horizonte, Brazil

Sharad Mathur Technical Center Engelhard Corporation Gordon, Georgia

A.R. Pratt Mining and Mineral Sciences Laboratories Natural Resources Canada (CANMET) Ottawa, Ontario, Canada

R. McEachern Potash Corporation of Saskatchewan Allan, Saskatchewan, Canada

Robert J. Pugh Chemical and Engineering Industries Section Institute for Surface Chemistry–YKI Stockholm, Sweden

Thomas P. Meloy West Virginia University Morgantown, West Virginia

Srinivasa Raghavan Department of Materials Science and Engineering University of Arizona Tucson, Arizona

J. Mielczarski Laboratoire Environment et Mineralurgie Vandoeuvre-les-Nancy, France Jan D. Miller Department of Metallurgical Engineering University of Utah Salt Lake City, Utah

John Ralston Ian Wark Research Institute University of South Australia Mawson Lakes Campus South Australia

vii

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page viii Tuesday, January 2, 2007 7:36 PM

K.H. Rao Division of Mineral Processing Lulea University of Technology Lulea, Sweden

X. Wang Department of Metallurgical Engineering University of Utah Salt Lake City, Utah

S.A. Ravishankar Minerals Processing Chemicals Division Cytec Industries Inc. Stamford, Connecticut

John Watt Division of Minerals CSIRO Minerals Melbourne, Victoria, Australia

Geoff Senior BHP Billiton Nickel West Perth, Western Australia

Asa T. Weber Dorr-Oliver Eimco Salt Lake City, Utah

W.M. Skinner Ian Wark Research Institute University of South Australia Mawson Lakes Campus South Australia

Mark C. Williams West Virginia University Morgantown, West Virginia James T. Woodcock CSIRO Minerals Clayton South, Victoria, Australia

Robert Snow Beneficiation and Mining Florida Institute of Phosphate Research Bartow, Florida

Ronald Woods School of Science Griffith University Nathan, Queensland, Australia

Ponisseril Somasundaran Henry Krumb School of Mines Columbia University New York, New York

Juan Yianatos Department of Chemical Engineering Santa Maria University Valparaiso, Chile

G.J. Sparrow CSIRO Minerals Clayton South, Victoria, Australia

Roe-Hoan Yoon Center for Advanced Separation Technologies Virginia Polytechnic Institute and State University Blacksburg, Virginia

Roger StC. Smart Applied Center for Structural and Synchrotron Studies University of South Australia Mawson Lakes Campus South Australia

Lui Zhang Akzo Nobel Chemicals Inc. Dobbs Ferry, New York

G. Strathdee Potash Corporation of Saskatchewan Allan, Saskatchewan, Canada

Patrick Zhang Beneficiation and Mining Florida Institute of Phosphate Research Bartow, Florida

Frank P. Traczyk Dorr-Oliver Eimco Salt Lake City, Utah

viii

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page ix Tuesday, January 2, 2007 7:36 PM

Preface The year 2005 marked the 100th anniversary of Sulman and Picard’s U.S. patent award (No. 793,808) that prescribed the use of air bubbles for flotation. 1905 was also the year when the Potter process was introduced to flotation in the minerals industry. The production of sphalerite concentrate at Broken Hill in Australia was the first major commercial application of froth flotation. Following that initial application, froth flotation quickly spread to the United States and the rest of the world, where it remains an essential separation step in the beneficiation of minerals and coal. Its scope is continually broadening to other applications such as environmental control, bitumen extraction from tar sands, and recycling. Recognizing its significance, a group of flotation researchers and practitioners met in 2001 to consider ways for commemorating this important anniversary. The idea was initiated by a group of individuals, including D.W. Fuerstenau from the University of California, Berkeley; M.C. Fuerstenau from the University of Nevada, Reno; and Roe-Hoan Yoon from Virginia Tech. They were joined by D.R. Nagaraj, Cytec Industries; J.A. Herbst, Metso Minerals; J.-P. Franzidis, Julius Kruttschnitt Mineral Research Centre; J.A. Ralston, Ian Wark Research Institute, University of South Australia; and G.J. Jameson, University of Newcastle. Two international initiatives were launched—a symposium and this commemorative volume. Managed by the Australasian Institute of Mining and Metallurgy, the Centenary of Flotation Symposium was held in June 2005 in Brisbane, Australia. It was a great success, attracting more than 450 delegates and 149 presentations from around the world. The conference fostered in-depth discussion of recent research and up-to-date descriptions of advanced plant practice. A CD of the conference proceedings is included with this volume. This commemorative volume, published by Society for Mining, Metallurgy, and Exploration, is a comprehensive resource detailing the state of the art of flotation. The book is the continuation of a distinguished series published by SME. The sequence began with Froth Flotation: 50th Anniversary Volume (1962), edited by D.W. Fuerstenau, to celebrate the first 50 years of flotation in the United States; followed by the A.M. Gaudin Memorial Volume (1976), edited by M.C. Fuerstenau. The continuing involvement of the Fuerstenau brothers in these important volumes over such a long time span is particularly noteworthy. The chapters in the book are written by experts in the various disciplines and cover all aspects of flotation, from fundamental research to industrial practice. Coverage includes the historical aspects of flotation; flotation fundamentals; flotation chemistry; flotation cells, modeling, and simulation; and flotation plant practice. The book is an invaluable reference for industry practitioners, researchers, and graduate students. Sincere appreciation is extended to all who have contributed to the various chapters. Despite its longevity, the field of flotation is quite active and rapidly changing. The editors and SME are fortunate to have contributions from so many leaders in the industry for this milestone project.

ix

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page x Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 1 Tuesday, January 2, 2007 7:36 PM

PART 1

Historical Aspects of Flotation

1

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 2 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 3 Tuesday, January 2, 2007 7:36 PM

A Century of Developments in the Chemistry of Flotation Processing Douglas W. Fuerstenau

A B S T R AC T

This chapter reviews some aspects of the significance of flotation during its early days, and particularly the development of the understanding of how flotation separations can be made by the utilization of chemical reagents that interact with mineral surfaces. The second quarter of the flotation century saw the development of most of the reagents and reagent schemes still used today in flotation technology. Most of this chapter is concerned with a review of the fundamentals of flotation chemistry research, particularly the surface chemistry on which flotation is based. The first decades of fundamental flotation research were oriented toward sulfide minerals, followed by extensive investigation of the flotation of oxide and silicate minerals, and then the sparingly-soluble salt minerals. More recent application of electrochemical and surface probe techniques brought attention again to the flotation chemistry of sulfide minerals. Topics presented here are necessarily limited to broader aspects of sulfide mineral surface chemistry and the role of oxidation in collection processes, the interfacial chemistry of oxide and silicate mineral flotation and the role of the electrical double layer and hydrocarbon chain association, and the influence of aqueous solution chemistry on the flotation of sparingly-soluble salt minerals. INTRODUCTION

No metallurgical process developed in the 20th century compares with that of froth flotation and the profound effect it had on the mineral industry. Most of the early developments in flotation processing originated in Australia between 1900 and 1910. In the bulk oil processes that preceded froth flotation, generally the separation was aided by levitation of the oil/mineral mass, either through the entrainment of air during mixing or by reduction of pressure to generate bubbles, or by the addition of sulfuric acid to generate carbon dioxide bubbles from carbonate minerals in the ore. Working independently as well as for Minerals Separation Ltd., A.H. Higgins in London and G.A. Chapman at Broken Hill, Australia, found that by reducing the oil content (oleic acid) to less than 1% and agitating the ore, the mineral-laden bubbles rose to the surface (Rickard 1916). Modern flotation is attributed to the resulting basic patent of Minerals Separation Ltd., where the aid of chemically generated gas bubbles was definitively discarded in favor of air bubbles (Sulman, Picard, and Ballot 1905). The first operations in Australia simply involved bulk flotation to recover the fine particles that were left behind in gravity concentration plants. Froth flotation as it is known today is the process that had its beginnings 100 years ago in Australia, but a graphite flotation process preceded it by nearly three decades. As stated by Sutherland and Wark in 1955: The brothers Bessel (1877) patented a true flotation process for the concentration of graphite ores.… The modern flotation process differs little in principle from the 3

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 4 Tuesday, January 2, 2007 7:36 PM

4

HISTORICAL ASPECTS OF FLOTATION

Bessel process.… However, the work of the Bessel Brothers was forgotten, and the modern process was evolved as the result of the work of many later investigators. A German patent was issued to Gebrueder Bessel in Dresden, and using only 305 words, this 1877 patent outlined a process for the flotation of graphite from ores using 1% to 10% of a nonpolar oil and listing 16 or more sources for the oil. After the ground ore was mixed with the oil, this mass was added to water and the slurry then raised to boiling temperature. According to the patent, the graphite flakes attached to bubbles, rose to the surface, and were skimmed off to make the separation. From an ore containing 40% graphite, the Bessel brothers produced a concentrate containing more than 90% graphite in an operation near Passau (Graichen et al. 1977). The market of their product was for the production of graphite crucibles for smelting. In an attempt to reduce costs, in 1886 they patented another gas-generating method for the process by adding acid with carbonates or metals (Gebrueder Bessell 1886). About that same time, Ceylon graphite was discovered to be of higher quality, which led to the demise of the Bessel graphite operation, and subsequently to the disappearance of their process from the technical world. In 1911, James M. Hyde installed the first flotation operation in the United States at Basin, Montana, for the Butte and Superior Copper Company (Rickard 1916). Within 2 months, Minerals Separation filed suit in the U.S. District Court in Montana for infringement against Patent No. 835,120. This sparked the beginning of litigation in the early days of modern flotation. Litigation affected the widespread adoption of flotation processing, which is reflected in a paper by Barker (1928), who wrote: Although flotation was known to be a successful process prior to 1912, Utah Copper Co.’s ores were not entirely treated by this process until 1923. Experiments had been conducted, of course, prior to that time, and in February, 1917, the first unit of the Arthur plant was changed over from gravity concentration to flotation…. The reasons for the delay in adopting flotation at these plants were, first, that it was decided to await the outcome of the litigation with the Butte & Superior Mining Co., which began with an injunction served on the plant on Oct. 3, 1911. This litigation continued for years. After conversion of Utah Copper Co.’s operations to total flotation processing, the cutoff grade in mining was reduced and their reserves were enormously increased. Data gathered by the U.S. Bureau of Mines shows the growth of flotation in the United States, this growth being related to the development of selective flotation reagents and to the increasing demand for mineral products (Varley 1928; Merrill and Pennington 1962; Cooper 1980). Table 1 summarizes ore tonnages treated by flotation in the United States for some representative years. The increase in ore tonnage processed by flotation in 1923 as compared with that processed in 1919 resulted from the introduction of chemical flotation reagents. Similarly, the marked increase in concentration ratio resulted from the advent of selective flotation brought about by the introduction of these new chemical reagents, as will be discussed later. In the early years, essentially only sulfide ores were treated by flotation, but subsequently, processing other kinds of ores resulted from the development of new reagents and reagent schemes. The huge increase in flotation processing in the United States by 1960 resulted not only from increased copper ore production but also from extension to other commodities, particularly phosphate and potash ores, as shown in Table 2. By 1980 there was a very significant increase in copper ore (due to lower grade) as well as in phosphate and iron ore flotation. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 5 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

TABLE 1

5

Magnitude of the flotation industry in the United States for selected years

Year 1919 1923 1926 1960 1980

Ore Treated, Mt 24.08 34.29 46.16 179.86 404.34

Concentrates Produced, Mt 2.82 1.93 3.04 19.50 71.93

Concentration Ratio 8.6 17.7 15.2 10.8 5.6

TABLE 2 Types of ore treated by flotation and concentrates produced in the United States (in million metric tons) 1926 Type of Ore Copper Lead-zinc Gold-silver Iron Phosphate Potash Coal Feldspar-mica glass sand Misc. industrial minerals

Treated 39.89 5.57 0.48

0.23

1960

Concentrates 2.17 0.84 0.03

0.02

Treated 133.38 7.43 0.12 1.39 19.03 10.87 3.73 1.67 2.23

1980

Concentrates 4.82 0.49 0.003 0.54 6.37 2.83 2.54 1.06 0.83

Treated 211.61 11.39 0.10 37.88 108.70 12.93 11.70 11.58 0.58

Concentrates 4.67 0.84 0.005 21.48 26.63 2.99 6.86 8.51 0.37

In 1928, A.T. Tye wrote a landmark paper in which he described in detail not only how selective flotation success was achieved in treating the problem ore at Cananea (Mexico) but also the benefits to the Cananea Smelter of lowering the pyritic iron contamination in the flotation concentrates. In 1923, with a combined gravity and bulk flotation flowsheet using coal tar and pine oil as reagents, the grade of the concentrate was only 4.4% Cu, but by 1925 with selective flotation using xanthate, pine oil, and lime under very controlled conditions necessitated by the soluble salts in the water, the flotation concentrates averaged 17.7% Cu. Copper recovery by flotation increased from 87.4% to 91.2%, but overall recovery in the smelter increased from 91% to 97% because of lower copper losses in the reduced amount of slag. Further economic benefits resulted because much of the smelter could be shut down as a result of the reduced tonnage of smelter feed. It is of interest that in discussion of this 1928 paper, G. Oldright suggested the promise of treating copper concentrates hydrometallurgically instead of smelting them. Mining geologist P. Billingsly (1928) expressed how flotation greatly expanded the role of the exploration geologist: The mining geologist searches for materials which the metallurgist can utilize, and only such; and whenever an advance in metallurgy opens the gates for new materials, the geologist’s problem is correspondingly modified…. The metallurgist has been the geologist’s best friend, and the geologist in turn has been able to help convert the metallurgist’s ideas into the concrete form of an increased ore supply. Many authors of papers in Rickard’s edited classic 1916 monograph, The Flotation Process, asked questions about and speculated on the underlying phenomena involved in the flotation process. The overall objective of this chapter, therefore, is to show how many of those © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 6 Tuesday, January 2, 2007 7:36 PM

6

HISTORICAL ASPECTS OF FLOTATION

questions have been answered through research carried out during the last nine decades of the flotation century. This chapter will primarily discuss the behavior of some typical reagents and the surface chemistry involved in producing hydrophobic surfaces on minerals, leaving factors that affect flotation kinetics to other presentations. F L O TAT I O N R E A G E N T D E V E L O P M E N T

The success of any flotation separation depends on the range of chemical reagents added to the system to control the surface behavior of minerals in the ore. Early flotation reagents for sulfide mineral flotation were an almost unlimited range of various oils: coal tar derivatives, crude petroleum, wood tars, and pine oils. Oleic acid could not be used where the gangue minerals were calcareous. The coal tar derivatives contained sulfur compounds that probably possessed a certain amount of affinity for the sulfide minerals. Metallurgists were struggling to make separations between lead and zinc, copper minerals and pyrite, and oxide minerals. Flotation entered a new era in 1921 when Perkins patented the slightly-soluble thiocarbanilid as the first nonoily chemical collector for sulfide mineral flotation. James Bean’s (1971) recollections illustrate the significance of this to a mill operator: Thiocarbanilid for the first time gave the laboring metallurgist something that he could add which would improve the collection of the sought-for mineral without, at the same time, increasing the frothing to an uncontrollable degree. That this was no small triumph was demonstrated practically to me while I was flotation operator at the Arthur mill of Utah Copper Company which at the time (1922) was using Utah Copper’s own particular concoction of Barrett Oils and sulfur stewed up together. Late on a sleepy afternoon an operator unduly increased the “oil” being fed, hoping to lower mill tailing, but when the rougher froth got through two cleaning steps neither the launders nor the floors could hold the resulting froth and it literally ran out of the windows over a length of perhaps 40 feet and to a depth of 3 or 4 inches. Years later I could still mark the area as I passed by on the highway. Flotation reagents fall into six broad types: frothers, collectors, modifiers, activators, depressants, and flocculants (natural and synthetic polymers). The frother is added to control bubble size and froth stability. Collectors are surface-active organic reagents that impart hydrophobicity to minerals when they adsorb at mineral surfaces. The function of all other reagents is to attain optimal conditions for selective separation of the minerals in an ore. Activators are chemicals that enhance collector adsorption onto a specific mineral, whereas depressants are reagents that prevent collector adsorption or prevent bubble attachment to unwanted mineral surfaces. Modifiers constitute a broad range of inorganic and organic compounds that modulate the flotation environment. Flocculants are added for assisting dewatering of the flotation concentrates and are used in the selective flocculation/flotation processing of nonmagnetic taconites. The great step forward that revolutionized the industry came with the 1925 patent of Keller for water-soluble xanthates as sulfide mineral collectors, followed by the patent of Whitworth (1926) for dithiophosphates. Table 3 provides a brief glimpse of the amount and kinds of reagents used in the United States in two different eras: 1925–1926 and 1980 (Varley 1928, Cooper 1980). In 1925, various oils were still used as the collector with a large consumption of sulfuric acid to attempt selective flotation. In 1926, the change to xanthate collectors took hold, the use of oily collectors dropped sharply,

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 7 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

TABLE 3

7

Reagents used for flotation in the United States (in metric tons) Reagent Amount

Ore Treated Frothers Collectors Oils Chemicals Modifiers Acids Alkalis Other Activators Depressants Flocculants

1925 41,259,000 2,195

1926 41,616,000 2,935

1980 440,361,000 12,489

8,818 1,875

2,665 1,896

115,218 108,883

18,157 1,695 NA* 3,210 754 NA

2,061 75,701 NA 4,962 1,104 NA

35,169 413,055 28,735 3,925 33,389 18,069

*NA = Not available.

and that of alkalies dramatically increased in order to achieve the high pH necessary for selective sulfide flotation. In 1925–1926, sulfidizing agents to treat oxidized lead and copper ores accounted for about three-fourths of the activator consumption. By 1980, the total tonnage of ore treated was nearly ten times that in 1926. The application of flotation to processing nonmetallic ores resulted in higher reagent consumption because of amines, soaps, and sulfonates being used as collectors along with various depressants. Oil consumption was again high because of its use in phosphate and coal flotation. Along with the quest for suitable organic chemicals having the ability to collect the desired mineral in the froth, early flotation operators also tried to find agents to aid or inhibit mineral floatability. Their discoveries, associated with the use of inorganic compounds in flotation, made possible the remarkable success achieved at present in the separation of sulfide minerals from each other, and in the concentration of oxides, silicates, and salt-type minerals. A chronological account outlining some of these findings is presented in Table 4, which also lists the flotation function of each chemical reagent. As this brief historical survey shows, most of the reagents used or known today were introduced during the first half-century of flotation. Reagent development had a great deal to do with improvement in the effectiveness of flotation. The invention of Dow Chemical Company’s Z-200, a dialkyldithionocarbamate by G.H. Harris and B.C. Fischback (1954), is undoubtedly the most significant sulfide flotation reagent development since the invention of xanthate as a flotation collector by Keller and dithiophosphates by Whitworth shortly thereafter. The impact of Z-200 on sulfide ore flotation, and particularly copper ore flotation, can be illustrated with data for 1979 as an example (Harris, personal communication). In 1979, according to Harris, 4,500,000 kg of this reagent (and its reproduction by other producers) were sold worldwide. At a reagent consumption of 0.01–0.02 kg/t, there is an increase in copper flotation recovery of +2 percentage points. With the treatment of 300 Mt of copper sulfide ores worldwide at a grade of 0.7% Cu, this means that the invention of Z-200 gave the world an additional 40 million kg of copper in 1979 alone.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 8 Tuesday, January 2, 2007 7:36 PM

8

HISTORICAL ASPECTS OF FLOTATION

TABLE 4 Year 1901

1905 1909 1910 1912 1913 1921 1922 1923 1924 1925 1926 1928 1931 1934 1935 1952 1954 1965 1985

Milestones in the development of flotation reagents Chemical Reagent Sulfuric acid Salt cake (NaHSO4) Oils Sodium sulfide Ketones, aldehydes Alkalis Sodium dichromate Sulfur dioxide Copper sulfate Thiocarbanilid Cyanide Alkali sulfides Soluble sulfites Soaps Alkali xanthates Dithiophosphates Sodium silicate Sodium carbonate Starch Alkyl sulfates Amines Polypropylene glycols Thionocarbamates Hydroxamates Alkoxycarbonyl adducts

Function Gas-bubble generator Gas-bubble generator Collectors for sulfide minerals Activator for oxidized heavy-metal minerals Soluble frothers Sphalerite depressants Galena depressant Sphalerite depressant Sphalerite activator Slightly soluble chemical collector Sphalerite and pyrite depressant Sphalerite and pyrite depressants Sphalerite depressant Collectors for nonsulfide minerals Soluble collectors for heavy-metal minerals Collectors Depressant pH regulator Depressant Nonmetallic mineral collectors Cationic collectors Water-soluble frothers (polyethers) Sulfide mineral collectors (copper) Chelating agent for collector of Cu, Fe oxide Collectors/modifiers for sulfides and nonsulfides

B R I E F C H R O N O L O G Y O F F L O TAT I O N R E S E A R C H

Many early efforts at understanding flotation were directed toward explaining differential flotation in terms of the relative occlusion of gases, which would be driven out to nucleate bubbles, thereby giving rise to selective flotation. In 1916, bubbles were considered to be at the heart of flotation science, and Rickard (1916) postulated how progress in flotation would be made: “…we know that the key to the flotation process is to be found not in the oil, the acid, or the apparatus, but in the bubbles. The man who understands the physics of a soap bubble has mastered the chief mystery of flotation.” As important a component as they are in the process, bubbles usually play an inert role in flotation and merely provide a means for levitating the desired mineral particles into a froth layer. Although industrial operators and reagent manufacturers devoted effort toward finding cheaper chemicals that might act as frothing agents and might alter froth characteristics, through the years bubbles have never received the attention from flotation researchers speculated on by Rickard. However, in 1934 Gaudin commented: Developments in flotation have been so rapid that one of the essential factors at play—namely, the chemical effects of dissolved gases—has received scant attention. Recent theories have shown that gases are of extreme importance in many instances. It is not unlikely that control of flotation can be exercised through control of the gases.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 9 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

9

In a 1915 paper, Ralston suggested that flotation might result from the electrical attraction between negatively charged air bubbles and positively charged mineral particles, and this postulate was actively debated for some years before eventually being discarded (Ralston 1916). However, today it is known that potential differences between bubbles and particles enter into the kinetics of bubble-particle attachment. The role of interfaces in flotation had been considered by Sulman by 1912 (see Rickard 1916) when he published the results of an investigation carried out at Minerals Separation to determine the magnitude of the contact angle that various minerals needed before they were wetted and would sink when brought into contact with a water surface. About these phenomena, Ralston wrote in 1915 of adsorption changing contact angles and the properties of interfacial films: A glance at Clerk Maxwell’s famous paper on capillarity upon which Reinders’ work is based, will suggest immediately the explanation of a contact angle, and that it is the result of a certain equilibrium of interfacial tensions of air, water, and solid.… There can be no doubt that there is a close parallelism between the angle of hysteresis of the contact angle and the ability of a mineral to float.… To go into this a little farther, we ought to consider the properties of the surface layers of the substances involved.… One important property of this film is that it will often take up dissolved substances in different proportion from the amounts in which they are taken up in the bulk solution, and there always is a definite equilibrium between the two.… The properties of these interfacial films have been found to be greatly modified by small amounts of dissolved substances. The importance of the study of interfacial films becomes obvious. The first direct application of thermodynamics to systems similar to flotation was that of von Reinders (1913). Based on Maxwell’s capillarity equations, von Reinders deduced how fine solid particles would be distributed between oil and water phases. For example, using γ to represent the interfacial tensions at the oil–water (ow), solid–water (sw), and solid–oil (so) interfaces, von Reinders showed that the solid will disperse in the aqueous phase if γso > γow + γsw. Analogous relations give conditions under which the solid will disperse in the oil phase or concentrate at the oil–water interface. The three interfacial tensions are interrelated with contact angles by the Young equation. Ralston suggested that von Reinders’ relations might explain how interfacial tensions control flotation. In 1917, Taggart and Beach fairly lucidly applied these concepts directly to flotation. Several decades would elapse before thermodynamics would become a fairly widely used tool for the analysis of flotation phenomena. In 1917, Anderson suggested that adsorption might play a dominant role in flotation. Anderson discussed the Gibbs adsorption equation in relation to frother adsorption at the air-water interface and, interestingly, stated: “An electric charge on an adsorbed substance probably would considerably influence the amount adsorbed.” In 1920, Langmuir showed that oleic acid created large contact angles on cleaved calcite and galena but only small angles on clean glass and cleaved mica. Oleic acid was irreversibly adsorbed on calcite and galena but not on glass and mica. He suggested further research with other kinds of reagents on clean mineral surfaces. In 1928, Taggart described the results of adsorption tests on sulfide minerals that related the structure of the adsorbate to its ability to act as a flotation collector. He wrote that powdered sulfide minerals abstracted 90% of the thiocarbanilid in a solution and captive-bubble experiments showed the sulfide to be hydrophobic. The © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 10 Tuesday, January 2, 2007 7:36 PM

10

HISTORICAL ASPECTS OF FLOTATION

adsorbed thiocarbanilid could be leached off with ethyl alcohol. Similar experiments with thiourea (which has the same formula as thiocarbanilid but without the phenyl groups attached to the nitrogen) showed that sulfide minerals adsorb thiourea almost as much as thiocarbanilid but there is no collecting action. With carbanilid, there was no removal from solution. These experiments led to Taggart’s formulation of the definition of the molecular structure needed for a soluble flotation collector, namely, that it must possess both a polar group that binds it to the surface and a nonpolar group that can orient away when adsorbed at a mineral–water interface. NHC6H5 S

C

NHC6H5 O

NHC6H5 Thiocarbanilid

C

NH2 S

NHC6H5 Carbanilid

C NH2

Thiourea

What might be considered to be the first adsorption isotherm of a soluble flotation collector on a mineral are the results published by Taggart, Taylor, and Knoll in 1930 for the abstraction of potassium ethyl xanthate (KEX) by ground galena as a function of reagent concentration in solution. It would be some time before adsorption isotherms could reliably be determined quantitatively and several years until methods were developed for determining specific surface area, for radioactively marking adsorbates, for spectrophotometrically measuring reagent concentrations in solution, and for quantitatively analyzing infrared absorption spectra. Because nearly all of the early flotation operations involved sulfide ores, the behavior of sulfide minerals received nearly all of the initial research attention. Although such early researchers as Fahrenwald, Sulman, and Taggart carried out a number of experiments to elucidate flotation phenomena, the founder of the scientific basis of flotation was A.M. Gaudin. The first systematic research that opened the way toward understanding the chemistry of the flotation process was the extensive, dedicated investigation initiated in 1926 at the University of Utah under Gaudin, using high-purity single minerals in a miniature flotation cell (50 g of pure 100 × 600 mesh cleaned samples) that had been developed at the Utah Engineering Experiment Station by Gates and Jacobsen (1925). In 1928, Gaudin described their laboratory approach: It is a generally recognized scientific principle that to investigate a certain set of phenomena one variable must be allowed to vary at one time while other variables are kept strictly constant. Therefore, to obtain consistent results in flotation research, pure minerals having a definite size should be used either by themselves or as artificial mixtures. These minerals should have an especially clean surface, cleaned in standard fashion, and the test should be run in a standard machine cleaned in standard fashion, for a standard length of time after a standard preagitation period at a definite temperature. All reagents should preferably be added in solution to eliminate the necessity for conditioning. Distilled water should be used throughout. With these guiding principles, this early work by Gaudin and his colleagues was the beginning of the modern approach to research in flotation chemistry. In the author’s opinion, Gaudin was indeed the father of fundamental flotation research as it is known today. Figure 1 illustrates the quality of flotation experiments conducted with carefully cleaned mineral samples and high-purity reagents (Gaudin et al. 1928). In Gaudin and colleagues’ original paper, the flotation recovery of 100 × 600 mesh galena was presented as a linear function of the fatty acid addition in pounds per ton. By recalculating the published results © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 11 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

Flotation Recovery, %

100

C12

Galena 0.2 kg/t Terpineol

C11

C10

C9

C8

11

C7

80

60

C6 40

20

C6 C7 C8 C9 C10 C11 C12

Caproic Acid Enanthic Acid Caprylic Acid Pelargonic Acid Capric Acid Undecanoic Acid Lauric Acid

0 0.01

0.1

1.0

10.0

Fatty Acid Addition, mol/t

Adapted from Gaudin et al. 1928.

FIGURE 1 Flotation of 100 × 600 mesh galena with fatty acids of different alkyl chain lengths ranging from 7 to 12 carbon atoms, with the reagent addition in mol/t (mol wt of C11 lauric acid is 200)

in terms of moles per metric ton and replotting those data in a semilogarithmic manner as shown in Figure 1, their results show the very systematic effect of the number of carbon atoms in the alkyl chain expected by the Traube rule. Because the carbon atom in the carboxyl head group is not part of the alkyl chain, lauric acid (mol wt 200) is given as an 11-carbon reagent in Figure 1. This systematic chain-length effect indeed substantiates the validity and care taken in their work. Interestingly, Gaudin never continued using mini-scale flotation cells in his research after he moved from the University of Utah to the Montana School of Mines. Flotation processing technology did not come into being as a result of an intensive fundamental research effort, but, in a manner similar to the development of so much of the other technology used in the processing of raw materials, it was developed over the years through much empirical and intuitive work on complex ores. Fundamental understanding of flotation resulted from careful experimentation with well-controlled systems, later followed by a firm grounding in physicochemical principles, including thermodynamics, surface and colloid chemistry, and electrochemistry. Major headway in understanding the flotation chemistry of sulfide mineral flotation started shortly before 1930, and that of nonmetallic mineral flotation shortly before 1950. Prior to about 1950, most of the fundamental investigations were directed toward the flotation chemistry of sulfide mineral separations. To achieve the desired separations from complex ores, the early research (1925– 1935) was mainly centered on interactions between mineral surfaces and sulfhydryl flotation reagents. The leading researchers, chronologically, in that era were Taggart and Gaudin in the United States, and Wark in Australia. The key issues were the mechanism of interaction between the reagent and the mineral surface (by Taggart and by Gaudin), identification of species responsible for flotation (by Gaudin), and the assessment of chemical conditions for floatability (by Wark ). About mid-century most of the research shifted to oxides, particularly quartz, corundum, hematite, rutile, and silicates. In the last quarter of the flotation century, much attention was directed toward the flotation chemistry of the sparingly-soluble salt © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 12 Tuesday, January 2, 2007 7:36 PM

12

HISTORICAL ASPECTS OF FLOTATION

TABLE 5 Selected experimental techniques that provided significant advances toward understanding flotation chemistry Major Experimental Techniques Single-mineral laboratory flotation mini-scale 50-gram scale 5-gram scale Modified Hallimond tube flotation

Captive-bubble contact angle determination

Adsorption density/isotherm

Electrokinetics (zeta potentials)

Infrared spectroscopy Ex situ In situ

Researchers J.F. Gates and L.K. Jacobsen A.M. Gaudin M.C. Fuerstenau A.F. Hallimond E.W. Ewers D.W. Fuerstenau H.S. Choi and I. Iwasaki A.F. Taggart, T.C. Taylor, and C.R. Ince I.W. Wark and A.B. Cox R.H. Ottewill J.S. Laskowski S. Chander and D.W. Fuerstenau A.M. Gaudin P.L. deBruyn, I. Iwasaki, and G.A. Parks P. Somasundaran and D.W. Fuerstenau J.M. Cases S.C. Sun and A.M. Gaudin A.S. Buchanan and D.J. O’Connor D.W. Fuerstenau M.E. Wadsworth and A.S. Peck J. Leja and G.W. Poling J.D. Miller J.A. Mielczarski J.D. Miller I.W. Wark and A.B. Cox

Electrochemistry Rest potential

J.C. Nixon and S.G. Salamy

Polarization: voltammetry

J.T. Woodcock and M.H. Jones

Impedance spectroscopy

R. Woods W.J. Trahar S. Chander and D.W. Fuerstenau P.A. Richardson

minerals, particularly apatite, calcite, dolomite, and bastnaesite. Problems of energy supply gave rise to research on coal flotation and coal desulfurization. With the advent of newer electrochemical techniques, major effort resumed in the last quarter of the flotation century to extensive investigation of sulfide mineral flotation phenomena. Although numerous experimental methods have been applied to investigating flotation phenomena, several techniques have been widely used and have been responsible for the greatest progress. These are summarized in Table 5, together with the names of several of those who developed or applied these techniques to the study of chemical phenomena involved in flotation. Numerous other techniques have been devised and utilized through the years to study flotation phenomena, but they are not included here because they may not have yielded definitive results or may not have had the impact or widespread use of the seven techniques given in Table 5. Examples of some of these techniques (and some of researchers who used them) include vacuum flotation (R. Schuhmann and B. Prakash), bubble-pickup (S.R.B. Cooke), induction time measurement (I. Sven-Nilsen; V.A. Glembotsky; R.H. Yoon), film flotation (M.C. Williams and D.W. Fuerstenau), microcalorimetry (O. Mellgren), and radiography (I.N. Plaksin). There has been worldwide interest in surfactant adsorption behavior at solid–water interfaces in recent years, resulting in many new tools having been

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 13 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

13

used to probe the detailed nature of adsorbed surfactant and polymer films. For example, P. Somasundaran and his graduate students at Columbia University have extensively used such newer molecular-level-information-yielding techniques as absorption, emission, magnetic resonance, and scattering spectroscopic techniques (fluorescence, electron spin resonance, excited state resonance, Raman, etc.) along with adsorption, flotation, flocculation, and electrokinetic studies to gather information about the microscopic properties of the adsorbed surfactant and polymer films. X-ray photoelectron spectroscopy, or XPS (A.N. Buckley and R. Woods), has been used to identify chemical species at mineral surfaces. Secondary ion mass spectrometry, or SIMS (D.R. Nagaraj), has recently been utilized to clearly show the nature of complexes adsorbed at mineral surfaces. Atomic force microscopy has been applied to the study of the nature of adsorbed surfactant films (R.H. Yoon; T.W. Healy; J.D. Miller). Starting in the 1950s, two of the relatively simple techniques listed in Table 5 were widely adapted to the study of flotation chemistry effects. When it became understood that any ion that strongly adsorbs at a mineral–water interface is reflected in its effect on the zeta potential, the use of zeta potential measurements in flotation surface chemistry spread rapidly, and particularly so because of the simplicity of electrophoresis techniques. The modified Hallimond tube permitted study of flotation response without changing the solution composition (because no material is removed as a mineral-laden froth from the device during an experiment); this permitted direct correlation with the solution chemistry of the system. Almost all of the experimental investigations on flotation chemistry carried out during the first half-century involved the use of a single experimental technique, such as flotation testing, contact angle measurement, identification of surface species, determination of adsorption isotherms, and so forth. However, using a number of different experimental techniques to probe the behavior of the same system led to being able to make correlations among various types of interfacial phenomena in flotation systems, and this led to a more complete understanding of the surface chemical processes involved. An example of such a correlation is given in Figure 2, which presents the zeta potential, adsorption density, contact angle, and flotation response of quartz with dodecylammonium acetate (DAA) as collector (D.W. Fuerstenau, Healy, and Somasundaran 1964). Here, two-phase mineral–water interfacial phenomena (adsorption density and zeta potential) correlate well with three-phase behavior (contact angle and flotation response). The first such correlation was published in 1957 for the DAA–quartz system at constant collector concentration with pH as the variable; later results for the same system at constant pH but with collector concentration as the variable are somewhat easier to explain and are therefore given in Figure 2. (The reasons for the sharp breaks in the curves that occur at hemimicelle concentration [HMC] will be discussed in a later section.) Major advances, particularly starting in the 1950s, were achieved through better understanding and application of the fundamental principles of surface and colloid chemistry, particularly electrical double-layer phenomena, to flotation systems. In part, this was strongly influenced by Professor J.Th.G. Overbeek’s year at the Massachusetts Institute of Technology (MIT) with the mineral engineering group of Gaudin, and disseminated worldwide by the generations of students that followed. Detailed analysis of the thermodynamic stability of minerals and reagents, speciation of complexes in aqueous solution, and solubility phenomena have also helped expand the understanding of different types of flotation systems. All of this, combined with application of the many new techniques for probing mineral–water interfaces at the molecular level, led to much of the research in the second © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 14 Tuesday, January 2, 2007 7:36 PM

14

HISTORICAL ASPECTS OF FLOTATION

5 Quartz pH 6–7 Contact Angle Adsorption Density Zeta Potential Flotation

80 +40

0.85

2

0.90

60 0 40

Zeta Potential, mV

0.80

Flotation Recovery, %

3

+80

Cosine θ

Adsorption Density, μmol/m 2

4

Hemimicelle Concentration

–40 1

0.95

20 –80

0 10–6

10–5

10–4

10–3

DAA Concentration, mol/t

Adapted from D.W. Fuerstenau, Healy, and Somasundaran 1964.

FIGURE 2 Correlation of adsorption density, contact angles, and zeta potentials with the flotation of quartz at pH 6–7 as a function of DAA concentration

half of the flotation century being devoted to elucidating the detailed principles of mineral– reagent interactions in flotation. In the sections that follow, some of the major advances in understanding the flotation chemistry of various mineral systems will be briefly reviewed. S U L F I D E M I N E R A L F L O TAT I O N C H E M I S T R Y

Because flotation was first applied to the recovery of sulfide minerals from ores, all of the early research was conducted on sulfide minerals, particularly galena, sphalerite, chalcocite, chalcopyrite, and pyrite. The first systematic investigations on sulfide mineral flotation were the pure mineral flotation experiments of Gaudin and his associates at the University of Utah (Gaudin et al. 1928). Their initial research was concerned with the behavior of galena. Gaudin (1932) stated that “…pure, unoxidized galena floats readily without the addition of a collecting agent, a frother alone being required. This can be ascertained by grinding pure galena particles in water under anaerobic conditions, and floating immediately.… In practice galena particles are more or less oxidized during grinding and classification, requiring varying amounts of collecting agents.” For fatty acids as collector, Figure 1 illustrates the quality of their results. Most of their work was conducted with xanthates and other sulfhydryl reagents as the collector, and Figure 3 presents the results of Gaudin et al. (1928) for the flotation of galena with xanthates of different chain length, but again with the xanthate additions being recalculated in terms of moles per metric ton, rather than pounds per ton, and plotted semilogarithmically. The amount of xanthate required for complete flotation with xanthates of two or three methylene groups is extremely low, merely about 0.1 mol/t of mineral, showing an extremely high affinity for the surface that is not strongly dependent on chain length if the collector has three or four carbon atoms. Although no specific numbers are available, nearly all of the © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 15 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

15

Flotation Recovery, %

100

80

60

40

Methyl Ethyl Propyl Butyl

20 0.05 kg/t Terpineol 0.5 kg/t Sodium Carbonate 0 0.01

0.1

1.0

10.0

Potassium Alkyl Xanthate Addition, mol/t

Adapted from Gaudin et al. 1928.

FIGURE 3 Flotation of 100 × 600 mesh galena with alkyl xanthates of different alkyl chain lengths with reagent additions expressed in mol/t (mol wt of KEX is 160)

added xanthate would have been abstracted from solution. In the case of fatty acid flotation, there probably was considerable residual carboxylate left in solution. Given that the surface of galena readily oxidizes, the oxidation products must enter into the adsorption process. In 1934, Taylor and Knoll conducted a careful set of experiments to quantitatively determine the exchange process involved in the uptake of ethyl xanthate by galena, using an iodometric titration technique to determine the xanthate concentration in solution. Taking one set of measurements as an illustration of their findings, with all concentrations being expressed as equivalent to 25 mg KEX per liter of solution, the original concentration of xanthate in solution was 200.0 mg/L, the amount of xanthate ion abstracted was 58.3 mg, and the stoichiometric equivalent of reduced sulfur-oxygen ions emitted was 13.8 mg, 16.1 mg sulfate ions emitted, and 27.2 mg carbonate ions—or a total stoichiometric equivalent of 57.1 mg. Clearly, xanthate uptake by galena was exactly balanced by an exchange with oxidation product ions at the surface. In 1934, Wark and Cox presented some data on the contact angle of an air bubble on galena as a function of the concentration of KEX in solution. Their data given as milligrams of collector per liter have been converted to moles per liter (mol wt = 160) and are plotted in Figure 4. The results tend toward the maximum contact angle of 60°, after increasing sharply to about 50° at concentrations below 20 or so micromoles per liter. The 1928 results from Gaudin et al. were recalculated in terms of micromoles per kilogram of 100 × 600 mesh galena and are also plotted in Figure 4. This plot shows that about 400 μmol/kg of galena is required to achieve 90% recovery. Assuming that most of the added xanthate was adsorbed, in 1957 Gaudin estimated that roughly monolayer adsorption was achieved at this ethyl xanthate addition. However, in that same year, Bogdanov et al. (1957) published a paper that presented a summary of extensive work conducted in Russia on the adsorption of different reagents on various minerals using a number of radioactively marked adsorbates, together with their effect on flotation response. Their results for the flotation recovery of galena as a function of the percentage of monolayer coverage of ethyl xanthate are also plotted in Figure 4. These experiments show the strong affinity of a sulfhydryl collector for the surface © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 16 Tuesday, January 2, 2007 7:36 PM

16

HISTORICAL ASPECTS OF FLOTATION

Contact Angle: μmol/L Solution 60

80 40 60

40

Contact Angle, degrees

Flotation Recovery, %

100

20 Flotation Contact Angle Adsoprtion

20

0 100

0 0

20

40

60

80

Collector Addition, [ μmol/kg galena][0.1] Adsorption Density, % monolayer

FIGURE 4 Contact angle on galena as a function of the concentration of KEX expressed in mol/L (data from Wark and Cox 1934), the flotation of galena as a function of the adsorption density of ethyl xanthate expressed in terms of monolayer fraction (data from Bogdanov et al. 1957), and the flotation of galena as a function ethyl xanthate addition expressed in terms of μmol/kg (data from Gaudin et al. 1928)

Flotation Recovery, %

100

80

60

40 Chalcocite Pyrite

20

0.015 kg/t KAX 0.10 kg/t Terpineol 0 0

2

4

6

8

10

12

14

pH

Adapted from Gaudin 1929.

FIGURE 5 collector

Effect of pH on the flotation of 100 × 600 mesh chalcocite and pyrite with KAX as

of sulfide minerals and also show that experiments conducted with pure systems under controlled conditions can exhibit agreement among different measures of mineral-collector interaction. Regulation of pH has been the most important method for regulating flotation chemistry. In 1929, Gaudin first published the results of his measurements of the flotation of a © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 17 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

17

variety of minerals as a function of pH. Figure 5 presents the flotation recovery of chalcocite and pyrite as a function of pH with 0.015 kg/t of potassium amyl xanthate (KAX) and 0.10 kg/t terpineol as frother. The results clearly show that there is sharp decrease in the flotation of pyrite as the pH is increased above 6 and that chalcocite remains fully floatable under these conditions until the pH exceeds about 13. This was definitive work showing the depressant role of pH in the flotation of sulfide minerals. Gaudin on several occasions commented that not patenting the use of pH control for selective flotation was one of his oversights. In the early 1930s at the University of Melbourne, I.W. Wark (personal communication) initiated an important program to understand the flotation chemistry of sulfide minerals. Wark’s research group spent a prolonged effort, essentially a year or more, in first refining the contact angle measurement technique of Taggart, Taylor, and Ince (1930) and sample preparation so that reliable and reproducible results could be obtained. In 1934, Wark and Cox published the first of a remarkable set of papers in which they presented their classic diagrams showing the relationship between collector concentration and pH for conditions of incipient flotation, and for the behavior of a wide variety of modifiers and depressants with various collectors. Figure 6 presents one of their critical pH diagrams for three sulfide minerals—namely, pyrite, galena, and chalcopyrite—with sodium diethyldithiophosphate as collector. In each case, flotation should occur under conditions to the left of the curve. Diagrams such as these provide a means for predicting conditions under which flotation separations can be made. If one considers that hydroxyl ions adsorb competitively with collector ions, that the amount of collector adsorbed under conditions of incipient flotation is constant, and also that the standard free energy of adsorption is constant, then each line in Figure 6 must be characterized by [X–]/[OH–] being constant. These critical pH curves were a major contribution to early flotation theory and they show, for example, the pH and collector concentration at which flotation does or does not take place. In discussion of the 1934 Wark and Cox paper where KEX was used as the collector, Barsky (1934) pointed out for their experiments that [X–][H+] was constant along their critical pH curves and that the results could be interpreted as xanthic acid [HX] being constant along these curves. Gaudin (1957) interpreted the results in terms of ion exchange between adsorbed X– and OH– for surface sites. Wark and co-workers (Sutherland and Wark 1955) also measured contact angles of various thiol collectors having a range of carbon atoms in their nonpolar groups. For example, they found the contact angle of collectors having an ethyl group on the nonpolar chain to be 60° on all sulfides. This included xanthate, mercaptan, dithiophosphate, disubstituted dithiocarbamate, and others. Methyl xanthate and disubstituted dithiocarbamate produced contact angles of 50°. For nearly 25 years, there was spirited and ongoing debate about the mechanism of collection in sulfide mineral systems. Gaudin was a strong proponent of adsorption as the means of collector uptake by minerals. In 1927 he wrote, “The mechanism by which xanthates float other sulfides than galena may involve an adsorption of xanthate ions without further reaction.” On the other hand, Taggart was convinced that collectors coated mineral surfaces by chemical reaction. In 1930, Taggart, Taylor, and Knoll wrote, “All dissolved reagents which, in flotation pulps, either by action on the to-be-floated or on the not-to-befloated particles affect their floatability, by function of the reason of chemical reactions of well recognized types between the reagent and the particle affected.” Taggart’s shortcoming was his belief that the chemical theory of flotation was all-inclusive, even with regard to oils on naturally hydrophobic minerals, and for collectors that do not form insoluble products © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 18 Tuesday, January 2, 2007 7:36 PM

18

HISTORICAL ASPECTS OF FLOTATION

700

Pyrite

Galena

Chalcopyrite

Collector Concentration, mg/L

600

500

400

300

200

100

0 2

3

4

5

6

7

8

9

10

11

pH

Adapted from Wark and Cox 1934.

FIGURE 6 Critical pH curves for the flotation of pyrite, galena, and chalcopyrite with sodium diethyl dithiophosphate as collector

such as amines on minerals. In reality, Taggart’s chemical theory of collection is merely exchange adsorption (as was shown by Taylor and Knoll [1934]). Overall, Wark was another advocate of the adsorption theory of collector uptake. In 1934, Wark and Cox wrote, “We find there is a strong connection between adsorption of xanthates and the solubility of the heavy-metal xanthates, but we are unable to decide if this is an identity.” In 1950, Cook and Nixon were as forceful in promoting the concept that sulfide mineral flotation takes place by neutral molecule adsorption as Taggart had been in his promoting the idea of chemical reaction. They wrote, “Assuming a complete or nearly complete monolayer of ‘ions’ on the mineral particles, one would obtain a bulk concentrate with so much charge that it would explode with greater violence than an equal weight of nitroglycerine!” M.A. Cook, an expert in explosives and an outstanding solution physical chemist, did not think in terms of the electrical double layer because in all cases of ion adsorption, counterion adsorption or exchange adsorption keeps the system electrically neutral. Note that Cook’s neutral molecule theory is the same idea that Barsky (1934) had presented in his discussion of the critical pH curves of Wark and Cox in 1934. There are many examples where the collector indeed appears to adsorb in its neutral molecule form. In 1967, Steininger showed that the upper pH limit for the flotation of sphalerite with a wide variety of thiol collectors was a function of their pKa values. Such results indicate that the chemisorption of the neutral molecule may indeed have a role in flotation in this mineral–collector system. Raghavan and Fuerstenau (1975) demonstrated that the neutral hydroxamic acid molecule appears to be the active adsorbing species in the hematite–hydroxamate system. However, as will be subsequently discussed, when a cationic amine collector hydrolyzes to the neutral molecule species with oxide minerals, flotation ceases. In 1957, Nixon wrote, “Prominent theories could be reconciled by the electrochemical approach.” In 1984, Woods summarized sulfide flotation as follows: “Electrochemical investigations of the interaction of the thiol collectors with sulfide minerals have demonstrated that each of the three anodic processes—chemisorption, reaction © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 19 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

1,014

1,212

1,110

100

19

50 1,110– 1,210

1,020

1,112

B

1,140

0 50

990 987

50

0 1,300 cm–1

D

1,115

0

1,020

1,195

50

1,110

C 0

1,140

Absorption, %

986

0 50

A

1,200

1,100

1,000

E

900

800

700

Adapted from Leja, Little, and Poling 1962–1963.

FIGURE 7 Infrared spectra showing adsorption of ethyl xanthate onto an evaporated PbS film: (a) bulk lead ethyl xanthate, solid on Nujol mull; (b) freshly evaporated PbS film after atmospheric oxidation; (c) PbS film treated in aqueous solution of ethyl xanthate; (d) after prolonged washing in ether; and (e) after washing in pyridine

to form a metal collector compound, and the formation of a dithiolate—plays a role in creating hydrophobic surfaces.” New instrumentation permitted identification of species at the surface and quantification of energies involved in surface reactions. In a seminal study, Leja, Little, and Poling (1962–1963) applied infrared spectroscopy to demonstrate the nature of collector species at mineral surfaces. Figure 7 presents their classic infrared spectra showing the adsorption of ethyl xanthate onto an evaporated lead sulfide (PbS) film. The top curve (a) in Figure 7, taken from their work, shows the infrared spectrum of bulk lead ethyl xanthate, and the second spectrum (b) is for a lead sulfide film that has been oxidized in the atmosphere. After exposing that film to xanthate in solution, they obtained the spectrum (c) that is virtually identical to that of lead ethyl xanthate, showing that indeed a chemical compound is formed at the surface. Washing with ether (d) removed some of the surface lead xanthate, but it took a strong solvent, pyridine, to completely remove the xanthate, returning the spectrum (e) back to that of a lead sulfide (oxidized) surface. Infrared spectroscopy has become a widely used tool to study the nature of adsorbed films in flotation systems. The energetics of the interaction of xanthate with galena was carefully determined by Mellgren (1966) using microcalorimetry techniques. First, Mellgren reacted lead sulfate with xanthate. Then he reacted xanthate with galena that had lead sulfate on its surface and again measured the heat that evolved. Mellgren’s measurements of the heat of reaction for these two cases gave identical results; namely, that the enthalpy is –22 kcal/mol Pb2+ in each case. These measurements clearly indicate that the uptake of xanthate by oxidized galena is energetically equivalent to the chemical exchange reaction forming lead ethyl xanthate from lead sulfate. He conducted similar studies with lead carbonate. Mellgren also observed that © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 20 Tuesday, January 2, 2007 7:36 PM

20

HISTORICAL ASPECTS OF FLOTATION

at high pH, galena ceases to float because basic lead compounds form on the surface in preference to lead xanthate. Thus, depression in this system may also occur by a chemical reaction mechanism, that is, by the formation of a surface compound that is more insoluble than the surface collector salt. As early as 1931, Kamienski measured the rest potential of a galena in xanthate solution. In 1934, Wark and Cox presented a direct correlation of the potential of a copper electrode and the prevention of xanthate adsorption on chalcopyrite. However, about two decades elapsed before electrochemical methods were again used in flotation research for investigating mechanisms of thiol collector–sulfide mineral interaction. Salamy and Nixon (1953) found that the concentration of xanthate determines the rest potential of the mercury electrode, which corresponds to the xanthate–dixanthogen reversible potential. They concluded that the adsorbing species at the metal–solution interface is dixanthogen, the oxidation product of the xanthate ion. Definitive papers on xanthate flotation of pyrite, published independently by M.C. Fuerstenau, Kuhn, and Elgillani (1968) and by Majima and Takida (1968), showed dixanthogen to be the surface species responsible for flotation in this system. The former researchers used infrared spectroscopy and oxidation potential (Eh) measurements, and the latter conducted polarization experiments in basic medium with a pyrite electrode. It is interesting to note that Gaudin and Wilkinson wrote in 1933 that “Pyrite, or ferric iron derived from it by oxidation, changes xanthate to dixanthogen; the dixanthogen can be extracted from the mineral surface provided oxidation of the dixanthogen is prevented.” Their observation lay dormant for the next 35 years. Numerous electrochemical investigations have been made in sulfide mineral flotation systems, particularly conducting linear potential sweep voltammetry for various systems. Chander and Fuerstenau (1975) first combined this technique with the measurement of contact angles. Woods (1984), Woods and Richardson (1986), and Chander (1985) presented thorough reviews of the electrochemistry of sulfide mineral flotation. The cathodic step usually involves the reduction of oxygen: O 2 + 2H 2 O + 4e – = 4OH –

(EQ 1)

and is coupled with an anodic step involving oxidation of either the collector or the mineral. The products of the anodic reaction depend on the mineral and the collector used, and the pretreatment of the mineral surface. They have been identified as chemisorbed collector, dithiolates, and metal–collector compounds. As suggested by Woods and Richardson (1986), the reactions for the anodic oxidation of the collector can be written as follows: A. Charge transfer chemisorption of a thiol ion (X–): X – = X ( ads ) + e – B. Oxidation of a thiol ion to the disulfide: 2X – = X 2 + 2e – C. Formation of a thiol compound with a metal component of the mineral: MS + 2X – = MX 2 + S + 2e –

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 21 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

21

This mixed potential model does not actually contradict earlier theories of sulfide mineral flotation. Gaudin’s and Wark’s adsorption models would be equivalent to Reaction A, whereas Taggart’s chemical reaction theory would be represented by the anodic process given by Reaction C, which can be a two-step reaction involving oxidation of the mineral and metathesis. In a classic paper, Allison et al. (1972) presented the results of a detailed study of the products resulting from the interaction of sulfide minerals with thiol collectors. Table 6 presents a summary of their identification of surface products in ethyl xanthate–sulfide mineral systems and a correlation of those results with the rest potential of the minerals. These results clearly indicate that, with the exception of covellite, sulfide minerals that display a rest potential below the corresponding reversible potential for the oxidation of the collector to its disulfide (dixanthogen, in this case) will react with thiols, forming metal tholates. Those sulfide minerals whose rest potentials are above this value oxidize the thiol to its disulfide, and this product is now accepted as the collecting species in these systems. Similar results were reported for the diethyl dithiocarbamate–sulfide systems. A necessary condition for sulfide mineral floatability is the thermodynamic stability of the hydrophobic entity formed at the surface. If this compound forms at a rate that is too slow compared to the rate at which the sulfide mineral oxidizes, however, the mineral will not float because surface oxidation may result in the formation of oxide-type materials that may not only impede electron transfer from the collector to the mineral but are also extremely hydrophilic. If the species is decomposed by oxidation, strong oxidizing conditions would be detrimental to flotation. Knowledge of the Eh of the system is therefore vital in sulfide mineral flotation. It is possible to predict the oxidizing and reducing conditions necessary for a substance to be thermodynamically stable by constructing the appropriate Eh–pH diagram. Chander (1985) grouped sulfide minerals into two types—reversible and passivated— based on their electrochemical behavior in aqueous solutions. The reversible group of sulfide minerals includes galena and chalcocite. The potentials for oxidation–reduction reactions can be predicted if metastability of elemental sulfur is considered. In the passivated group of sulfide minerals in which Chander included pyrite, chalcopyrite, and bornite, the reactions are irreversible. The surfaces of passivated sulfides are normally covered with a layer of oxidation products. The potential of such surfaces cannot be predicted thermodynamically but are so-called mixed potentials. Chander also pointed out that the collecting species for sulfides with sulfhydryl collectors such as xanthate can also be divided into two categories. He observed that metal-xanthate salts form at the surface of reversible sulfides, TABLE 6

Rest potential and reaction product of sulfide minerals with KEX solutions

Mineral Sphalerite Stibnite Galena Bornite Chalcocite Chalcopyrite Molybdenite Pyrite Arsenopyrite

Rest Potential, V –0.15 –0.13 +0.06 +0.06 +0.14 +0.16 +0.21 +0.22 +0.22

Reaction Product* NPI NPI MX MX NPI (MX) X2 X2 X2 X2

Adapted from Allison et al. 1972. *MX = metal xanthate; NPI = not positively identified; X2 = dixanthogen. Potential for X 2 + 2e – → 2X – at 0.625 mM KEX = 0.13 V.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 22 Tuesday, January 2, 2007 7:36 PM

22

HISTORICAL ASPECTS OF FLOTATION

100 0.02 mM KEX (0.014 mM for Cu2S) pH 9.2

Flotation Recovery, %

80

Cu2S

60

Cu5FeS4

CuFeS2

FeS2

40

20

0 –0.6

–0.4

–0.2

0.0

0.2

Potential, V (vs. SCE)

Adapted from Richardson and Walker 1985.

FIGURE 8 Relationship between flotation recovery and conditioning potential for chalcocite, bornite, chalcopyrite, and pyrite with KEX as collector at a concentration of 0.02 mM with the exception of 0.0144 mM for chalcocite (potentials given versus the standard calomel electrode, or SCE)

whereas dixanthogen is the oxidation product on passivated sulfides, although this remains an area for further study. Using in-situ electrodes in microflotation cells, Gardner and Woods (1974) and Chander and Fuerstenau (1975) were the first to independently and simultaneously demonstrate that the electrochemical potential can control the flotation of sulfide minerals. Using an electrochemical microflotation cell that incorporated a packed bed of conducting sulfide particles as the working electrode to correlate interfacial electrochemical reactions with flotation response, Richardson and Walker (1985) investigated the xanthate flotation of bornite, chalcocite, chalcopyrite, and pyrite as a function of the electrochemical potential. The flotation response of these sulfides, shown in Figure 8, is strongly dependent on the conditioning potential and occurs only under moderately oxidizing conditions. At a certain reducing potential, depending on the mineral, flotation ceases. Identification of the products of the sulfide mineral–ethyl xanthate interaction was conducted by linear sweep voltammetry and ultraviolet (UV) spectroscopy. They found that surface hydrophobicity appears to be metal xanthates or surface analogs of metal xanthates formed by chemisorption on chalcocite and bornite, dixanthogen on pyrite because the potential at which flotation begins is that at which xanthate is oxidized to dixanthogen, and both metal xanthate and dixanthogen (at higher potentials) on chalcopyrite. Trahar (1984) used chemical means to control the pulp potential and observed similar behavior in the flotation of sulfide minerals. In much of the early flotation literature, there are discussions of how low the adsorption density of collector is for flotation. A subsequent detailed study of the flotation of chalcocite and pyrite using the electrochemical flotation cell, together with determination of the amount of xanthate collector adsorbed, was carried out by Gebhardt and Richardson (1987). Their results for the flotation of these two sulfides individually are plotted in Figure 9. Collection of chalcocite occurs by chemisorption. The plots given in Figure 9 indicate that © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 23 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

23

100 14

Flotation Adsorption Chalcocite Pyrite –5

Flotation Recovery, %

5 × 10

M KET, pH 9.2 10

60 8 Monolayer 6

40

4

Adsorption Density, μmol/m 2

12

80

20 2

0 –0.8

0 –0.6

–0.4

–0.2

0.0

0.2

Potential, V (vs. SCE)

Adapted from Gebhardt and Richardson 1987.

FIGURE 9 Flotation recovery and ethyl xanthate adsorption as a function of potential for single-component chalcocite and pyrite beds at pH 9.2 in an electrochemical flotation cell at a collector concentration of 0.05 mM

when complete flotation of chalcocite occurs, the adsorption density of xanthate is in the range of their calculated monolayer, which would be the situation for any chemisorption process. Flotation of pyrite does not begin until the potential for dixanthogen formation occurs, and the dixanthogen coating that hydrophobizes pyrite could exceed monolayer coverage. The authors found that separation of mixtures of the two minerals by controlled potential could be achieved but under conditions different from that for the minerals alone. Chalcocite dissolution products had a deleterious effect on the chalcocite-pyrite separation by activating pyrite so that it floated below the potential at which xanthate is oxidized to dixanthogen, but dissolution products from pyrite had no effect on the flotation of chalcocite. A C T I VAT I O N A N D D E P R E S S I O N O F S U L F I D E A N D OX I D I Z E D MINERALS

With the exception of sphalerite and pyrrhotite (Sutherland and Wark 1955), sulfide minerals respond well to flotation with short-chain xanthate collectors. Even methyl xanthate will float galena, as can be seen from the plots given in Figure 3. Although there is no flotation with ethyl xanthate at low concentrations, Gaudin (1930) showed that sphalerite exhibits some response with amyl xanthate and significant flotation with heptyl xanthate, reaching 100% recovery at a collector addition of 0.2 kg/t. M.C. Fuerstenau, Clifford, and Kuhn (1974) conducted a detailed study of the flotation of sphalerite with short-chain (C2 to C6) xanthates and found a regular relation between flotation and collector concentration in solution. Flotation increased sharply once the necessary collector concentration was reached, indicating zinc xanthate precipitation enters into the process. Fifty percent recovery was achieved at an ethyl xanthate concentration in solution of 10 mM, whereas only 0.1 mM hexyl xanthate was necessary to achieve the same degree of flotation. Flotation response correlates exactly with the solubility products of the various zinc xanthate. Most importantly, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 24 Tuesday, January 2, 2007 7:36 PM

24

HISTORICAL ASPECTS OF FLOTATION

upon the addition of copper salts, sphalerite readily responds to flotation with low additions of short-chain xanthates. As discovered by Leslie Bradford in 1913 in Australia, this means of enhancing the floatability of sphalerite has become the classic example of activation in flotation technology, and since then, the inorganic salt, copper sulfate, has proved to be indispensable for effectively concentrating zinc sulfide ores by flotation. The general consensus on the activation of sphalerite by Cu(II) ions has been that the process involves an ion exchange mechanism in which cupric ions are exchanged for zinc ions according to the following reaction, driven by the extremely low solubility product of copper sulfide (CuS): ZnS ( s ) + Cu 2+ = CuS ( s ) + Zn 2+

(EQ 2)

In his 1930 paper, Gaudin commented that from a geological standpoint, the action of copper sulfate and similar salts on sphalerite had long been known, with some experiments having been conducted on the reaction as far back as 1911 by Rogers. The coating on sphalerite was shown to be covellite. In 1930, Gaudin reported on the results of some experiments conducted to determine the rate of copper uptake by sphalerite and found copper uptake to be rapid initially, followed by a slow process. He correctly attributed this to slow diffusion as the copper sulfide coating thickens. About 20 years later, Gaudin again became interested in this problem and with radioactive 64Cu showed that there indeed is a 1:1 Cu–Zn exchange, that the first two or three layers rapidly exchange, and that the reaction rate then continues by solid-state diffusion (Gaudin, Fuerstenau, and Mao 1959). Mellgren and colleagues (1973) measured the heat of adsorption of copper onto sphalerite under controlled deoxygenated conditions and found the heat of adsorption to be virtually identical to the heat of formation of CuS. If copper sulfate is added under alkaline conditions, hydrolyzed cupric species form and are rapidly adsorbed. Because CuS is far more stable than copper hydroxide, the adsorbed copper hydroxy compounds subsequently transform to CuS and thereby activate the sphalerite. In 1930, Gaudin also conducted some preliminary experiments on the deactivation of sphalerite with sodium cyanide, and this early interest in deactivation reactions were taken up again by Mao (Gaudin, Fuerstenau, and Mao 1959), who quantitatively investigated this phenomenon with 64Cu and several copper-complexing agents. Through contact-angle measurements and flotation experiments, Wark and Wark (Sutherland and Wark 1955) examined the activation behavior of sphalerite with a wide range of metal salts. Their results indicate that successful activators are salts of metals which, alone or with their sulfides, readily respond to flotation with thiol collectors. Salts of lead, cadmium, silver, and mercury fall within this category. D.W. Fuerstenau and Metzger (1960) investigated the adsorption of Pb(II) on sphalerite and clearly showed that the addition of zinc sulfate can prevent lead activation of sphalerite. In flotation plants, the oxidation of galena provides sufficient lead ions to activate sphalerite. It is for this reason that the addition of zinc sulfate became standard practice in lead-zinc flotation mills. In the flotation separation of complex sulfide ores, selective depression of different sulfide minerals is desired. Depressants include such inorganic reagents as hydroxyl ions, sodium sulfide, sodium cyanide, and alkali sulfites, and their mechanism of action has been the basis for much discussion through the years (Sutherland and Wark 1955; Gaudin 1957). In his 1985 paper, Chander interpreted depressant mechanisms in terms of whether the mineral is a reversible or passivated sulfide (MS). This is schematically shown in Figure 10. As pointed out by Chander, if the hydrophobic entity at the surface is X2, depression of the © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 25 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

Reversible Sulfides

25

Passive Sulfides

1.0

1.0 X2

MX

Eh, V

Eh, V

X

0.0

–

MS

0.0

MS

–1.0

–1.0 0

7

14

pH

Flotation Species Depressant Species

Metal/collector compounds MX, MX2

0

7

14

pH Oxidation product of collector X2

(a) Mineral oxidation product layer MO, M(OH)n, etc.

(a) Mineral oxidation product layer MO, M(OH)n, etc.

(b) Hydrophilic layer of mineral– depressant compound, MD

(b) Hydrophilic MD layer

(c) Removal of surface MX, MX2 by oxidation, reduction, or hydrolysis

(c) Removal of surface X2 by reduction

Adapted from Chander 1985.

FIGURE 10 Conditions for the flotation and depression of reversible and passivated sulfide minerals with thiol collectors

passivated sulfide mineral would occur if the mineral oxidizes to form a hydrophilic oxide or hydroxide layer, if the depressant decomposes X2 by reduction, or if the depressant reacts chemically or electrochemically to form a hydrophilic metal-depressant (MD) coating. If MX is the hydrophobic entity, depression of a reversible sulfide mineral would occur if the mineral oxidizes to a hydrophilic coating, if MX decomposes by oxidation or by reduction, if MX decomposes by hydrolysis, or if the depressant reacts chemically or electrochemically to form a hydrophilic coating. Each of the sulfide mineral–collector–depressant systems can be interpreted in terms of the foregoing mechanisms. Oxidized heavy-metal minerals such as anglesite (lead sulfate) and cerussite (lead carbonate) require large quantities of sulfhydryl collectors before they respond to flotation. Comparison of xanthate collector required for the flotation of chalcocite with that for malachite is quite striking. To achieve 50% recovery of 100 × 600 mesh chalcocite with potassium heptyl xanthate (mol wt = 230), a collector addition of 0.0022 mol/t is required, whereas for similar flotation of 100 × 600 mesh malachite, 4.4 mol/t is required (Gaudin 1957). The results of that same investigation permit comparison for KEX (mol wt = 160) at 20% recovery, namely 0.062 mol/t for chalcocite and 11.2 mol/t for malachite. These huge differences show the disparity between collector adsorption on chalcocite and chemical exchange reaction on malachite. Fleming (1952a) quantitatively studied the metathetical exchange of xanthate for carbonate with cerussite. Gaudin (1957) discussed the results of a study of the uptake of amyl xanthate by cerussite conducted years earlier, where a coating several hundreds of ions thick was formed. He commented that because the new phase does not have much crystal-chemical resemblance to the old solid phase (cerussite), the connection between the substratum and coating is fragile. To overcome the high collector consumption, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 26 Tuesday, January 2, 2007 7:36 PM

26

HISTORICAL ASPECTS OF FLOTATION

these oxidized minerals have long been sulfidized before adding the collector, a process that converts the mineral surface to a sulfide. The most common activator for this purpose is sodium sulfide (Gaudin 1957). The interaction of sodium sulfide with cerussite was investigated in detail by Fleming (1952b), who observed this to be a metathetical reaction with an equilibrium constant of about 3 × 1016. Sulfidization is similar to copper activation of sphalerite in that, initially, the uptake is rapid and the process then continues more slowly by diffusion as the sulfide layer increases in thickness. In contrast to the Cu–ZnS (copper–zinc sulfide) system, sulfidized coatings often do not adhere strongly (as noted previously by Gaudin for the collector coating). Herrera-Urbina, Sotillo, and Fuerstenau (1999) conducted detailed flotation and pulp potential measurements on the sulfidization reaction and flotation behavior of cerussite with KAX as collector. A summary of their complex findings can be seen in Figure 11. The effect of adding sodium sulfide to the system manifests itself in consecutive steps. The first addition of sulfide ions precipitates the lead ions in solution as PbS particles that coagulate onto the cerussite. After all the dissolved lead has been precipitated, the sulfide concentration in solution reaches that for depression of PbS flotation but is not yet high enough to initiate the metathetical exchange between carbonate and sulfide at the cerussite surface. Upon further addition of sodium sulfide, the sulfidization reaction begins to take place and flotation sharply increases. The surface reaction is rapid until the PbS coating reaches about 7 monolayers, at which point the residual sulfide ions in solution again reach the concentration for depression of PbS flotation, causing a pronounced cessation in flotation of the sulfidized cerussite. At this point, the concentration of sulfide ions (as measured by an ion-specific electrode) in solution builds up sharply while the PbS layer on cerussite continues to thicken, but at a reduced rate due to diffusion control. Thus, in industrial flotation practice, the addition of sodium sulfide must be closely controlled, which can be accomplished by measurement of pulp potential. 100

25

Cerussite 20

Flotation Monolayers Residual Sulfide 60

15

40

10

20

5

0 10–6

10–5

10–4

10–3

Number of Monolayers, Residual Sulfide, M × 10 5

Flotation Recovery, %

80

0.05 mM KAX 5 mM KNO3, pH 9.5

0 10–2

Sodium Sulfide Addition, mol/L

Adapted from Herrera-Urbina, Sotillo, and Fuerstenau 1999.

FIGURE 11 Flotation recovery of cerussite at pH 9.5 with 0.05 mM KAX as collector, the residual aqueous sulfide in solution, and the number of sulfide monolayers as a function of added sodium sulfide in 5 mM potassium nitrate open to the atmosphere

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 27 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

27

G A S E S I N F L O TAT I O N

Many people involved in flotation considered gas bubbles to be the crux of flotation (Rickard 1916). In 1928, Adams published the results of a detailed study conducted to ascertain whether the gas constituting the bubbles had any effect on the relative floatabilities of minerals. He carried out extensive investigations with the common sulfide minerals plus graphite and sulfur using a range of gases that included air, hydrogen, oxygen, nitrogen, and carbon dioxide. He concluded from his experiments that the floatability of minerals depends on the nature of the gas being used. However, in their extensive contact-angle investigation, Wark and Cox (1934) were somewhat surprised to find virtually the identical contact angle of 60° with ethyl xanthate collector for a range of gases, including air, oxygen, hydrogen, nitrogen, carbon dioxide, and sulfur dioxide. In the 1950s, Plaksin extensively studied the effect of nitrogen gases and dissolved gases on the flotation of various minerals, particularly the effect of the oxygen in water (Plaksin 1959; Klassen and Mokrousov 1963). Flotation in closed cells with nitrogen as the flotation gas is used industrially in the final separation stage of copper-molybdenum concentrates, the main savings being reduction in sodium sulfide consumption. For the flotation of an auriferous pyrite that oxidizes so readily that it rapidly loses its floatability, a nitrogen flotation system was installed at the Lone Tree mine in Nevada (Simmons et al. 1999). It was not until 1958, in a study of the effect of chemical reagents on the motion of air bubbles in water, that D.W. Fuerstenau and Wayman published the first results which clearly showed that the adsorption of organic surface-active agents (frothers) retarded the velocity of bubbles. Previous investigations had shown that bubbles in distilled water, before they begin to distort from spherical shape, rise faster than solids of the same specific gravity due to circulation within the bubble. By adding a frother (terpineol) to distilled water, the rise velocity of aged bubbles was found to slow down to that expected if they had been solid due to the surface tension gradient at the surface set up by the adsorbed frother molecules. This retardation was similar to that observed earlier for air bubbles in tap water. In distilled water in the absence of terpineol, solutions containing about 20 mg/L of potassium hydroxide (KOH), KCl, or KEX did not have any appreciable effect on the rise velocity of bubbles. However, air bubbles in industrial flotation pulps full of surface-active materials would be like bubbles in tap water and would not exhibit the rapid rise characteristic of bubbles in distilled water. The primary role of gas bubbles in flotation, of course, is to attach to hydrophobic mineral particles, levitate them to the surface, and maintain a froth layer that lasts long enough for effective recovery of the desired mineral particles. This requires organic compounds (frothers) that create a surface tension gradient at the air–water interface. Long-chained surfactants such as amines, sulfonates, and fatty acids function not only as collectors but also as frothers. Typical frothers are alcohols (ROH), either straight-chain with 6 to 9 carbon atoms or branched-chain compounds with 6 to 16 carbons. Pine and eucalyptus oils represent typical cyclic alcohols, with the active molecule being terpineol. In more recent decades, alkoxy-substituted paraffin-type frothers (such as triethoxy butane) contain no hydroxyl groups but gain their polarity from ether linkages. Another class of frothers includes the hydroxylated polyglycol ethers (King 1982). Interest in frothers has increased because of overall economic factors. For example, Klimpel and Hansen (1988) carried out considerable work on the effect of frother structure on the selectivity and recovery of minerals and found that with increasing branching, the maximum particle size that can be recovered decreases while at the same time selectivity increases. More research is needed in this area. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 28 Tuesday, January 2, 2007 7:36 PM

28

HISTORICAL ASPECTS OF FLOTATION

O X I D E A N D S I L I C AT E M I N E R A L S A N D T H E E L E C T R I C A L D O U B L E L AY E R

The immersion of a mineral into an aqueous solution produces a region of electrical inhomogeneity at the solid–solution interface; namely, an excess (+ or –) charge apparently fixed at the solid surface is exactly balanced by a diffuse region of ions of equal but opposite charge (termed counterions). This is the electrical double layer. Useful monographs that discuss the electrical double layer are those of Hunter (1981), King (1982), and Stumm (1992). Electrical double-layer phenomena have a limited role in the collection of sulfide minerals but are very important in nonmetallic mineral flotation, particularly with oxides and silicates. Figure 12 is a simple model of the electrical double layer at a mineral–water interface, showing the charge on the solid surface and the counterions extending as a diffuse layer out into the aqueous phase. This figure also shows the drop in potential across the double layer. The closest distance of approach of hydrated counterions to the surface, δ, is called the Stern plane. The total double-layer potential, or surface potential, is ψo, and the potential at the Stern plane is ψδ. From the Stern plane out into the bulk of the solution, the potential drops exponentially to zero. The electrokinetic or zeta potential, ζ, is the potential just outside the Stern plane where the diffuse layer is able to slip relative to the solid surface. In the case of ions that directly interact with surface sites, either chemically or by some other strong specific adsorption force, the adsorbed ions may lie closer to the surface at a plane called the inner Stern plane. However, these discussions simply refer to adsorption in the Stern plane. Several different parameters that quantify the electrical double layer are useful in interpreting flotation behavior, particularly the selective adsorption of collectors. This includes such factors as the magnitude of the surface charge, the point of zero charge (PZC) of the mineral, interfacial potentials, thickness of the electrical double layer, specific adsorption of collectors, and ion exchange phenomena (D.W. Fuerstenau and Healy 1972). The surface charge in systems of importance to flotation may arise in a number of ways. For example, the surface charge on an oil droplet or an air bubble may result from the adsorption of long-chained surfactant ions at the oil–water or air–water interface. In the case of layer-silicate minerals (such as clays and micas), because of substitution of Al3+ for Si4+ in the silica tetrahedra and Mg2+ for Al3+ in the octahedral layer of the crystal lattice, the surfaces of these crystal faces carry a constant negative charge that is independent of Stern Layer Surface Charge, σo

Counterions, σd δ

ψo

Shear Region

ψδ

Solid

Potential

ζ

δ

1/κ

Distance

NOTE: Shown are the surface potential, the Stern layer potential, the zeta potential at the slipping or shear plane, and the potential decrease to zero-out into the bulk solution. The distance 1/κ is the center of gravity of the diffuse layer of counterions.

FIGURE 12 Simple schematic of the electrical double layer showing the surface charge on the solid and the counterions adsorbed in the diffuse layer, with their closest distance of approach being the Stern plane

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 29 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

29

solution conditions. For two of the layer-silicate minerals, talc and pyrophyllite, the charge is exactly balanced within sheets that constitute the layer structure, the faces of these crystals are uncharged, and the minerals are naturally hydrophobic. In the case of salt-type minerals such as barite (barium sulfate), fluorite (calcium fluoride), argentite (silver sulfide), iodyrite (silver iodide), and so forth, the surface charge arises from the preference of one of the lattice ions for the solid relative to the aqueous phase. Equilibrium is attained when the electrochemical potential of these ions is constant throughout the system. Those particular ions that are free to pass between both phases and therefore establish the electrical double layer are called potential-determining (PD) ions. For oxide and silicate minerals, hydrogen and hydroxyl ions have long been considered to be PD ions (although there remains a difference in opinion as to how pH controls the surface charge on oxides), and adsorption/dissociation of surface hydroxyls is generally assumed to be the mechanism of developing a surface charge: M – OH + H + → M – OH 2+ M – OH + OH – → M – O – + H 2 O

(EQ 3)

If H+ and OH– are the potential-determining ions (PD+Z and express PD–Z are the PD ions of valence z in more general terms) for an oxide mineral, then the surface charge (σo) is simply given by σ o = zF ( Γ H + – Γ OH – )

(EQ 4)

where F is the Faraday constant and the adsorption density (the quantity in the brackets) of the PD ions is in moles per square centimeters. This can be measured by titration of a suspension of the mineral in water and is generally a complex function of the ionic strength and the activity of PD ions (pH for oxides) in solution. If the adsorption of the positive PD ion exceeds that of the negative PD ion, the surface of the mineral is positively charged, and vice versa if the adsorption of the negative PD ion exceeds that of the positive PD ion. If adsorption of counterions occurs only because of electrostatic interaction, then the diffuse layer charge, σd, is oppositely equal to the surface charge, σo. Such counterions are called indifferent ions. The relationships among surface charge, diffuse layer charge, and ψδ are given for a symmetrical electrolyte with ions of valence z (where z = z+ = z–) by the Gouy– Chapman equation, as modified by Stern (Hunter 1981): σ d = – σ o = ( 8εε o RTC ) 1 ⁄ 2 sin h ( zFψ δ ⁄ 2RT )

(EQ 5)

where ε is the dielectric constant of the liquid, εo is the permittivity constant, R is the gas constant, T is the temperature, and C is the concentration of indifferent electrolyte in solution. If the net adsorption density of PD ions is zero, the surface of the mineral is uncharged and the solid is at its PZC. With regard to flotation behavior, the single most important parameter that describes the electrical double layer of a mineral in water is its PZC, which is determined by a particular value of the activity, a, of the PD cation of valence z. The surface potential is considered to be zero at the PZC, and in the case of oxides (if the Nernst equation is assumed to be valid), its value at any other pH is given by RT ψ o = -------- ln [ a ( H + ) ⁄ a ( H + ) PZC ] = 0.059 [ pH PZC – pH ] volt zF © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 6)

flotation0.book Page 30 Tuesday, January 2, 2007 7:36 PM

30

HISTORICAL ASPECTS OF FLOTATION

The importance of the PZC is that the sign and magnitude of the surface charge has a major effect on the adsorption of all other ions, and particularly those charged oppositely to the surface. Typical values of the PZC for some of the oxide minerals are as follows (D.W. Fuerstenau and Healy 1972; King 1982): quartz (SiO2), pH 2; cassiterite (SnO2), pH 4.5; zirconia (ZrO2), pH 4; rutile (TiO2), pH 5.6; natural hematite (Fe2O3), pH 4.8–6.7; synthetic hematite, pH 8.6; corundum (Al2O3), pH 9.1; and magnesia (MgO), pH 12. The PZCs of silicate minerals are affected by crystal chemistry and selective leaching of metal cations or silica from the surface. Typical values of the PZC of some silicate minerals are as follows (D.W. Fuerstenau and Raghavan 1980; King 1982): kyanite (Al2SiO5), pH 7.8; zircon (ZrSiO4), pH 5.8; olivine [(Mg,Fe)2SiO4], pH 4.1; almandine [Fe3Al2(SiO4)3], pH 5.8; beryl [Be3Al2(Si6O18)], pH 3.4; spodumene [LiAl(SiO3)2], pH 2.6; rhodonite (MnSiO3), pH 2.8; talc [Mg6(Si8O20)(OH)4], pH 3.6; muscovite [K2Al4(Al2Si6O20)(OH,F)4], pH 1; and orthoclase [K(AlSi3O8)], pH 1.8. A useful method for the study of adsorption phenomena in solid–liquid systems is the measurement of electrokinetic potentials that result from the interrelation of mechanical fluid dynamic forces with interfacial potentials. In making electrokinetic measurements, the liquid phase is caused to move relative to the solid phase by the application of a mechanical force (streaming potential) or by an electric field (electrophoresis). More recently, electroacoustophoresis has also been used to evaluate zeta potentials, particularly with concentrated suspensions. From the appropriate theory, one evaluates the potential at the slipping plane, which generally is considered to be just outside the Stern plane. The potential at the slipping or shear plane is the zeta potential, ζ, and is often assumed to approximate the Stern plane potential, ψδ, although ζ < ψδ, as can be seen in Figure 12. The determination of zeta potentials has been a powerful tool in delineating flotation adsorption phenomena. Early on, not everyone believed in discussing electrokinetic phenomena in terms of zeta potentials. Because of his concern with regard to the equations connecting electrophoretic mobilities to zeta potentials, the distinguished surface and colloid chemist, Victor LaMer (1967), commented: It is for these reasons that I feel strongly that no scientific purpose is served by converting mobilities into zeta potentials until the more complicated connecting equations have been verified. Of course if you have something to sell, zeta potential is a much better advertising catch word than is electrophoretic mobility. The natives are mystified and admire with great awe the black box which gives the results on the dials.… This above shows that much of the recent ‘hullabaloo’ about zeta potentials is meaningless. The success of interpreting adsorption phenomena in terms and zeta potentials and all of the examples given by Hunter (1981) proved LaMer to be wrong. When counterions adsorb in the Stern plane through forces in addition to simple electrostatics, such counterions are considered to be specifically adsorbed. Flotation collectors for oxide minerals, for example, are such ions. Such phenomena as covalent bond formation, hydrophobic bonding, hydrogen bonding, solvation effects, and so forth, lead to specific adsorption. Because of their surface activity, the charge in the Stern plane can exceed the surface charge, and the sign of ψδ is reversed. Actually, indifferent and specifically adsorbed ions may lie in different planes because of ionic size and hydration. Chemisorbed ions may lie closer to the surface because they are dehydrated, in what is termed the inner Stern plane. These discussions will not differentiate between an inner and outer Stern plane. Stern first © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 31 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

31

suggested specific adsorption of counterions in terms of a Langmuir-type adsorption equation. For adsorption densities less than about 30% coverage, Grahame (1947) derived a Boltzmann-type relation, now generally called the Stern–Grahame equation, expressed here in terms of adsorption-free energies and adsorption densities, rather than in terms of charge and potentials as performed by Grahame: o ⁄ RT ) Γ δ = 2rC exp ( – ΔG ads

(EQ 7)

where Γδ is the adsorption density in moles per square centimeters, ΔGadso is the standard free energy of adsorption, r is the effective radius of the adsorbed ion, and C has to be the bulk concentration in moles per cubic centimeter. Another form of the relation expressing adsorption in the Stern plane is given as fractional coverage: C Φ o ⁄ RT ) -----------= ---------- exp ( – ΔG ads 55.5 1–Φ

(EQ 8)

In this equation, C is the concentration of adsorbate in moles per liter, and 55.5 is the number of moles of water in a liter. If ions are adsorbed at the Stern plane only because of electrostatic interactions, then the standard free energy of adsorption is given by o o ΔG ads = ΔG elec = zFψ δ

(EQ 9)

When an ion exhibits surface activity, such as the case for a flotation collector, then the standard free energy of adsorption has additional terms: o o ΔG ads = zFψ δ + ΔG spec

(EQ 10)

where ΔGspeco represents the specific interaction terms. These can be comprised of various contributions: o o o o o ΔG spec = ΔG chem + ΔG ho + ΔG hpb + ΔG solv + ΔG hpb* + ....

(EQ 11)

where the individual terms represent changes in the standard free energy due to chemical bonding, hydrogen bonding, hydrophobic bonding, and solvation effects, respectively. The o term ΔG hpb* represents the specific adsorption phenomena through surfactant chain interaction with a hydrophobic solid, such as talc or graphite, and would be absent for a hydrophilic mineral, such as quartz. Depending on the mechanisms involved in the interaction of the collector with the mineral surface, the contributions to the change in adsorption free energy can be essentially zero or have a finite value. This approach will be used in explaining the behavior of various types of collectors on nonmetallic minerals. There are two parts to the usual ionic collector—the charged head group and the hydrocarbon chain—and both can give rise to specific adsorption effects in the Stern layer. P H Y S I S O R P T I O N O F F L O TAT I O N C O L L E C T O R S

As first clearly defined by Taggart, the molecular structure of chemical flotation collectors requires that it has a polar group and a nonpolar hydrocarbon chain. The polar head group may react chemically with metal sites at the mineral surface or it may adsorb merely because of the sign of its electrical charge. The former type of interaction is chemisorption, as exemplified

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 32 Tuesday, January 2, 2007 7:36 PM

32

HISTORICAL ASPECTS OF FLOTATION

by xanthates on sulfide minerals and gold, whereas the latter type of interaction is one of physical adsorption. In this case, such factors as the size of the head group, its electrical charge, and whether the head group hydrolyzes are important. Of particular importance is the alkyl chain of the collector given that it enters into the process because of hydrophobic bonding phenomena. When surfactant ions or molecules in solution become sufficiently concentrated, they aggregate through association of their hydrocarbon chains into clusters containing about 50 to 80 surfactant ions called micelles. The charged heads orient outward to the solution so that the chains are effectively removed from water. The driving force is as a consequence of the desire for water molecules to hydrogen-bond with themselves, which results in a free energy decrease of about 1 kT per CH2 group (or 1 RT or 0.6 kcal/mol) removed from water (where k is the Boltzmann constant, R is the gas constant, and T is the absolute temperature). Thus, the number of ions in the micelle and the concentration at which micelles form, the critical micelle concentration or CMC, depends on the number of carbon atoms in the hydrocarbon chain of the surfactant. Based on his findings of how the adsorption of dodecylammonium ions affected the zeta potential of quartz, in 1953 D.W. Fuerstenau first proposed that similar phenomena can take place at the mineral–water interface. In their 1955 paper, Gaudin and Fuerstenau termed these surface aggregates hemimicelles because the charged heads would be oriented toward the mineral surface (at least until the zeta potential is reversed). As the adsorption density increases in the Stern plane, the adsorbed surfactant ions come sufficiently close together that they begin to associate into twodimensional aggregates similar to micelle formation in bulk solution. When hemimicelles begin to form, the free energy of adsorption is given by o o ΔG ads = zFψ δ + ΔG hpb = zFψ δ + Nφ

(EQ 12)

where N is the number of CH2 groups in the hydrocarbon chain, and φ is the free energy change on removal of one mole of CH2 groups from water. Through these same chain association effects, organic molecules (such as alcohols) can co-adsorb with adsorbed organic ions. The plots given in Figure 2 illustrate the effect of hemimicelle formation on a number of interfacial properties of quartz in aqueous solutions of DAA at pH 6–7, conditions under which the quartz surface itself is negatively charged. At about 10–4 M DAA, there are abrupt changes in the amount of aminium ions adsorbed and the zeta potential becomes sharply positive. This is because the hydrophobic bonding contribution to the adsorption free energy dominates, that is, Nφ > zFψδ . It is also seen that the onset of rapid flotation occurs under conditions where hemimicelles begin to form, clearly indicating that good flotation depends on strong collector adsorption in the Stern plane. If the driving force for the adsorption of physisorbing surfactants were only electrostatic, flotation with such collectors would be limited. Careful delineation of surfactant adsorption phenomena under controlled ionic strength was first conducted by Somasundaran and Fuerstenau (1966), and continued by Wakamatsu and Fuerstenau (1968). Combining zeta potential measurements with adsorption phenomena in the alumina–sodium dodecylsulfonate system clearly showed the existence of a three-step isotherm, which later was shown to be a four-step isotherm when surfactant concentrations were taken above that used in flotation, namely, all the way up to the CMC. This so-called S–F isotherm can be illustrated in terms of the results obtained for the adsorption of sodium dodecylsulfonate on alumina at pH 7.2 (Figure 13). The plots © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 33 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

33

9

7

12

pH

6 5

14

PZR

4

pH for C12 No. C for pH 7.2

3

Number of Carbon Atoms

10

8

16

2

18

10.0

III 80

II Contact Angle, degrees

1.0 Adsorption Density, μmol/m 2

–40

70

I 0.1

60 –20

50 40 30

0

20 10 0.01

Zeta Potential, mV

Contact Angle Adsorption Density Zeta Potential

20

0 40

Alumina 2 mM Ionic Strength pH 7.2 0.001 10–5

10–4

10–3

10–2

Sodium Dodecylsulfonate Concentration, M NOTE: Because the maximum contact angle is reached at the PZR, the dependence of the PZR on alkylsulfonate chain length at pH 7.2 and on pH for dodecylsulfonate are given in the upper portion of this figure.

Adapted from D.W. Fuerstenau and Wakamatsu 1973.

FIGURE 13 For alumina at pH 7.2 and 2 mM ionic strength, the effect of sodium dodecylsulfonate on the collector adsorption density, zeta potential, and equilibrium contact angle as a function of reagent concentration, showing three distinct adsorption regions

given in Figure 13 show the three regions of adsorption, which can be interpreted in terms of the Stern–Grahame equations. In Region I, the sulfonate ions adsorb individually by electrostatic interaction and ion exchange with chloride ions, and the zeta potential is therefore constant given that excess adsorption in the Stern plane is absent. Region II is characterized by a sharp change in the zeta potential and a sharp increase in the adsorption of sulfonate ions due to hemimicelle formation through the onset of the hydrophobic bonding contribution to specific adsorption. The boundary between regions II and III occurs precisely at the concentration where the zeta potential reverses (the PZR), at which point the electrical repulsion in the Stern layer begins to act against the hydrophobic bonding forces that are responsible for continued sulfonate adsorption. In Figure 13, the equilibrium contact angle (virtually identical to the liquid-receding contact angle) of an air bubble on alumina is also plotted as a function of the equilibrium sulfonate concentration in solution (after the results presented by D.W. Fuerstenau and Wakamatsu 1973). At the PZR, in all cases the contact angle reached its maximum and remained such as the concentration of sulfonate was © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 34 Tuesday, January 2, 2007 7:36 PM

34

HISTORICAL ASPECTS OF FLOTATION

C18

C16

C14

C12

C10

C8

C6

C4

Quartz

100

Flotation Recovery, %

pH 6–7 RNH3 Acetate 80

60

40

20

0 10–7

10–6

10–5

10–4

10–3

10–2

10–1

100

Alkylammonium Acetate Concentration, M

Adapted from D.W. Fuerstenau, Healy, and Somasundaran 1964.

FIGURE 14 Flotation recovery of quartz as a function of the concentration of alkylammonium acetates of various chain lengths at pH 6 to 7

increased. One interpretation is that after the PZR, with continued adsorption, some of the surfactant ions adsorb in reverse orientation. This behavior was observed for sodium dodecylsulfonate at different pH values and also for alkylsulfonates of different chain lengths at pH 7.2. These results are summarized in the upper portion of Figure 13, which shows the concentration where the maximum contact angle (the PZR) is reached for sulfonates of different chain lengths at pH 7.2 and for dodecylsulfonate at different pH values. This shows that the amount of collector can be reduced by increasing the alkyl chain length or by reducing the pH (alumina is positively charged below pH 9). Because the formation of hemimicelles depends on the length of the hydrocarbon chain, the flotation of quartz with alkylammonium salts should exhibit a regular dependence on chain length, similar to the well-known Traube rule. Figure 14 presents the flotation of quartz with amine collectors ranging from 4 to 18 carbon atoms. Because of the Nφ term, collectors having a longer hydrocarbon chain (greater N) adsorb more strongly and function effectively as flotation reagents at more dilute concentrations. It is interesting to note that a 4-carbon amine requires about 1 molar residual concentration for complete flotation of quartz, which is in sharp contrast to such strongly chemisorbing sulfhydryl collectors as ethyl xanthate on sulfide minerals. The role of the hydrocarbon chain in galena flotation with carboxylic acids is apparent by the results of Gaudin et al. (1928), given in Figure 1. The role of double bonds of flotation collectors has been the subject of numerous investigations and that will be presented when chemisorption phenomena is discussed later in this chapter. E L E C T R O S TAT I C M O D E L O F F L O TAT I O N

During the period from 1953 to 1956, D.W. Fuerstenau began to develop the concept that flotation collectors which physically adsorb must function as counterions in the electrical double layer, and that oxide mineral flotation with physisorbed anionic collectors should be © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 35 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

35

150 NaCl, M 10–5 10–4 10–3 10–2 10–1

PZC

Zeta Potential, mV

100

50

0

–50

Corundum 100

Flotation Recovery, %

80

60

40

Corundum 4 × 10–5M, C12 RNH3Cl RSO4Na RSO3Na

20

0 0

2

4

6

8

10

12

14

pH NOTE: The upper curve gives the zeta potential of corundum as a function of pH for a range of sodium chloride concentrations showing the PZC. The lower curve represents the flotation response of corundum with 0.04 mM SDS, sodium dodecylsulfonate, and DAC as a function of pH, showing the dependence of flotation on the PZC and the collector.

Adapted from Modi 1956; Modi and Fuerstenau 1960.

FIGURE 15

Dependence of the flotation of corundum on surface charge

appreciable only at pH values below the PZC and with cationic collectors only at pH values above the PZC. These concepts, which have been termed the electrostatic model of flotation, are briefly summarized here. Experiments to confirm these ideas were conducted by H.J. Modi as part of his doctoral thesis at MIT (Modi 1956). He determined the PZC of corundum by electrokinetics and conducted Hallimond tube flotation experiments with a variety of physisorbed collectors as a function of concentration and pH. Figure 15 presents the results of the very first experiments carried out to test these concepts, which were first presented in Modi’s doctoral thesis in 1956 and then published by Modi and Fuerstenau in 1960. The upper part of Figure 15 presents the zeta potential of corundum (synthetic sapphire), determined by streaming potential measurements as a function of pH with various additions of sodium chloride as indifferent electrolyte. All the curves intersect and cross at zero zeta potential at about pH 9, which is the PZC of this material. Flotation experiments were conducted at a solution concentration of 4 × 10–5 M with three different high-purity collectors: dodecylammonium chloride (DAC), sodium dodecylsulfate (SDS), and sodium dodecylsulfonate. The lower part of Figure 15 clearly shows that corundum responds to © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 36 Tuesday, January 2, 2007 7:36 PM

36

HISTORICAL ASPECTS OF FLOTATION

PZC Corundum Goethite Ilmenite Quartz

SDS DAC

pH 9.1 pH 6.7 pH 5.6 pH 2.4

10–4 M

Flotation Recovery, %

100

80

60

40

20

0 –3

–2

–1

0

1

2

3

4

pH–pHPZC

Adapted from D.W. Fuerstenau and Herrera-Urbina 1989.

FIGURE 16 Effect of the PZC on the flotation of several different oxide minerals with physisorbing anionic (SDS) and cationic (DAC) collectors

anionic collectors at pH values below the PZC where corundum is positively charged, and to the cationic collector at higher pH values where corundum is negatively charged. The pKa of dodecylamine is 10.4; at pH 10.4, the DAC in solution will be 50% aminium ions and 50% amine molecules. Under these conditions, flotation is maximal because of coadsorption of aminium ions and amine molecules. As the pH is raised to about 12, flotation drops sharply and ceases at pH 12.6. Under these conditions, there are insufficient aminium ions to bind the collector to the surface. This upper pH limit for flotation with primary amine collectors is virtually universal. Experiments with sodium dodecyl xanthate showed that this reagent also functions as a physisorbed collector for corundum in a manner similar to any other anionic long-chained surfactant. After returning to the University of Minnesota upon completing his doctorate at MIT in 1957, Iwasaki and several co-workers carried out similar detailed research on the flotation of a number of iron ore minerals with high-purity reagents (Iwasaki, Cooke, and Columbo 1960; Iwasaki, Cooke, and Choi 1960). The flotation response of oxide and silicate minerals to these types of collectors is characteristically similar to that presented in Figure 15, as can be seen from the plots given in Figure 16, which is a composite drawing showing the flotation response of four different oxides whose PZCs range from pH 2 to pH 9 with DAC and SDS as collectors. Two factors are involved in the electrostatic model of flotation: adsorption on a surface oppositely charged to the collector and a hydrocarbon chain sufficiently long to help anchor the physisorbed collector. In an excellent investigation of the flotation of iron oxides with 12- and 18-carbon collectors, Iwasaki, Cooke, and Choi (1960) showed that the flotation response of hematite with 12-carbon alkyl amines and sulfates is that predicted by the PZC. Their results, given in Figure 17, show that the 18-carbon surfactants continue to function as collectors at about © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 37 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

37

+4

Mobility, μm/sec/V/ cm

+2 0 –2

Hematite –4

100

Flotation Recovery, %

RNH3 RSO4 80

10–4 M C12

C12

60

C18

40

C18

20

0 0

2

4

6

8

10

12

14

pH

Adapted from Iwasaki, Cooke, and Choi 1960.

FIGURE 17 Influence of hydrocarbon chain length on the flotation of hematite as a function of pH with alkyl sulfate and aminium collectors having 12 (open symbols) and 18 (solid symbols) carbon atoms. The upper portion of this figure gives the electrophoretic mobility of the hematite showing that its PZC occurs at pH 6.7

3 pH units above or below the PZC, respectively. This means that hydrophobic bonding phenomena must be sufficient to overcome the electrical repulsion. Because each CH2 group contributes about 1.1 kT to the energy of hydrophobic bond formation, the increased contribution to the free energy of adsorption on lengthening the alkyl chain from 12 to 18 carbons is about 6.6 RT/mol. As an approximation, the potential at the surface increases 177 mV (7 RT) when the pH is changed by 3 units. The potential in the Stern plane would be less, but that potential is equivalent to 7 RT. Unfortunately, similar systematic flotation studies were never conducted with intermediate chain lengths in this system. C H E M I S O R P T I O N O F C O L L E C TO R S O N OX I D E S A N D S I L I C AT E S

If counterions are adsorbed only through such forces as electrostatic attraction and hydrophobic bonding (association between the hydrocarbon chains), the process is termed physical adsorption or physisorption. If the surfactant forms covalent bonds with metal atoms in the surface, then the process is called chemical adsorption or chemisorption. As already discussed, examples of physical adsorption in mineral–water systems include alkylammonium ions on quartz and other oxide minerals, and alkyl sulfates and sulfonates on alumina. Conditions can be such that lattice ions are displaced from their lattice positions by the adsorbate, giving rise to surface reaction (such as the uptake of xanthate by cerussite). Examples of © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 38 Tuesday, January 2, 2007 7:36 PM

38

HISTORICAL ASPECTS OF FLOTATION

chemisorption of collectors on oxide and nonmetallic minerals include the oleate–hematite, hydroxamate–hematite, hydroxamate–MnO2, and oleate–fluorite systems. Infrared spectroscopic studies have helped confirm the existence of calcium oleate, ferric oleate, ferric hydroxamate, and Mn-hydroxamate species on the surface, indicating chemisorption by covalent bond linkages. Chemisorption is attractive for several reasons: (1) flotation selectivity should be enhanced if the collector ions/molecules bind to surface sites on a single mineral; (2) there is reduced reagent consumption because of lower residual solution concentrations; and (3) fine particles should respond to flotation if the mineral surface is not highly charged and if the collector is not adsorbed on the air bubbles because of the collector having a shorter hydrocarbon chain or to residual collector concentrations in solution being more dilute. In nonmetallic mineral flotation, a commonly used collector is oleic acid, which has received considerable attention through the years. The results of an excellent and interesting investigation of oleic acid interaction with a variety of oxide and complex silicate minerals was presented by Polkin and Najfonow in 1964. In addition to flotation tests, they determined the amount of collector adsorbed with 14C-marked oleate, delineated chemisorption reactions with infrared spectroscopy, conducted leaching studies to remove surface metal ions, ascertained the effect of reagents and leaching on the zeta potentials of the minerals, and considered various regulating or modifying reagents to increase selectivity. Figure 18 presents their results for the effect of pH on the flotation recovery of nine different minerals with 1 kg/t of oleic acid as collector. This figure shows that the flotation response of these minerals to oleate collector is approximately the same (except for albite, a feldspar). Experiments with radioactively marked oleate showed the formation of durable multilayers on the eight minerals that readily float. They found that pretreatment of some of the minerals with acids provided a means for achieving selective flotation because various polyvalent metal 100

Oleic Acid 90

1kg/t 5

80

Flotation Recovery, %

70 6,7 60 1 50 2 40 5 6,7

30 20

8 9

3

10 3

1

2

4

9

0 1

2

3

4

5

6

7

8

9

10

11

12

13

pH

Adapted from Polkin and Najfonow 1964.

FIGURE 18 Similarity in the influence of pH on the flotation of a wide variety of minerals with 1 kg/t oleic acid: (1) columbite, (2) zircon, (3) tantalite, (4) ilmenite, (5) rutile, (6) garnet, (7) tourmaline, (8) albite, and (9) perovskite

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 39 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

39

ions were leached or removed from the surface of some of the minerals, which thereby removed the means by which oleate attached to the mineral. The metal ions at the mineral surface have a major role in chemisorption of oleate collectors, which will be discussed later in this chapter. The role of chemisorption in the flotation of hematite has received considerable attention through the years. Reagents, such as fatty acids and soaps, chemisorb on hematite, as is readily seen by flotation and collector adsorption occurring several pH units above a mineral’s PZC. Also, the shift in the zeta potential as a function of pH in the presence of soaps clearly shows strong adsorption until the pH is 2 or 3 units above the PZC. The first detailed study of chemisorption in the oleate–hematite system was the infrared (IR) spectroscopic investigation by Peck, Raby, and Wadsworth (1966). They found the appearance of a carbonyl band with maximum absorbance between 1,520 and 1,530 cm–1 that was the result of the formation of a surface cation–collector anion compound, with the cation bonded to the mineral structure and the anion to the cation. They also conducted flotation experiments as a function of pH, using 25 g of –65 mesh specular hematite with 6.3 mg (2.2 × 10–5 mol) of oleic acid as collector in 275 mL of water, giving a starting collector concentration of 8 × 10–5 M (the residual solution concentration is unknown). Their results are plotted in Figure 19 as a function of pH, the peak in absorbance is 0.30. Both the maximum in flotation and IR absorbance occur at pH 7.9. With a titration method, Peck and colleagues determined the PZC of their specular hematite sample to be pH 7.7. They proposed that the reactions of hematite with oleic acid can be expressed by the following equation: M-OH + HOl → M-OH···HOl M-OH···HOl → M-Ol + H 2 O

(EQ 13)

where HOl is oleic acid, M-OH are uncharged surface hydroxyls, and M-Ol are mineral surface sites with chemisorbed collector. 100

80

Hematite Oleate

Relative Acid-Soap Concentration Relative Infrared Absorbance

Flotation Recovery, %

Flotation Infrared Absorbance Flotation Acid Soap

60

40

20

0 4

5

6

7

8

9

10

11

pH

FIGURE 19 Correlation of hematite flotation recovery and infrared absorbance of adsorbed oleate as a function of pH (data from Peck, Raby, and Wadsworth 1966) and correlation of the flotation of hematite with the concentration of oleate acid-soap concentration (data from Kulkarni and Somasundaran 1980)

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 40 Tuesday, January 2, 2007 7:36 PM

40

HISTORICAL ASPECTS OF FLOTATION

Noting that there is almost a universal pH at which a wide range of minerals appear to respond to flotation with oleic acid, Kulkarni and Somasundaran (1980) proposed that it is the acid soap which is responsible for the strong adsorption of oleic acid or oleate soap in slightly alkaline environments. They carried out detailed analysis of the composition of oleic acid solutions and conducted a wide variety of adsorption and flotation experiments with hematite. Titration experiments showed the PZC of their hematite sample to be pH 7.1. Also in Figure 19, their results for the Hallimond tube flotation of hematite with 3 × 10–5 M potassium oleate and 8 × 10–5 M potassium nitrate are plotted together with the calculated concentration of the acid-soap dimer. The maximum in their plot corresponds to 4.8 × 10–8 M acid-soap dimer concentration. From the results of a series of investigations on chemisorption in nonmetallic mineral flotation, M.C. Fuerstenau and his colleagues (M.C. Fuerstenau and Palmer 1976) found correlations between the flotation response and the pH at which metal ions at the surface of the mineral hydrolyze. Figure 20 presents the results obtained by Palmer, Fuerstenau, and Aplan (1975) for influence of pH on the flotation of chromite with sodium oleate as collector. Chromite ideally is FeO·Cr2O3, but isomorphous substitution of Mg(II) for Fe(II) and Fe(III) for Cr(III) generally occurs in nature. The chromite used by Palmer and colleagues assayed 41.7% Cr(III), 8.0% Al2O3, 3.7% Fe(III), 7.1% Fe(II), 8.% Mg(II), plus some minor amounts of other elements. The two flotation peaks in the vicinity of pH 8 and pH 11 match the hydrolysis peaks of FeOH and MgOH, respectively. The peak at about pH 4 is most likely due to physisorption of oleate anions on positively charged chromite, given that the PZC occurs at about pH 7. Cr and Al probably do not participate in the surface hydrolysis reactions since Cr and Al are coordinated octahedrally, whereas the divalent cations are coordinated tetrahedrally with oxygen. Reaction with chemically hydrolyzed cations at the surface must differ from the chemisorption concept of Peck, Raby, and Wadsworth (1966) in that the hydrolyzed cation is probably dislodged from its lattice site before reacting with the oleate ion. Hydrolysis of the metal ion could free the metal ion from its lattice site and make it available for surface reaction. The interaction of both chrysocolla and hematite with K octylhydroxamate leads to a 100

Flotation Recovery, %

Chromite 80

60

40

1 × 10–4 M 5 × 10–5 M

20

Oleate 0 0

2

4

6

8

10

12

14

pH

Adapted from Palmer, Fuerstenau, and Aplan 1975.

FIGURE 20

Flotation of chromite as a function of pH and oleate concentration

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 41 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

41

change in color of the mineral surface, indicating that the surface hydroxamate product is quite thick and suggesting that a surface reaction takes place, building up multilayers of product. If a reagent chemisorbs, it can only reach monolayer coverage because the chemisorbing species interacts with a surface site. If the adsorbing species combines with a lattice ion, the species dislodges the ion from the lattice; this process is termed adsorption with surface reaction. A surface reaction can take place by reagents reacting with hydrolyzed species, as shown in the following reactions. Chemisorption M 2+ + OH – → M 2+ OH – M 2+ + X – → M 2+ X – Adsorption with surface reaction M 2+ + OH – → M 2+ + X – →

| ( MOH ) + | ( MX ) +

Surface reaction | ( MOH ) + + OH – → | ( MOH ) + + X – → | ( MX ) + + X – →

|M ( OH ) 2 | ( MX ) + |MX 2

T H E N AT U R E O F T H E H Y D R O C A R B O N C H A I N I N FAT T Y A C I D F L O TAT I O N

The role of double bonds of flotation collectors has been the subject of numerous investigations. Probably the most detailed investigation of the nature of the hydrocarbon (other than chain length) in flotation has, perhaps, been that of Cooke, Iwasaki, and co-workers at the University of Minnesota for fatty acid–iron ore flotation systems. In particular, they were concerned with the degree of unsaturation in the chain of 18-carbon fatty acid collectors and carried out extensive investigations with elaidic, oleic, linoleic, linolenic, and stearic acid as collectors for iron oxide minerals (Iwasaki, Cooke, and Choi 1960). For hematite, the degree of effectiveness at room temperature followed the order: elaidic > oleic > linoleic > linolenic. At room temperature (25°C), contact angles on hematite with 3 × 10–5 M fatty acid at pH 6 were found to be in degrees: stearic, 81; elaidic, 90; oleic, 86; linoleic, 80; and linolenic, 75. Because stearic acid has limited solubility at room temperature, the researchers also measured contact angles at 70°C, where stearic acid has appreciable solubility, and obtained the following results: stearic, 103; elaidic, 91; oleic, 88; linoleic, 81; and linolenic, 80. The greater the degree of unsaturation in the alkyl chain, the greater the degree with which water molecules interact with the chains and, hence, the less is their surface activity. Therefore, in a chemisorbing system, the hydrocarbon chain also plays a significant role. A very systematic study of the double bonds in the flotation of rutile was conducted by Purcell and Sun (1963). This included determination of zeta potentials by means of streaming potential measurements and flotation response with a Hallimond tube. Because of the © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 42 Tuesday, January 2, 2007 7:36 PM

42

HISTORICAL ASPECTS OF FLOTATION

thoroughness and rigor of their measurements, Figure 21 was prepared from their results for 0.1 mM reagent concentration to illustrate how double bonds in oleate, linoleate, and linolenate affected flotation and the zeta potential of rutile. The PZC of their rutile sample was pH 6.7. Chemisorption of the soaps causes the zeta potential to become much more negative than in sodium chloride solutions. As the pH is raised, linolenate joins the sodium chloride (NaCl) zeta potential curve at about pH 8, and the linoleate and oleate curves join the NaCl curve at still higher pH values. It is at this point that the electrical repulsion of the charged surface overcomes the adsorption tendency of the soap ions. Linolenate with three double bonds can interact with water molecules more frequently and hence is repelled from the surface at a lower pH. The lower portion of Figure 21 shows that the effect of pH on flotation response of rutile with these three collectors correlates exactly. Interestingly, in highly 100

Rutile NaCl, 0.1 mM Sodium Oleate Sodium Linoleate Sodium Linolenate

Zeta Potential, mV

50

0

–50

–100

100

Flotation Recovery, %

80

60

40

Rutile 20

Sodium Oleate, 0.1 mM Sodium Linoleate Sodium Linolenate

0 0

2

4

6

8

10

12

pH

Adapted from Purcell and Sun 1963.

FIGURE 21 Flotation of rutile in 0.1-mM sodium oleate, linoleate, and linolenate as a function of pH to show the influence of double bonds. The upper portion shows the zeta potential of rutile in 0.1-mM solutions of these chemisorbing collectors and also NaCl, which shows that the PZC of this rutile sample occurs at pH 6.7.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 43 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

43

acidic solutions, the flotation response of rutile to the three collectors is identical, with flotation sharply decreasing to zero at about pH 1. The three soaps form fatty acids that behave the same. At this point, brief mention is made of a planned attempt to utilize hydrocarbon chain configuration to effect flotation separations. Schulman and Smith (1953) found that a branched-chain fatty acid collector permitted separation of a cobalt mineral from a copper mineral, but not with a straight chain collector. Specifically, after first investigating force/ area curves for monolayers with metal salts in the substrate, they found that both chalcopyrite and carrolite float together with caprylic acid (a carboxylic acid with a 7-carbon alkyl chain), whereas with 2-ethyl hexanoic acid, only chalcopyrite floats. R O L E O F E X P E R I M E N TA L M E T H O D O L O G Y I N F L O TAT I O N FINDINGS

Figure 20 also shows that the collector concentration used in conducting an experiment can mask results. For example, the dual peaks in the alkaline pH region for chromite flotation are very apparent when the collector concentration is 5 × 10–5 M, but at 1 × 10–4 M, oleate hydrophobicity is great enough to swamp the reduced floatability at pH 10. Figure 22 illustrates how flotation time with the Hallimond tube can lead to different insights into the flotation behavior of a mineral. In their experiments, Iwasaki, Cooke, and Choi (1960) floated 100 × 150 mesh hematite with 10–4 M oleic acid for 5 minutes in a Hallimond tube. As can be seen from the plot shown in Figure 22, this long period of flotation yields 100% recovery between the lower and upper pH limits. M.C. Fuerstenau, Harper, and Miller (1970) floated 65 × 100 mesh hematite for 45 seconds in a Hallimond tube. They investigated conditioning time and found 10-minute conditioning yielded somewhat enhanced recovery over that obtained after 3 minutes of conditioning. Their results for 10-minute conditioning with 10–4 M potassium oleate are also shown in Figure 22. 100

Flotation Recovery, %

80

60

40

Hematite Oleate 20

Iwasaki M.C. Fuerstenau Somasundaran

0 0

2

4

6

8

10

12

14

pH

FIGURE 22 Illustration of how time for Hallimond tube flotation can accentuate or mask various aspects of the results, based on the flotation of hematite by various investigators (5 minutes by Iwasaki, Cooke, and Choi [1960]; 45 seconds by M.C. Fuerstenau, Harper, and Miller [1970]; and 10 seconds by Kulkarni and Somasundaran [1980])

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 44 Tuesday, January 2, 2007 7:36 PM

44

HISTORICAL ASPECTS OF FLOTATION

By using this shorter flotation time, the decrease in hematite flotation between pH 5 and pH 6 is very pronounced, followed by an increase again as the pH is decreased to below 5. This second flotation region is due to the physisorption of oleate ions on the positively charged mineral. On the other hand, Figure 22 also includes the results of Kulkarni and Somasundaran (1980), who floated hematite with 3 × 10–5 M potassium oleate as collector in a modified Hallimond tube for a flotation time of 10 seconds. This reduced collector concentration and very short flotation time yields only the optimum pH for flotation. Thus, in conducting flotation chemistry research, close attention should be given to the effect of the range of such variables as reagent concentration, conditioning, and flotation time in order to be certain that various effects are not masked. C O M PA R I S O N B E T W E E N P H Y S I C A L A N D C H E M I C A L ADSORPTION PROCESSES

In cases involving the physical adsorption of flotation collectors, one observes that flotation is strongly controlled by the PZC of the mineral. In other words, flotation takes place when the collector is ionic and when the mineral and collector are oppositely charged. This means that electrokinetic measurements can quite readily delineate conditions for flotation with physisorbing collectors. On the other hand, with chemisorbing collectors, the active species can be an ion (as already shown for oleate) or can be a neutral molecule, as will be shown for hydroxamic acid (a chelating agent). Moreover, when an anionic collector chemisorbs, it can adsorb on a negatively charged mineral surface until the surface potential is made sufficiently negative to prevent adsorption (by increasing the pH above the PZC in the case of oxides). A detailed investigation of the behavior of manganese dioxide provides a comparison of the complicated nature of physical and chemical adsorption in flotation processing (D.W. Fuerstenau and Pradip 1984). The PZC of this manganese dioxide (gamma MnO2) occurs at pH 5.6. Figure 23 (top, middle, and bottom) shows the effect of pH on the Hallimond tube flotation response of this oxide at three concentrations of three different collectors, namely sodium dodecylsulfonate, potassium octyl hydroxamate, and sodium oleate, respectively. The results shown in Figure 23 (top) indicate that the flotation of manganese dioxide with the anionic sulfonate behaves as expected for a physisorbing collector. Flotation only occurs when the pH is decreased below about pH 6, which happens in conditions where the mineral carries a positive surface charge and hence adsorbs the anionic sulfonate ions as counterions. Lower pH values (higher positive surface charge) are necessary for initiating flotation at lower collector concentrations. Figure 23 (middle) shows the effect of pH on flotation with a chelating agent (hydroxamate), which strongly coordinates with manganese ions at the mineral surface. Although chelating agents have been investigated as flotation reagents for 60 years or more (Gutzeit 1946; Marabini, Cases, and Barbaro 1989; Somasundaran and Nagaraj 1984; D.W. Fuerstenau, Herrera-Urbina and McGlashan 2000), hydroxamates have received, by far, more attention than any other single chelating agent. Coordination of the nitrogen-oxygen atoms to manganese takes place as shown schematically:

Mn OH +

O

C

R

HO

N

H

O

C

R

O

N

H

Mn

+ H2O

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 45 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

45

100

MnO2

PZC

Flotation Recovery, %

80

10–6 M SDS 10–5 10–4

60

40

20

0 100

MnO2

Flotation Recovery, %

80

1 × 10–5 M HXm 1 × 10–4 3 × 10–4

60

40

20

0 100

MnO2 Sodium Oleate

Flotation Recovery, %

80

5 × 10–5 M 1 × 10–4 5 × 10–4

60

40

20

0 0

2

4

6

8

10

12

14

pH

Adapted from D.W. Fuerstenau and Pradip 1984.

FIGURE 23 Influence of pH and collector type on the flotation of manganese dioxide at the various reagent additions (top: physisorbing SDS; middle: chemisorbing potassium octyl hydroxamate; bottom: chemisorbing/physisorbing sodium oleate)

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 46 Tuesday, January 2, 2007 7:36 PM

46

HISTORICAL ASPECTS OF FLOTATION

In this chemisorbing system, flotation takes place at pH values where the surface of the mineral is highly negatively charged. The maximum in floatability occurs at about pH 9, similar to the case for hematite, which is approximately the pKa of hydroxamic acid. Thus, optimum flotation appears to take place where the neutral molecules and hydroxamate anions are about equal, indicating that both coadsorption and surface hydroxylation may be responsible for the pronounced chemisorption at pH 9. As the pH is raised above 9, flotation decreases because the surface is so highly negatively charged that the hydroxamate anions are repelled. The flotation response of manganese dioxide with oleate collector is particularly complicated, similar to the behavior of hematite, with two flotation peaks being present in this system. The peak in the alkaline region occurs under conditions where the solid surface is highly negatively charged (PZC = pH 5.6); therefore, strong chemical adsorption forces must be operative, typical of oleate–mineral systems previously discussed. Under these conditions, the carboxyl ion may be chemically binding to Mn surface sites or the acid-soap dimer could be extremely surface-active. In the vicinity of pH 5, there is little flotation of manganese dioxide with oleate. However, as the pH is lowered even more, a second flotation maximum occurs. This must now be the result of physical adsorption of oleate anions on a positively charged surface. Since the pKa of oleate is about pH 4.7, the actual adsorbing species must be oleate ions together with oleic acid molecules. At pH values less than about 3, most of the oleate has been transformed to molecular oleic acid, which is not the reactive species, and flotation ceases (similar to the case for amines at high pH). As can be seen from the plots for oleate in Figure 23, the total concentration of oleate in the system must be sufficiently high to provide enough oleate ions for adsorption at low pH values. Liquid oleic acid droplets form at low pH values and higher oleate concentrations, but that will not be taken into account in this chapter. M.C. Fuerstenau, Harper, and Miller (1970) compared the flotation of finely ground hematitic ore using octylhydroxamate as collector with using oleic acid as collector. In a cited example, the ore was ground to 70% minus 15 μm, and an addition of 0.2 kg/t of hydroxamate collector resulted in a final concentrate recovery of 86% at a grade of 64% Fe. The ability of a chelating agent such as hydroxamate to successfully float fine particles has much to do with the fact that the bubbles will not be highly charged. A C T I VAT I O N I N N O N M E TA L L I C M I N E R A L F L O TAT I O N

Activation in nonmetallic mineral (oxides and silicates) flotation is the result of strong adsorption in the Stern plane of multivalent species that can reverse the sign of the zeta potential and cause the formation of a triple layer: the first layer will be the charge on the surface of the mineral itself; the second layer is the oppositely charged Stern layer; and the third layer is collector counterions charged similarly to the mineral surface. Thus, activation of a negatively charged oxide for flotation with an anionic collector requires the strong adsorption of inorganic cations as activator. Activation in these systems does not produce a new surface, as in the case of copper activation of sphalerite or the sulfidization of oxidized lead or copper minerals. Industrially, calcium or magnesium salts are used for the activation of a mineral such as quartz for oleic acid flotation. Hydrolyzed multivalent metal ions are strongly adsorbed, and this affinity for a mineral surface has long been recognized. The papers of James and Healy (1972) provide a method for quantification of these adsorption processes. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 47 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

47

In 1928, Gaudin et al. were the first to find that pH most markedly affects the flotation of quartz activated by ferric and cupric ions when oleate is used as a collector. For quartz activated with ferric chloride, sodium oleate as the collector, and terpineol as the frother, they found that marked flotation occurs in the pH range of about 5 to 7. In 1934, Kraeber and Boppel (in Sutherland and Wark 1955) found that pure quartz was not floated by a sulfonated castor oil but that it responded well if treated by a number of different heavy-metal salts over specific pH ranges for each activating metal salt. Hydrolysis of the activating cation undoubtedly is important in the process. In the case of Fe(III) activation, they found a rather wide pH range for quartz flotation for their system, pH 2 to 9. In 1966, Mackenzie conducted a detailed investigation of the effect of ferric chloride on the zeta potential of quartz, and for an addition of 5.7 × 10–5 M ferric chloride, he observed a marked increase in the zeta potential at about pH 3, reaching a maximum at about pH 5, and then reversing sign at pH 7.3. Hergt et al. (in Sutherland and Wark 1955) used contact angles to delineate the critical bubble contact region for Fe-activated quartz with sodium hexadecylsulfate as collector, and they observed the lower pH required for contact to be about pH 3.3 and the pH at which contact ceased to be about 7.5. Schuhmann and Prakash (1950) presented the results of a comprehensive investigation of activation in the soap flotation of quartz, with vacuum flotation being their main research tool. For ferric chloride as activator and oleic acid as collector, they found that the flotation range was between pH 3 and pH 12. Perhaps with the vacuum flotation test procedure used by Schuhmann and Prakash, the ferric hydroxide precipitate itself was responding to flotation and carrying with it the quartz particles. Subsequently, M.C. Fuerstenau and associates (e.g., in M.C. Fuerstenau and Palmer 1976) conducted detailed systematic investigations of hydrolyzing phenomena involved in activation phenomena. Figure 24 presents the results of M.C. Fuerstenau and Palmer (1976) for the flotation of quartz, with a sulfonate of mol wt 450 as collector, as a function of pH for various activating metal ions. For clarity, only the initial flotation edge is shown in this pH of Hydroxo Complex Formation FeOH2+

AlOH2+

PbOH+

MnOH+

MgOH+

CaOH+

100

Flotation Recovery, %

80

60

Fe3+

Al3+

Pb2+

4

6

Mn2+

Mg2+

Ca2+

40

20

0 0

2

8

10

12

14

pH

Adapted from M.C. Fuerstenau and Palmer 1976.

FIGURE 24 Minimum flotation edges for the flotation of quartz as a function of pH with 0.1 mM sulfonate collector and 0.1 mM metal ion activators

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 48 Tuesday, January 2, 2007 7:36 PM

48

HISTORICAL ASPECTS OF FLOTATION

figure. As can be seen, there is a distinct correlation of flotation response with the pH of the hydroxy complex formation. However, in their work, M.C. Fuerstenau and Palmer found both the upper and lower flotation edge to be steep, in contrast to the work previously discussed. With their system, the pH range for 90% flotation recovery with 10–4 M sulfonate collector and 10–4 M metal salt as activator was as follows: Fe(III), pH 2.9–3.8; Al, pH 3.8– 8.4; Pb, pH 6.5–12.0; Mn(II), pH 8.5–9.4; Mg, pH 10.9–11.7; Ca, pH 12.0 and greater. Their experimental techniques clearly delineate the activation pH for optimum floatability. In silicate mineral flotation, activation by anions has been important. Specifically, fluoride has been widely used as an activator in the cationic flotation of feldspar from quartz, and as a depressant in the anionic flotation of beryl and spodumene (Smith 1963). By measuring contact angle on quartz and microcline as a function of pH in the presence and absence of sodium fluoride with DAC as collector, Smith showed that there is a specific pH range in which microcline (a feldspar) is activated and quartz is depressed, as can be seen from the results given in Figure 25. In the absence of fluoride, quartz and feldspar behave identically with the cationic collector. Although the contact angles on quartz are not affected by the addition of fluoride, those on microcline change significantly. The activation

Quartz

60

Microline

10–2 M NaF No NaF 4 × 10–5 M DAC

50

Contact Angle, degrees

40

30

20

10

0 0

2

4

6

8

10

pH

Adapted from Smith 1963.

FIGURE 25 Contact angles on quartz and microcline (feldspar) in aqueous dodecylamine as a function of pH in the presence and absence of NaF, showing the activation of feldspar at low pH

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 49 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

49

of microcline can only take place through strong specific adsorption of fluoride ions on the aluminum sites at the feldspar surface, such that the surface carries a high enough negative charge to attract the physisorbed cationic collector ions. The high insolubility of aluminum fluoride suggests that fluoride ions may strongly chemisorb on the aluminum sites at the surface. Shergold, Prosser, and Mellgren (1968) found that inorganic anions function as activators for the flotation of hematite at pH 1.5 with 0.2 mM dodecylamine. At concentrations greater than 3.5 mM, bubble-pickup tests showed firm bubble–particle attachment with sodium fluoride (NaF) and NaCl but no attachment with sodium nitrate (NaNO3), sodium thiocyanate (NaCNS), or sodium acetate (NaCH3COO), and weak attachment with sodium sulfate (Na2SO4). Batch 1-kg flotation tests with a hematite ore and with synthetic hematite–quartz mixtures showed excellent flotation separation with hydrochloric acid (HCl) and also with sulfuric acid (H2SO4) at pH 1.5, with or without a small addition of ferric chloride (FeCl3). Hematite recovery was about 90% at almost 100% hematite grade. Evidently, a highly negative surface must be produced by adsorption of the activating anions or surface complexes for the cationic collector to adsorb. In the case of oxide and silicate minerals, because collector ions function as counterions in the double layer, their adsorption density will depend on competition with any other counterions in solution. Thus, the presence of excessive amounts of dissolved salts can inhibit flotation because inorganic ions similarly charged to the collector can then act as a depressant. In the case of the flotation of goethite with quaternary amine salts at pH 11, adding 0.03 M NaCl will reduce flotation to about nil (Iwasaki, Cooke, and Colombo 1960). Onoda and Fuerstenau (1965) carried out a detailed study of the depression of quartz flotation with DAA as collector and showed that Ba2+ and Na+ both inhibit flotation, the effect being considerably greater with the divalent salt, as can be seen from the plots given in Figure 26. Ion exchange as related to flotation would be controlled by the phenomena involved in the Stern–Grahame equation. In the absence of specific adsorption, the Quartz 100

60

80

60 40

40

Zeta Potential, –mV

Flotation Recovery, %

80

0.1 mM DAA pH 6.5

20 Flotation ζ-Potential

20 NaCl BaCl2 0 10–7

10–6

10–5

10–4

10–3

10–2

10–1

0 100

Added Salt Concentration, M

Adapted from Onoda and Fuerstenau 1965.

FIGURE 26 Effect of adding barium chloride and sodium chloride on the flotation and zeta potential of quartz in 0.1-mM DAA solutions at pH 6.5, showing the depression of flotation by ionic competition

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 50 Tuesday, January 2, 2007 7:36 PM

50

HISTORICAL ASPECTS OF FLOTATION

exchange is controlled directly by the ratio of their concentrations in the bulk solution. In the case of one ion being multivalent, the valence effect given in that relation is relevant. Most importantly, if one or both of the ions exhibit specific adsorption, then the magnitude of the specific adsorption free energy must be taken into account. The concentration of DAA is 0.1 mM, which is just below that at which hemimicelles begin to form. However, barium ions can reverse the zeta potential at 0.03 M, indicating specific adsorption and meaning that both specific adsorption and valency enter into these depression phenomena. Onoda and Fuerstenau presented a tabulation of the experimental and calculated concentrations of sodium and barium ions that would lead to depression. As an example, to reduce the recovery to 70%, the concentration of barium chloride needed was 0.15 mM, whereas the concentration was 6 mM for sodium chloride. The calculated concentration ratio of Na+/Ba2+ was 62 and the experimental ratio 40, which is reasonably close for such simplified calculations. N AT U R A L F L O TA B I L I T Y

Most minerals are hydrophilic and require collectors for flotation. Because sulfide minerals readily oxidize when exposed to air, they are hydrophilic under the usual conditions encountered in processing. In the case of oxide and silicate minerals, all but two are hydrophilic. As for the sparingly-soluble salt minerals, all are hydrophilic because of broken ionic bonds that form their surface. However, the silver halides exhibit some natural hydrophobicity. The Bessel brothers in 1877 were the first to utilize natural floatability in their process for upgrading graphite ores. It was A.M. Gaudin who, in his texts of 1932 and 1957, postulated that the natural floatability or nonpolar character of certain minerals was the result of not breaking primary bonds upon forming their surfaces. This condition would be met with crystals held together by dispersion forces (van der Waals bonds). Examples would be molecular crystals such as sulfur, which consists of S8 rings held in a crystal by dispersion or van der Waals forces, as well as paraffin. Most of the nonpolar minerals are sheet crystals in which their crystal chemistry results in individual layers that are electrically neutral, with dispersion forces acting between the sheets to hold them together. The faces of such crystals are nonpolar, but the edges would be polar given that primary ionic or covalent bonds are broken in forming edge surfaces. Examples of such minerals are graphite and two of the layer silicates, talc and pyrophyllite, which on their cleavage plane present uncharged siloxane rings. Two sulfide minerals exhibit natural hydrophobicity, namely stibnite and molybdenite, also a layer mineral. As pointed out by Gaudin, boric acid (H3BO3) has a layer structure in which all potential hydrogen-bonding OH are internally satisfied and not available for hydrogen bonding with water molecules. It is the strong tendency of water molecules to hydrogen-bond with each other that provides the energy for water to be displaced from nonpolar surfaces by an air bubble or oil droplet. Again, Taggart was at odds with Gaudin over the concept of natural floatability. In 1934, Taggart, del Guidice, and Ziehl wrote, “It may seem odd, at this date, to resurrect so old a friend as the inherent floatability of minerals, and would be so had not a recent writer unearthed the ancient fossil for us and dressed it up in modern appearing clothes.… Consequently, we dissent vigorously and finally from any idea of inherent natural floatability.” The fact that natural hydrophobicity occurs when the cohesive energy of water is greater than the dispersion forces interacting between water and a solid was not known at that time. Taggart went to great lengths to prove that natural floatability did not exist. It is interesting that Fowkes and Harkins in 1940 (Harkins being an extremely meticulous surface chemist) © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 51 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

51

published their measurements of contact angles using a carefully designed tilting plate apparatus rather than the captive-bubble technique. They reported contact angles for water on the following hydrophobic solids: Ceylon graphite, 85.7°; talc, 87.8°; stibnite, 84.2°; and paraffin, 108° to 111°. They obtained the same value of the contact angle whether the liquid was advancing or receding before the measurement was made, and commented that, “The ‘advancing’ and ‘receding’ angles obtained by nearly all investigators other than those in this laboratory are due to improper preparation of the surface and poor techniques in making measurements.” The contact angles with paraffin oil were found to be zero and these solids considered to be oleophilic by Fowkes and Harkins. The addition of such compounds as butyl alcohol, butyl amine, butyric acid, propionic acid, and acetic acid lowered the contact angles on paraffin and graphite. Laskowski (1986) has reviewed fundamental aspects of the relation between natural hydrophobicity and floatability. Using an analysis involving concepts of the work of adhesion, Laskowski and Kitchener (1969) concluded that all solids would be hydrophobic if they did not carry polar or ionic groups on their surface. It is the high cohesive energy of water due to its hydrogen bonding that gives rise to hydrophobicity (and to the formation of micelles in solution and hemimicelles at a mineral surface). The role of the flotation collector is to cover the polar sites on mineral surfaces that are formed by the breakage of primary bonds to prevent hydrogen bonding of water to surface sites. In the flotation of talc, graphite, or molybdenite, the addition of a neutral oil is used to enhance the hydrophobicity of the mineral. In many instances, depression of these minerals is desired, and the standard depressants are hydrophilic polymers that adsorb and inhibit bubble attachment. The flotation of coal has become important over the last few decades, but it is a naturally floatable material whose surface is very susceptible to oxidation that can severely reduce its hydrophobicity. Figure 27 is presented to illustrate the flotation response of a naturally floatable mineral, talc, without the addition of a collector as a function of pH. Although not shown, the isoelectric point (IEP) of this talc sample occurs at pH 2. At pH 1, the zeta potential is +20 mV, and in the pH range of 4 to 8, the zeta potential is about –30 mV and then becomes more negative, to about –50 mV at pH 10 and above. The greater magnitude of the negative zeta potential is responsible for the decrease in flotation observed above pH 10. The induction time correlates well with flotation response. Naturally floating minerals, such as talc, graphite, and molybdenite, are depressed industrially by the addition of hydrophilic polymers. Such polymers are adsorbed at the surface of a hydrophobic mineral by hydrophobic bonding phenomena (by adsorbing, they effectively increase the hydrogen bonding of water molecules near the interface). The adsorbed hydrophilic polymer prevents bubble attachment because water molecules now can hydrogen-bond to the polymer. Figure 27 shows the effect of 8.1 mg/L of dextrin on the flotation response of talc. Independent of pH, this small amount of added dextrin reduces the flotation recovery to 40%. Talc can also be depressed by hydrolyzing trivalent cations, as shown by M.C. Fuerstenau, Lopez-Valdivieso, and Fuerstenau (1988) in a detailed electrokinetic and flotation study with Fe(III), Al(III), and Cr(III). As the pH is increased, the cations hydrolyze and sharply change to reverse the sign of the zeta potential but do not affect flotation, apparently because the hydroxo complex species adsorb onto the polar edges of the talc particles. However, upon further increase of the pH, the metal hydroxide precipitates, the zeta potential becomes positive, and flotation ceases. As the pH is increased further, the zeta potential of the precipitated hydroxide becomes negative again, and talc once more responds to flotation. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 52 Tuesday, January 2, 2007 7:36 PM

52

HISTORICAL ASPECTS OF FLOTATION

100

0

Talc 0.002 M KNO3

40

80 60 120 40 160

Induction Time, μsec

Flotation Recovery, %

80

20 Flotation Without Collector Induction Time Without Collector Flotation with Dextrin

200

0 0

2

4

6

8

10

12

240 14

pH

Adapted from D.W. Fuerstenau and Huang 2003.

FIGURE 27 Influence of pH on the flotation of talc and induction time in 2 mM potassium nitrate without the addition of a collector, and the depression of talc by the addition of 8.1 mg/L dextrin

Depression is due to heterocoagulation of the hydroxide onto the nonpolar face of the talc particles. The driving force for this must be the displacement of water molecules from the talc face, thereby allowing water molecules that were at the interface to resume hydrogen bonding with each other and with the hydroxide coating on the talc. Once the zeta potential of the hydroxide particles becomes negative again, the hydroxide particles redisperse, once more providing a nonpolar talc face. S PA R I N G LY - S O L U B L E S A LT M I N E R A L S

The first systematic research on the flotation of salt-type minerals was conducted by Gaudin and Martin (1928) on a wide range of carbonates, namely, calcite, magnesite, rhodochrosite, siderite, malachite, and azurite. They found that aliphatic fatty acids are effective collectors for these minerals and that there is a pronounced systematic chain-length effect. In general, the carboxylic acid needed to have at least 7 carbon atoms (heptylic acid), although chains as short as propionic collected azurite and malachite. The industrial workhorse collector for salt-type minerals is oleic acid, which interacts with the mineral surface by chemical exchange. Gaudin and Martin (1928) conducted experiments at higher temperatures with longer-chained fatty acids and found marked increase in flotation by raising the temperature from 25°C to 70°C. An increase with rising temperature is a direct indication of activated chemical reaction taking place—the chemical exchange reaction of carboxylate with carbonate ions in the crystal lattice. More recent measurements of flotation and zeta potentials by Somasundaran and Agar (1967) showed that DAC and SDS are physically adsorbed by calcite, at least until solubility products are exceeded, as evidenced by plots similar to those shown in Figure 16. The flotation of sparingly-soluble salt minerals such as apatite, barite, calcite, and fluorite appears to be controlled by chemical interaction of the carboxylate collector with mineral © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 53 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

53

cations, and collector interaction appears to be controlled by solubility criteria. Many researchers have conducted IR spectroscopy studies on all of these salt-type minerals, and the results generally show the expected metal–carboxylate bond from chemical exchange, chemical reaction, or chemisorption when oleate is used as the collector. Some physical adsorption has been observed at lower pH values. However, with these types of systems, various researchers have found that collector uptake may go far beyond a monolayer and that a new soap phase may form at the surface. Thus, the collection process perhaps should be considered as one of surface reaction rather than one of adsorption. In fact, Kitchener (1984) wrote the following about the flotation of salt-type minerals: The main problem over soaps is to identify the form of the product, which, in this case, seems very unlikely to conform to the naïve monolayer model. There is no doubt that, given a chance, calcium minerals, for example, would go on reacting with sodium oleate almost indefinitely. This is not reversible physical adsorption; Atademin has shown that supposed ‘adsorption isotherms’ for such systems are almost certainly abstraction-by-precipitation curves. Industrial processes for the recovery of salt-type minerals from oxide and silicate gangue minerals are quite straightforward. However, separating salt-type minerals from each other is complex and difficult. For example, several are calcium salts that interact quite similarly with the collector, or they have slightly different solubilities such that dissolved anions (or cations) can react with the surface of the less soluble mineral, causing a surface transformation that leads to reduced selectivity. Flotation separations of these minerals are effected by utilizing a number of modifying agents that make insoluble inorganic compounds with the alkaline-earth cations in the minerals, including silicate, fluoride, phosphate, and dichromate, or by the addition of organic molecules such as tannins and starches that coat the surface with a hydrophilic layer of material. Pugh and Stenius (1985) presented results of a detailed study of the electrokinetic behavior, solubility, and flotation of fluorapatite, calcite, and fluorite with sodium oleate. Figure 28 presents their results for the flotation of these three minerals as a function of sodium oleate concentration at pH 10. This figure shows that the amount of oleate required as collector follows the order fluorite < apatite < calcite. Fa et al. (2003) also determined the flotation response of fluorite and calcite as a function of sodium oleate concentration and obtained fairly similar results, namely, to obtain 50% flotation recovery of fluorite, 3 × 10–6 M oleate was required and 4 × 10–5 M oleate for calcite. Using molecular modeling, Pradip and Rai (2002) carried out computations to model the interactions of oleic acid with calcium minerals and calculated the interaction energies for oleic acid with these calcium mineral surfaces to be –52.6, –46.8, and –40.2 kcal/mol for fluorite, fluorapatite, and calcite, respectively. They also calculated the interaction energies for water with these minerals, which is lower in each case, indicating that oleic acid will replace water at the mineral surfaces. Their calculated interaction energies give the same order as the observed flotation response. Fa and colleagues suggest that the lower floatability of calcite is due to the low density of calcium sites at the carbonate surface. Aqueous solutions of these three minerals are complex, because all of the ions involved are subject to hydrolysis, depending on pH. Fa et al. (2003) listed the solubility products of these three minerals as follows: fluorapatite [Ca10(PO4)6F2], 6.3 × 10–137; fluorite (CaF2), 5.0 × 10–11; and calcite (CaCO3), 4.6 × 01–9. Their measured solubilities of calcium ions in solution after 15 minutes were 2.5 × 10–5, 1.3 × 10–4, and 1.5 × 10–4 M for fluorapatite, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 54 Tuesday, January 2, 2007 7:36 PM

54

HISTORICAL ASPECTS OF FLOTATION

100

Flotation Recovery, %

80

Interaction Energy Calculation by Molecular Modeling:

Fluorite –52.6 kcal/mol

Apatite –46.8 kcal/mol

Calcite –40.2 kcal/mol

Fluorite Apatite Calcite pH 10

60

40

20

0 10–6

10–5

10–4

10–3

Sodium Oleate Concentration, mol/L

Adapted from Pugh and Stenius 1985; Pradip and Rai 2002, 2003.

FIGURE 28 Flotation recovery of fluorite, apatite, and calcite as a function of the concentration of sodium oleate at pH 10

fluorite, and calcite, respectively. Fa and colleagues report high adsorption densities of oleate on these minerals, the amount being related to their rate of dissolution and solubility, namely 11, 100, and 300 μmol/m2 for apatite, fluorite, and calcite, respectively. Because monolayer coverage is 6 μmol/m2, any oleate uptake above that amount cannot be chemisorption but must be the calcium oleate soap resulting from surface reaction. The solubility product of calcium oleate is 3 × 10–16, which indicates that calcium soap will be precipitated on addition of sodium oleate in alkaline solutions. Free and Miller (1996) investigated the precipitation and transport of precipitated calcium oleate soap to the fluorite surface. Because the fluorite surface would have a coating of chemisorbed oleate, this process is one of coagulation and not really heterocoagulation, with hydrophobic bonding phenomena playing a significant role. In their paper, Fa et al. (2003) showed that colloidal particles of calcium oleate soap coagulate onto the surface of fluorite and make it readily floatable. A higher concentration of calcium oleate colloids was required to initiate calcite flotation. In flotation systems involving slightly soluble salt minerals, a major complication is that of the conversion of the surface of a mineral to that of another mineral or compound. As an illustration of surface conversion, consider the use of soda ash on the surface properties of barite. Equilibrium is controlled by the following reaction: BaSO 4 + HCO 3–( aq ) = BaCO 3 ( s ) + H (+aq ) + SO 42–( aq )

(EQ 14)

There are several ways to demonstrate that the surface of barite behaves as barium carbonate (BaCO3) rather than barium sulfate (BaSO4) in the presence of sodium carbonate (Pradip and Fuerstenau 1991). Figure 29 shows that the addition of sodium carbonate for pH regulation in a flotation separation involving barite and calcite causes barite to behave as © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 55 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

55

3

Electrophoretic Mobility, μm/sec per V/cm

2

Reaction Onset BaSO4 to BaCO3

BaSO4 (Barite)

1

BaCO3 (Powder)

0

–1

–2

–3

24 Hours Equilibration 11

Equilibrium pH

10

BaCO3

9

Ba

SO

Na2CO3 Solution

4

8

+[

HC

O

– 3]

=〈

Ba

CO

3〉

+[

H+ ]+

7 BaSO4 6 10–6

10–5

10–4

10–3

[SO = 4]

10–2

10–1

Initial Sodium Carbonate Concentration, mol/L

Adapted from Pradip and Fuerstenau 1991.

FIGURE 29 Surface transformation of barite to barium carbonate by the addition of sodium carbonate, as shown by zeta potential measurements and solution equilibrium pH

though it were BaCO3, as indicated by the electrokinetic behavior of the mineral. This surface transformation is controlled by bulk thermodynamics, as would be expected when any of these processes are a surface chemical reaction. Finally, in this system, the flotation response of the barite under these conditions must be that of a carbonate. When working with single minerals of calcite, azurite, and malachite, the results of Gaudin and Martin (1928) suggest that it should be possible to separate the copper minerals from calcite with heptylic acid as the collector, but they found that no separation from calcite could be achieved. Sutherland and Wark (1955) stated that this is probably one of the first examples of cross-activation to be found in nonsulfides. Gaudin and Martin aptly commented: “It is indeed very remarkable that azurite and malachite, two minerals which are very similar in chemical composition, and crystallographically, can be separated by flotation © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 56 Tuesday, January 2, 2007 7:36 PM

56

HISTORICAL ASPECTS OF FLOTATION

with one of the lower fatty acids, while the carbonates of copper cannot, by any means now known, be separated from calcite, by this same reagent.” Similar to the foregoing statement, in a complex mixture of salt-type minerals such as calcite and apatite, dissolution of the minerals can put carbonate, phosphate, and fluoride ions into solution—ions that can all chemisorb onto the different minerals and compete with the collector. Likewise, calcium ions can precipitate the collector if the solubility of the calcium soap is exceeded (the KSP of calcium oleate is 10–15.6). Using the known equilibria of the various species that would be involved with apatite and calcite in water, Anathapadmanabhan and Somasundaran (1984) constructed a diagram showing the amount of Ca2+ in solution from apatite, calcite (closed and open to the atmosphere), apatite, and calcite supernatants, and also calcium oleate. Any condition where the concentration of Ca2+ exceeds that of calcium oleate can cause collector precipitation and hence depression, particularly when the mineral is in equilibrium with the solution before the collector is added. The researchers also conducted detailed experiments on the flotation of apatite and calcite in water, in their supernatants, and with added nitrate, carbonate, and phosphate salts. Figure 30 presents their results for the oleate flotation of calcite in water and in supernatants of calcite and apatite. In the case of calcite flotation in water, conditions were such that little calcite would have dissolved during the experiments. This figure shows that supernatants of apatite and even that of calcite depressed the flotation of calcite in the pH region of about 6 to 13. Turbidity measurements after the addition of oleate to the supernatants, but at slightly lower oleate concentration without added potassium nitrate, are also given in Figure 30 and show the precipitation of calcium oleate from these solutions. Added calcium nitrate depressed calcite flotation similar to that shown in Figure 30 for the effect of supernatants. For this system, depression results from the bulk precipitation of the collector as calcium 100

100

80

Turbidity

80

0.1 mM Potassium Oleate 60

Apatite Supernatant Calcite Supernatant

40

Calcite Flotation

70 0.2 mM Potassium Oleate 20 mM Potassium Nitrate 60

Water Apatite Supernatant Calcite Supernatant

20

0 0

2

4

Turbidity, % transmittance

Flotation Recovery, %

90

6

8

10

12

50 14

pH

Adapted from Ananthapadmanabhan and Somasundaran 1984.

FIGURE 30 Effect of apatite and calcite supernatants on the flotation of calcite with potassium oleate as collector and also the turbidity of the supernatants upon the addition of oleate at various pH values showing collector depletion by bulk calcium oleate precipitation

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 57 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

57

oleate, probably before any appreciable amount adsorbed onto the mineral surface. By increasing the oleate concentration, flotation was fully restored. The standard, yet rather inefficient, process for the flotation of bastnaesite ore involves six stages of conditioning with steam, soda ash, sodium fluosilicate, sodium lignin sulfonate, and tall oil from carbonate and sulfate gangue, producing a low-grade concentrate. In an attempt to improve the separation of bastnaesite, (La,Ce)CO3F, from barite and calcite, conditions were found under which bastnaesite can be floated quite effectively with hydroxamate as the collector (Pradip and Fuerstenau 1983, 1991). This appears to result because hydroxamate chelates more strongly with rare-earth ions than with Ba2+ and Ca2+. From their results, the collector interaction with barite is probably one of adsorption, reaching only a close-packed monolayer. In the case of calcite, uptake initially appears to be that of adsorption but eventually changes to multilayer uptake. On the other hand, the rare-earth chelates with hydroxamate are so stable that a surface layer which is equivalent to 5 or 6 monolayers is formed. Again, this can no longer be considered an adsorption process, but must be one of surface reaction, where ions are pulled from lattice sites and form multilayers of a metal hydroxamate compound at the surface. If the rate of metal dissolution and diffusion through the boundary layer is faster than diffusion of the collector to the surface, bulk precipitation may occur. As previously discussed, hydroxylation of the cations in the mineral surface may assist surface reaction phenomena by first providing some surface atom movement. Readsorption of hydrolyzed species may participate in surface reactions. Flotation experiments conducted with potassium octyl hydroxamate as collector at pH 9–9.5 showed that 50% flotation recovery for bastnaesite, calcite, and barite is achieved at the respective initial collector concentrations of 0.12 mM, 0.30 mM, and 0.80 mM (Pradip and Fuerstenau 1991). Recent computations by Pradip and Rai (2003) show that the interaction energies for hydroxamate with bastnaesite and barite are –66 and –33 kcal/mol, respectively, in accordance with the strong uptake of hydroxamate by the rare-earth mineral. S O M E E X A M P L E S O F P R A C T I C A L M I N E R A L S E PA R AT I O N S

The recovery of copper, lead, and zinc from a complex sulfide ore can be achieved in various ways. For an ore that might contain galena, sphalerite, and chalcopyrite with such gangue minerals as pyrite, carbonates, and quartz/silicates, the first step involves the joint flotation of chalcopyrite and galena at pH 6–7 with xanthate collector and a small amount of sodium cyanide to depress pyrite and zinc sulfate to depress any sphalerite activated by heavy-metal ions in solution. Copper sulfate is then added to the tailings from this first step to activate the sphalerite, and sodium cyanide and lime are added to bring the pH to 10.5 to ensure depression of the pyrite. With the addition of more xanthate, sphalerite is then floated. If the pyrite contains gold, for example, it could subsequently be recovered from the tailings. Separation of galena and chalcopyrite in the bulk concentrate can be achieved by depressing galena with sulfur dioxide (SO2) or with sodium dichromate at weakly acidic pH values. Another procedure is to float the galena after depressing the copper sulfides with sodium cyanide at pH 8–9. As an example of oxide mineral flotation, with iron ores, the usual problem is separation of hematite (PZC, pH 7) from quartz (PZC, pH 2). Hematite can be floated away from quartz with a sulfonate at pH 2–4, or with sodium oleate at pH 6–8. Quartz can be floated away from hematite with an amine at pH 6–7, or at pH 11–12 with sodium oleate as collector after activating the quartz with calcium ions and depressing the hematite with starch. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 58 Tuesday, January 2, 2007 7:36 PM

58

HISTORICAL ASPECTS OF FLOTATION

Although not used industrially in iron ore processing, hematite can be floated away from quartz with hydroxamate at pH 8.5, or with an amine at pH 1.5 in the presence of hydrochloric or sulfuric acid as a hematite activator. In the case of a pegmatite containing spodumene (PZC, pH 2), muscovite (IEP, pH 1), iron-bearing oxide/silicates (PZC, pH 6), feldspar (PZC, pH 2), and quartz (PZC, pH 2), separations can be made sequentially by utilizing differences in silicate mineral surface chemistry and crystal chemistry. First, spodumene is floated at pH 7.6 with oleic acid as collector through chemisorption on aluminum surface sites. Then by adding an amine collector at pH 2–2.5, the layer-silicate mineral muscovite is floated by making use of its fixed negative surface charge. After that, the iron silicate impurities are floated at pH 3 with a sulfonate as collector. Finally, with hydrofluoric acid (HF) as an activator, the feldspar can be floated with an amine at pH 3, leaving a marketable pure quartz as the tailing. The simplest large-tonnage separation of a sparingly-soluble salt mineral is exemplified by the flotation of apatite from quartz. In a typical Florida phosphate plant, both anionic and cationic flotation are used to produce an acceptable product. After desliming, the phosphate mineral is floated at pH 9–9.5 with fatty acid and fuel oil extender. Subsequently, the concentrate is acid-blunged to remove the collector coating and then refloated with an amine at pH 7–8 to remove silica impurities. A major challenge is the increased dolomite content of phosphate ores, namely, to effectively prevent the dolomite (Ca,Mg)CO3 from floating with the apatite. S U M M A RY

Basic flotation research conducted over the last several decades has answered questions posed by Rickard in 1916 as to why minerals float. By simultaneously using more than one technique to study the surface chemistry and flotation response of pure minerals with purified chemical reagent systems, the fundamental mechanisms by which sulfide, oxide, silicate, naturally-floatable, and even sparingly-soluble salt minerals respond to flotation is now fairly well understood. As outlined in this review, some systems are better understood than others. Because collector–mineral interactions appear to be more interesting, more research has been directed toward the behavior of collectors than depressants. How activators and inorganic depressants function is fairly well understood, but fundamental knowledge of how and why such organic depressants as quebracho, starch, gum guars, and so forth, attach to mineral surfaces is lacking. Systems involving mixtures of sparingly-soluble salt minerals are subject to complex solution chemistry where species from one mineral may dissolve and adsorb/precipitate onto the surface of another mineral. Furthermore, minerals made up of ions, such as carbonates, phosphates, sulfates, and sulfides, appear to react with collectors, consuming reagent and forming precipitates that may adsorb (coat) more than one mineral, lowering grade. The chemistry of frothers and the role of frothers in determining selectivity have not received adequate attention. Real ores do not behave as pure minerals. Mineral grains may have different chemical compositions (trace elements and locked particles), surfaces smeared with coatings of a softer mineral in the ore, and highly active surfaces (that may change with time) due to flaws produced during comminution. More research should be directed toward the study of mineral mixtures and the behavior of actual ores—but conducted with an aim toward quantifying what is going on and not just ore testing.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 59 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

59

As speculated by Kitchener (1984), hypothetically, there should be a systematic way to plan the flotation processing of an ore by establishing the surface chemistry of all the constituent minerals along with their responses to the reagents relevant to the process, then by determining the likely interferences between the various species, and, finally, by planning the best conditions for securing a large difference in hydrophobicity between the reagenttreated minerals. Many of the reagents and reagent schemes that have been successfully used before and after the invention of chemical collectors were found by trial and error, and often with great ingenuity. However, with decreasing grade, decreasing grain size, and increasing complexity of ores with the passage of time, there is an ongoing need for more selective and effective flotation reagents and reagent schemes. In some cases, finding reagents that adsorb rather than react with a mineral may lead to reduced reagent consumption. The rational design of new reagents may result from understanding the selectivity of interaction of flotation reagents with interfaces in terms of identifying the underlying molecular recognition mechanisms. BIBLIOGRAPHY

Adams, A.S. 1928. Gas sorption in flotation. Page 216 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. Allison, S.A., L.A. Gould, M.J. Nicol, and A. Granville. 1972. A determination of the products of reaction between various sulfide minerals and aqueous xanthate solution, and a correlation of rest potentials. Met. Trans. 3:2613. Ananthapadmanabhan, K.P., and P. Somasundaran. 1984. Role of dissolved mineral species in calciteapatite flotation. Miner. Metall. Process. 1(1):36. Ananthapadmanabhan, K.P., P. Somasundaran, and T.W. Healy. 1977. Chemistry of oleate and amine solutions in relation to flotation. Trans. AIME 266:2003. Anderson, R.J. 1917. The flotation of minerals. Trans. AIME 55:527. Barker, E.E. 1928. Flotation and the Utah Copper Mine flotation practice. Page 19 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. Barsky, G. 1934. Discussion to the Wark and Cox paper, Principles of flotation, I. Trans. AIME 112:236. Bean, J.J. 1971. Tale of tales. World Min. 59. Billingsly, P. 1928. How flotation has broadened the geologist’s viewpoint. Page 33 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. Bogdanov, O.S., V.Y. Hainman, A.K. Podnek, and N.A. Jarvis. 1957. Investigation of the action of modifying agents in flotation. Page 479 in Progress in Mineral Dressing. Stockholm: Transactions of the International Congress on Mineral Dressing. Bradford, L. 1913. Australian Patent 8,123. Chander, S. 1985. Oxidation/reduction effects in depression of sulfide minerals—a review. Miner. Metall. Process. 2(1):26. ———. 2003. A brief review of pulp potentials in sulfide flotation. Int. J. Miner. Process. 72:141. Chander, S., and D.W. Fuerstenau. 1975. Effect of potential on the flotation and wetting behavior of chalcocite and copper. Trans. SME 258:284. Cook, M.A., and J.C. Nixon. 1950. Theory of water-repellent films on solids formed by the adsorption from aqueous solutions of heteropolar compounds. J. Phys. Chem. 54:445. Cooper, F.D. 1980. Mining and quarrying trends in the metals and nonmetal industries. In Minerals Yearbook. Volume 1. Washington, DC: U.S. Bureau of Mines. Fa, K., T. Jiang, J. Nalaskowski, and J.D. Miller. 2003. Interaction forces between a calcium dioleate sphere and calcite-fluorite surfaces and their significance in flotation. Langmuir 19:10253. Fleming, M.G. 1952a. Effects of alkalinity on the flotation of lead minerals. Trans. AIME 193:1231.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 60 Tuesday, January 2, 2007 7:36 PM

60

HISTORICAL ASPECTS OF FLOTATION

———. 1952b. Effects of soluble sulphide in the flotation of secondary lead minerals. Page 521 in Recent Developments in Mineral Dressing. London: The Institution of Mining and Metallurgy. Fowkes, F.M., and W.D. Harkins. 1940. J. Am. Chem. Soc. 62:3377. Free, M.L., and J.D. Miller. 1996. The significance of collector colloid adsorption phenomena in the fluorite-oleate flotation system as revealed by FTIR-IRS and solution chemistry analysis. Int. J. Miner. Process. 48:197. Fuerstenau, D.W. 1953. Streaming potential studies on quartz. Sc.D. thesis, Massachusetts Institute of Technology. ———. 1985. Chemistry of flotation. Pages 7–29 in Principles of Flotation—the Wark Symposium. Edited by M.H. Jones and J.T. Woodcock. Symposium Series 40. Parkville: Australasian Institute of Mining and Metallurgy. Fuerstenau, D.W., and T.W. Healy. 1972. Principles of flotation. Pages 92–131 in Adsorptive Bubble Separation Techniques. Edited by R. Lemlich. New York: Academic Press. Fuerstenau, D.W., T.W. Healy, and P. Somasundaran. 1964. The role of the hydrocarbon chain of alkyl collectors in flotation. Trans. AIME 229:321. Fuerstenau, D.W., and R. Herrera-Urbina. 1989. Mineral separation by flotation. Surfactant Based Separation Processes. Edited by J.F. Scamehorn and J.H. Harwell. New York: Marcel Dekker. Fuerstenau, D.W., R. Herrera-Urbina, and D.W. McGlashan. 2000. Studies on the applicability of chelating agents as universal collectors for copper minerals. Int. J. Miner. Process. 58:15. Fuerstenau, D.W., and P. Huang. 2003. Interfacial phenomena in talc flotation and depression. Proc. XXII Int. Miner. Process. Cong. 2:1034-1043. Cape Town: South African Institute of Mining and Metallurgy. Fuerstenau, D.W., and P.H. Metzger. 1960. Activation of sphalerite with lead ions in the presence of zinc salts. Trans. AIME 217:119. Fuerstenau, D.W., and Pradip. 1984. Mineral flotation with hydroxamate collectors. Reagents in the Mineral Industry. London: The Institution of Mining and Metallurgy. Fuerstenau, D.W., Pradip, and R. Herrera-Urbina. 1992. The surface chemistry of bastnaesite, barite and calcite in aqueous carbonate solutions. Colloids Surf. 68:95. Fuerstenau, D.W., and S. Raghavan. 1976. Some aspects of the thermodynamics of flotation. Pages 21–65 in Flotation—A.M. Gaudin Memorial Volume. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. ———. 1980. The crystal chemistry, surface properties and flotation behavior of silicate minerals. Pages 368–415 in Proceedings of the XII International Mineral Processing Congress, Sao Paulo. Nacional Pulicaconoses e Publicidade. Fuerstenau, D.W., and T. Wakamatsu. 1973. Effect of alkyl sulfonates on the wettability of alumina. Trans. AIME 254:123. Fuerstenau, D.W., and C.H. Wayman. 1958. Effect of chemical reagents on the motion of single air bubbles in water. Trans. AIME 212:430. Fuerstenau, M.C., K.L. Clifford, and M.C. Kuhn. 1974. The role of zinc xanthate precipitation in sphalerite flotation. Int. J. Miner. Process. 1:307. Fuerstenau, M.C., R.W. Harper, and J.D. Miller. 1970. Hydroxamate vs. fatty acid flotation of iron oxide. Trans. AIME 247:69. Fuerstenau, M.C., M.C. Kuhn, and D.A. Elgillani. 1968. The role of dixanthogen in xanthate flotation of pyrite. Trans. AIME 241:148. Fuerstenau, M.C., A. Lopez-Valdivieso, and D.W. Fuerstenau. 1988. Role of hydrolyzed cations in the natural hydrophobicity of talc. Int. J. Miner. Process. 23:161. Fuerstenau, M.C., and B.R. Palmer. 1976. Anionic flotation of oxides and silicates. Pages 148–196 in Flotation—A.M. Gaudin Memorial Volume. Volume 1. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Gardner, J.R., and R. Woods. 1974. An electrochemical investigation of contact angle and of flotation in the presence of alkylxanthates. I. Platinum and gold surfaces. Aust. J. Chem. 27:2139. Gates, J.F., and L.K. Jacobsen. 1925. Page 38 in Some Flotation Fundamentals and Their Practical Application. Bulletin 16. University of Utah Engineering Experiment Station.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 61 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

61

Gaudin, A.M. 1927. Flotation mechanism, a discussion of the functions of flotation reagents. AIME Technical Publication 4. New York: American Institute of Mining and Metallurgical Engineers. ———. 1928. Flotation mechanism, a discussion of the functions of flotation reagents. Page 50 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. ———. 1929. The influence of hydrogen-ion concentration on recovery in simple flotation systems. Min. Metall. 10:19. ———. 1930. Effect of xanthate, copper sulfates and cyanides on flotation of sphalerite. Trans. AIME, Mill. Methods 417. ———. 1932. Flotation. New York: McGraw-Hill. ———. 1934. Flotation’s future beset with difficult problems. Eng. Min. J. 135(1):29. ———. 1957. Flotation. 2nd edition. New York: McGraw-Hill. Gaudin, A.M., and D.W. Fuerstenau. 1955. Streaming potential studies—quartz flotation with cationic collectors. Trans. AIME 202:958. Gaudin, A.M., D.W. Fuerstenau, and G.M. Mao. 1959. Activation and deactivation studies with copper on sphalerite. Trans. AIME 214:430. Gaudin, A.M., H. Glover, M.S. Hanson, and C.W. Orr. 1928. Flotation fundamentals, Part I. Technical Paper No. 1. University of Utah and U.S. Bureau of Mines. Gaudin, A.M., and J.S. Martin. 1928. Flotation fundamentals, Part III. Technical Paper No. 5. University of Utah and U.S. Bureau of Mines. Gaudin, A.M., and W.D. Wilkinson. 1933. Surface actions of sulfur bearing organic compounds on some finely ground sulfide minerals. J. Phys. Chem. 37:833. Gebhardt, J.E., and P.E. Richardson. 1987. Differential flotation of a chalcocite-pyrite particle bed by electrochemical control. Miner. Metall. Process. 4:140. Gebrueder Bessel. 1877. Verfahren zur Reinigung von Graphit [process for the purification of graphite]. German patent 42, Class 22. ———. 1886. German patent 39,369. Grahame, D.C. 1947. The electrical double layer and the theory of electrocapillarity. Chem. Rev. 41:441. Graichen, K., J. Hanisch, H. Schubert, K.D. Steiner, C. Tanneberger, and E. Waechtler. 1977. Die Gebrueder Bessel und die Anfange der flotativen Aufbereitung. Kolloquium 100 Jahre Flotation. Berkakademie Freiberg, Germany. Gutzeit, G. 1946. Chelate-forming compounds as flotation reagents. Trans. AIME 169:272. Harris, G.H., and B.C. Fischback. 1954. U.S. Patent 2,691,635. Herrera-Urbina, R., F.J. Sotillo, and D.W. Fuerstenau. 1999. Effect of sodium sulfide addition on the pulp potential and amyl xanthate flotation of cerussite and galena. Int. J. Miner. Process. 55:157. Hunter, R.J. 1981. Zeta Potential in Colloid Science. London: Academic Press. Iwasaki, I., S.R.B. Cooke, and H.S. Choi. 1960. Flotation characteristics of hematite, goethite and activated quartz with 18-carbon aliphatic acids and related compounds. Trans. AIME 217:237. Iwasaki, I., S.R.B. Cooke, and A.F. Colombo. 1960. Flotation characteristics of goethite. Report of Investigations 5593. Department of the Interior. Washington, DC: U.S. Bureau of Mines. James, R.O., and T.W. Healy. 1972. Adsorption of hydrolysable metals ions at the oxide-water interface, Parts I, II, and III. J. Colloid Interface Sci. 40:42. Kamienski, B. 1931. So-called flotation. Przem. Chem. 15:201. (1932. Chem. Abstr. 26:53 [translated from Russian]). Keller, C.H. 1925. U.S. Patent 1,554,216. King, R.P. 1982. Principles of Flotation. Johannesburg: South African Institute of Mining and Metallurgy. Kitchener, J.A. 1984. The froth flotation process: Past, present and future—in brief. Page 3 in The Scientific Basis of Flotation. Edited by K.J. Ives. NATO ASI Series. The Hague: Matinus Nojhoff. Klassen, V.I., and V.A. Mokrousov. 1963. An Introduction to the Theory of Flotation. Translated by J. Leja and G.W. Poling. London: Butterworths. Klimpel, R.R., and R.D. Hansen. 1988. Page 385 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 62 Tuesday, January 2, 2007 7:36 PM

62

HISTORICAL ASPECTS OF FLOTATION

Kulkarni, R.D., and P. Somasundaran. 1980. Flotation chemistry of hematite-oleate system. Colloids Surf. 1:387. LaMer, V.K. 1967. Discussion. Pages 242–243 in Principles and Applications of Water Chemistry. Edited by S.D. Faust and J.V. Hunter. New York: John Wiley & Sons. Langmuir, I. 1920. The mechanism of surface phenomena in flotation. Trans. Faraday Soc. 15:62. Laskowski, J.S. 1986. The relationship between floatability and hydrophobicity. Page 189 in Advances in Mineral Processing. Edited by P. Somasundaran. Littleton, CO: SME. Laskowski, J.S., and J.A. Kitchener. 1969. The hydrophilic-hydrophobic transition on silica. J. Colloid Interface Sci. 29:670. Leja, J., L.H. Little, and G.W. Poling. 1962–1963. Xanthate adsorption studies using infrared spectroscopy, 2. Evaporated lead sulfide, galena and metallic lead substrates. Trans. Inst. Min. Metall. 72:414. Mackenzie, J.M.W. 1966. Zeta potential of quartz in the presence of ferric iron. Trans. AIME 235:82. Majima, H., and M. Takida. 1968. Electrochemical studies of xanthate-dixanthogen system on pyrite. Trans. AIME 241:431. Marabini, A., J. Cases, and M. Barbaro. 1989. Chelating reagents as collectors and their mechanism. Page 35 in Challenges in Mineral Processing. Edited by K.V.S. Sastry and M.C. Fuerstenau. Littleton, CO: SME. Mellgren, O. 1966. Heat of adsorption and surface reactions of potassium ethyl xanthate on galena. Trans. AIME 235:46. Mellgren, O., R.J. Gochin, H.L. Shergold, and J.A. Kitchener. 1973. Thermochemical measurements in flotation research. Page 451 in Proceedings of the 10th International Mineral Processing Congress. London: The Institution of Mining and Metallurgy. Merrill, C.W., and J.W. Pennington. 1962. The magnitude and significance of flotation in the mineral industries of the United States. Page 55 in Froth Flotation—50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Modi, H.J. 1956. Electrokinetic properties and flotation behavior of corundum. Sc.D. thesis, Massachusetts Institute of Technology. Modi, H.J., and D.W. Fuerstenau. 1960. The flotation of corundum—an electrochemical interpretation. Trans. AIME 217:381. Nixon, J.C. 1957. Discussion. Page 369 in Proceedings of the 2nd International Congress on Surface Activity. Volume 3. Edited by J.H. Schulman. London: Butterworths. Onoda, G.Y., and D.W. Fuerstenau. 1965. Amine flotation of quartz in the presence of inorganic electrolytes. Pages 301–306 in Proceedings of the 7th International Mineral Processing Congress. New York: Gordon and Breach Scientific Publishers. Palmer, B.R., M.C. Fuerstenau, and F.F. Aplan. 1975. Mechanisms involved in the flotation of oxides and silicates with anionic collectors. Part II. Trans. AIME 258:261. Peck, A.S., L.H. Raby, and M.E. Wadsworth. 1966. An infrared study of the flotation of hematite with oleic acid and sodium oleate. Trans. AIME 235:301. Perkins, C.L. 1921. U.S. Patent 1,364,304. Plaksin, I. 1959. Interaction of minerals with gas and reagents in flotation. Min. Eng. 11:319. Polkin, S.I., and T.V. Najfonow. 1964. Concerning the mechanism of collector and regulator interaction in the flotation of silicate and oxide minerals. Pages 307–318 in Proceedings of the VII International Mineral Processing Congress. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Pradip, and D.W. Fuerstenau. 1983. The adsorption of hydroxamate on semi-salt minerals. Part I. Adsorption on barite, calcite, and bastnaesite. Colloids Surf. 8:103. ———. 1991. The role of inorganic and organic reagents in the flotation separation of rare-earth ores. Int. J. Miner. Process. 32:1. Pradip, and B. Rai. 2002. Design of tailor-made surfactants for industrial applications using a molecular modeling approach. Colloids Surf. A 205:139. ———. 2003. Molecular modeling and rational design of flotation reagents. Int. J. Miner. Process. 72:95.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 63 Tuesday, January 2, 2007 7:36 PM

DEVELOPMENTS IN THE CHEMISTRY OF FLOTATION PROCESSING

63

Pugh, R., and P. Stenius. 1985. Solution chemistry studies and flotation behavior of apatite, calcite and fluorite minerals with sodium oleate collector. Int. J. Miner. Process. 15:193. Purcell, G., and S.C. Sun. 1963. Significance of double bonds in fatty acid flotation—an electrokinetic study; —a flotation study. Trans. AIME 226:6, 13. Raghavan, S., and D.W. Fuerstenau. 1975. The adsorption of aqueous octylhydroxamate on ferric oxide. J. Colloid Interface Sci. 50:319. Ralston, O.C. 1916. Why do minerals float? Page 175 in The Flotation Process. Edited by T.A. Rickard. San Francisco: Mining and Scientific Press. Richardson, P.E., and G.W. Walker. 1985. The flotation of chalcocite, bornite, chalcopyrite and pyrite in an electrochemical-flotation cell. Page 198 in XVth International Mineral Processing Congress, Tome II. Orleans, France: Bureau de Recherches Geologiques et Minieres. Rickard, T.A, editor. 1916. Page 7 in The Flotation Process. San Francisco: Mining and Scientific Press. Salamy, S.G., and J.C. Nixon. 1953. The application of electrochemical methods to flotation research. Page 503 in Recent Developments in Mineral Dressing. London: The Institution of Mining and Metallurgy. Shergold, H.L., A.P. Prosser, and O. Mellgren. 1968. New region of floatability in the hematitedodecylamine system. Trans. IMM 77:C166. Schuhmann, R., and B. Prakash. 1950. Effect of BaCl2 and other activators on soap flotation of quartz. Trans. AIME 187:591. Schulman, J.H., and T.D. Smith. 1953. Selective flotation of metals and minerals. Pages 393–413 in Recent Developments in Mineral Dressing. London: The Institution of Mining and Metallurgy. Simmons, G.L., J.N. Orlich, J.C. Lenz, and J.A. Cole. 1999. Implementation and start-up of N2TEC flotation at the Lone Tree mine. Page 183 in Advances in Flotation Technology. Edited by B.K. Parikh and J.D. Miller. Littleton, CO: SME. Smith, R.W. 1963. Effect of fluoride addition on contact angle in the system microclinedodecylamine solution-nitrogen. Proceedings, South Dakota Academy of Science 42:60. Somasundaran, P., and G.E. Agar. 1967. The zero point of charge of calcite. J. Colloid Interface Sci. 24:433. Somasundaran, P., and D.W. Fuerstenau. 1966. Mechanisms of sulfonate adsorption at the aluminawater interface. J. Phys. Chem. 70:90. Somasundaran, P., and D.R. Nagaraj. 1984. Chemistry and applications of chelating agents in flotation and flocculation. Page 209 in Reagents in the Mineral Industry. Edited by M.J. Jones and R. Oblatt. London: Institution of Mining and Metallurgy. Steininger, J. 1967. Collector ionization in sphalerite flotation with sulfhydryl collectors. Trans. AIME 238:251. Stumm, W. 1992. Chemistry of the Solid-Water Interface. New York: John Wiley & Sons. Sulman, H.L., H.F.K. Picard, and J. Ballot. 1905. British Patent 7,803, April 12; duplicated as U.S. Patent 835,120, May 29. Sutherland, K.L., and I.W. Wark. 1955. Principles of Flotation. Melbourne: Australasian Institute of Mining and Metallurgy. Taggart, A.F. 1928. Flotation reagents. Page 40 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. Taggart, A.F., and F.E. Beach. 1917. An explanation for the flotation of minerals. Trans. AIME 55:547. Taggart, A.F., G.R.M. del Giudice, and O.A. Ziehl. 1934. The case for the chemical theory of flotation. Trans AIME 112:348. Taggart, A.F., T.C. Taylor, and C.R. Ince. 1930. Experiments with flotation reagents. Trans. AIME, Mill. Methods 285. Taggart, A.F., T.C. Taylor, and A.F. Knoll. 1930. Chemical reactions in flotation. Trans. AIME, Mill. Methods 217. Taylor, T.C., and A.F. Knoll. 1934. Action of alkali xanthates on galena. Trans. AIME 112:382. Trahar, W.J. 1984. The influence of pulp potential in sulphide flotation. Pages 117–135 in Principles of Flotation—the Wark Symposium. Edited by M.H. Jones and J.T. Woodcock. Symposium Series 40. Parkville: The Australasian Institute of Mining and Metallurgy.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 64 Tuesday, January 2, 2007 7:36 PM

64

HISTORICAL ASPECTS OF FLOTATION

Tye, A.T. 1928. Effect of preferential flotation at Cananea mill and smelter. Page 151 in Flotation Practice. New York: American Institute of Mining and Metallurgical Engineers. Varley, T. 1928. Reports of Investigations, Serial No. 2852. Washington, DC: U.S. Bureau of Mines. von Reinders, W. 1913. Die Verteilung eines suspendierten Pulvers oder eines Kolloid gelosten Stoffes zwischen zwei Losungsmitteln. Kolloid Zeitrschrift 13:235. Wakamatsu, T., and D.W. Fuerstenau. 1968. The effect of chain length on the adsorption of sulfonates at the solid-water interface. Advances in Chemistry Series 79. Columbus, OH: American Chemical Society. Wark, I.W., and A.B. Cox. 1934. Principles of flotation, I, II, and III. Trans. AIME 112:189, 245, 267. Whitworth, F.T. 1926. U.S. Patent 1,553,232. Woods, R. 1984. Electrochemistry of sulfide flotation. Pages 91–115 in Principles of Flotation—the Wark Symposium. Edited by M.H. Jones and J.T. Woodcock. Symposium Series 40. Parkville: The Australasian Institute of Mining and Metallurgy. Woods, R., and P.E. Richardson. 1986. The flotation of sulfide minerals—electrochemical aspects. Page 154 in Advances in Mineral Processing. Edited by P. Somasundaran. Littleton, CO: SME.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 65 Tuesday, January 2, 2007 7:36 PM

History of Flotation Technology A.J. Lynch, J.S. Watt, J.A. Finch, and G.E. Harbort

A B S T R AC T

The development of flotation as a major industrial process occurred during three main time periods. During 1860 to 1900, small-scale attempts were made in industry to float or agglomerate the valuable minerals in ores and wash away the waste minerals. From 1900 to 1925, the economic necessity to concentrate fine sulfide particles led to immense research efforts for floating zinc and lead minerals at Broken Hill, Australia (1901 to 1915), and copper minerals at the huge mines in the western United States (1911 to 1925). During this time, flotation became an industrial technology and provided much of the copper that made the widespread distribution of electricity possible. Two major advances occured after 1960. X-ray and radioisotope on-stream analysis systems were developed, which gave rapid information about assays of process streams and made accurate process control possible, and new flotation machines were introduced. There were high-volume columns in which the pulp and air bubbles moved in countercurrent flow, and high-energy cells in which the pulp was aerated with very small bubbles prior to separation. The high-energy cells provide much higher flotation rates than the columns. This chapter presents developments in flotation technology that occurred during these periods. THE NEED FOR A NEW PROCESS

“A new metallurgical process never springs fully developed from the brain of one person, but is the result of patient investigation, application, and improvement by many minds, during many years” (Hoover 1914, p. 2). Flotation did not happen in isolation; it was one of many inventions in the second half of the 19th century that brought a seminal change to mining and mineral processing technology and greatly increased mineral production. This was an exciting period in the mineral industry as the Industrial Revolution was gaining momentum and was causing rapid increases in the consumption of minerals and metals (see Table 1). In 1850 the mineral industry had been technically stagnant for more than 200 years, the last major innovations being the use of water power to drive crushing and grinding machinery in the 16th century and the amalgamation process and blasting by black powder in the 17th century. The industry was ill-equipped then to handle the problems presented by the rising demands for minerals and metals, but it was transformed by new technology during 1850–1900 and moved from the era of black powder, hand carts, stamp TABLE 1

World production of copper, lead, zinc, and coal, 1850–1900

Commodity Copper, kt Lead, kt Zinc, kt Coal, Mt

1850 55 130 65 75

1875 130 320 165 233

Source: Habashi 1994.

65

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

1900 525 850 480 660

flotation0.book Page 66 Tuesday, January 2, 2007 7:36 PM

66

TABLE 2

HISTORICAL ASPECTS OF FLOTATION

Production rates and profits at Broken Hill Proprietary Company, 1886–1902

Mining, profits Kilotons mined Dividends and bonuses, $A × 1,000

1886 10

1888 80

1890 170

1892 300

1894 590

1896 440

1898 400

1900 520

1902 660

50

370

1,000

800

580

420

280

180

110

Pb, % 69.2 30.8 100.0

Proportion Ag, % Zn, % 49.8 12.8 50.2 87.2 100.0 100.0

Source: Lynch 1987.

TABLE 3

Typical milling statement, Broken Hill, 1900

Product Lead concentrate Tailing Crude ore

Ton 11,141 37,936 49,077

Pb, % 60.6 7.8 19.9

Assay Ag, oz 19.6 5.8 8.9

Zn, % 10.4 20.8 18.5

Source: Woodward 1965.

TABLE 4 Recoveries of copper by gravity concentration in mills processing porphyry copper ores Milling Year Tons milled per day Average copper in ore, % Average copper in concentrate, % Copper recovery, %

Utah* 1913 25,000 1.25 17.31 63.95

Chino† 1915 7,357 2.16 21.55 66.59

Ray‡ 1915 7,805 1.67 19.29 64.11

Nevada§ 1915 8,442 1.54 7.77 70.18

Source: Hines and Vincent 1962. *Utah Copper Company, Utah. †Chino Copper Corporation, New Mexico. ‡Ray Consolidated Copper Company, Arizona. §Nevada Consolidation Copper Company, Nevada.

mills, and sluices to the era of dynamite, steam shovels, ball mills, and Wilfley tables. Even with all the improvements, there was still a serious problem—fine particles could not be concentrated efficiently by gravity machines, and fine-grained ores were replacing coarse-grained ores as the source of many metals. The problem can be illustrated by referring to what happened at the Broken Hill Proprietary Company (BHP) in Australia. Table 2 shows how dividends and bonuses at BHP declined per ton of mined ore from 1896. BHP started operating at Broken Hill in 1886, and, initially, profits were very high because miners extracted the surface ore that was rich in coarse-grained silver and lead minerals, but earnings plummeted when this ore was exhausted, and the fine-grained primary sulfides had to be mined. This was a problem because there were high losses of silver and lead when the fine particles from the mills were concentrated in gravity machines, and zinc was lost almost entirely. Table 3 shows a typical milling statement in 1900. By 1900 the early years of prosperity had given way to pessimism, and employment had fallen by 30%. The economics were simple—find a new process or abandon the mines. Heavy investments were made in magnetic separation and in the unproven flotation process, and it was flotation that provided the answer. The growth of flotation from ideas described in patents into a remarkable industrial process was described as “…one of the outstanding achievements in twentieth century technology…” (Klassen and Mokrousov 1963, p. xiv). The same problem occurred with porphyry copper ores some years later in the western United States. There were high metal losses using gravity concentration, as shown in Table 4, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 67 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

67

and a process had to be found to reduce these losses and increase copper recovery by 20%. Again, flotation was the answer. The research and development programs that established flotation as the major concentration process at Broken Hill and in the western United States overlapped, the main activity at Broken Hill occurring from about 1900 to 1915 and in the western United States from about 1911 to 1920. Minerals Separation Company (MSC), which was active in both areas, ensured that there was transfer of technology, although at the cost of bitter patent lawsuits. At Broken Hill the major results of the program were as follows: • Froth flotation, which was developed as an industrial process for concentrating sulfides and was used to extract zinc from millions of tons of slime tailings. By 1907, the annual production of zinc concentrate had risen to 236,251 tons. • Differential froth flotation, which was developed during 1910–1915 for making separate lead and zinc sulfides Progress must have seemed slow to investors in the Broken Hill mines, but metallurgists had to find out how to float minerals on water and how to control the many variables that made the process successful. It was not an easy task. “The propensity of minerals to float was a valuable discovery, but scores of more elusive and more important discoveries had to be made before the process could earn a profit. ‘It is to manipulation, learned empirically in the laboratory and mill, that the flotation process owes its metallurgic success,’ wrote T.A. Rickard” (Blainey 1968, p. 70; Rickard 1932). In the United States, froth flotation was first used in a zinc mill in Montana in 1911, and its success gave companies the incentive to investigate the process for the concentration of copper sulfides. Its potential to improve the economics of copper milling was realized in 1915 when a 15,000-tons-per-day flotation plant was built at the Inspiration Company, and the recovery of copper was increased to 80%. Not surprisingly, flotation circuits swept the copper industry within a few years. How flotation developed as a great industrial process will be discussed in this chapter. It is necessarily brief, and more information about the early years of flotation is given in the bibliography, in particular, Hoover (1914), members of the Broken Hill Branch of the Australasian Institute of Mining and Metallurgy (1930), Hines and Vincent (1962), Crabtree and Vincent (1962), Fuerstenau (1999), and Megraw (1918). E A R LY I D E A S , 1 8 6 0 – 1 9 0 0

The first hint that differences in surface properties could be used to separate minerals appeared in a patent awarded to William Haynes (Haynes 1860). The process claimed that sulfides in a powdered ore could be agglomerated by oil and the nonsulfides could be removed by washing. There is no evidence that the idea was tested in a plant. The first commercial flotation plant was built by the Bessel brothers in Dresden, Germany, in 1877 to clean graphite ore (Graichen et al. 1977). Adolph Bessel graduated from the University of Gottingen in 1855 and joined a factory that made refractories and crucibles in Grobalmerode. In 1864 this factory was moved to Dresden, close to the Polytecnic and the Bergakademie Freiberg. Adolph and his brother became its owners in 1866. Because the quality of the graphite used in the crucibles was poor, they developed a process for cleaning it that involved mixing graphite ore with a small amount of oil, adding water, and boiling the mixture to float the graphite to the surface of the pulp. Their process yielded a concentrate containing 90% graphite from 40% graphite in the feed. Bessel patents of July 2, 1877, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 68 Tuesday, January 2, 2007 7:36 PM

68

HISTORICAL ASPECTS OF FLOTATION

and May 12, 1886, contained all the essential features of froth flotation including the use of nonpolar oils to enhance process kinetics (Bessel 1877, 1886). The first patent referred to bubbles being generated by boiling and the second to CO2 bubbles being generated by the reaction of lime with acid. In 1878 the Wohler Gold Medal was awarded to Adolph Bessel for the invention patented in 1877. The first flotation plant to process sulfide ores was based on Carrie Everson’s work, although it was not a commercial success as was Bessel’s plant. The circumstances were unusual. Everson was born in 1842 in Massachusetts and studied medicine before marrying Dr. W. Everson in 1864. He invested in mining shares which failed, and this led to Carrie taking an interest in mineralogy so she could understand the reason for the failure. About 1878 she started to experiment with ways to concentrate sulfide minerals and eventually patented a process in 1885 for separating sulfides from gangue by mixing powdered ore with a small amount of oil in an acid solution and floating the sulfides in a scum (Everson 1885). Flotation must have been due to entrained air. The myth is that Everson’s patent originated in observations made while washing geologists’ sample bags; the less romantic reality is that she was a good scientist who would carry out experiments in a laboratory and was prepared to test the results in practice. Everson’s process was successful in small plants but not on a larger scale (Megraw 1918), perhaps because the ores were unsuitable—sulfide flotation did not reveal its secrets easily. Unfortunately, she did not have the financial resources to continue her research, and she became a teacher to earn a living. Later assessments of Everson’s work were “…if the invention had been a less startling innovation, it would probably have received more attention from engineers and metallurgists, and the application of the idea would probably in that case have taken place many years before it did,” (Hoover 1914 p. 6) and “as a metallurgist she was a quarter of a century in advance of her profession” (Megraw 1918 p. 7). The year 1885 was important in the history of flotation because of the patents by the Bessel brothers and Carrie Everson. The same year, a patent was awarded to Hezekiah Bradford in the United States for a film flotation process in which powdered sulfide ore was placed gently onto the surface of water and the sulfides adhered to the surface while other minerals sank (Bradford 1885). It is likely that these inventors made their discoveries independently, and although their efforts had little immediate technical impact, their patents showed that the potential significance of flotation-type processes was becoming recognized. In 1898 Francis Elmore patented a process for concentrating sulfide minerals by adding oil to pulverized ore in water, agglomerating the sulfides and buoying them to the surface of the water, and washing away the gangue particles (Elmore 1898). He proved its value at the Glasdir mine in Wales, and his work was discussed at an Institution of Mining and Metallurgy meeting in London in 1900 (Hoover 1914). The process was widely applied and can be regarded as the first successful process for floating sulfides, although it was not froth flotation as it is currently known. Entrained air was an unrecognized but important factor. By 1900 it seems that only the Bessel brothers had deliberately used gas bubbles to accelerate flotation rates, but in 1901 an engineer in Italy, Alcide Froment, patented the use of gas bubbles to float sulfide particles (Froment 1902). Froment also used sulfuric acid and limestone to generate bubbles although he recognized that gas of any kind would be suitable (Hoover 1914). This is the background to the events that occurred at Broken Hill during 1901–1915. Engineers there would have known of the process in which sulfides could be agglomerated by oil and gangue washed away, so it is not surprising that they became interested in the technique. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 69 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

69

F R O M A S TA R T L I N G I N N OVAT I O N T O A N I N D U S T R I A L PROCESS, 1901–1925 Zinc and Lead at Broken Hill, 1901–1915

In 1901 the immediate problem at Broken Hill was how to extract zinc from the dumps. As a measure of their size and potential, in 1904 the dumps contained more than 7 million tons of tailings that assayed at about 4% lead, 140 g/ton silver, and 15% zinc. A process to extract zinc would revive the mines and restore prosperity. The zinc problem at Broken Hill was, therefore, the question of the hour. From 1901 engineers at Broken Hill worked on several flotation processes and machines (Woodward 1965). • Froth flotation was independently investigated by Guillaume Delprat, general manager of BHP, and by Charles Potter, who was a brewer in Melbourne. They each patented a process in which the zinc mineral in gravity plant tailings was floated by carbon dioxide generated by adding acid to hot pulps that contained carbonate minerals (Potter 1902; Delprat 1902). Neither used oil; probably the tailings contained a sufficient amount to make flotation occur. Potter’s process had a short life, but the BHP process worked on gravity plant tailings from 1902 to 1923, the acid consumption being 26 lb per ton and the pulp temperature being 82°–88°C (180°–190°F). About 90,000 tons of zinc concentrate was made annually from 300,000 tons of tailings. • A film flotation process that was similar to Bradford’s 1886 invention was patented by Auguste de Bavay in 1904. In this process a pulp that had been deslimed, acidified, and oiled flowed down a corrugated cone dipped at an angle into water. The hydrophobic sulfides floated on the water, whereas other particles were wetted and sank. It worked at Broken Hill from 1905 to 1917, and at its peak produced 80,000 tons of zinc concentrate annually from 300,000 tons of tailings. • Vacuum flotation, a form of froth flotation, was patented by Francis Elmore in 1904 and was used in the Zinc Corporation plant for 6 years. In this process, a small amount of oil was added to an acidified tailings pulp, and the sulfide particles were floated with bubbles generated by applying a vacuum of 600 mm of mercury to the pulp and precipitating the dissolved air. At its peak, it produced 80,000 tons of concentrates annually from 250,000 tons of tailings. • The Minerals Separation Company, which had been formed in England in 1903 to specialize in ore dressing problems, came to Broken Hill in 1904 to test a process patented by Arthur Cattermole (1902) for which it had purchased the rights. In this process, a small amount of oil—insufficient to give a buoyant effect—caused the sulfides to agglomerate and sink, and other minerals were washed away. This process was a failure because the granules of sulfides that still required further concentration tended to break on the concentrating tables. Another test proved successful, however, wherein an even smaller amount of oil was added and the pulp was violently agitated to entrain air because the sulfides were carried into a froth and removed in a spitzkasten. Staff at MSC and the Central Mine at Broken Hill developed this concept into stirred flotation cells that were used in series. • Sketches of early flotation cells are shown in Figures 1 and 2. All the concentrates made by flotation at Broken Hill contained 47%–49% Zn; recoveries were more than 80%, and working costs varied only by 10%. A measure of the success of flotation

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 70 Tuesday, January 2, 2007 7:36 PM

70

HISTORICAL ASPECTS OF FLOTATION

To Vacuum Pump Separating Cone

Acid 9 ft

Worm-wheel

Feed

Oil Acid Froth Launder 15 ft

Feed Mixer

Baffles

Tailing Tailing Potter-Delprat Cell

Concentrate

Elmore Vacuum Cell

Feed Regulating Wheel for Discharge Valve

Corrugated Cone

Water Level Overflow Launder

Froth Launder

Agitation Box

Water Level Feed

Frothing Chamber

Discharge

Tailing

Minerals Separation Cell

De Bavay Cone

Source: Truscott 1923.

FIGURE 1

Flotation cells at Broken Hill, 1902–1910

Feed Concentrate a f e a — Thickener b — Feed tank c — Pump d — Separating cone e, f, g — Pipe and valves for air and frother k, l — Concentrate, tailings outlets Lyster Cell

c

g

h

k d

Impellor l

Air Inlet

Owen Cell

Source: Hoover 1914.

FIGURE 2

Cells built for differential flotation at Broken Hill, 1911–1913

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 71 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

71

was that in 1913, 10 years after the start of investigations, more than 11 Mt of material had been floated, and 3 Mt of zinc concentrate produced (Woodward 1965; Hoover 1914). While these cells were being developed and used at Broken Hill, a flotation cell was patented in 1903 by H.L. Sulman and H.F.K. Picard of London, in which particles were floated by air bubbles formed in the cell by compressed air flowing through holes in an immersed, perforated tube (Sulman and Picard 1903). It would be some years before pneumatic cells were used in plants. By 1908 bulk flotation of zinc concentrates was working well, and it was time to develop a differential flotation process for primary sulfides. Three approaches were investigated (Woodward 1965): 1. In 1910, E.J. Horwood of BHP roasted to 400°–500°C a concentrate made by bulk flotation to “deaden” galena by oxidizing its surface to lead sulfate. The blende was unaltered and was refloated to make salable concentrate. 2. In 1912, F.J. Lyster, mill superintendent at the Zinc Corporation, observed that the natural flotation rates of galena and blende were different and devised a process to make separate concentrates. Galena was collected during gentle flotation of an alkaline pulp to which eucalyptus oil had been added as a frother; blende was then floated from the deleaded pulp. Lyster recognized the importance of air control and devised a cell in which air was added to a pulp, and the mixture was passed through a pump before entering a tank in which the froth separated. Differential flotation was achieved by controlling the air flow rate. A subaeration cell was developed by T.M. Owen at Broken Hill South in 1913 for the same purpose, and in an improved form, it became the standard cell. 3. Leslie Bradford at BHP activated and depressed minerals selectively by adding chemicals to the pulp. In 1913 he patented his discoveries: that copper sulfate activated sulfide minerals and that sulfur dioxide depressed blende during galena flotation. Some months later, John Myers in the United States independently discovered that copper sulfate was an effective activator. These discoveries, in particular the selective activation and depression of minerals, changed flotation from an inflexible bulk process into a process that could be used for the production of individual mineral concentrates. By 1916 the urgent problem of finding a new concentration process for the Broken Hill ore had been solved; the prosperity of the town was ensured, and the turbulence and excitement associated with testing new ideas diminished, at least for a time. Copper in the United States, 1911–1920

Film flotation was the first flotation process used successfully in the United States. Machines designed by Arthur Macquisten in Scotland were used in Nevada in 1906 and in Idaho in 1911. The principle of operation was that deslimed and oiled sands flowed through a rotating drum designed to continually lift the particles and gently present them to the surface of the pulp (Figure 3). Sulfides adhered to the pulp surface and were collected. Macquisten cells worked well on sands; the Nevada plant produced a 20% copper concentrate from a 2.5% copper feed, and the Idaho plant, which operated for 10 years, produced a 45% zinc concentrate and a high-grade lead concentrate (Truscott 1923; Crabtree and Vincent 1962).

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 72 Tuesday, January 2, 2007 7:36 PM

72

HISTORICAL ASPECTS OF FLOTATION

6 ft

Water Level

Feed

Roller

Concentrate

Tailing Source: Truscott 1923.

FIGURE 3

Macquisten film flotation cell

In 1911 froth flotation was used in the United States for the first time (Hines and Vincent 1962). “James M. Hyde installed the first froth flotation plant in the U.S., about August 1, 1911, at the Basin Reduction Company plant, Basin, Montana, which was then under lease to the Butte and Superior Copper Company (Hines and Vincent 1962, p. 11). …It would have been difficult however to pick an ore as suitable as the Black Rock ore for treatment by froth flotation with the knowledge available in 1911. The ore contained only 1% pyrite, 1.1% Pb as galena with which the silver was associated, 17–20% Zn, and 0.25% Cu (p. 19).” The events relating to this plant have been fully described by Hines and Vincent (1962); suffice it to say that an important technical innovation was the use of rougher and cleaner cells in closed circuit with cleaner tailing returning to the rougher. James Hyde clearly understood the importance of cleaning concentrates to remove entrained gangue particles although the 50-ton mill at Basin was his first experience with froth flotation. During the next year, the company built mills with capacities of 200 and then 1,200 tons per day to verify that flotation would work, and with their success, flotation was poised to take off. The Basin ore was a good ore to start with, but the real prize would be the copper ores. Growth in the use of copper in the United States during the late 1800s and early 1900s was 5.8% annually, and even the rich deposits in Upper Michigan, Montana, and Arizona could not support this indefinitely. Lower-grade deposits had to be mined eventually. Daniel Jackling set the pattern in 1903 when he was given the task to build a 300-metric-tons-per-day mill to process 2% copper ore and persuaded the owners to mine 5,000 metric tons per day, which required a total change in the mining and milling systems. The new mill was ready for operation in 1907, and even at 60% recovery, it made large profits on 1% ore. Its capacity was soon doubled and redoubled. Others followed his lead, and large mills were built in Nevada and Arizona to process low-grade ores that were also very profitable at high copper prices although the recoveries were low. The Inspiration Company was a leader in developing flotation for porphyry copper ores. In 1911 it owned part of a huge deposit in Arizona, but this ore gave poor results with simple gravity concentration. The investment required to make it profitable could not be made by its backer, W.B. Thompson, and it was purchased by the Anaconda Company. Dr. Louis Ricketts became consulting engineer. “He did not think the mill Thompson’s engineers had planned would recover enough of the copper. To the horror of the stockholders, he threw away one million dollars’ worth of mill construction and spent a year and another million dollars experimenting. Then, he built the first mill that used the new flotation process. The result of Dr. Rickett’s delay was that this company caught the high copper price of 1915 with the most successful mill that had ever been built” ( Joralemon 1973, p. 243). In © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 73 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

Feed 9 ft Water Level

73

Overflow Lip Float

Porous Canvas Mat

Concentrate

Tailing

Feed Air Box

Air at 4.5 lb

Cross Section

Inclination 1:20

Air Mat

Tailing Complete Cell

Source: Truscott 1923.

FIGURE 4

Pneumatic Callow cell modified by Inspiration Copper

late 1912, Dr. Ricketts carried out flotation tests in the MSC laboratory and obtained 87% recovery of copper in a 15% concentrate from a 2% ore. Then he built a 50-tons-per-day mill in early 1913 and a 600-tons-per-day mill in early 1914 to verify the results. His testwork was comprehensive and included studies of different flowsheets and flotation machines, new reagents, and the effect of fine grinding. In 1915 a 15,000-tons-per-day mill was built, and 80% of the copper in the feed was recovered. It was the first mill in which flotation was applied to ore instead of gravity tailings. Two of the innovations in the new mill were ball mill–classifier circuits for grinding the ore directly to flotation feed size and flotation machines that used compressed air. Both of these had an effect on the control of the process: the closed grinding circuits controlled particle size and minimized the production of coarse, composite particles; and the compressed air cells controlled bubble size. The cells built by J.M. Callow (see Figure 4) were chosen by the Inspiration Company in preference to subaeration cells, and their success led to their wide use in concentrators for many years. “Flotation spread at a rapid rate; by 1914, 42 mining companies were operating or experimenting with the flotation process. The list increased in 1915 to include most of the principal copper and lead mines” (Hines and Vincent 1962, p. 29), and in 1918, 25 million short tons of copper ore were concentrated by flotation. A spin-off from the new technology was a growing business in flotation chemicals for enhancing or retarding the flotation rates of specific minerals, and by 1916 many companies were making reagents. By 1925 thousands of patents had been awarded for flotation chemicals, the most important being to Cornelius Keller and Carle Lewis in 1923 of MSC for the use of xanthates as collectors for sulfides. Xanthates took much of the guesswork out of sulfide flotation because they increased flotation rates of sulfides considerably; they were soluble in water, and their addition rates could be controlled. It is not known how Keller and Lewis came to discover the collecting properties for sulfides; perhaps it was because sodium ethyl xanthate was used in making rubber and as a defoliant and herbicide, or the initial tests might have been conducted because a bottle of xanthate was readily available on the shelf. The result of their discovery was that xanthate-lime-pine oil circuits were soon in common use, and by 1925 xanthate had transformed flotation into a process that was stable and reliable because it could be added in controllable amounts. With the success of copper flotation, the “startling innovation” proposed by Carrie Everson had become a reality. Because she lived until 1914, she saw the start of large-scale flotation and would have known that her remarkable, but unrewarded, efforts during 1885– 1892 had not been in vain.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 74 Tuesday, January 2, 2007 7:36 PM

74

HISTORICAL ASPECTS OF FLOTATION

Zinc and Lead at Cominco in Canada, 1917–1922*

Flotation started in Canada in 1917 at the Sullivan mine of Consolidated Mining and Smelting Company (Cominco). The metallurgist responsible was Ralph Diamond who had joined the Anaconda Company in Anaconda upon graduation from the University of Toronto in 1913. In May 1914 he was invited to lead a project to study a new “secret” process. The first flotation plant in North America had been built at the Superior Mill in Butte, Montana, and arrangements were made for Diamond to learn laboratory flotation testing there. The Superior Mill used the Hyde process on ore from the Black Rock mine. Following this assignment, Diamond commenced testwork on Anaconda slimes under the direction of George Chapman of MSC. Chapman had been one of the pioneers in froth flotation in Australia between 1904 and 1906. Later, in 1914, Diamond worked under J.M. Callow at the Inspiration Test Mill in Miami, Arizona, on MS cells in which ore was being treated at a rate of about 16,000 tons per day. It is interesting that the Anaconda plant used the experimental Callow pneumatic cell. The Anaconda slimes flotation plant started operation using standard Hardinge mills—likely an association that resulted in the use of Hardinge mills at Sullivan in 1922. The Anaconda Company had taken out a license under MSC. The processes used for copper slimes and copper sands and zinc ore had been developed by MSC, largely under the direction of George Chapman. Diamond remained in charge of flotation research for Anaconda until February 1917. At that time, Diamond went to Utah to install and operate a flotation plant for the Ohio Copper Company near Bingham, Utah, for the treatment of a partially oxidized copper ore. While in Utah, Diamond was contacted by Selwyn Blaylock, assistant general manager of the Cominco, regarding work in the development of the electrolytic treatment of zinc ores. This proposal was declined but resulted in a proposition to instigate testwork on the application of the froth flotation process to the refractory ore from the Sullivan deposit. The Sullivan deposit presented two new challenges to existing froth flotation practice. These were, firstly, a remarkably fine association between the valuable galena and blende minerals and the gangue iron sulfide present. Secondly, the ratio of iron sulfide, mainly pyrrhotite together with a small amount of pyrite, to the lead and zinc sulfide minerals was significantly higher than in ores previously studied. A sample of ore was sent to Diamond in Utah for preliminary testing. This work was sufficiently encouraging that Diamond joined the Consolidated company in June of 1917 and continued flotation work on the Sullivan ore at Trail. By the end of 1918, a successful three-stage differential flotation process had been demonstrated on a 600-tons-per-day test mill. The 1918 annual report of the company includes the following statement: “Important improvements in metallurgical practice in regard to treatment of the complex ores of the Sullivan Mine have added many years to the life of that property and have made it one of the most valuable mineral deposits in America if not in the whole world.” Cominco had taken out an MSC license in 1917. This was imperative despite the high royalty rates as MSC controlled most of the important patents. Furthermore, much of the general flotation knowledge that prevailed among millmen in those early days stemmed from the MSC reservoir of knowledge although MSC’s work on Sullivan ore was negative throughout. MSC never had an employee stationed in Trail to work on the Sullivan deposit problems, not even for a few days. MSC had a laboratory in San Francisco, but trained men were at a premium, and at that time, and for the next few years, they were simply deluged with ore samples for testing, requests for information, and help in the field. They had just *This section was written by Mike Fairweather. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 75 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

75

three or four and occasionally five men at any time who were available for all laboratory and fieldwork. Most of the ores submitted were simple and reacted favorably to the existing acid or neutral circuit flotation configurations. Some of these represented great reserves, potential large operations, and large royalty producers. With many, major construction was soon under way. All ores as received were simply subjected to certain standard tests, and those presenting difficulties were set aside. There were plenty of simple ores to fully occupy their time. The Sullivan was a very complex ore and did not react to standard tests. Moreover, MSC knew little at that time of the potential of the Sullivan ore body, and Cominco, as a company, did not occupy a prominent place in the mining world. In those early years, MSC was not impressed with the possibilities of copper sulfate, and this reagent was not included in testwork in 1916. It was around 1920 that they became aware of its importance. Sullivan ore was a special problem, and copper sulfate meant more to its successful treatment than was the case with 99 out of 100 ores. Coal in Europe

“It was not until 1920 that the experimental work of Bury, Broadbridge and Hutchinson directed attention to the possibilities of the use of froth flotation for coal cleaning. The first froth flotation plants for coal cleaning were erected the same year in Spain and in France. The first British plant was erected in 1922, and during the following year, plants were erected in Germany and in Belgium” (Chapman and Mott 1928, p. 385). Flotation was the only process available in 1920 to clean the fine fraction (–0.5 mm) of coking coal. Laboratory studies of coal flotation had started in the United States in 1915, and it was found to be very effective, but there was little incentive to use the process because coal was mined from thick seams, and relatively few fines were made during mining and transport. The high cost of dewatering coal concentrate was also a deterrent. In Europe, coal mining was mainly carried out in thin seams underground, and mechanized mining was used, which generated a high proportion of fines, so there was a higher incentive to use flotation. Consequently, coal flotation was first used there in 1920, and by 1925, flotation was cleaning about 1 Mt of coal annually. In 1927, 36 plants were using MSC cells, 14 being in Spain and 12 in Germany. Coal flotation took much longer to become popular than sulfide flotation because it could be used to clean only a small part of the mined coal, and its product was low in value. It did grow, however, and in 1933, there were about 60 coal flotation plants in Europe and 1 in the United States (Aplan 1999). It is interesting that the Elmore vacuum process, which had only limited success on sulfide ores, was widely used in England for many years to float minus-1/8-in. coal because the power required was low, the froth was easy to dewater, and coarse particles were floated. Technology Transfer and Litigation

The growing interest in flotation was reflected by the fact that the Engineering and Mining Journal published 3 articles about the process in 1901, 10 articles in 1902, and 32 in 1903 (Hoover 1914). There were many inventions and patents, and inevitably, there was litigation between inventors who developed virtually identical processes at different mines and who wished to claim damages for breach of patents. An important issue was license fees. These encouraged the sale and transfer of technology if they were reasonable but often led to piracy and secrecy if they were excessive. At Broken Hill the royalty cost in flotation was up to 30% of the operating cost, so it is not surprising that companies sought to make changes in the process to avoid paying royalties. Litigation often continued for years and must have © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 76 Tuesday, January 2, 2007 7:36 PM

76

HISTORICAL ASPECTS OF FLOTATION

rewarded many lawyers with high incomes. Whether litigation helped or hindered flotation is debatable, but there is no doubt that the expensive court cases are a reminder that legal action should be a last resort in metallurgical disputes. To close this section on the formative years of flotation, a comment should be made on the contribution of MSC. It was formed in London to apply oil agglomeration processes to the Broken Hill ore to concentrate sulfides, but by 1904, it had found that floating sulfides in a scum with a little oil gave better results than agglomerating them with excess oil. So the company turned to flotation. Then came stirred cells, froth flotation, and eventually xanthates—the company was involved in all three processes. The company employed skilled and experienced engineers who successfully promoted the flotation process and sold MSC flotation cells in many countries. Not surprisingly, MSC became involved in many lawsuits as staff left the company and worked on flotation as consultants or employees for other companies. There is no doubt that staff of MSC contributed much to the growth of flotation and that the rather aggressive tactics of the company helped the transfer of the new flotation technology considerably. Y E A R S O F C O N S O L I D AT I O N , 1 9 2 5 – 1 9 6 0

By 1925 efficient subaeration cells were available, and chemicals were being used for the selective activation and depression of minerals. This was just as well, given that technical progress was slow during the next 25 years, which were dominated by economic depression and war. But the demand for minerals produced by flotation continued to increase. The scope of flotation also expanded during this time, two examples being an oil flotation process that was patented in 1928 to recover phosphates from previously discarded fines and a process that was developed in 1937 to recover potash from salt-saturated brines. To meet the increased demand for flotation, plant sizes increased. But caution was the order of the day in plant design, and large plant capacities were obtained by use of many small units that were known to be reliable rather than by use of new large units that promised economy of scale but were untested. One example was the Morenci concentrator built in 1942, which had 432 cells, each with 2.2 m3 of volume to float 45,000 tons per day of ore. A by-product of the early years of flotation was the interest of many senior engineers in research because they had seen the immense rewards that it could bring. The result was that companies established research units in several universities in the 1920s to work on flotation fundamentals, and these contributed much to the understanding of flotation. Particular mention is made of the groups led by A.F. Taggart at Columbia University (New York), A.M. Gaudin at the Massachusetts Institute of Technology (MIT), and I.W. Wark at the University of Melbourne (Victoria). A D VA N C E S I N F L O TAT I O N T E C H N O L O G Y, 1 9 6 0 – 2 0 0 0

The production of minerals by flotation increased rapidly from 1950, as shown in Table 5 for copper and zinc, and during this time there were many improvements in the process. The focus was on reagents, flotation machines, and the control of circuits. Discussion in this section will be limited to flotation machines and to on-line analysis of pulp streams, which is necessary for accurate control. The most commonly used flotation machines can be divided into three groups based on flotation rates:

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 77 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

TABLE 5

77

World mine production of copper and zinc, 1950–2000

Metal Copper, Mt Zinc, Mt

1950 2.3 2.0

1960 4.1 2.8

1970 6.3 5.7

1980 7.6 6.1

1990 9.0 7.0

2000 11.0 8.9

Source: International Zinc Association 2003; International Copper Study Group 2004.

1. Flotation columns in which feed pulp enters the collection zone at the top of a column, flows downward, and contacts air bubbles that have been generated by spargers at the base of the column. The column tank itself is both the primary collection zone and the disengagement zone. Columns are considered to be low-intensity machines, and the flotation rate constants are low. 2. Mechanically agitated flotation machines in which a rotating mechanism is used to keep solids in suspension and to create bubbles by shearing air that is either applied to the machine under pressure or induced. They are considered to be mediumintensity devices with flotation rate constants 1.2 to 1.5 times that of a column. 3. High-intensity flotation machines that consist of an external aeration/contacting mechanism by which pulp is brought into intense contact with fine bubbles. The external contactor can use either pressurized air or air entrained into a fluid jet. The contactor is the primary collection zone, and the tank is the disengagement zone. These machines have flotation rate constants 2 to 4 times greater than that of the mechanical flotation cells. Only mechanically agitated machines were used until the 1960s, after which columns were introduced, followed by high-intensity machines. These type of machines will be discussed in the following sections. Mechanical Cells

Mechanical flotation cells have changed little in their principle of operation since they were invented in 1912, although many aspects of their operation have been improved. Significant literature is available on their design and development, with excellent recent reviews by Arbiter (1999), Weber et al. (1999), and Yianatos (2003). Their maximum size has been increased greatly during the last 35 years to reduce the capital cost of equipment and the floor area required per metric ton of ore floated in high-capacity plants. Increase in maximum cell volume from 1960 to 2004 is shown in Table 6. Pneumatic Flotation

These machines introduce air through diffusers or aerators (or, as a general term, spargers) rather than by mechanical dispersion. Flotation columns are the current principal representatives of this class. Their history contains the colorful characters, claims, and counterclaims that have marked flotation from the earliest days. The first machine with obvious column geometry, tall relative to its side dimensions, appears to have been invented by Norris (1907; Figure 5a). The novelty, however, seemed less to do with geometry and more with introducing air using pressurized water to overcome the limitations of vacuum release flotation at high altitude. Towne and Flinn (1919) patented a column depicting a countercurrent-flow slurry descending against a rising swarm of bubbles (Figure 5b). Towne and Flinn made some prescient observations: that the addition

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 78 Tuesday, January 2, 2007 7:36 PM

78

HISTORICAL ASPECTS OF FLOTATION

Maximum size of Outokumpu mechanical flotation cells (m3), 1950–2000

TABLE 6 1970 16

1976 38

1980 60

1995 100

1997 160

2002 200

2004 300

Source: S. Ronkainen, personal communication.

Wash Water

1

2 1 – Feed 2 – Concentrate 3 – Tailing 4 – Air

2 2

1 1

4 3

3 4 A. 1907 Norris column (Norris 1907). Air injected into pulp. Not tested in plants.

FIGURE 5

4 3

C. 1963 Boutin and Tremblay B. 1919 Towne and Flinn column (Taggart 1927; column (Boutin and Tremblay 1963; Rubenstein 1995). An Rubenstein 1995). Air air sparger and wash water entered through porous medium. Unsuccessful in were used—a simple and successful column. plants because of sanding.

Evolution of flotation columns

of oil increased “bubble-forming capacity” and that attached “oil-jacketed” particles stabilized froth. They also noted that particles not attached to bubbles in the froth “settle back towards the water or pulp column”; that is, they were describing entrainment and dropback. Depending on the perspective, either of these candidates represents the “first” flotation column (Figure 5c; Rubenstein 1995; Jameson 2002). Initially, interest in pneumatic flotation was short-lived and not revived until the second half of the 20th century. The problems that developed, which were articulated in the patent issued to Hollingsworth (1968), included sanding of coarse particles in the absence of mechanical agitation, plugging of porous diffusers, and channeling in scaled-up versions. It was not so much solving these problems (many still exist, in fact) that prompted the rapid commercial expansion after about 1980 but rather the use and control of wash water into the froth—the “key feature which permits high upgrading” (Finch and Dobby 1990, p. 3). If wash water is the “key,” it is of interest as to who first claimed this innovation. Searching column patents has identified four that incorporated wash water: Bennett and Dell (1963), Boutin and Tremblay (1964), Hukki (1967), and Hollingsworth (1968). For wash-water use, the patents reference Bennett and Dell back to 1958, Boutin and Tremblay to 1963, Hukki to 1965, and Hollingsworth to 1965. Therefore, the honor of first mention appears to go to Bennett and Dell. The use of water sprays was already known to increase grade by washing away of gangue particles, which are mechanically entrained into the froth (Klassen and Mokrousov 1963). But application was fitful. Why was it more successfully exploited in columns? It was certainly not because wash water “can kill the froth” in mechanical cells (Fuerstenau and Han 2003, p. 297). Column geometry, small cross-section to volume compared to most mechanical cells, does economize on wash water. But, most importantly, column operators learned how much to add. Column Flotation of Canada Ltd. (the first column flotation company, © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 79 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

79

founded to develop and market the invention of Boutin and Tremblay) promoted the addition of sufficient wash water to generate a so-called “positive bias,” that is, a net flow of water downward across the froth–pulp interface, which could be controlled by maintaining a tails flow higher than the feed flow (ideally, this refers to flow of water, but in reality, it is usually flow of slurry). Operating with a positive bias helped ensure that entrainment was countered. Wheeler (1983), president of the company after 1963, openly discussed this strategy during a column seminar at McGill University in Montreal and within days, one of the attendees, Roger Amelunxen, was back at Gibraltar Mines successfully operating a “homemade” column. Many column designs have been proposed in the West and in Russia (Rubenstein 1995), but the one that survived is that of Boutin and Tremblay. The deciding factors were probably the simple design, basically an open vessel, coupled with the positive bias strategy. The invention germinated in trials using columns for a solvent-in-pulp process (P. Boutin, personal communication). Entrainment of slurry in the rising droplet wake was solved by the addition of clear aqueous phase just below the interface. The inventors were quick to realize that flotation offered more potential. Given the role of “oil” in the early history of flotation, it is interesting to see it as a source of inspiration here. The inspiration continues with a recent proposal to combine flotation and solvent extraction (Chen et al. 2003). The commercial road was not smooth. Some 17 years after founding Column Flotation of Canada Ltd., the first industrial installation was recorded at Les Mines Gaspé (Cienski and Coffin 1981). This may seem a long gestation period but is about the norm for the minerals business (Napier-Munn 1997). After 1981, progress was rapid. A scale-up methodology was developed (Dobby and Finch 1986) that was first used to design the columns at Mount Isa Mines (Espinosa-Gomez et al. 1989). By the early 1990s, three additional companies were marketing flotation columns based on the Boutin–Tremblay design, which was becoming known as the conventional or Canadian column. The main suppliers today (in no particular order) are MinnovEX, CPT, Dorr-Oliver-Eimco, Cisa, Control International, Multotech, RIF, Metso, and Dual Extraction. One estimate is that there are some 3,000 columns installed worldwide, about 30%–40% being homemade, ranging from the minerals industry to the offshore oil industry (the flotation capacity in the oil industry may actually exceed that in the minerals industry). The success of columns helped usher in other novel developments in flotation machines (Finch 1995). Some were aimed at overcoming the well-known problem of pneumatic cells—reliable gas injection. Jet-type spargers is one outcome. Dispersing air into slurry and injecting the mix into the column is another. The patent of Hollingsworth (1968) ventured this possibility, the Microcel being a commercial version (Yoon, Adel, and Luttrell 1992). The high retention time in some columns (tens of minutes on occasion) is eliminated in the Jameson cell, which exploits air dispersion/slurry contacting in a downcomer where retention time is just seconds ( Jameson 1988). The Voith Sulzer cell developed for de-inking recycled paper shares some design features (Finch and Hardie 1999). Research focused on understanding mechanisms has also led to a range of sensors finding application in mechanical cells (Gomez and Finch 2002), such as optimizing gas distribution to banks of cells (Cooper et al. 2004). The conventional cell manufacturers were not idle during this period. From the “virtual columns” of the early 1990s (Finch 1995), a range of competing tank cell designs is now available, which, when viewed from a certain angle, reveal their inspiration.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 80 Tuesday, January 2, 2007 7:36 PM

80

HISTORICAL ASPECTS OF FLOTATION

Tank

Air Feed

Flotation Tank

Jet Chamber Nozzle Body

Baffle

Tailing

Deaeration Zone

Source: Cusack 1968.

FIGURE 6

Davcra flotation cell

FIGURE 7

XPM-series flotation machine

Flotation columns have taken their place among the options in cell selection. Overused in the initial flurry perhaps, they have retained a place in many circuits for most mineral commodities. High-Intensity Flotation Cells

High-intensity cells use very small bubbles produced by forced air (e.g., Davcra, Bahr, and Microcel cells) or induced air (e.g., Jameson and XPM cells). The Bahr cell, which was invented in 1974, was a column fed by a high-pressure air–pulp mixture, and this cell started a trend to combine the best ideas of columns and high-intensity cells into single machines. The first reported high-intensity cell was the Davcra cell (shown in Figure 6) devised by Bill Davis at the Zinc Corporation in Broken Hill and tested there in 1966. The principle was that most particle–bubble interactions are time independent, and recovery depended mainly on the characteristics of the intensely mixed zone (Davis 1966). The cell worked by air and feed slurry being injected into the tank through a dispersion nozzle, with energy being dissipated via collision with a vertical baffle, as shown in Figure 6 (Cusack 1968). The Davcra cell was used for some years in plants for floating sulfide minerals and coal. In China, Professor Daiwei Wu and colleagues developed jet flotation cells and started using them in an industrial plant in 1967. The XPM flotation machine is similar to a mechanical flotation machine, but the rotating mechanisms are replaced by jets of pulp and air that form “aeration-agitation” zones. Part of the pulp within each cell is drawn along with air into a circulating pump, and the mixture is pressurized and squirted from a conical jet, as shown in Figure 7. Froth forms rapidly and is removed by scrapers. These cells are used in 14 Chinese coal-preparation plants (Wu and Ma 1998). The largest cells are now 23 m3. For the past 30 years, extensive development of high-intensity flotation machines has been conducted in Germany with much of the original work undertaken by Professor Bahr at the Technical University of Clausthal. The Bahr cell shown in Figure 8 was one of the initial flotation devices developed from this work. The Bahr cell uses aerator units in which compressed air flows through small openings via channels into the pulp. The aerator units were located beneath the main flotation tank and entered the tank vertically (Cordes 1997). Since its development, many derivations and variations of the Bahr cell have been produced under a variety of names including Ekoflot, Pneuflot, Allflot, and Imhoflot. The Jameson cell, shown in Figure 9, was developed by Graeme Jameson of the University of Newcastle (Australia) and the technology was developed for commercial application by © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 81 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

81

Air

Air

Feed

Tailing

Concentrate

Feed

General arrangement. The tank contains several feed pipes with their aeration zones.

Typical feed pipe showing the aeration zone.

Source: Truscott 1923.

FIGURE 8

Bahr cell Pulp Feed

Induced Air

Downcomer

Nozzle Free Jet

Plunging Jet Recirculating Eddy

Mixing Zone

Submerged Jet Pipe Flow Zone Concentrate Disengagement Zone Pulp Mixture

Cell

FIGURE 9

Tailings

Jameson cell

Mount Isa Mines Limited. Its principles of operation have been discussed by numerous authors including Jameson (1988); Jameson et al. (1988); and Evans, Atkinson, and Jameson (1995). The high-intensity contacting zone is the downcomer. Feed pulp is pumped into the downcomer through an orifice plate, creating a high-pressure jet. The plunging jet of liquid shears and then entrains air, which has been naturally aspirated. Because of a high mixing velocity and a large interfacial area, there is rapid contact and collection of particles. One unique feature of the Jameson cell is the operation of the downcomer under a vacuum, which results in a high-intensity contact residence time that varies from 1 to 10 seconds. Since its invention in 1986, there have been 225 Jameson cells installed in variety of coal, metalliferous, and industrial mineral applications. The Microcel was developed by Yoon, Adel, and Luttrell (1992) at the Virginia Polytechnic Institute. In the Microcel flotation machine, the slurry is mixed with small bubbles in the microbubble generator, which is outside the column, and separation occurs inside the column. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 82 Tuesday, January 2, 2007 7:36 PM

82

HISTORICAL ASPECTS OF FLOTATION

Wash Water Distributor

Froth Product

Slurry Air

Feed Inlet Slurry Manifold

Wash Water Inlet Pressure Transducer Frother Inlet Inlet from Pump

Bubble Generators

Tailing Outlet Control Valve Outlet to Pump

Microbubble Suspension

Source: Yoon, Adel, and Luttrell 1992.

FIGURE 10

Microcel

The microbubble generator is a high-intensity bubble-contacting zone formed by a static in-line mixer; slurry is drawn from the base of the column and pumped through the generator; pressurized air is injected into the mixers, and the high resultant shear forces create fine bubbles (see Figure 10). The operation of the Microcel has been described in several publications, for example, Phillips et al. (1997) and Brake (1998). There are more than 100 installations in mineral and coal plants worldwide. O N - S T R E A M A N A LY S I S

A seminal change in flotation technology occurred when on-stream analysis (OSA) systems were developed. These enabled the metal contents of streams to be measured on-line and circuit grades and recoveries to be calculated every few minutes. In the days before OSA, flotation was an art, and results depended on the operator’s skills in observing the froth, using the panning dish, and controlling air and reagents manually. After OSA became available, flotation could be monitored accurately and controlled. Its development in the 1960s was timely; cell size was about to grow rapidly, and there was no alternative sensor for process control. OSA by itself was not enough. In the early 1960s, digital computers became available, and it was the OSA–computer link that changed flotation into an advanced technology. Setting the Scene

The problem with flotation for many years was that there was a long delay between taking samples from circuits and obtaining the assays, so the results were of historical value only and were not useful for circuit control. Better control could only be achieved if the delay was reduced to minutes, but rapid analysis of elements such as lead and copper in circuit streams was not possible until particles could be analyzed in the pulps rather than by sampling, drying, and use of conventional wet methods. In the 1950s, it was realized that X-ray techniques offered the most promising approach, and there seemed to be only one likely candidate: X-ray fluorescence (XRF) analysis by wavelength dispersive methods based on X-ray tube and (Bragg) crystal spectrometer. In the 1960s, an alternative approach was developed of using radioisotope X-ray techniques based on gamma-ray preferential absorption, or XRA, and energy dispersive (XRF) analysis. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 83 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

83

Developments Required

In 1960 the X-ray and radioisotope approaches required completely different developments before a practical OSA system could be realized. • The X-ray tube and crystal spectrometer system was relatively complex and expensive but was known to give accurate results because the wavelength dispersive system could resolve fluorescent X-rays from adjacent atomic number elements. The equipment had to be mounted in a central location of the plant, where it would sequentially analyze continuous samples taken from each process stream. This involved accurate sampling of each process stream, long runs of pipelines to the central analyzer, pumping, sample splitting, constant head tanks, and flow cells. • The radioisotope techniques of analysis involved the use of relatively inexpensive source-detector systems. Hence, one or more head units could be placed near each plant process stream, with only a short sample by-line between process stream and analysis system being required so that construction problems were simpler. The two radioisotope techniques were at different stages of development: 1. XRA techniques: These were suitable for high atomic number elements such as lead, uranium, tungsten, and bismuth. The problem was that there were only a few radioisotope sources emitting gamma rays of suitable energy, and even fewer were commercially available. 2. XRF techniques: These were essential for medium atomic number elements such as iron, nickel, copper, zinc, and tin. The problem in this case was that the detectors that were available could not resolve fluorescent X-rays from adjacent atomic number elements. X-ray Tube Systems

Plant tests were carried out in the late 1950s using laboratory X-ray tube systems for off-line measurement of dry plant samples and on-line measurement of plant pulps. The systems worked well. The first on-line system was tested during 1959–1960 in the 36,000-tons-perday concentrator of the Anaconda Copper Corporation in Butte, Montana. The X-ray system was satisfactory, but problems occurred with sample handling and presentation, which took months to solve. For example, “Sufficient wood is present to completely stop all flow through the X-ray head. Various types of screens were tried before finding a satisfactory solution. This was typical of the type of mechanical problem that plagued and delayed the final process control by X-ray analysis” (Lucy, Fulmore, and Holderreed 1963, p. 682). The system presented assays of 13 streams every 20 minutes. Another system was built in 1962 by the Research and Instrumentation Division of Rhoanglo Mine Services Ltd. in Northern Rhodesia (now Zambia). It was installed in the Bancroft concentrator and presented copper assays of 6 streams every 8 minutes (Barlin and Keys 1963). In 1962 the Outokumpu Group in Finland started what became a successful program of X-ray tube on-stream analyzers. In that year, the company established an Institute of Physics and brought the Pyhasalmi multimetal mine into operation. The mine became a large-scale laboratory in which instruments developed by the institute were tested, and this was an important factor in the success of the program. The objectives of the institute were to develop electromagnetic and XRF technology, and Professor Pekka Rautala became its director. Ore from the Pyhasalmi mine was floated to produce lead, copper, zinc, and pyrite concentrates, so there was a broad scope for experimentation. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 84 Tuesday, January 2, 2007 7:36 PM

84

HISTORICAL ASPECTS OF FLOTATION

The institute developed a 14-stream, wavelength dispersive X-ray analyzer, which was installed in the Pyhasalmi plant in 1968, and its success led to the installation of a second analyzer in 1970 (Lahteenmaki, Miettunen, and Saloheimo 1999). Assays of each stream were available to the operators every 6–7 minutes; the cost per assay was about 2 cents, and the analyzers had a high availability that was usually more than 99%. Using this analyzer, Outokumpu developed its own process control and management system and installed the first at the Kotalahti nickel concentrator in Finland in 1973. The analyzers have been continually improved, and current models can be installed in the process area instead of in a separate room. By 2004 more than 400 Outokumpu X-ray tube analyzers had been sold for use in mineral concentrators worldwide (M. Kongas, personal communication). Radioisotope Systems

Radioisotope systems offered the possibility of simpler and less expensive OSA compared with the X-ray tube systems. The plant system could be built up in stages as needs arose. Short sample by-lines would reduce the problems of pipe blockage caused by coarse particles and wood chips, and wear on detector windows would also be reduced. The critical link between the mineral industry and expertise in radioisotope X-ray techniques was established in 1962 when North Broken Hill Ltd. (NBH) approached the Australian Atomic Energy Commission (AAEC) with their requirements for OSA. The approach by Conzinc Riotinto of Australia (CRA) to the AAEC in 1965 resulted in the highly productive collaboration between physicists at the AAEC led by John Watt and metallurgists at CRA led by Bruce Rawling. Gamma-ray Transmission for High Atomic Number Elements

Gamma-ray transmission was the first technique developed for OSA using radioisotopes. In 1957 the AAEC was investigating a reactor system based on uranium powder suspended in liquid sodium, and the suspension was simulated by tungsten powder in water which was pumped around a 25-mm cross-sectional loop. The concentrations of tungsten over a cross section of pipe were determined by scanning the gamma-ray beam (thulium 170, 84 keV gamma rays) over the pipe cross-section. This was the first radioisotope OSA system of slurries (Watt and Lawther 1958). In 1962 NBH asked the AAEC whether it was possible to continuously determine the lead concentration of their flotation feed slurry on-line in a 150-mm-diameter steel pipe. Calculations showed that this should be possible by combining measurements of gamma-ray transmission at two different gamma-ray energies, about 200 and 662 keV. The AAEC overcame the lack of a suitable 200-keV radioisotope by developing a novel source based on Compton scattering of higher-energy gamma rays yielding an output of about 225-keV. This dual-energy gamma-ray transmission technique was successfully tested in the NBH plant in 1964 and 1966 (Ellis et al. 1967). The radioisotope system was installed on-line at the NBH concentrator in 1968. This was the world’s first permanent installation of a radioisotope OSA system in a mineral concentrator. In December 1965, Bruce Rawling asked John Watt about measuring both lead and zinc in sample by-lines from various process streams in a CRA concentrator at Broken Hill. Watt’s response was that lead in flotation feed would be accurately determined by dualenergy gamma-ray transmission based on radioisotopes Gd-153 (100 keV) and Cs-137 (662 keV). For tailings, a correction would have to be made for matrix variations by a further transmission measurement with gamma rays of suitable energy. These predictions were later © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 85 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

85

confirmed by calculation and experiments on samples of solids taken from plant streams (Watt 1967; Ellis et al. 1969). Philips Industries Pty. Ltd. provided laboratory equipment to CRA for sample by-line trials, and in 1968 CRA staff proved that lead could be determined accurately in a sample by-line from the main flotation feed stream (Hinckfuss and Rawling 1968). XRF for Medium Atomic Number Elements

The discussions between Rawling and Watt in 1965 crystallized thoughts on the urgency of developing radioisotope XRF techniques for OSA for medium atomic number elements such as copper and zinc. Australian mineral companies had to be contacted to obtain comprehensive information about their requirements, and samples had to be collected over extended periods of time from various process streams in several concentrators so that the AAEC could determine whether the techniques provided sufficient accuracy to meet these requirements. The AAEC contracted Australian Mineral Development Laboratories (Amdel) in 1966 to undertake the survey of mineral company requirements for OSA, and this was completed in mid-1967. Over the period from 1966 to 1968, mineral companies supplied the AAEC with suites of about 25 samples from each of several process streams in their plants that were taken over a period of at least 6 weeks. During 1966–1968 the AAEC undertook extensive development of radioisotope XRF assemblies and techniques (Watt and Gravitis 1973; Watt 1983). In 1966, XRF measurements were made on samples of lead/zinc ore, taken from widespread locations throughout the CRA mine, with excellent results of zinc in the range 0–34 wt % being determined to 0.6 wt % (1σ) (Watt 1967). Measurements on the samples taken from several streams in each of six concentrators also gave promising results (Ellis et al. 1969) with one exception: copper in the iron-rich tailings from Tennant Creek, which was later solved with the development of the detector-radiator assembly (Watt 1972). The success of laboratory measurements on samples was followed up with on-stream trials of radioisotope XRF systems at five mineral concentrators, undertaken during 1968–1971 by AAEC, Amdel, and plant staff (Fookes et al. 1971). Overall, these trials were very successful and led to improvements in radioisotope X-ray assemblies and techniques. During 1967 Douglas Hinckfuss of CRA proposed replacing XRF measurements on a sample by-line with measurements by probes directly immersed into the plant process stream (Hinckfuss 1972). The probe was a casing containing the radioisotope source and detector assembly. The immersion probe, shown in Figure 11, was a key development to the radioisotope OSA system because it overcame the need for use of sample by-lines and dramatically reduced window wear (Fookes et al. 1973). There was now a complete contrast in approaches to OSA by the radioisotope and the X-ray tube systems. Joint CRA–AAEC trials Density Probe Pb Probe

Pulp In

Zn Probe To Adjacent Flotation Cells Existing Slurry Vessel

Source: Cutmore et al. 1993.

FIGURE 11

Radioisotope probes immersed in a plant mineral slurry

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 86 Tuesday, January 2, 2007 7:36 PM

86

HISTORICAL ASPECTS OF FLOTATION

at Broken Hill using immersion probes demonstrated excellent results for lead and zinc determined in plant process streams (Fookes et al. 1973). Stump and Roberts (1974) demonstrated better control of grinding and flotation at the New Broken Hill Consolidated concentrator based on using radioisotope probes and computer control, as well as excellent accuracies of the OSA for lead and zinc in actual plant installations. From 1967 the AAEC also used silicon solid-state detectors for laboratory XRF measurements on mineral samples from plants. These detectors had good X-ray energy resolution and proved to be very suitable for the analysis of samples from process streams (Gravitis, Greig, and Watt 1974). These detectors were then not sufficiently stable for use in industrial plants. In the early 1980s, Amdel incorporated these detectors into immersion probes for plant use in tailing streams. Commercial System

The radioisotope XRF and XRA assemblies developed by the AAEC, incorporated into the casing of the immersion probe developed by CRA, became the basis of the commercial OSA system. Philips Industries Ltd. was selected as the licensee to design and manufacture the commercial system hardware, and Amdel was selected to undertake the feasibility studies, installation, and calibration. Amdel installed the first three plant-analysis systems in concentrators in 1973. The development of the radioisotope on-stream analysis system had been a productive 10-year project that contributed much to flotation technology. Amdel took over the system manufacture in 1978. Thermo Electron Corporation took over the Amdel Instrumentation Division, including staff, in 1999 and continues to manufacture the radioisotope OSA system in Australia and market it worldwide. By 2003 about 170 radioisotope X-ray systems had been sold worldwide for mineral processing operations. Automatic Control of Flotation Circuits

By 1975 the problem of long delays between taking samples in plants and receiving assays had been solved; flotation circuit performance could be assessed every few minutes, and automatic control systems could be developed for flotation circuits that would take into account changes in mineral contents and floatabilities. Multistream X-ray tube systems and radioisotope systems were being used to monitor and control flotation circuits in several concentrators. The objective of automatic control was and still is to operate each flotation circuit at the point on its optimum grade-recovery curve that gave the best economic results. Some of the approaches tried in early control systems were • Controlling reagents by the feed grade and/or the concentrate grade • Automatic raising and lowering of concentrate diverter trays on rougher banks to maintain stable performance of cleaner banks • Controlling reagents by making incremental changes and searching for optimum circuit performance These early systems were designed for specific cases, because no two ores are identical, and they usually improved circuit performance by reducing variations in concentrate grades and increasing recoveries of valuable minerals. They also improved the skills of operators who were given much more information about the circuit by computer-based data logging systems than was available through observation.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 87 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

87

The problem with control systems in those early days was that if the ore changed significantly, it was difficult to determine on-line the new grade-recovery curve and the best operating point, so the control target was difficult to define. There has been some progress on this problem, mainly through the use of various mathematical techniques. Software to control flotation circuits is commercially available. O T H E R U S E S O F F L O TAT I O N

Over the decades, flotation has found uses in areas far removed from mine sites. It is used for the removal of solids in wineries, breweries, butter and cheese factories, dairies, and sewerage plants, among others. It is commonly used as a means for removing oil and contaminants from water in smelters and refineries, as well as for removing algae and other organic contaminants from water. Other areas in which flotation can be found are the de-inking of recycled paper, treatment of abattoir and sawmill effluent, sugar milling and refining, wool scouring, the production of vegetable oil and margarine, paint manufacturing, and paper processing. POSTSCRIPT

The Centenary of Flotation Symposium celebrated the life of a remarkable process and the achievements of scientists and engineers in many countries who overcame early problems and made flotation an advanced technology. Theodore Hoover wrote of the difficulties of developing “concentration upside down,” as flotation was described (Ingalls 1907), into an acceptable and standard process: “The only previous authentic case where substances heavier than water have been made to float was the occasion of Elisha’s miracle with the axe (2 Kings 6) and mining and metallurgical engineers are not great believers in miracles” (Hoover 1914). But the engineers were never daunted. Flotation was very difficult to operate in its early years because of the variable nature of the ores, yet from its inception, engineers have continued to extend its limits of application. The variety and extent of its uses today could not have been imagined 100 years ago when thin scums of zinc concentrates were being floated on top of hot, acid pulps. Progress since then has been due to the cumulative achievements of a legion of engineers working on flotation processes in many countries. There will be many advances yet in flotation technology as its capabilities for separating valuable and waste materials are used more extensively. It is hoped that this brief history gives incoming flotation engineers and scientists some of the fascinating background of the process with which they will be working. BIBLIOGRAPHY

Anon. 1963. The flotation column. Can. Min. J. 84:55–56. Aplan, F.F., 1999. The historical development of coal flotation in the United States. Pages 276–278 in Advances in Flotation Technology. Edited by B.J. Parekh and J.D. Miller. Littleton, CO: SME. Arbiter, N. 1999. Development and scale-up of large flotation cells. Pages 345–352 in Advances in Flotation Technology. Edited by B.J. Parekh and J.D. Miller. Littleton, CO: SME. Barlin, B., and N.J. Keys. 1963. Concentration at Bancroft. Min. Eng. 15(9):47–52. Bennett, A.J.R., and C. Dell. 1963. Improvements in or relating to methods of and apparatus for separating and/or concentrating particles in liquid suspensions. GB Patent GB926,172. May 15. Bessel, G. 1877. Berlin Patent 42. July 2. ———. 1886. Berlin Patent 39,369. May 12. Blainey, G. 1968. Page 70 in The Rise of Broken Hill. Australia: Macmillan.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 88 Tuesday, January 2, 2007 7:36 PM

88

HISTORICAL ASPECTS OF FLOTATION

Boutin, P., and R.J. Tremblay. 1963. Froth flotation method with counter-current separation. U.S. Patent 3,339,730. ———. 1964. Method and apparatus for the froth flotation of ores. GB Patent GB970,841. September 23. Boutin, P., and D.A. Wheeler. 1967. Column flotation development using an 18 inch pilot unit. Can. Min. J. 88:94–101. Bradford, H. 1885. U.S. Patent 345,951. June 22. Brake, I.R. 1998. The development and commissioning of a new Microcel column flotation circuit for BHP Coal’s Peak Downs coal preparation plant. Pages 767–776 in XIII International Coal Preparation Congress, Brisbane Australia. Edited by A.C. Partridge and I.R. Partridge. New South Wales: Australian Coal Preparation Society. Broken Hill Branch, Australasian Institute of Mining and Metallurgy. 1930. The development of processes for the treatment of crude ore. Accumulated dumps and slimes at Broken Hill, NSW. Proc. Aust. Inst. Min. Metall. 80:379–444. Cattermole, A.E. 1902. British Patent 26,295. November 28. Chapman, W.R., and R.A. Mott. 1928. The Cleaning of Coal. London: Chapman and Hall. Chen, F., J.A. Finch, P.A. Distin, and C.O. Gomez. 2003. Air-assisted solvent extraction. Can. Metall. Q. 42(3):277–280. Cienski, T., and V.L. Coffin. 1981. Column flotation operation at Mines Gaspé molybdenum circuit. Pages 240–262 in Proceedings of the 13th Annual Meeting of the Canadian Mineral Processors. January. Cooper, M., D. Scott, R. Dahlke, J.A. Finch, and C.O. Gomez. 2004. Impact of air distribution profile on banks in a Zn cleaning circuit. Pages 525–540 in Proceedings of the 36th Annual Meeting of the Canadian Mineral Processors of CIM. Cordes, H. 1997. Development of pneumatic flotation cells to their present day status. Miner. Process. J. 2:69–82. Crabtree, E.H., and J.D. Vincent. 1962. Historical outline of major flotation developments. Pages 39– 54 in Froth Flotation 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Cusack, B.L. 1968. The Development of the Davcra flotation cell. Pages 481–487 in Broken Hill Mines—1968. Edited by M. Radmanovitch and J.T. Woodcock. Monograph Series AusIMM. Melbourne: The Australasian Institute of Mining and Metallurgy. Cutmore, N.G., W.J. Howarth, B.D. Sowerby, and J.S. Watt. 1993. On-line analysis for the mineral industry. Pages 189–198 in Proceedings of the AusIMM Centenary Conference, Adelaide. Melbourne: The Australasian Institute of Mining and Metallurgy. Davis, W.J.N. 1966. The development of a mathematical model of the lead flotation circuit at the Zinc Corporation Ltd.—Discussion and contributions. Proc., Aust. Inst. Min. Metall. 220(December):79–85. De Bavay, A.J.F. 1904. British Patent 18,660. August 29. Delprat, G.D. 1902. British Patent 26,279. November 28. Diamond, R.W. 1961. A Detailed Account of the Development of the Treatment by Flotation of the Ore of the Sullivan Mine, Kimberley, B.C. Private Report 1961. The Consolidated Mining and Smelting Company of Canada. Dobby, G.S. 1984. A fundamental flotation model and flotation column scale-up. Ph.D. thesis, McGill University, Montreal, PQ. Dobby, G.S., and J.A. Finch. 1986. Flotation column scale-up and modelling. CIM Bull. 79(889):89–96. Ellis, W.K., R.A. Fookes, V.L. Gravitis, and J.S. Watt. 1969. Radioisotope X-ray techniques for on-stream analysis of slurries: Feasibility studies using solid samples of mineral products. Int. J. Appl. Radiat. Isotopes 20:691–701. Ellis, W.K., R.A. Fookes, J.S. Watt, E.L. Hardy, and C.C. Stewart. 1967. Determination of lead in ore pulps by a technique using two gamma-ray absorption gauges. Int. J. Appl. Radiat. Isotopes 18:473–478. Elmore, F.E. 1898. British Patent 21,948. October 18. ———. 1904. British Patent 17,816. August 16. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 89 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

89

Espinosa-Gomez, R., N.W. Johnson, J.D. Pease, and P.D. Munro. 1989. The commissioning of the first flotation columns at Mount Isa Mines Limited. Pages 293–302 in Processing of Complex Ores. Edited by G.S. Dobby and S.R. Rao. New York: Pergamon Press. Espinosa-Gomez, R., J.B. Yianatos, J.A. Finch, and N.W. Johnson. 1988. Carrying capacity limitations in flotation columns in column flotation ’88. Pages 143–148 in SME Annual Meeting. Phoenix, Arizona. Edited by K.V.S. Sastry. Littleton, CO: SME. Evans, G.M., B.W. Atkinson, and G.J. Jameson. 1995. The Jameson Cell. Pages 331–363 in Flotation Science and Engineering. Edited by K.A. Matis. New York: Marcel Dekker Everson, C.J. 1885. U.S. Patent 348,157. August 29. Finch, J.A. 1995. Column flotation: A selected review—part IV: Novel flotation devices. Miner. Eng. 8(6):587–602. Finch, J.A., and G.S. Dobby. 1990. Column Flotation. Oxford: Pergamon Press. Finch, J.A., and C.A. Hardie. 1999. An example of innovation from the waste management industry: De-inking flotation cells. Miner. Eng. 12(5):467–475. Fookes, R.A., V.L. Gravitis, D.A. Hinckfuss, N.W. Stump, and J.S. Watt. 1973. Plant trials of radioisotope immersion probes for on-stream analysis of mineral process streams. Trans. Inst. Min. Metall. C 82(796):C21–C25. Fookes, R.A., V.L. Gravitis, J.S. Watt, G. Wenk, and L.R. Wilkinson. 1971. On-stream analysis for copper, zinc, tin, and lead in plant mineral slurries using radioisotope X-ray techniques. Pages 21–23 in Proceedings of Symposium on Automatic Control Systems in Mineral Processing Plants, Brisbane, May 17–20. Melbourne: Australasian Institute of Mining and Metallurgy. Froment, A. 1902. British Patent 12,778. June 4. Fuerstenau, D.W., editor. 1962. Froth Flotation 50th Anniversary Volume. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Fuerstenau, M.C. 1999. Froth flotation: The first ninety years. Pages xi–xxxiii in Advances in Flotation Technology. Edited by B.K. Parekh and J.D. Miller. Littleton, CO: SME. Fuerstenau, M.C., and K.N. Han, editors. 2003. Principles of Mineral Processing. Littleton, CO: SME. Gomez, C.O., and J.A. Finch. 2002. Gas dispersion measurements in flotation machines. CIM Bull. 95(1066):73–78. Graichen, K., J. Hanisch, H. Schubert, K.D. Steiner, C. Tanneberger, and E. Wachtler. 1977. Dei Gebruder Bessel und die Anfange der Flotativen Aufbereitung. Neue Bergbautechnik 7.Jg Heft 10 Oktober. Gravitis, V.L., R.A. Greig, and J.S. Watt. 1974. X-ray fluorescence analysis of mineral samples using solid state detector and radioisotope X-ray source. Proc. Aust. Inst. Min. Metall. 249:1–4. Habashi, F. 1994. Page 270 in A History of Metallurgy. Metallurgie Extractive Quebec, Enr. Quebec: Librairie des Presses de l’Universite Laval. Hinckfuss, D.A. 1972. Immersible fluorescence probe. Australian Patent 39,964/72. Hinckfuss, D.A., and B.S. Rawling. 1968. The development and application of an on-stream analysis system for lead at the Zinc Corporation, Limited. Pages 475–479 in Broken Hill Mines—1968. Edited by M. Radmanovitch and J.T. Woodcock. Monograph Series AusIMM. Melbourne: The Australasian Institute of Mining and Metallurgy. Hines, P.R., and J.D. Vincent. 1962. The early days of froth flotation. Pages 11–38 in Froth Flotation 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Haynes, W. 1860. British Patent 488. February 23. Hollingsworth, C.A. 1968. Flotation apparatus for concentration of minerals. U.S. Patent 3,371,779. Hoover, T.J. 1914. Pages 2–41, 183–194 in Concentrating Ores by Flotation. 2nd edition. London: The Mining Magazine. Hukki, R.T. 1967. Froth flotation apparatus. GB Patent GB105,891,415. February. Imhof, R. 1991. Device for carrying out pneumatic flotation. German Patent 4,116,645.0. ———. 1993. Five years of Ekoflot: Pneumatic flotation on the march. Aufbereit. Tech. 34(5):263– 268. Ingalls, W.R. 1907. Concentration upside down. Eng. Min. J. (October). © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 90 Tuesday, January 2, 2007 7:36 PM

90

HISTORICAL ASPECTS OF FLOTATION

International Copper Study Group. 2004. Home page. www.icsg.org/home.htm. Accessed May 2006. International Zinc Association. 2003. Zinc production and demand 1960–2002. In Zinc Guide. www.iza.com/zgd_org/HTM/056.htm. Accessed February 17, 2004. Jameson, G.J. 1988. A new concept in flotation column design. Pages 281–289 in Column ’88— Proceedings of an International Symposium on Column Flotation. Edited by K.V.S. Sastry. Littleton, CO: SME. ———. 2002. The froth phase in column flotation. Pages 161–168 in Flotation and Flocculation, From Fundamentals to Applications. Proceedings, Strategic Conference and Workshop, Hawaii, July 28–August 3. Edited by J. Ralston, J.D. Miller, and J. Rubio. Adelaide, Australia: Snap Printing. Jameson, G.J., M. Belk, N.W. Johnson, R. Espinosa-Gomez, and J.P. Andreaditis. 1988. Mineral flotation in a high intensity column. Pages 507–510 in Chemeca 88, 16th Australian Conference on Chemical Engineering. Sydney. Barton, Australia: Institution of Engineers. Joralemon, I.B. 1973. Page 243 in Copper: The Encompassing Story of Mankind’s First Metal. California: Howell-North Books. Klassen, V.I., and V.A. Mokrousov. 1963. An Introduction to the Theory of Flotation. London: Butterworths. Lahteenmaki, S., J. Miettunen, and K. Saloheimo. 1999. 30 years of on-stream analysis at the Pyhasalmi Mine. Presented at SME Annual Meeting, Denver, CO, March 1–3. Lucy, W., T.G. Fulmore, and F.L. Holderreed. 1963. Copper analysis of pulp streams in the Anaconda Copper Concentrator by X-ray fluorescence. Pages 679–689 in Proceedings of the 6th International Mineral Processing Congress. Edited by A. Roberts. London: Pergamon Press. Lynch, A.J. 1987. Pages 475–479 in Leslie Bradford Golden Jubilee Oration 4. Edited by J.T. Woodcock. AusIMM Monograph Series. Melbourne: The Australasian Institute of Mining and Metallurgy. Megraw, H.A. 1918. Pages 5–8 in The Flotation Process. 2nd edition. New York: McGraw-Hill. Napier-Munn, T. 1997. Invention and innovation in mineral processing. Miner. Eng. 10:757–774. Norris, D.H. 1907. Apparatus for separating the metallic particles of ores from the rocky constituents thereof. U.S. Patent 873,586. December 10. Phillips, D.I., R. Yoon, G.H. Luttrell, L. Fish, and T.A. Toney. 1997. Installation of 4-meter diameter Microcel flotation columns at LadyDunn preparation plant. Pages 115–132 in 14th International Coal Preparation Exhibition and Conference, Lexington, Kentucky. Potter, C.V. 1902. U.S. Patent 776,145. January 14. Rickard, T.A. 1932. Pages 398–403 in A History of American Mining. New York: McGraw-Hill. Rubenstein, B. 1995. Column flotation: Processes, designs, and practices. Pages 2–4 in Process Engineering for the Chemical, Metals and Minerals Industries. Volume 2. Edited by T.J. Veasey. Basel, Switzerland: Gordon and Breach Science Publishers. Stump, N.W., and A.N. Roberts. 1974. On-stream analysis and computer control at the New Broken Hill Consolidated Limited concentrator. Trans. AIME 256:143–148. Sulman, H.L., and H.F.K. Picard. 1903. British Patent 20,419. September 22, 1903; U.S. Patent 793,808. July 4, 1905. Taggart, A.F. 1927. Page 809 in Handbook of Ore Dressing. New York: John Wiley & Sons. Towne and Flinn. 1919. U.S. Patent 1,295,817. Truscott, S.J. 1923. Pages 393–416 in A Textbook of Ore Dressing. London: Macmillan. Watt, J.S. 1967. Recent developments in low energy X- and gamma-ray sources and applications in Australia. Pages 663–695 in Proceedings of the 2nd Symposium on Low-energy X- and Gamma-ray Sources and Applications. Volume 2. Austin, Texas, March 27–29. Report ORNL-11C-10. Edited by P.S. Baker and M. Gerrard. Oak Ridge, TN: Oak Ridge National Laboratory. ———. 1972. Radioisotope detector-radiator assemblies in x-ray fluorescence analysis for copper and zinc in iron-rich minerals. Int. J. Appl. Radiat. Isotopes 23(6):257–264. ———. 1983. On-stream analysis of metalliferous ore slurries. Int. J. Appl. Radiat. Isotopes 34(1):309–331.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 91 Tuesday, January 2, 2007 7:36 PM

HISTORY OF FLOTATION TECHNOLOGY

91

Watt, J.S., and V.L. Gravitis. 1973. Radioisotope X-ray fluorescence techniques applied to on-stream analysis of mineral process streams. Pages 199–205 in Automatic Control in Mining, Mineral and Metal Processing, IFAC International Symposium, Sydney, August 13–17. National Conference Publications No. 1973. Australia: Institution of Engineers. Watt, J.S., and K.R. Lawther. 1958. Measurement of concentration of tungsten suspensions and density of liquid sodium by gamma-ray absorption. Section 5 in Proceedings of the Symposium on the Peaceful Uses of Atomic Energy in Australia. Sydney: Australian Atomic Energy Commission. Weber, A., C. Walker, L. Redden, D.S. Lelinski, and S. Ware. 1999. Scale-up and design of large scale flotation equipment. Pages 353–370 in Advances in Flotation Technology. Edited by B.K. Parekh and J.D. Miller. Littleton, CO: SME. Wheeler, D.A. 1983. Column Flotation Seminar, McGill University, Montreal, PQ. May. ———. 1988. Historical view of column flotation development. Pages 3–4 in Column Flotation ’88. Edited by K.V.S. Sastry. Littleton, CO: SME. Woodward, O.H. 1965. Pages 78–99 in A Review of the Broken Hill Lead-Silver-Zinc Industry. 2nd edition. Edited by K.P.W. Parsons. Melbourne: Broken Hill Mining Managers Association– Australasian Institute of Mining and Metallurgy. Wu, D., and L. Ma. 1998. XPM-Series jet flotation machine. Pages 737–745 in XIII International Coal Preparation Congress, Brisbane, Australia. Edited by A.C. Partridge and I.R. Partridge. New South Wales: Australian Coal Preparation Society. Yianatos, J.B. 2003. Design, modelling and control of flotation equipment. Pages 59–68 in XXII International Mineral Processing Congress, Cape Town, 2003. Edited by Lorenzen, Bradshaw, Aldrich, Eksteen, Wright, and Thom. Marshalltown, SA: South African Institute of Mining and Metallurgy. Yoon, R., G.T. Adel, and G.H. Luttrell. 1992. Apparatus and process for the separation of hydrophobic and hydrophilic particles using microbubble column flotation together with a process and apparatus for the generation of microbubbles. U.S. Patent 5,167,798.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 92 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 93 Tuesday, January 2, 2007 7:36 PM

PART 2

Flotation Fundamentals

93

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 94 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 95 Tuesday, January 2, 2007 7:36 PM

Some Aspects of Flotation Thermodynamics D.W. Fuerstenau and S. Raghavan

A B S T R AC T

This chapter presents a brief summary of some of the thermodynamic aspects of flotation processes. Thermodynamic considerations that control interfacial and wetting behavior in mineral– water–air systems are discussed. Particular attention is given to the thermodynamics of collector adsorption. INTRODUCTION

The application of the principles of thermodynamics to flotation systems has contributed significantly toward understanding some of the underlying foundations of flotation processes. When applied to physical and chemical systems, such as flotation, thermodynamics includes several topics: 1. The conditions under which chemical substances, or different physical states of the same substance, exist in equilibrium 2. Whether, under certain specified conditions, a chemical reaction or a phase change will take place spontaneously. 3. The relation between the interchange of heat and other forms of energy when a chemical reaction or phase displacement occurs 4. The effect of temperature on chemical reactions and phase equilibria 5. The principles underlying the methods of measurement of those properties whose values are required for quantifying the foregoing Thermodynamics is used to predict whether or not a change will tend to occur and yet reveals nothing about the rate at which the change will take place. Though thermodynamics cannot actually report to a flotation engineer about what the mineral recovery will be at a given temperature or under given solution conditions, it can help the engineer make some predictions as to how the flotation response may change with temperature, type of collector, type of mineral, and so on. One criticism that has been leveled against the practical need for studying the thermodynamics of flotation is that thermodynamics is concerned mainly with equilibrium processes and the fact that, in the time span during which flotation takes place, the system may not be in equilibrium. Interestingly, Wada (1960) actually defined flotation as a thermodynamic process in which gas–liquid–solid interfaces participate in separating finely divided solids from one another. In this chapter, an attempt is made to briefly review various thermodynamic investigations and approaches undertaken to gain an insight into those interfacial processes that play a dominant role in the flotation process. This review is not considered to be exhaustive, but only summarizes some of the highlights. 95

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 96 Tuesday, January 2, 2007 7:36 PM

96

FLOTATION FUNDAMENTALS

T H E R M O DY N A M I C S O F S U R FA C E S

The enthalpy of the surface on a unit area basis, HS, is defined by (EQ 1)

HS = ES + (PV)0 = GS + T · SS

where ES is the total surface energy, PV is the pressure volume, GS is the Gibbs surface free energy, T is the temperature, and SS is the surface entropy per square centimeter of surface. Because the PV term is negligible for a surface, the surface energy and surface enthalpy are equivalent. The surface tension, γ, and surface free energy, GS, are defined by ∂G G S = γ = ⎛ -------⎞ ⎝ ∂A⎠ T, P, n

(EQ 2)

where A is the interfacial area, and n is the number of moles present in the system. These quantities are usually given as ergs per square centimeter or dynes per centimeter, which are identical in magnitude. Actually, only in a one-component system are GS and γ identical, but for the purpose of this review, no distinction will be made between them. Because the surface entropy at constant pressure is given by ∂G dγ S S = – ⎛ --------S-⎞ = – ------⎝ ∂t ⎠ p dT

(EQ 3)

the relation between total surface energy and surface tension is dγ E S = γ – T ------dT

(EQ 4)

For most liquids, the surface tension decreases linearly with temperature (Adamson 1967). In the case of water at 20°C, γ = 72.75 ergs/cm2 and dγ/dT = –0.16, from which one evaluates the total surface energy of water to be 120 ergs/cm2. In the case of octane, a typical liquid hydrocarbon, γ = 21.80 ergs/cm2 at 20°C and dγ/dT = –0.10, from which ES is calculated to be 51.1 ergs/cm2. In the case of solids, the evaluation of surface energies is less straightforward, both experimentally and theoretically. Some typical values of the surface tension (and surface energies) of selected materials are given in Table 1. These can only be considered approximate, but they enable one to see how widely solids do differ. Clearly, the type of chemical TABLE 1

Approximate surface energies of some materials at room temperature γ, ergs/cm2 25 110 230 450 1,000 1,800 1,900 5,600

Material Paraffin Graphite Halite (NaCl) Fluorite (CaF2) Magnesia (MgO) Gold Alumina (Al2O3) Diamond

ES, ergs/cm2

280 1,090

5,600

Source: Adamson 1967.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 97 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

97

bonds that hold a crystal together have a marked effect on the magnitude of the surface energy of solids. Finally, it should be added that mineral–water interfacial tensions are generally much lower than the surface energies, particularly for oxides. T H E R M O DY N A M I C S O F A D S O R P T I O N

What happens in the case of a system with several components when one or more of the components accumulates, that is, adsorbs, at an interface? The basis for the thermodynamics of the adsorption of dissolved substances was unambiguously laid down by Gibbs. The Gibbs adsorption equation relates the interfacial tension between two phases to the temperature, T, of the system; the chemical potentials of various species, μ1, μ2…μi, in the bulk; and the surface excess or adsorption density of the various species, Γ1, Γ2…Γi, at the interface; and has the following form (Defay and Prigogine 1966): dγ = – S S dT – ∑ Γ i dμ i

(EQ 5)

i

There are several ways to define Γi, but the simplest way mathematically is to use the convention proposed by Gibbs, namely, that the adsorption of the solvent is zero and that Γi is the excess surface concentration. Excess surface concentration means the excess at the interface over that which would be expected if the solution phase were uniform up to a hypothetical plane surface that divides the two bulk phases in a heterogeneous system. If one measures adsorption densities from differences in bulk concentrations in dilute systems (as are usually encountered in flotation systems), then the Gibbs convention is most applicable. Overall, the Gibbs adsorption equation is highly relevant to flotation through its application to phenomena involving frother systems, wettability, the development of surface charge at mineral–water interfaces, and so on, because it quantitatively expresses the change in surface tension due to the adsorption of surface-active materials. By defining the adsorption of the solvent (component 1) as zero, Equation 5 can be modified as i

dγ = – S S dT –

∑ Γi( 1 ) dμi

(EQ 6)

i=2

where Γi(1) refers to the relative adsorption of component i at the dividing surface, such that Γ1 = 0. Because flotation processes are often carried out at constant temperature, Equation 6 can be simplified to i

dγ = – ∑ Γ i( 1 ) dμ i

(EQ 7)

i=2

Recalling that dμi = RT d ln ai, where ai is the activity of species i in the bulk aqueous solution, R is the gas constant, and T is the temperature, Equation 7 can be rearranged to give ∂γ 1 Γ i( 1 ) = – -------- ⎛ ------------⎞ ⎝ RT ∂ ln a i⎠ T, μj

;i ≠ j

(EQ 8)

The adsorption density, Γ(1), will be positive if ( ∂γ ⁄ ∂ ln a i ) T, μ ;i ≠ j is negative, and j vice versa. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 98 Tuesday, January 2, 2007 7:36 PM

98

FLOTATION FUNDAMENTALS

The adsorption density can be calculated from the slope of the surface tension-versuslog activity (or concentration) curve if the surface tension of the particular interface under consideration can be determined experimentally. To illustrate the use of the Gibbs equation in flotation, first consider aqueous solutions of frothers, which are surface-active agents that have a tendency to concentrate at the air–water interface. The Gibbs equation, when applied to a dilute aqueous solution of a frother such as an alcohol (ROH) in water, has the following form (de Bruyn and Agar 1962): dγ = ( – RT )Γ ROH d ln a ROH

(EQ 9)

≈ ( – RT )Γ ROH d ln C ROH

The dependence of surface tension on the frother concentration and the adsorption density of the frother at the air–water interface are shown in Figure 1 for the water–butyl alcohol system. The lowering of the surface tension of water due to the addition of alcohol shows that the alcohol is positively adsorbed at the water–air interface. For example, at an activity of 0.712 (0.854 molality), the adsorption density of butyl alcohol is 6.03 × 10–10 mol/cm2. At this concentration, the area per molecule is 27.4 sq Å. Close-packed films of longchained carboxylic acids and alcohols exhibit areas per molecule of 21.6 sq Å, suggesting that the adsorbed butyl alcohol on water is monomolecular. For proper frothing, there must 80

Surface Tension, dynes/cm

72.8

60

Butanol/Water 20°C

40

A 20

120

6

Γ, mol/cm 2 × 10 10

100 80

4

60 2

40 20

Area per Adsorbed Molecule, sq. Å

140

B 0 –3

0 –2

–1

0

log a2

Source: de Bruyn and Agar 1962.

FIGURE 1 Adsorption of butanol at the aqueous solution/gas interface: (a) Surface tension and (b) adsorption density and area per adsorbed molecule versus log butanol activity

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 99 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

99

be a surface tension–concentration gradient such as that shown in Figure 1 (i.e., dγ/dC ≠ 0), so that momentarily deformed films can withstand the shock. The surface tension lowerings produced by the positive adsorption of a homologous series of n-aliphatic alcohols at the water–air interface are most conveniently compared by plotting the surface tension of the solution as a function of the logarithm of the activity (or concentration) of the alcohol (Defay and Prigogine 1966). In the concentration range corresponding to the lower parts of a series of such curves, the slope is independent of the alcohol considered, showing that the same number of molecules are adsorbed per unit area of the surface. Further, the concentration ratio of two neighboring homologs is constant and approximately equal to 3.2 for the same Δγ. In other words, the surface activity is approximately tripled for each additional –CH2– group in the molecule. This is called Traube’s rule. Collectors are heteropolar organic compounds whose main function is to adsorb at the mineral–water interface, but they also tend to adsorb at the mineral–air and water–air interfaces. They differ from the frothers in the sense that they are generally electrolytes. Taking into consideration the aqueous dodecyl sulfonate system, the change in the surface tension of the solution–air interface is given by dγ = – Γ R – dμ R – – Γ Na + dμ Na +

(EQ 10)

where R– refers to the dodecyl sulfonate anion (C12H25SO3–). In the absence of any other electrolytes where Na+ and R– would be the only species that are adsorbing, it can easily be shown that dγ = – 2RT Γ R – d ln C NaR

(EQ 11)

If the organic electrolyte is a weak electrolyte, such as the salt of a fatty acid, then depending on the pH of the solution, the adsorption of neutral collector molecules must also be considered. Probably the most marked lowering of surface tensions due to adsorption is that exhibited in solid–vapor systems. For example, water vapor adsorption can lower the surface tension of oxides by several hundred ergs per square centimeter. Adsorption phenomena at the bubble–mineral interface appear to have a significant role in flotation. T H E R M O DY N A M I C S O F W E T T I N G

Bubble–particle contact is one of the key factors controlling the process of froth flotation. In this section, the thermodynamic aspects of the bubble–particle contact will be reviewed and critically analyzed (Adamson 1967; Gaudin 1957). The general thermodynamic condition for three-phase contact is defined by Young’s equation for the system depicted schematically in Figure 2. γ SG = γ SL + γ LG cos θ

(EQ 12)

where γSG, γSL , and γLG are the tensions of the solid–gas, solid–liquid, and liquid–gas interfaces, respectively, and θ is the contact angle. The change in the free energy accompanying the replacement of a unit area of the solid–liquid interface by solid–gas interface is given by Dupre’s equation, namely: ΔG = γ SG – ( γ SL + γ LG ) © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 13)

flotation0.book Page 100 Tuesday, January 2, 2007 7:36 PM

100

FLOTATION FUNDAMENTALS

Liquid γLG Gas

θ

Solid

γSG

γSL

FIGURE 2 Schematic representation of the equilibrium contact between an air bubble and a solid immersed in a liquid. The contact angle θ is the angle between the liquid–gas and the liquid–solid interfaces, measured through the liquid.

Combination of Dupre’s equation with Young’s equation yields the following expression for the free energy change, namely: ΔG = γ LG ( cos θ – 1 )

(EQ 14)

Thus, for any finite value of the contact angle, there will be a free energy decrease upon attachment of a mineral particle to an air bubble. The theoretical relationship of Dupre’s equation expresses the maximum possible decrease in the free energy of the system resulting from the bubble–particle contact, which can be realized only when there are no other energy-consuming effects, such as deformation of the bubble. Thus, the geometry of the system is not taken into account in Dupre’s equation. Further, Young’s equation (Equation 12) is valid in an ideal system where all gravitational effects are absent and the system is at equilibrium ( Johnson 1959). Often, considerable hysteresis exists in measured contact angles because of surface roughness, contamination, nonequilibrium adsorption effects, etc. (Adamson 1967). If there is hysteresis, that is, if a liquid-advancing contact has a different value than a liquidreceding angle, equilibrium is not attained in the system. Under these conditions, the use of Young’s equation is not valid. Figure 3 presents equilibrium, advancing and receding contact angles on alumina in sodium dodecyl sulfonate solutions at pH 7.2 as measured by Wakamatsu and Fuerstenau (1973). As can be seen in this figure, a pronounced hysteresis effect exists in this system under the conditions of this investigation. Leja and Poling (1960) conducted an interesting theoretical study on the attachment of air bubbles to flat solids, both in the presence and absence of gravitational effects, and their results will be summarized at this stage. Typically, because the solid surface is not of the same contour as that of the air bubble, work must be expended by the system in deforming the air–liquid interface during attachment, and the actual free energy of adhesion per unit area, WP, is smaller than the theoretically available value, WA (equal to ΔG given by Equation 13). In the absence of other forms of energy (gravitational and kinetic), the work of deformation must be performed solely by the interfacial energy. The deformation is then governed by the shape of the solid surface and, for a perfectly flat surface, the magnitude of the contact angle, θ, determines the extent of deformation that the interfacial energy pool, WA , is capable of performing. Leja and Poling (1960) suggest that gravitational and kinetic energies (due to the motion of particles and air bubbles) affect the energy expended in deformation of the air– liquid interface, either decreasing or increasing the amount of deformation to be performed by the interfacial energy pool. So when part of the deformation work is performed by gravitational or kinetic energy, hysteresis of the practical contact angle is observed, even when © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 101 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

100

Contact Angle, degrees

80

S A

S A

S A

Liquid Receding

Equilibrium

Liquid Advancing

101

60

Advancing 40

Equilibrium

Alumina Ionic Strength 2 × 10–3N pH 7.2 24±1°C

20 Receding 0 10–6

10–5

10–4

10–3

Concentration of Dodecyl Sulfonate, mol/L

Source: Wakamatsu and Fuerstenau 1973.

FIGURE 3 Equilibrium, and receding and advancing contact angles on alumina in aqueous dodecyl sulfonate solutions at pH 7.2

other effects, such as surface roughness, surface contamination, and so forth, are excluded. According to Leja and Poling, an advancing contact angle indicates that a greater portion of WA is being converted to the actual adhesion, WP, than under the condition of equilibrium. On the other hand, a finite receding contact angle indicates a lower WP. Their conclusion was that the contact angles, which were determined experimentally with fairly large bubbles or drops, should not be used in Young’s equation unless a suitable correction in their magnitude is made to account for gravitational effects, a correction that becomes particularly significant at the threshold of hydrophobic character (i.e., with small contact angles). A limited amount of work has been carried out on the temperature dependence of contact angles. The temperature coefficient of the contact angle can provide thermodynamic information about wetting processes. For example, knowledge of the temperature coefficient of contact angle provides a means of calculating the heat of immersion, which is the enthalpy change upon immersing a clean, dry solid into a liquid. It can be shown (Adamson 1967) that d cos θ – ΔH imm = E LG cos θ – Tγ LG -------------dT

(EQ 15)

where ELG is the total surface energy of the liquid–gas interface. For an excellent account of the temperature dependence of contact angles on low-energy surfaces, refer to an article by Neumann (1974). T H I N F I L M P H E N O M E N A A N D F L O TAT I O N

The well-known classical condition for the possibility of a bubble–particle contact in a given liquid medium has been specified by Equation 13 where ΔG = γ SG – γ LG – γ SL < 0 © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 102 Tuesday, January 2, 2007 7:36 PM

102

FLOTATION FUNDAMENTALS

This relation assumes that there is no liquid layer between the bubble and the particle after contact. This equation ignores the possibility of the existence of a thin liquid layer between the bubble and the solid even after attachment. The thinning of this layer is believed to control flotation (Laskowski 1974). Laskowski has made an attempt to modify this equation for the presence of a liquid layer by making the following substitutions: 1. GS(h) for γSG where GS is the free energy of a column of liquid layer (of height h) and solid of unit cross section, and 2. GS(h) = γSL + γLG(h). γLG(h) is the surface tension of the liquid–gas interface expressed as a function of the thickness of a liquid layer h. These substitutions lead to ΔG ( h ) = γ LG ( h ) – γ LG

(EQ 16)

where ΔG(h) expresses the energy barrier in the transition from no contact to bubble contact, which must be overcome for attachment to occur (Laskowski 1974). Derjaguin (1932) suggested that the thermodynamic properties of thin films are different from those of the bulk phase and introduced the parameter “disjoining pressure,” Π, as a measure of the corresponding change in the thermodynamic properties. In effect, it is the change of free energy with thickness and is given by the expression ∞

γ = γ o + ∫ Π dh

(EQ 17)

h

where γ is the specific-surface free energy of the thin liquid film, γo is the specific-surface free energy of an infinitely thick liquid film, h is the thickness of the film, and Π is the disjoining pressure. If p is the vapor pressure in equilibrium with a flat thin film, and po is the vapor pressure in equilibrium with the bulk liquid, then Derjaguin and Shcherbakov (1961) showed p ∂G RT ln ----- = V m ⎛ -------⎞ ⎝ ∂h ⎠ A, T po

(EQ 18)

where Vm is the molar volume of the liquid and the term ∂G/∂h is the disjoining pressure, Π. If Π is positive, the thin film is stable. (For curved liquid–air interfaces, the above expression has been suitably modified by Padday [1970]). Depending on the liquid and the surface, Π can be positive or negative and can change sign with film thickness (Clifford 1975). For liquid–solid systems with a finite contact angle, Π must be negative for certain film thicknesses (Clifford 1975). A complete understanding of the mechanism of attachment of particles to air bubbles in flotation should be based on the analysis of changes in the surface free energies of thin films of water between the solid and the air bubbles. Rehbinder (1949) made some theoretical calculations to determine the excess free energy of thin films. In the thinning of the water layer on the approach of a solid surface to an air bubble, the excess free energy of the layer, according to Rehbinder, changes in relation to the initial hydration of the surface. When the surface is highly hydrated (threshold hydrophilicity), the free energy of the liquid film increases continuously during the thinning of the liquid film, thus preventing spontaneous attachment of the air bubble to the surface. When the hydration of the surface is low, the

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 103 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

103

free energy of the thin film does not increase on initial thinning of the film, thus favoring spontaneous attachment of a bubble to a mineral. The previous two cases are represented schematically in Figure 4. Curve 2 in this figure is most typical of practical flotation systems. In the case of strongly hydrophobic solids where water vapor adsorption is far lower than a monolayer, the film would certainly be unstable and would rupture spontaneously. The disjoining pressure can be due to many effects. It is usual to consider that these effects are additive (Sheludko 1967) and that the disjoining pressure can be broken down into a number of components (Padday 1970): Π = Π LL + Π SL + Π el + Π h + Π p + Π i

(EQ 19)

Free Energy of a Thin Hydrated Layer, γ

where ΠLL is the disjoining pressure of the liquid film in the absence of a solid, ΠSL is the disjoining pressure due to the effect of the solid on the liquid film, Πel is an electrostatic term to account for double-layer effects, Πh is the contribution from hydrogen bonding, ΠP is from polar interactions, and Πi is the change in free energy due to any other specific interactions between the liquid layer and the surface. Chander and Fuerstenau (1972) have made an attempt to explain the natural floatability of molybdenite using this approach. In 1960, Eigeles and Volova commented that “the experimental data on collector action suggest that there may exist forces which are not taken into account by the present-day theory of coagulation of hydrophobic colloidal sols but exert a major influence on the kinetics of flotation adhesion.” They added that “this unaccounted factor is what causes sharp acceleration of the thinning of the boundary layer and a pronounced increase of its fluidity.” This led to their conducting a detailed investigation of the effect of temperature on induction time, in which they observed a regular decrease in induction time with increasing temperature. From their experimental observations, they evaluated an apparent activation energy for film thinning and attachment processes. The additional force suggested for bubble–particle

1

2

3 Δγ

Distance from the Surface

Source: Rehbinder 1949.

FIGURE 4 Change in the free energy of a water film between a gas bubble and solid surfaces with differing hydrations: (1) maximum surface hydration, (2) moderate surface hydration, and (3) very weak surface hydration

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 104 Tuesday, January 2, 2007 7:36 PM

104

FLOTATION FUNDAMENTALS

attachment processes in flotation by Eigeles and Volova was first observed and measured by Israelachvili and Pashley in 1982 between mica surfaces coated with cetyltrimethylammonium bromide, using the surface force apparatus (Israelachvili and Pashley 1988). This force, which is caused by the disruption of normal hydrogen bonding in water near a hydrophobic surface, is now called the hydrophobic force and would be an additional component in Equation 19. Yoon (2000) has made extensive measurements of hydrophobic forces and applied the concept to bubble–particle interaction in flotation. T H E E N E R G E T I C S O F F L OATA B I L I T Y

Consider the conditions for floatability, namely, a finite contact angle of an air bubble on the solid to be floated, in terms of work of adhesion of water to the solid (Wad ) and the work of cohesion of the liquid (Wco). Because Wco = 2γLG (the surface tension of the liquid, which is usually only slightly smaller than that of pure water) and Wad = γSG + γLG – γSL , assuming that the contact angle is given by Young’s equation (cos θ = [γSG – γSL]/γLG), it can be shown that Wad – Wco = γLG (cos θ – 1). Hence, flotation is possible if Wad < Wco (≤144 dynes/cm). It should be remembered that Wad is correctly defined as the work required to separate liquid water from a solid–water interface, leaving behind an adsorbed layer in equilibrium with saturated water vapor (Mellgren et al. 1974). After detailed experimentation with methylated silica, Laskowski and Kitchener (1968) analyzed hydrophobicity (floatability) in terms of these relations and concluded that all solids would be hydrophobic if they did not carry polar or ionic groups on their surface. They also emphasized that hydrophobicity arises from the exceptionally large cohesive energy of water that is due to hydrogen bonding between water molecules. The magnitude of Wad depends on γSL which reflects the extent of interaction between the solid and water. Only a few solids are naturally hydrophobic and therefore respond to flotation without adding a collector. These solids include such materials as graphite, sulfur, talc, molybdenite, and stibnite (Gaudin 1957). Hydrophobic materials such as graphite, in principle, do not exhibit significant polar interactions with water, and the energetics of the solid–liquid interface are mainly controlled by dispersion forces. Less hydrophobic surfaces have some polar interactions with water, and hydrophilic solids such as oxides exhibit strong polar interactions with water and are covered with hydroxyl groups (e.g., silica). Because the function of the flotation collector is to render the surface more hydrophobic by eliminating or shielding the polar sites of the solid, investigation of the energetics of the solid–water interface, both in the presence and absence of a collector, should help one to comprehend the concomitant phenomena of flotation. Wad = (γLG(cos θ + 1)) can readily be obtained for hydrophobic solids given that θ > 0. But for hydrophilic solids, θ = 0 (Young’s equation is no longer valid under these conditions), and consequently, evaluation of Wad is not easy. However, two methods (Mellgren et al. 1974) for measuring solid–water interactions have been widely used: (1) adsorption of water vapor on solids and (2) calorimetric heats of immersion in liquid water. In the first method, the adsorption density of water vapor is determined as a function of water vapor pressure from the dry state up to (or close to) saturation. The free energy of saturation of the solid with water (ΔGsat) is given by the expression

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 105 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

105

Po

ΔG sat = RT ∫ Γ d ln p ⁄ p o

(EQ 20)

o

where Γ is the adsorption density (expressed in moles per area), p is the water vapor pressure, and po is the saturation pressure. With hydrophilic solids, the major contribution to ΔGsat comes from the most strongly bound water (i.e., at low p/po) and the net value is much greater than 144 ergs/cm2. Some reported values are –316 ergs/cm2 for quartz, –264 ergs/cm2 for calcite, and –240 ergs/cm2 for barite (Mellgren et al. 1974). The water-vapor adsorption method has also been used to study changes in the wettability of surfaces induced by the adsorption of collectors. Hall, Lovell, and Finkelstein (1970) determined the effect of adsorbed oleate on calcite and fluorite on the ΔGsat value. Their results gave a ΔGsat value of –300 ergs/cm2 for clean fluorite and –50 ergs/cm2 for oleate-covered fluorite. Heats of immersion can provide some information on the energetics for systems in cases of zero contact angles. On a unit area basis, the heat of immersion of a clean solid, ΔHimm, is given by ΔH simm = H SL – H S ≈ E SL – E S

(EQ 21)

The hydrophilic character of a surface is clearly revealed by a large exothermic heat of immersion, which indicates a strong interaction with water molecules. Specifically, calorimetric heat measurements suggest that the first layer of water on polar solids is very strongly bound. When approximately three monolayers of water vapor are preadsorbed on the solid before immersion, the heat value falls to about 120 ergs/cm2, which is the surface enthalpy of normal liquid water (Mellgren et al. 1974). Therefore, water molecules beyond the first few layers behave similar to bulk water. Griffiths (1973) analyzed the results of Wade and Hackerman (1964) on the effect of the outgasing temperature on the heat of immersion of alumina (Al2O3) in water and concluded that about one-half of the heat of immersion of alumina is due to the physically adsorbed water with the remainder being due to the reformation of surface hydroxyl groups lost during heat treatment. To gain more insight into the nature of physically adsorbed water layers, Griffiths studied the temperature coefficient of the heat of immersion of Al2O3 in water and found it to have a near-zero value. He concluded that the water structure outward from the surface does not gradually grade into that of bulk water. In his words, what exists “is an inner film of strongly bound water that discontinuously ends with near normal water structure (at least energetically) continuing on outward from the surface” (Griffiths 1973). Numerous heats of immersion measurements of minerals have been reported in the literature (Healy and Fuerstenau 1965; Zettlemoyer 1968), and Table 2 presents a brief summary of some typical values. Although values of ΔHimm are high for hydrophilic and low for hydrophobic surfaces, there is no well-defined “critical heat of immersion” for development of contact angles. Hence, the two approaches discussed in this section are still of limited value in flotation research. One study that relates the heat of immersion of oxides to the point of zero charge (which plays an important role in selective flotation and collector choice) is worth mentioning as a final point. Healy and Fuerstenau (1965) found a quantitative linear relation between the heat of immersion and the pH of the point of zero charge for a number of inorganic oxides. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 106 Tuesday, January 2, 2007 7:36 PM

106

TABLE 2

FLOTATION FUNDAMENTALS

Heats of immersion and contact angles of some materials in water* ΔHimm, ergs/cm2 –Δ 6 26 260–370 400–570 550 530 650–900

Solid Teflon Graphon Quartz SnO2 (cassiterite) TiO2 (rutile) Fe2O3 (hematite) Al2O3 (alumina)

Contact Angle, degrees 108 82 0 0 0 0 0

*The oxides were activated at 200°C.

C R I T E R I O N F O R F L O TAT I O N I N T E R M S O F C O L L E C T O R ADSORPTION

Starting with the work of de Bruyn, Overbeek, and Schuhmann (1954), there has been a growing interest in whether collector adsorption at the S–L, S–G, or L–G-interface is the main factor controlling flotation behavior. Consider a collector x that is present at all three interfaces. On the assumptions that the collector adsorption at each interface follows the Gibbs equation and the contact angle of an air bubble on the solid is given by Young’s equation, the following relations can be written: dγ LG = – RT Γ xLG d ln a x

(EQ 22a)

dγ SG = – RT Γ xSG d ln a x

(EQ 22b)

dγ SL = – RT Γ xSL d ln a x

(EQ 22c)

γ SG – γ SL = γ LG cos θ

(EQ 22d)

where Γxij represents the adsorption density of x at interface ij, ax is the activity of the collector in the bulk solution, γij is the tension of the interface ij, and θ is the contact angle of the air bubble on the solid. Assuming ax = Cx, which is the concentration x, by differentiating Equation 12 with respect to ln Cx at constant P and T and applying relations 22a–c, the following result is obtained (Smolders 1961). dθ γ LG sin θ --------------- = RT ( Γ xSG – Γ xSL – Γ xLG cos θ ) d ln C x

(EQ 23)

From Equation 23, the following results can be deduced: 1. If the angle of contact increases with the increase of solute concentration (dθ/dln ax is positive), ΓxSG must be greater than ΓxSL + ΓxLG cos θ. 2. In cases where the contact angle does not vary with concentration (dθ/dln ax = 0), ΓxSG must be equal to ΓxSL + ΓxLG cos θ. 3. If the contact angle decreases with increasing concentration (which may happen because of the reverse orientation of the second layer of ionic surfactants at the solid–liquid interface), ΓxSG must be less than ΓxSL + ΓxLG cos θ.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 107 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

107

As early as 1954, de Bruyn, Overbeek, and Schuhmann stressed the first result obtained from Equation 23, namely, that ΓxSG > ΓxSL (for θ < 90°). In an attempt to apply the analysis of de Bruyn, Overbeek, and Schuhmann to a simple flotation system, Aplan and de Bruyn (1963) undertook an investigation of the adsorption of hexyl mercaptan onto gold and the flotation of gold particles from an aqueous solution of hexyl mercaptan using nitrogen saturated with both water and mercaptan vapor as the gaseous phase. They found that excellent flotation occurs at 3% to 30% surface coverage (low ΓSL), substantially below an equilibrium concentration where an abrupt change in the adsorption density from solution takes place. In addition, they observed ΓSG to be greater than ΓSL in the flotation range. In 1968, Somasundaran, in an attempt to explain the effect of collector chain length on flotation at concentrations below those necessary for hemimicelle association of the hydrocarbon chains at the S–L interface, stressed the need for looking at adsorption conditions on the bubble and at the S–G interface, in addition to the S–L interface. From the results of his investigation of the quartz-dodecylamine solution–gas system, he found that the amount of collector adsorption at the S–G interface was approximately equal to the adsorption at the S–L interface. Sandvik and Digre (1968), without directly measuring ΓSG and ΓSL , showed that the adsorption of a collector on a solid surface (silica) increased in the presence of gas bubbles. They proposed a bubble-transfer hypothesis, which envisages the collector as being transferred from the bubble surface to the solid. Their hypothesis in fact stresses the importance of adsorption at the L–G interface. Pope and Sutton’s (1972) work did not confirm this hypothesis given that they observed a decrease in collector adsorption density at the solid after flotation. Finch and Smith (1972) found that the flotation recovery of quartz and magnetite (in the presence of dodecylamine) as a function of pH can be correlated with the variation in the surface pressure [γsolvent – γsolution], of the L–G interface with pH. They concluded that this correlation implies that the greater the adsorption of the collector at the bubble surface (higher the surface pressure), the better the flotation. In a later investigation on the quartz–dodecylammonium acetate and the magnetite–dodecylammonium acetate systems, Finch and Smith (1975) found a decrease in the tenacity of the bubble–solid attachment with decreasing surface tension of the liquid–air interface, thus casting a doubt on their earlier conclusion. Bleier, Goddard, and Kulkarni (1976) later reported good correlation between the flotation recovery of quartz in the presence of amines and the decrease in the surface tension of the water–air interface. It should be stressed that the interface of primary importance to flotation is the solid– liquid interface. The collector simply must adsorb at this interface in order to reduce molecular attractions between the solid and the liquid. T H E E L E C T R I C A L D O U B L E L AY E R AT M I N E R A L – WAT E R I N T E R FA C E S

Because adsorption phenomena at mineral–water interfaces are controlled in most cases by the electrical double layer, one must be concerned with factors responsible for the charge on the solid surface and with the behavior of ions that adsorb as counterions to maintain electroneutrality (Kruyt 1952). Figure 5 presents a schematic representation of the electrical double layer of counterions extending out into the liquid phase. This figure also shows the drop in potential across the double layer, neglecting the potential because of dipole effects. The closest distance of approach of counterions to the surface (δ) is the Stern plane. Depending on whether ions remain hydrated or are dehydrated upon adsorption, there can

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 108 Tuesday, January 2, 2007 7:36 PM

108

FLOTATION FUNDAMENTALS

Surface Charge, σo

Counterions, σd δ

ψo

Solid

Potential

ψδ

δ

Distance

FIGURE 5 Schematic representation of the electrical double layer and the potential drop across the double layer at a solid–water interface

exist an inner and an outer Stern plane (Grahame 1947); for purposes of this review, though, no distinction shall be made between the inner and outer Stern plane. The surface potential is ψo, at the Stern plane it is ψδ ; and from the Stern plane into the bulk of the solution, the potential drops to zero. In the case of the ionic solids such as barite (BaSO4), fluorite (CaF2), silver iodide (AgI), and silver sulfide (Ag2S), the surface charge arises from a preference of one of the lattice ions for sites at the solid surface as compared with the aqueous phase. Equilibrium is attained when the electrochemical potential ( μ = μ + vF φ where φ is the Galvani potential in the phase) of these ions is constant throughout the system. Those particular ions that are free to pass between both phases and therefore establish the electrical double layer are called potential-determining ions. In the case of AgI, the potential determining ions are Ag+ and I–. For a solid such as calcite, CaCO3, the potential-determining ions are Ca2+ and CO32–, and also H+, OH–, and HCO3– because of the equilibria between these latter ions and CO32–. Similarly for apatite, the potential-determining ions are Ca2+, PO43–, and OH–, with the other hydrolysis products also functioning in this role because of the complex equilibria involved in this system. For oxide minerals, which will be discussed in more detail in the next section, hydrogen and hydroxyl ions have long been considered to be potential-determining (Fuerstenau and Healy 1972). In the layer silicate minerals such as clays and micas, because of the substitution of Al3+ for Si4+ in the silica tetrahedra or Mg2+ for Al3+ in the octahedral layer of the crystal lattice, the surfaces of these crystal faces carry a negative charge that is independent of solution conditions. The surface charge, σs, on a solid in water is determined by the adsorption density of potential-determining ions on the solid surface. In the case of a 1–1 valent salt, σs is given by σS = F ( ΓM+ – ΓA– )

(EQ 24)

where F is the Faraday constant, ΓM+ is the adsorption density in moles per square centimeter of the potential-determining cation, and ΓA– is that of the potential-determining anion. For an oxide, M+ and A– can be considered as H and OH–, respectively, and for AgI, M+ and A– are simply Ag+ and I–. By means of a simple titration procedure (Kruyt 1952; Parks and de Bruyn 1962), the magnitude of the surface charge can be determined if the surface area of the solid–liquid interface is known. The single most important parameter describing the surface is the condition under which the surface charge, σs, is zero. The activity of potentialdetermining ions at which this occurs is called the point of zero charge, or PZC. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 109 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

TABLE 3

109

PZC for some ionic solids

Material Barite, BaSO4 Calcite, CaCO3 Fluorapatite, Ca5(PO4)3(F,OH) Fluorite, CaF2 Hydroxyapatite, Ca5(PO4)3(OH) Scheelite, CaWO4 Silver chloride, AgCl Silver iodide, AgI Silver sulfide Ag2S

PZC pBa 6.7 pH 9.5* pH 6* pCa 3 pH 7* pCa 4.8 pAg 4 pAg 5.6 pAg 10.2

Source: Fuerstenau 1971. *The activities of the other potential-determining ions can be calculated from the hydrolysis equilibria and solubility data.

Assuming that potential differences due to dipoles, and so forth, remain constant, the total double-layer potential or the surface potential, ψo, is considered to be zero at the PZC. The value of the surface potential at any activity of 1–1 valent potential-determining electrolyte is given by aM+ RT ψ o = -------- ln ---------------------F ( a M + ) PZC

(EQ 25)

where R is the gas constant, T is the temperature in degrees Kelvin, and ( a M + ) PZC is the activity of the potential-determining cation at the PZC. Table 3 presents PZC values for a number of ionic (salt-type) solids that have been investigated. Calcite, fluorite, and barite are positively charged in their saturated solution at neutral pH, whereas the other materials listed are negative, except for hydroxyapatite, which is uncharged at the pH shown. The importance of the PZC is that the sign of the surface charge has a major effect on the adsorption of all other ions and particularly those ions charged oppositely to the surface because these ions function as the counterions to maintain electroneutrality. In contrast to the situation in which the potential-determining ions are special for each system, any ions present in the solution can function as the counterions. If the counterions are adsorbed only by electrostatic attraction, they are called indifferent electrolytes. As has been well established (Kruyt 1952), the counterions occur in a diffuse layer that extends from the surface into the bulk solution. The “thickness” of the diffuse double layer is 1/κ, where κ is given by κ = ( 8πF 2 v 2 C ⁄ εRT ) 1 ⁄ 2

(EQ 26)

where v is the valence of the ions (for a symmetrical electrolyte), and ε is the dielectric constant of the liquid. For a 1–1 valent electrolyte, for example, 1/κ is 1,000 Å in 10–5 M, 100 Å in 10–3 M, 10 Å in 10–1 M solutions. The charge in the diffuse double layer σd given by the Gouy–Chapman relation (Kruyt 1952) as modified by Stern (for a symmetrical electrolyte) is σ d = – σ S = – [ ( 2εRT ⁄ π )C ] 1 ⁄ 2 sinh ( vFψ δ ⁄ RT )

(EQ 27)

Further, if the potential does not change appreciably, this equation shows that the adsorption density of counterions should vary as the square root of the concentration of

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 110 Tuesday, January 2, 2007 7:36 PM

110

FLOTATION FUNDAMENTALS

added electrolyte, as de Bruyn (1955) has found for the adsorption of dodecylammonium acetate on quartz at low concentrations. On the other hand, some ions exhibit surface activity in addition to electrostatic attraction and adsorb strongly in the Stern plane because of such phenomena as covalent bond formation, hydrophobic bonding, hydrogen bonding, solvation effects, and so forth. Because of their surface activity, the charge of such surface-active counterions adsorbed in the Stern plane can exceed the surface charge. In this case σS = –( σδ + σd )

(EQ 28)

where σδ is the charge due to adsorption in the Stern plane, δ. Flotation collectors generally function as surface-active counterions. Electrokinetic phenomena, which involve the interrelation between mechanical and electrical effects at a moving interface, have found widespread use in colloid and surface chemistry. The two electrokinetic effects that have been most widely used are electrophoresis and streaming potential measurements. Electrokinetic results are generally expressed in terms of the ζ-potential, which is the potential at the slipping plane when liquid is forced to move relative to the solid; only those ions in the diffuse layer outside of the slipping plane are involved in the electrokinetic process. Thus, while knowledge of the ζ-potential at some single condition may be of certain value, determination of the change in ζ-potentials as solution conditions are varied is extremely useful. From these changes, modes of adsorption of various kinds of ions can be ascertained if one makes the useful assumption that the slipping plane and the Stern plane coincide (Kruyt 1952). This approximation seems permissible because the potential differences between the plane δ and the slipping plane are small compared with the total potential differences across the double layer. It should be pointed out that the case in which there is no ambiguity is when ψδ = 0, because ζ must be zero. T H E E L E C T R I C A L D O U B L E L AY E R O N OX I D E M I N E R A L S

Because oxides constitute the most important class of nonmetallic minerals, they will be given considerable detail in this section. Given that oxide minerals form hydroxylated surfaces when in contact with water vapor, a hydroxylated surface should be expected when the solid is in equilibrium with an aqueous solution. Adsorption-dissociation of H+ from the surface hydroxyls can account for the surface charge on the oxide (Yopps and Fuerstenau 1964; Zettlemoyer 1968): MOH ( surf ) ↔ MO (–surf ) + H (+aq )

(EQ 29)

MOH ( surf ) + H (+aq ) ↔ MOH 2+( surf )

(EQ 30)

Parks and de Bruyn (Parks and de Bruyn 1962; Parks 1967) have postulated a different mechanism for the charging of oxide surfaces, involving partial dissolution of the oxide and formation of hydroxyl complexes in solution, followed by adsorption of these complexes: M 2 O 3 ( solid ) + 3H 2 O ↔ 2M ( OH ) 3 ( aq )

(EQ 31)

3 – m + ( 3 – m )OH – M ( OH ) 3 ( aq ) ↔ M ( OH ) m ( aq ) ( aq )

(EQ 32)

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 111 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

111

3 – m ↔ M ( OH ) 3 – m M ( OH ) m ( aq ) m ( surf )

(EQ 33)

The formation of a surface charge by either of these mechanisms, or even by direct adsorption of H+ and OH–, would result in an equivalent change in the pH of the solution. Titration of a suspension of an oxide in water at high ionic strength (with an indifferent electrolyte such as sodium chloride, or NaCl) can yield the surface charge by assuming that all added H+ or OH– change the solution pH or are adsorbed at the surface. Under these conditions, given that the counterions are the indifferent ions, σ S ≡ F ( Γ H + – Γ ( OH ) – )

(EQ 34)

Figure 6 presents the results of such titration of synthetic ferric oxide (hematite) with hydrogen and hydroxyl ions in the presence of nitrate (KNO3) as supporting electrolyte (Parks and de Bruyn 1962). This figure clearly shows that the surface charge on ferric oxide reverses its sign at pH 8.6 and that it increases in absolute magnitude with increasing ionic strength and increasing concentration of potential-determining ion. The intersection of the curves yields the PZC because the adsorption density is zero at this point. Interestingly, the adsorption isotherms given in Figure 6 are linear with pH at high ionic strengths and can be represented by an equation of the following form: ( Γ H + – Γ ( OH ) – ) = – F ( pH – pH PZC )

(EQ 35)

where pHPZC is the pH of the solution at the PZC of ferric oxide. (This results as a consequence of the capacitance of the Stern layer being constant, but this will not be discussed further here.) For an oxide, the surface potential would ideally be given by ( aH+ ) RT ψ o = -------- ln ---------------------- = 0.059 ( pH PZC – pH ) F ( a H – ) PZC

volts

(EQ 36)

The potential drop across the double layer at the alumina–water interface is shown schematically in Figure 7. As the ionic strength is increased, the double layer is reduced in thickness, and the potential at the Stern plane (δ) is reduced. Table 4 presents a tabulation of typical PZC values of several oxides. This table shows that the surfaces of oxides range from being acidic in nature to quite basic. Examples of how pH strongly affects adsorption at the surface of oxides will be shown later. At this point, consider the Gibbs equation for oxide minerals in aqueous media in the presence of an electrolyte, such as NaC1. The change in tension of the mineral–water interface due to the adsorption of H+, OH–, Na+, and Cl– is given by dγ = – Γ H + dμ H + – Γ OH – dμ OH – – Γ Na + dμ Na + – Γ Cl – dμ Cl –

(EQ 37)

At constant ionic strength, this equation simplifies to dγ = – ( Γ H + – Γ ( OH ) – )dμ H + = 2.3RT ( Γ H + – Γ OH – ) ⋅ d ( pH )

(EQ 38)

Thus, it can be seen that the surface tension of the mineral–water interface can be controlled by changing the pH at constant ionic strength. Figure 7 also shows schematically the

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 112 Tuesday, January 2, 2007 7:36 PM

112

FLOTATION FUNDAMENTALS

–40

1M

Ferric Oxide KNO3 21°C

10–1

10–2

Adsorption Density (ΓH+ – ΓOH–), μmol/g

–20

10–3 10–4

0

+20

+40 PZC

+60 5

6

7

8

9

10

11

pH

Source: Parks and de Bruyn 1962.

FIGURE 6 Adsorption density of potential-determining ions on ferric oxide as a function of pH and ionic strength using KNO3 as the indifferent electrolyte δ

+59

+118

10–1 10–3 M

Potential, mV

Potential, mV

+118

0 Distance –59

+59 0 –59

Distance 10–1 10–2 M

–118

–118 pH 7

pH 11

0

10–3 M 10–2

Δγ

10–1

7

9

11

pH

FIGURE 7 Electrical double layer at the alumina–water interface, showing the potential distance curves for 10–3 and 10–1 molar indifferent electrolyte at pH 7 and pH 11 (the PZC occurs at pH 9). The change in the solid–liquid interfacial tension with pH and ionic strength is also shown schematically.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 113 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

TABLE 4

113

PZC of some oxides*

Material SiO2, silica gel SiO2, α-quartz SnO2, cassiterite ZrO2, zirconia TiO2, rutile Fe2O3, hematite (natural) Fe2O3, hematite (synthetic) FeOOH, goethite Al2O3, corundum AlOOH, boehmite MgO, magnesia

PZC, pH 1–2 2–3 4.5 4 5.8–6.7 4.8–6.7 8.6 6.8 9.1 7.8–8.8 12

Source: Fuerstenau 1970. *These are typical results. The sources of oxide—its trace impurities, method of pretreatment, etc.—cause variations in observed values.

change in interfacial tension of alumina with pH for several salt concentrations. The maximum in the interfacial tension occurs at the PZC (i.e., at pH 9). In view of this discussion on surface charge, the Gibbs adsorption equation for an oxide can be rewritten as dγ = – σ S dψ o – Γ Na + dμ Na + – Γ Cl – dμ Cl –

(EQ 39)

or at constant ionic strength as dγ = – σ S dψ o

(EQ 40)

The greater effect of increased ionic strength on the interfacial tension lowering (Figure 7) simply results from the greater magnitude of σs. Effect of Temperature on the Electrical Double Layer

In order to understand temperature effects in the adsorption of ionic collectors, the effects that temperature has on the double layer at interfaces must be considered. Experimentally, the PZC of oxides has been found to decrease with increasing temperature (Parks 1960; Lai 1970; Tewari and McLean 1972). An example of this is shown in Figure 8, which presents electrophoretic mobility data for alumina and magnesia at three temperatures. The overall reaction and its equilibrium constant, K, for the charging of an oxide surface given by Equations 29 and 30 can be written as MO (–surf ) + 2H (+aq ) ↔ MOH 2+( surf )

(EQ 41)

( a MOH + ) 2 K = -------------------------------( a MO – ) ( a H + ) 2

(EQ 42)

From thermodynamics, it is known that ΔG o = ΔH o – T ΔS o = – RT ln K

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 43)

flotation0.book Page 114 Tuesday, January 2, 2007 7:36 PM

114

FLOTATION FUNDAMENTALS

or ( a MOH + ) H o ΔS o 2 - = Δ – ln ------------------------------------------ – -------2 RT R ( a MO – ) ( a H + )

(EQ 44)

Given that the number of positive and negative sites must be equal at the PZC, it can be seen that Δ H o ΔS o ln ( a H + ) = ----------- – -------2RT 2R

(EQ 45)

– ΔH o ΔS o ( pH ) PZC = -------------- + ----------4.6RT 4.6R

(EQ 46)

Assuming that the enthalpy does not vary over the temperature range considered, the enthalpy and entropy for the protonization of the surface sites on oxides can then be evaluated from the shift in the PZC with temperature. Table 5 gives the results of this calculation for several oxides (Lai 1970). These calculations show that the enthalpy for protonization of surface sites varies with the nature of the metal oxide, but the entropy is approximately constant. Because the process is exothermic, the PZC of oxides decreases with increasing temperature. +6 MgO 25°C 15°C 5°C

Electrophoretic Mobility, μm/sec per V/cm

+4

+2

0

–2

Al2O3 45°C 25°C 10°C Calculated

–4

–6 4

6

8

10

12

14

pH

Source: Lai 1970.

FIGURE 8 Electrophoretic mobilities of alumina and magnesia at different temperatures as a function of pH. The pH at which the mobility reverses sign under these conditions is the PZC.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 115 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

115

TABLE 5 Enthalpy and entropy of hydrogen ion adsorption on oxides, expressed for the adsorption/desorption of a single H+ Oxide SiO2, cassiterite TiO2, rutile Al2O3, alumina MgO, magnesia

ΔH° kJ/mol H+ –0.65 –6.05 –10.16 –13.59

(pH)PZC at 25°C 2.2 6.3 9.1 11.8

ΔS° entropy units/mol H+ 7.9 8.7 7.8 8.7

Source: Lai 1970.

–6 5° 25° Surface Charge, δs, μC/cm 2

–4

45° 65° 85°

–2

0

+2 85° Silver Iodide 0.1 M Ionic Strength

5° +4

+200

+100

0

–100

–200

–300

-400

Surface Potential, Ψ0, mV

Source: Lyklema 1966.

FIGURE 9 Temperature dependence of the surface charge on AgI particles, showing the decrease in surface charge with increasing temperature. All curves are normalized to the PZC at reference temperature.

In addition to shifting the PZC, temperature also affects the charge on mineral surfaces. In general, raising the temperature results in a decrease of σs. Figure 9 presents Lyklema’s (1966) results for the charge on silver iodide at temperatures ranging from 5°C to 85°C. The curves are all normalized to the PZC at the temperature of reference. Lyklema suggests that the decrease in surface charge with increasing temperature might be due to a gradual desorption of counterions from the double layer, with the double layer having a more diffuse character. Because the surface charge also is given by the expression: ε dψ σ S = – ------ ⎛ -------⎞ 4π ⎝ dx ⎠ x = 0

(EQ 47)

where ε is the dielectric constant of the medium, the decrease in ε with increasing temperature would account for some of the decrease. In the case of the double layer on oxides, based on Equation 47 one can calculate the charge at various temperatures and pH’s, and such calculations show that the charge decreases with increasing temperature.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 116 Tuesday, January 2, 2007 7:36 PM

116

FLOTATION FUNDAMENTALS

F R E E E N E R G Y O F C O L L E C TO R A D S O R P T I O N

Information on the thermodynamic aspects of collector adsorption can be obtained from experimentally measured adsorption isotherms. Analysis of adsorption isotherms to obtain thermodynamic data for collector adsorption has been completed using two different approaches. Fuerstenau, Harper, and Miller (1970) have successfully used the Stern–Grahame model of the electrical double layer to explain the adsorption of alkylammonium ions at the quartz–water interface and alkyl sulfonates at the alumina–water interface. These collector ions are considered to be adsorbed as counterions in the double layer. Cases (1970), on the other hand, has interpreted the results of his adsorption studies on biotite in terms of the theoretical isotherms established by Frumkin, Fowler, Hill, and Halsey. Both approaches will be briefly discussed in the next few paragraphs. First, equilibrium in heterogeneous systems is attained when the chemical potential of species i is equal in all phases. For collector species i, its chemical potential in bulk solution is given by μ i = μ io + RT ln a i

(EQ 48)

where μio is its standard chemical potential and ai is its activity in solution. The chemical potential of the same species at the surface, μiS, is μ iS = ( μ io ) S + RT ln a iS

(EQ 49)

where (μio)S is the standard chemical potential of this species at the surface and aiS is its activity in the surface. At equilibrium, given that μi = μiS, μ io – ( μ io ) S a iS ⁄ a i = exp ------------------------RT

(EQ 50)

This relation can be transformed into the well-known Stern–Grahame equation by making the following assumptions: a i = C(the bulk concentration) and a iS = Γ δ ⁄ 2r

(EQ 51)

where Γδ is the adsorption density in the Stern plane, and r is the effective radius of the adsorbed ion. The standard free energy of adsorption, ΔGadso, is defined as o ΔG ads = ( μ io ) S – μ io

(EQ 52)

Substituting these three relations (from Equations 51 and 52) into Equation 50 yields the Stern–Grahame equation (Grahame 1947; Fuerstenau 1970) o ⁄ RT ) Γ δ = 2rC exp ( – ΔG ads

(EQ 53)

Another approach to developing the Stern–Grahame equation can be made by applying the Law of Mass Action to a binary mixture of similarly sized molecules showing ideal behavior in both the liquid phase and in the adsorbed layer. This leads to x 1S x 2 ⁄ x 2S x 1 = K

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 54)

flotation0.book Page 117 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

117

where x1 and x2 refer to the mole fraction of solute and solvent, respectively, in the liquid phase, and x1S and x2S are mole fractions in the adsorbed state. Given that x2S = 1 – x1S, Equation 54 can be rewritten as o – ΔG ads x x x kS ⎞ ------------- = K ----1 = ----1 exp ⎛ ----------------⎝ ⎠ S RT x x 1 – x1 2 2

(EQ 55)

At small values of x1, Equation 55 becomes essentially identical to the usual form of the Stern–Grahame equation (and to the Langmuir equation). The Stern–Grahame approach is useful in that it allows one to take electrical effects into account when dealing with ionic collectors. If ions adsorb only through electrostatic interactions, then the standard free energy of adsorption is given by o o ΔG ads = ΔG elec = vFψ δ

(EQ 56)

where v is the valence of the adsorbed ion including sign. On oxide minerals, alkali cations appear to be surface-inactive, that is, only electrostatic interactions appear to be operative. Nitrate anions are surface-inactive on cassiterite (SnO2), rutile (TiO2), Al2O3, and hematite (Fe2O3). Although chloride ions are not surface-active on Al2O3, they appear to be specifically adsorbed on Fe2O3 (Fuerstenau 1970). When an ion exhibits surface affinity, it is considered to be specifically adsorbed, and the free energy of adsorption has additional terms: o o ΔG ads = vFψ δ + ΔG spec

(EQ 57)

o where ΔG spec represents specific interaction terms. Certain inorganic ions exhibit surface activity: Ba2+ on quartz, Ba2+ and SO4= on alumina, Ca2+ and SO4= on rutile, and hydrolyzed metal cations on a wide variety of solids (Fuerstenau 1970). Equation 57 shows that if o ΔG spec is finite, then the ion will be positively adsorbed even if ψδ is zero or has the same o sign as the adsorbing ion. An estimation of ΔG spec can be made for conditions when ψδ is zero (e.g., when the electrophoretic mobility of the particles is zero). For the adsorption of organic collectors of interest to flotation scientists, various o as follows: attempts have been made to split ΔG ads o o o o o ΔG ads = ΔG elec + ΔG chem + ΔG CH + ΔG solv +… 2

(EQ 58)

o o where ΔG elec is the electrostatic contribution to the total free energy, ΔG chem represents o the free energy due to the formation of covalent bonds with the surface, ΔG CH represents 2 the interaction due to association of hydrocarbon chain of adsorbed collector ions at the o interface (sometimes called hydrophobic bonding), and ΔG solv is the contribution of solvation effects on the polar head of the adsorbate (collector) and adsorbent (mineral) to adsorption. o It is often customary to lump the ΔG o terms other than ΔG elec together and call it o o o o ). o ΔG spec = ΔG chem + ΔG CH + Δ G (i.e., ΔG spec solv 2 Specific adsorption can be either physical or chemical in nature. If the ions are adsorbed only through such forces as electrostatic attraction and through hydrophobic bonding (van der Waals interaction between hydrocarbon chains), the process should be termed physical adsorption or physisorption. If the collector forms a covalent bond with metal ions in the surface of the mineral, then the process should be termed chemisorption. Much of the work that

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 118 Tuesday, January 2, 2007 7:36 PM

118

FLOTATION FUNDAMENTALS

has been carried out with collector adsorption at interfaces has been concerned with physical adsorption, such as in the quartz–alkylammonium (de Bruyn 1955; Gaudin and Fuerstenau 1955; Somasundaran, Healy, and Fuerstenau 1964) and the alumina–alkyl sulfonate systems (Somasundaran and Fuerstenau 1966; Wakamatsu and Fuerstenau 1968). Adsorption isotherms and electrokinetic measurements strongly suggest that at low concentrations, these collector ions are adsorbed as individual ions in the Stern plane but that at higher concentrations, they associate at the interface into two-dimensional aggregates, which have been termed hemimicelles (Gaudin and Fuerstenau 1955). Adsorption in the alumina–alkyl sulfonate system has been studied in considerable detail. Figure 10 presents the effect of the concentration of dodecyl sulfonate on the electrophoretic mobility (zeta potential), adsorption density, and contact angle on alumina at pH 7.2 and at an ionic strength of 2 × 10–3 M controlled by NaCl. From this figure, it can be seen that the adsorption isotherm can be divided into three distinct regions. At low concentrations, adsorption of sulfonate ions occurs by exchange with chloride ions in the double layer; during the exchange, the zeta potential (and ψδ) remains constant. In this region, only electrostatic adsorption potential is active. In the second region, the adsorbed ions begin to associate, with adsorption increasing markedly because of the enhanced adsorption potential as o ΔG CH becomes effective. The third region is reached when the zeta potential reverses. At 2 concentrations higher than this, the electrostatic interaction opposes the specific adsorption effects, resulting in a decrease in the slope of the adsorption isotherm. Using data published o for the adsorption by Somasundaran and Fuerstenau (1972), Dick (1972) evaluated ΔG ads of sodium dodecyl sulfonate on alumina at pH 6.9 and his results are plotted in Figure 11. The onset of hemimicelle formation is accompanied by the standard free energy of adsorpo again has tion becoming sharply more negative. When the zeta potential reverses, ΔG ads much less dependence on the bulk concentration of sulfonate. By means of the Stern–Grahame model of the double layer, the contribution of the cohesive energy per mole of CH2 groups to the adsorption potential can be quantitatively evaluated. If the standard free energy for removing 1 mol of CH2 groups from water is through association φ, then the total contribution is nφ if n is the number of CH2 groups in the chain. Thus, the contribution from hydrocarbon chain association to the adsorption process is o ΔG CH = nφ 2

(EQ 59)

The adsorption density, Γδ, of collector ions in the Stern plane in the absence of chemisorption will be given by o ⁄ RT ) = 2rC exp [ ( – vFψ – n φ ) ⁄ RT ] Γ δ = 2rC exp ( – ΔG ads δ

(EQ 60)

From the results of Wakamatsu and Fuerstenau (1968), φ has been evaluated to be about –1.0 RT (about 0.6 kcal/mol of CH2 groups) in agreement with values obtained from solubility data and micelle formation. Because of the role that CH2 groups have in controlling the adsorption free energy in these systems, collector chain length will significantly affect flotation behavior (Wakamatsu and Fuerstenau 1973). Lin and Somasundaran (1971) have considered the transfer of aqueous surfactants to various types of interfacial states and found that transfer energies can range from –0.6 RT (for micellization) to –2.0 RT (for evaporation).

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 119 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

Cosine θ Adsorption Density Electrophoresis

III 0.2

–40

II

–30 0.4

0.6 10–11

–20

Cosine θ

Adsorption Density, mol/cm 2

10–10

–10

0

I +10

Zeta Potential, mV

10–9

119

0.8

Point of Zeta Reversal

+20

10–12

+30

1.0

Alumina 2 × 10–3 N Ionic Strength pH 7.2 24±1°C

+40

+50

10–13 10–5

10–4

10–3

10–2

Equilibrium Concentration of Sodium Dodecyl Sulfonate, mol/L

Source: Wakamatsu and Fuerstenau 1973.

FIGURE 10 Adsorption density of sodium dodecyl sulfonate, the electrophoretic mobility, and the contact angle of alumina as a function of the equilibrium concentration of sodium dodecyl sulfonate at pH 7.2 and ionic strength 2 × 10–3 M controlled with NaCl –24

–22

45°C 25°C

ζ=0

ΔG°ads, kJ/mol

–20

–18

–16 Alumina 2 × 10–3 N Ionic Strength pH 6.9

–14

–12 10–5

10–4

10–3

Equilibrium Concentration of Sodium Dodecyl Sulfontate, mol/L

Source: Dick 1972.

FIGURE 11 Variation of the standard free energy of adsorption of sodium dodecyl sulfonate on alumina at 25°C and 45°C for the isotherms given in Figure 13

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 120 Tuesday, January 2, 2007 7:36 PM

120

FLOTATION FUNDAMENTALS

In many mineral–collector systems, chemisorption occurs. Chemisorption is often arbitrarily referred to as an adsorption process in which the adsorbate attaches to the surface of the adsorbent with a molar free energy of approximately 42 kJ or greater. In terms of the o concepts of Equations 57 and 58, chemisorption occurs when ΔG chem has some finite value. In flotation systems, chemisorption is of primary interest because selectivity may be obtained if there is a specific collector–mineral adsorption reaction rendering a single mineral or group of minerals to be air avid. In nonmetallic flotation systems, several examples of chemisorption may be cited. Hexanethiol chemisorbs on the surface of zincite (ZnO) and willemite (Zn2SiO4), forming a strong zinc–mercaptan bond. Oleate chemisorbs on fluorite (CaF2), forming strongly adsorbed films that are difficult to remove, and on hematite and numerous other minerals. Hydroxamates strongly chemisorb on chrysocolla and hematite. A different approach has been taken by Cases and his associates, who consider the adsorption process as a condensation process on either a homogeneous or a nonhomogeneous surface similar to that of gas adsorption. The general relation for adsorption isotherms when the adsorption is localized without the dissociation of adsorbed molecules onto a homogeneous surface is (Cases 1970; Predali and Cases 1974) f -+A kT ln x = – k ln W a – φ a + kT ln --------1–f

(EQ 61)

where x is the mole fraction of adsorbate in the bulk, k ln Wa is the sum of all entropic terms except the configurational entropy for a molecule in the adsorbed state (f is the fractional surface coverage), φa is the differential energy of desorption per molecule, k ln f/ (1 – f ) is the configurational entropy of an adsorbed molecule, k is the Boltzmann constant, T is the temperature, and A is a constant. On assuming φa to be independent of surface coverage, Equation 61 can be rearranged to give the familiar Langmuir isotherm: x f = ------------------------------------G 1 x + ------ exp ⎛ ------a-⎞ ⎝ kT⎠ A2

(EQ 62)

where A2 = exp (–A/kT), and Ga is the free energy of adsorption per molecule = (–φa – kTlnWa). Langmuir isotherms have been widely used for estimating the free energy of adsorption from adsorption isotherms. If φa is assumed to be a function of surface coverage, f, one can obtain the well-known Frumkin–Fowler isotherm. To derive an equation for thermodynamic equilibrium, Predali and Cases (1974) assumed (1) φa = φao + fω, where φao is the normal binding energy between one adsorbed molecule and the surface, and fω represents the lateral interactions between the adsorbed species in the adsorbed layer; (2) in the adsorbed layer, the number of coordinations and the lateral interactions are the same as in the plane of the lattice of the micelle, which leads to the expression φo = φoo + ω/2, where φo is the energy to dissolve a molecule from half-crystal position in the micelle and φoo is the half of normal binding energies in micelle per molecule from half-crystal position; (3) the entropic functions are the same in the adsorbed layer and in the micelle; and (4) k T ln xo = –k T ln Wo – φo + A where k ln Wo is the sum of all entropic terms for a molecule in the half-crystal position, and xo is the mole fraction of the adsorbate (collector) in solution at saturation (i.e., in equilibrium with micelles of collector). Predali and Cases obtained the following relation for thermodynamic equilibrium between an adsorbed layer and a dilute solution of the same substance: © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 121 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

121

f ω kT ln ---------- = ( φ ao – φ oo ) – ---- ( 1 – 2f ) + Δμ 1–f 2

(EQ 63)

where Δμ = k T ln x/xo and is a measure of undersaturation of collector in the aqueous solution. A plot of f versus Δμ for two cases, namely, ω < 4 kT (e.g., C8 alkylammonium chlorides) and ω > 4 kT (e.g., C12 surfactants) is given in Figure 12a. From the figure, it can be noted that when ω < 4 kT, the slope of the isotherms on homogeneous surfaces at f = 0.5 is finite. It is possible to calculate the lateral binding energy between hydrocarbon chains from the value of the slope at f = 0.5. On the other hand, when ω > 4 kT, the slope of the isotherms is infinite at f = 0.5. Predali and Cases (1974) consider the portion of the curve between M and N in Figure 12a to be related to a change of state because of the condensation of the layer. Cases (1970) studied the adsorption of dodecylammonium chloride on homogeneous surfaces and found the isotherm to exhibit a constant slope up to f = 0.015, followed by an infinite slope up to f = 0.985. This corresponds to a case where ω > 4 kT. Cases suitably modified Equation 63 for nonhomogeneous surfaces, and this modified equation predicts a constant slope for adsorption isotherms (f versus C plot) up to a value of f close to 1. Predali and Cases (1974) studied the adsorption of long chain (> C10) alkylammonium chlorides on biotite and found the adsorption isotherms to exhibit two welldefined branches. The first has a finite slope, and, according to their developed theory, this is due to the condensation on a nonhomogeneous surface. The second branch has an infinite slope, which is considered to be due to condensation on a homogeneous surface—that is, onto the first adsorbed layer (see Figure 12b). Though the theoretical isotherm of Frumkin, Fowler, Hill, and Halsey fits the experimental data, as claimed by Cases (1970), there is one serious assumption underlying this approach, that is, the entropic effects involved in the adsorption from solution are similar and equal to that involved in the adsorption from gaseous state. This assumption is highly questionable because water structure at a solid–liquid interface plays a key role in adsorption, and any water molecules released on adsorption of the collector would significantly increase the entropy of the system. Furthermore, the utility of the Cases approach is somewhat A

B Biotite–Alkylammonium Chlorides

Δμ* N II ω < 4 kT f

0.5

I ω > 4 kT

M

Fractional Surface Coverage, f

1.0

2.0 C18

C16

C14 C12

1.5

1.0

C10

0.5

0.0 Δμ

0.0 10–6

10–5

10–4

10–3

Equilibrium Concentration, mol/L

Source: Predali and Cases 1974.

FIGURE 12 (a) Schematic Frumkin–Fowler isotherm on a homogeneous surface, and (b) adsorption isotherms of alkylammonium chlorides of different chain lengths on biotite at pH 5.5 and 25°C

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 122 Tuesday, January 2, 2007 7:36 PM

122

FLOTATION FUNDAMENTALS

limited in that the introduction of electrical double-layer concepts to adsorption is a more convenient method of dealing with the interaction of charged collectors with the mineral. Finally, plotting results in the form of fractional surface coverage versus concentration masks the very important effects encountered at low surface coverages. Heat of Adsorption of Collectors

Collectors of interest in flotation can be classified into two types: (1) surfactants that chemisorb or chemically react at the surface (e.g., hydroxamic acids on hematite or chrysocolla, oleate on fluorite or calcite, xanthates on sulfide), and (2) long-chain ionic surfactants that adsorb physically as counterions in the electrical double layer. The second class of collectors is characterized by strong hydrocarbon chain–hydrocarbon chain interaction in the Stern layer (e.g., dodecyl sulfonate on alumina or hematite, dodecylammonium chloride on quartz). The interaction of long-chain carboxylates with oxides might be considered as an intermediate category. Measurements of the heat of adsorption of collectors may be able to shed some light on the following aspects of the collection process: 1. Nature of the interaction with the substrate (For example, by analogy with the heats of chemisorption and physisorption of gases, it may be expected that type 1 reactions will be more exothermic than type 2.) 2. Identification of the mode of attachment of flotation collectors Pioneering work on calorimetric determination of the heat of adsorption of collectors on sulfides has been conducted at the Royal School of Mines, London. Mellgren (1966) investigated the heat of xanthate adsorption on galena samples “as ground” or previously treated with potassium carbonate, sulfate, or thiosulfate solutions. The heat of ethyl xanthate adsorption on untreated galena (–83 kJ/mol of Pb2+) was equivalent to that obtained when xanthate was reacted with lead thiosulfate at neutral pH values. This was as expected because thiosulfate was found to be the oxidation product on the galena used. Mellgren measured rather erratic and large ΔH values at high pH values and concluded that secondary reactions were involved at these high pH values. Heat of adsorption measurements on samples treated with carbonate, sulfate, and thiosulfate gave values equivalent to those obtained when ethyl xanthate was reacted with lead carbonate, lead sulfate, or lead thiosulfate. These findings suggested that metathetical reactions between the xanthate and the oxidation products on the surface of galena were involved. The quantities of xanthate adsorbed on the carbonate- and sulfate-treated galena surfaces were equivalent to the quantities of thiosulfate released into the solution after the treatment. This led to the conclusion that an ion exchange mechanism might be involved in the adsorption process. Mellgren and Rao (1968) performed similar experiments with potassium diethyldithiocarbamate instead of xanthate and obtained similar results. Gochin (1972) carried out an elaborate thermochemical study of the flotation of sphalerite and pyrite along similar lines. In addition, he studied the heat of Cu2+ adsorption on sphalerite (ZnS) and iron sulfide (FeS) with a view toward understanding the activation of sphalerite by Cu2+, which takes place according to the following reaction: o ZnS + Cu 2+ = CuS + Zn 2+ ; ΔH 298 K, theor. = 63kJ ⁄ mol

Gochin’s measured values agreed quite well with the theoretical value stated previously. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 123 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

123

Two systematic studies on the heat of adsorption of type 2 collectors on oxides have been reported—one by Roy and Fuerstenau (1968) and the other by Mellgren et al. (1974). The heat of adsorption of type 2 collectors should consist of two parts: (1) heat due to interactions between the charged polar head groups and the charged surface sites and (2) heat due to the interaction between the hydrocarbon chains. Both types of interactions involve the removal of water in the vicinity of the surface. Roy and Fuerstenau (1968) studied the heat of immersion of alumina in aqueous solutions containing sodium dodecyl sulfonate at various pH values. They found that as the sulfonate ions begin to adsorb (associate) in the Stern layer, the heat of immersion begins to become more negative than the heat of immersion of Al2O3 in water, reaching a value of –1,100 ergs/cm2 at monolayer coverage. At higher coverages, the heat of immersion remained at about –1,100 ergs/cm2. They did not detect any measurable heat effect when the sulfonate ions occurred only in the diffuse part of the double layer. They have presented a model in which adsorption occurs by a process where a layer of water molecules exists between the solid surface and the adsorbed layer of sulfonate ions. By this model, they estimated an integral heat of adsorption of –5.0 kJ/mol. Mellgren et al. (1974), working with the hematite–dodecyl sulfate system, found that the heat of adsorption (–5.7 kJ/mol) remained constant from an adsorption density that is about one-sixth of the monolayer coverage right up to that of the monolayer coverage. They measured a heat close to the heat of micellization of sodium dodecyl sulfate. (Roy and Fuerstenau [1968] are of the opinion that only at multilayer coverages should the heat of adsorption be close to that of the heat of micelle formation.) Mellgren and colleagues found evidence for sites of different adsorption energy on powdered natural hematite, the proportions of which depend on the method of grinding (or heat treatment) of hematite. An indirect method for determining the heat of adsorption is to measure the amount of collector adsorbed (Γ) as a function of concentration (C) at a given solution pH at two or more different temperatures. From a Clausius–Clapeyron type of equation: o – ΔH ads ln C⎞ ⎛ ∂-----------= ----------------⎝ ∂T ⎠ RT 2

(EQ 64)

This heat of adsorption is the same as the isosteric heat of adsorption, Q st. The heat of adsorption is calculated from adsorption isotherms obtained at two different temperatures. For collectors such as amines, sulfonates, and so forth, which do not chemically react with the surface, Q st is the heat of “adsorption” in the real sense of the word. On the other hand, collectors such as salicylaldehyde (Rinelli, Marabini, and Alesse 1976) and xanthates (Mellgren et al. 1974) chemically react with the surface and form metal–collector complexes both on the surface and in the bulk. In those cases, Q st is not the real heat of adsorption but represents a heat intermediate between heat of reaction and heat of adsorption. From measurement of isotherms for the adsorption of sodium dodecyl sulfonate at two temperatures (Figure 13), Somasundaran and Fuerstenau (1972) evaluated the enthalpy and entropy of the adsorption process. Their results show that the adsorption decreases with increasing temperature, indicating the exothermic nature of the process. Ball and Fuerstenau (1971) used a different approach for evaluating the heat of adsorption. Taking the logarithm of the adsorption density given by the Stern–Grahame equation, they obtained o ⁄ RT ) ln Γ δ = ln 2r + ln C – ( ΔG ads

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 65)

flotation0.book Page 124 Tuesday, January 2, 2007 7:36 PM

124

FLOTATION FUNDAMENTALS

Adsorption Density, mol/cm 2

10–10

10–11

Alumina pH 6.9 2 × 10 M Ionic Strength (NaCl)

10–12

–3

25°C 45°C 10–13 10–6

10–5

10–4

10–3

10–2

Equilibrium Concentration of Sodium Dodecyl Sulfonate, mol/L

Source: Somasundaran and Fuerstenau 1972.

FIGURE 13 Adsorption density at 25°C and 45°C of dodecyl sulfonate on alumina as a function of its equilibrium bulk concentration at pH 6.9 and ionic strength 2 × l0–3 M

Differentiating this equation with respect to temperature yields o ⁄ RT ) o d ( ΔG ads – ΔH ads ∂ ln C d ln r ∂ ln Γ ---------------------------------- = ⎛ ------------⎞ + ----------- – ⎛ -------------δ-⎞ = ----------------⎝ ∂T ⎠ Γ δ dT ⎝ ∂T ⎠ C dT RT 2

(EQ 66)

If the radius of the adsorbing ion is assumed to be independent of temperature, o – ΔH ads ln C⎞ ⎛ ∂ ln Γ δ⎞ ⎛ ∂-----------– -------------= ----------------⎝ ∂T ⎠ ⎝ ∂T ⎠ C RT 2

(EQ 67)

o Under conditions of constant adsorption density, ΔH ads becomes the isoteric heat of adsorption and is given by

Q st ln C⎞ ⎛ ∂-----------= --------⎝ ∂T ⎠ Γ δ RT 2

(EQ 68)

The values of isoteric heat of adsorption can be calculated from the slope of the plot of o can be obtained as a function of temperature (log C)Γ against temperature. Because ΔG ads o from the adsorption data using Equation 65, ΔS ads can be computed easily as o o ΔS ads = – d ( ΔG ads ) ⁄ dT. Ball and Fuerstenau (1971) used this approach to calculate various thermodynamic quantities for the process of adsorption of dodecylammonium acetate on quartz. Their calculations indicate large entropic effects that have been attributed to the phenomenon of hydrophobic bonding.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 125 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

125

Effect of Temperature on Collector Adsorption

Investigations of the effect of temperature on the adsorption of collectors on mineral surfaces can be classified into the following broad categories: 1. Systems in which the collector exhibits physisorption characteristics 2. Systems wherein the collector chemisorbs onto the mineral 3. Systems in which the collector seems to chemically react at the mineral surface with ions derived from the mineral surface. This is actually a borderline category and, for all practical purposes, could be classified under category 2. For any adsorption process, the free energy change is given by ΔG ads = ΔH ads – T ΔS ads

(EQ 69)

If the process occurs spontaneously, then ΔG ads is negative. This can be achieved by the adsorption process being exothermic or by its being accompanied by a large entropy increase. Not much is known about entropy changes in collector adsorption processes. On the one hand, there will be an entropy decrease because of the more ordered structure of collector ions or molecules at the interface, but on the other hand, there may be a significant increase in entropy because of the release of structured water molecules from the solid– water interface and/or from the collector molecule/ion. Particularly for longer-chained collectors, this could result in ΔS ads having a large positive value. Type 1 systems are characterized by an exothermic heat and, as a consequence, adsorption decreases with increase in temperature. A good example of this type is the alumina– dodecyl sulfonate system investigated by Somasundaran and Fuerstenau (1972). Figure 13, which demonstrates their results, shows such a decrease in the adsorption density. Although no one has yet analyzed the results in such a manner, these particular kinds of isotherms are also complicated by a shift in the PZC and a reduction in surface charge as the temperature is increased. Another example is the work of Ball and Fuerstenau (1971) on the aqueous dodecylammonium acetate–quartz system. They determined the effect of temperature on the electrokinetic potential of quartz in the presence of dodecylammonium acetate (see Figure 14) and found it to decrease in absolute magnitude with increasing temperature, which points indirectly to a decrease in adsorption with increasing temperature. Of the chemisorbing systems investigated, the temperature dependence of the adsorption of soaps on mineral surfaces has been investigated in some detail. Falconer (1949), referring to a 1938 French patent, wrote in 1949 about the value of using high temperature for the conditioning of nonsulfide ores for flotation with soaps. Specifically, he mentioned conditioning above 35°C, and preferably above 60°C. In 1950, Cook and Last conducted an interesting investigation of the effect of conditioning temperature on the flotation of fluorite with oleic acid. Figure 15 presents their results on the flotation of a fluorite/calcite/barite ore. Below 40°C, the recovery of fluorite is very low, but raising the conditioning temperature causes the recovery to rise sharply, to 90%–95% at temperatures of 70°C or greater. Their explanation is that the oleic acid is physisorbed at room temperature and that chemisorption does not take place appreciably until the temperature is raised to 45°C–60°C. Because of an activation energy, raising the temperature causes the chemisorption reaction rate to proceed reasonably rapidly during the conditioning period.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 126 Tuesday, January 2, 2007 7:36 PM

126

FLOTATION FUNDAMENTALS

–40

pH 7 25°C 15°C 5°C

Zeta Potential, mV

–30

–20

–10

0

+10

1

2

3

4

5

6

7 8 9 10

Dodecylammonium Acetate Concentration, M × 104

Source: Ball and Fuerstenau 1971.

FIGURE 14 Zeta potential of quartz as a function of dodecylammonium acetate concentration at different temperatures 100

Flotation Recovery or Grade,%

80

60

40

Fluorite

20

Grade Recovery 0 0

20

40

60

80

100

Conditioning Temperature,°C

Source: Cook and Last 1950.

FIGURE 15 Effect of conditioning temperature on flotation recovery and concentrate grade for the flotation of a fluorite ore, which had been conditioned for 5 minutes with oleic acid, quebracho, and sodium carbonate

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 127 Tuesday, January 2, 2007 7:36 PM

Amount Adsorbed

SOME ASPECTS OF FLOTATION THERMODYNAMICS

127

Chemisorption

Physisorption Temperature

FIGURE 16 Schematic representation of an adsorption isobar, showing the transition from physisorption to chemisorption. Only in the transition region should the adsorption increase with increasing temperature.

In conditioning at elevated temperatures, activating the adsorption process for changing it from physisorption to chemisorption is illustrated schematically in Figure 16. The change from one type of process to the other could begin at about 40°C in these systems. Kulkarni and Somasundaran (1975) found that the adsorption of oleate on hematite increases as the temperature is raised from 25°C to 75°C at low ionic strengths, but the converse was observed at very high ionic strengths (due to salting-out effects at lower temperatures). Figure 17, after Somasundaran and Kulkarni (1977), shows that the flotation recovery at low ionic strength increases as the temperature is increased, with maximum recovery always occurring at about pH 8. This implies marked increase in collector adsorption with increasing temperature. Raghavan and Fuerstenau (1975) have investigated the effect of concentration, pH, and temperature on the adsorption of octylhydroxamate on hematite. Optimum adsorption always occurred at about pH 8.4; they considered the adsorbing species to be the hydroxamic acid molecule. Using Equation 55 in the following form: o

– ΔG ads ⎞ Cf - = -----------exp ⎛ ------------------------⎝ RT ⎠ 55.55 1–f

(EQ 70)

where f is the fraction of the surface covered by the collector, and C is the collector conceno to be approximately –32 kJ/mol and nearly indepentration in solution, they found ΔG ads dent of pH and surface coverage (for f < 0.8). For a chemically adsorbing species, the o observed value of ΔG ads is rather small. Their results on the effect of temperature, which are presented in Figure 18, show a marked increase in adsorption with increasing temperature. However, the isotherms for 41°C and 60°C given in Figure 18 exhibit adsorption densities that exceed monolayer (7 × 10–10 mol/cm2) coverage. This indicates possible formation of ferric hydroxamate by surface reaction. Thus, one effect of the temperature might simply be to increase the solubility of the mineral, thereby giving rise to a larger number of Fe ions for reaction at the surface. In an investigation of the flotation of iron oxide with octylhydroxamate and oleate as collector, Fuerstenau, Harper, and Miller (1970) found enhanced flotation with increasing temperature. Their conclusions were that the chemisorption process definitely involves increased mineral solubility as the temperature is raised. In © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 128 Tuesday, January 2, 2007 7:36 PM

128

FLOTATION FUNDAMENTALS

100 Hematite 3 × 10–3 M KOI

80

Flotation Recovery, %

100°C 60 94°C

75°C

40

60°C 20

25°C

0 3

4

5

6

7

8

9

10

Flotation pH

Source: Somasundaran and Kulkarni 1977.

FIGURE 17 Effect of temperature on the flotation of hematite with 3 × l0–5 M potassium oleate as collector 20

Amount Adsorbed, mol/cm 2 × 10 10

16

12 60°C 41°C

8

4

20°C

Hematite–Hydroxamate pH 5.5 2 × 10–3 M KNO3

0 0

2

4

6

8

10

12

Hydroxamate Equilibrium Concentration × 104, mol/L

Source: Raghavan and Fuerstenau 1975.

FIGURE 18 Effect of temperature on the adsorption density of potassium octylhydroxamate on hematite

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 129 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

129

an earlier study, while working with octylhydroxamate and various other chelating agents as collectors for chrysocolla, Peterson et al. (1965) clearly showed that raising the temperature can markedly increase flotation response. The optimum pH for chrysocolla flotation always occurred at about pH 6. In an interesting study of the flotation characteristics of pyrolusite, Fuerstenau and Rice (1968) found that flotation decreased slightly with a sulfonate when the temperature was raised from 23°C to 60°C. On the other hand, with oleate as collector for the same mineral, flotation increased markedly upon raising the temperature from 23°C to 60°C. However, the pH for optimum flotation was reduced from 10 at 23°C to about 8 at 60°C. On the other hand, Yousef, Arafa, and Malati (1971) found the adsorption of oleate on beta-manganese dioxide to be an exothermic process both in the absence and the presence of Ba2+ ions. Cooke, Iawasaki, and Choi (1960) conducted a detailed investigation of the effect of temperature on the flotation of hematite. With a series of collectors—stearic, elaidic, oleic, linoleic, and linolenic acids—they observed only a very slight increase in flotation and in contact angles by increasing the temperature from 25°C to 70°C, except for stearic acid, which exhibited a strong increase due to its increased solubility. They also found that contact angles of calcium-activated quartz at pH 11 decreased with increasing temperature, indicating diminished calcium adsorption. On the other hand, flotation under the same conditions increased with increasing temperature, but this was probably due to rather unusual frothing characteristics at pH 11. As has already been discussed at length, pH has a marked effect in flotation because of its effect on collector ionization, surface charge, and mineral solubility. Thus, any temperature study of flotation systems should also take into account the temperature dependence of the dissociation constant of water and pH (Harned and Owen 1958). In conclusion, the flotation process, as a consequence of its being a multiphase heterogeneous system, is quite complicated to analyze thoroughly by simple thermodynamic approaches. What this chapter attempted to do is bring out a few salient features of the application of thermodynamic concepts to the understanding of flotation. REFERENCES

Adamson, A.W. 1967. Physical Chemistry of Surfaces. 2nd edition. New York: Interscience Publishers. Aplan, F.F., and P.L. de Bruyn. 1963. Adsorption of hexyl mercaptan on gold. Trans. AIME 226:235. Ball, B., and D.W. Fuerstenau. 1971. Disc. Faraday Soc. 52:361. Bleier, A., E.D. Goddard, and R.D. Kulkarni. 1976. Structural effects of amine collectors on the flotation of quartz. Page 117 in Flotation. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Cases, J.M. 1970. Normal interaction between adsorbed species and adsorbing surface. Trans. AIME 247(2):123. Chander, S., and D.W. Fuerstenau. 1972. Natural floatability of molybdenite. Trans. AIME 251(1):62. Clifford, J. 1975. Properties of water in capillaries and thin films. Page 79 in Water. Volume 5. Edited by F. Franks. New York: Plenum Press. Cook, M.A., and A.W. Last. 1950. Fluorite Flotation II. Bulletin No. 47. Utah Engineering Experiment Station, University of Utah. Cooke, S.R.B., I. Iawasaki, and H.S. Choi. 1960. Flotation characteristics of hematite, goethite, and activated quartz with 18-carbon aliphatic acids and related compounds. Trans. AIME 217:76. de Bruyn, P., and G.E. Agar. 1962. Surface Chemistry of Flotation. Pages 91–138 in Froth Flotation. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 130 Tuesday, January 2, 2007 7:36 PM

130

FLOTATION FUNDAMENTALS

de Bruyn, P.L. 1955. Flotation of quartz by cationic collectors. Trans. AIME 202:291. de Bruyn, P.L., J.T.G. Overbeek, and R. Schuhmann. 1954. Flotation and the Gibbs adsorption equation. Min. Eng. 6:519 Defay, R., and I. Prigogine. 1966. Surface Tension and Adsorption. New York: John Wiley & Sons. Derjaguin, B.V. 1932. The rigidity of thin layers of water. J. Phys. Chem. 3:29. Derjaguin, B.V., and L.M. Shcherbakov. 1961. Effect of surface forces on phase equilibrium of multimolecular layers and on the contact angle. Colloid J. USSR 23(1):33. Dick, S.G. 1972. Surfactant adsorption at the oxide–water interface. M.S. thesis, University of Melbourne. Eigeles, M.A., and M.L. Volova. 1960. Kinetic investigation of the effect of contact time, temperature and surface condition on the adhesion of bubbles to mineral surfaces. Page 271 in Proceedings of the 6th International Mineral Processing Congress. London: Institution of Mining and Metallurgy. Falconer, S.A. 1949. Pretreatment of mineral surfaces for froth flotation. Trans. AIME 184:247. Finch, J.A., and G.W. Smith. 1972. Liquid–vapor interface in the study of particle–bubble attachment. Can. Metall. Q. 11(4):569. ———. 1975. Bubble–solid attachment as a function of bubble surface tension. Can. Metall. Q. 14(1):47. Fuerstenau, D.W. 1970. Interfacial processes in mineral/water systems. Pure Appl. Chem. 24(1):135. ———. 1971. The adsorption of surfactants at solid–water interfaces. Pages 143–176 in The Chemistry of Biosurfaces. Edited by M.L. Hair. New York: Marcel Dekker. Fuerstenau, D.W., and T.W. Healy. 1972. Principles of mineral flotation. Pages 92–131 in Adsorptive Bubble Separation Technique. Edited by R. Lemlich. New York: Academic Press. Fuerstenau, M.C., R.W. Harper, and J.D. Miller. 1970. Hydroxamate vs. fatty acid flotation of iron oxide. Trans. AIME 247(1):69. Fuerstenau, M.C., and D.A. Rice. 1968. Flotation characteristics of pyrolusite. Trans. AIME 241(4):453. Gaudin, A.M. 1957. Flotation. 2nd edition. New York: McGraw-Hill. Gaudin, A.M., and D.W. Fuerstenau. 1955. Quartz flotation with cationic collectors. Trans. AIME 202:66. Gochin, R. 1972. A thermochemical study of the flotation of sphalerite and pyrite. Ph.D. thesis, University of London. Grahame, D.C. 1947. The electrical double layer and the theory of electro-capillarity. Chem. Rev. 41:441. Griffiths, D.A. 1973. The effect of pH and temperature on the heat of immersion of alumina. M.S. thesis, College of Engineering, University of California, Berkeley. Hall, P.G., V.M. Lovell, and N.P. Finkelstein. 1970. Adsorption of water vapor on ionic solids containing preadsorbed sodium oleate. 1. Calcium fluoride. Trans. Faraday Soc. 66(pt 6):1520. Harned, H.S., and B.B. Owen. 1958. The Physical Chemistry of Electrolyte Solutions. New York: Reinhold. Healy, T.W., and D.W. Fuerstenau. 1965. Oxide–water interface: Interrelation of the zero point of charge and the heat of immersion. J. Colloid Sci. 20(4):376. Israelachvili, J.N., and R.M. Pashley. 1988. The hydrophobic interaction is long range, decaying exponentially with distance. Nature 300:341. Johnson, R.E., Jr. 1959. Conflicts between Gibbsian thermodynamics and recent treatments of interfacial energies in solid-liquid-vapor systems. J. Phys. Chem. 63:1655. Kruyt, H.R. 1952. Colloid Science. Volume 1. Amsterdam: Elsevier. Kulkarni, R.D., and P. Somasundaran. 1975. Oleate adsorption at hematite/solution interface and its role in flotation. Paper presented at the 104th AIME Annual Meeting. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Lai, R.W. 1970. Surface charge, adsorption of ionic surfactants and wettability of oxide minerals. Ph.D. thesis, College of Engineering, University of California, Berkeley. Laskowski, J. 1974. Particle–bubble attachment in flotation. Miner. Sci. Eng. 6(4):223. Laskowski, J.S., and J.A. Kitchener. 1968. The hydrophilic-hydrophobic transition on silica. J. Colloid Interface Sci. 29:670. Leja, J., and G.W. Poling. 1960. On the interpretation of contact angle. Page 325 in Proceedings of the 5th International Mineral Processing Congress. London: Institution of Mining and Metallurgy.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 131 Tuesday, January 2, 2007 7:36 PM

SOME ASPECTS OF FLOTATION THERMODYNAMICS

131

Lin, I.J., and P. Somasundaran. 1971. Free-energy changes on transfer of surface-active agents between various colloidal and interfacial states. J. Colloid Interface Sci. 37(4):731. Lyklema, J. 1966. Electrical double layer on silver iodide: Influence of temperature and application to sol stability. Disc. Faraday Soc. 42:81. Mellgren, O. 1966. Heat of adsorption and surface reactions of potassium ethyl xanthate on galena. Trans. AIME 235:46. Mellgren, O., R.J. Gochin, H.L. Shergold, and J.A. Kitchener. 1974. Thermochemical measurements in flotation research. Pages 451 in Proceedings of the 10th International Mineral Processing Congress. Edited by M.J. Jones. London: Institution of Mining and Metallurgy. Mellgren, O., and S.R. Rao. 1968. Heat of adsorption and surface reactions of potassium diethyldithiocarbamate on galena. Trans. IMM 77( June):C65. Neumann, A.W. 1974. Contact angles and their temperature dependence: Thermodynamic status, measurement, interpretation, and application. Adv. Colloid Interface Sci. 4(2–3):105. Padday, J.F. 1970. Cohesive properties of thin films of liquids adhering to a solid surface. Special Disc. Faraday Soc. 1:64. Parks, G.A. 1960. A study of the surface of ferric oxide in aqueous systems. Ph.D. thesis, Massachusetts Institute of Technology. ———. 1967. Aqueous surface chemistry of oxides and complex oxide minerals: Isoelectric point and zero point of charge. Adv. Chem. Ser. 67:121. Parks, G.A., and P.L. de Bruyn. 1962. Zero point of charge of oxides. J. Phys. Chem. 66:967. Peterson, H.D., M.C. Fuerstenau, R.S. Rickard, and J.D. Miller. 1965. Chrysocolla flotation by the formation of insoluble surface chelates. Trans. AIME 232(4):389. Pope, M.I., and D.I. Sutton. 1972. Collector adsorption during froth flotation. Powder Tech. 5(2):101. Predali, J.J., and J.M. Cases. 1974. Thermodynamics of the adsorption of collectors. Page 473 in Proceedings of the 10th International Mineral Processing Congress. Edited by M.J. Jones. London: Institution of Mining and Metallurgy. Raghavan, S., and D.W. Fuerstenau. 1975. Adsorption of aqueous octylhydroxamate on ferric oxide. J. Colloid Interface Sci. 50(2):319. Rehbinder, P.A. 1949. General course in colloidal chemistry. Moscow University. Rinelli, G., A.M. Marabini, and V. Alesse. 1976. Flotation of cassiterite with salicylaldehyde as collector. Page 549 in Flotation. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Roy, P., and D.W. Fuerstenau. 1968. Heat of immersion of alumina into aqueous sodium dodecyl sulfonate solutions. J. Colloid Interface Sci. 26(1):102. Sandvik, K.L., and M. Digre. 1968. Adsorption of amine on quartz through bubble interaction. Trans. IMM 77( June):C61. Sheludko, A. 1967. Thin liquid films. Adv. Colloid Interface Sci. 1(4):391. Smolders, C.A. 1961. Contact angles; wetting and dewetting of mercury. II. Theory of wetting. Rec. Trav. Chim. 80:650. Somasundaran, P. 1968. Relation between adsorption at different interfaces and flotation behavior. Trans. AIME 241(1):105. Somasundaran, P., and D.W. Fuerstenau. 1966. Mechanisms of alkyl sulfonate adsorption at the alumina–water interface. J. Phys. Chem. 70(1):90. ———. 1972. Heat and entropy of adsorption and association of long-chain surfactants at the alumina–aqueous solution interface. Trans. AIME 252(3):275. Somasundaran, P., T.W. Healy, and D.W. Fuerstenau. 1964. Surfactant adsorption at the solid–liquid interface: Dependence of mechanism on chain length. J. Phys. Chem. 68(12):3567. Somasundaran, P., and R.D. Kulkarni. 1977. Effect of reagentizing temperature and ionic strength and their interactions in hematite flotation. Trans. AIME 262(2):120. Tewari, P.H., and A.W. McLean. 1972. Temperature dependence of point of zero charge of alumina and magnetite. J. Colloid Interface Sci. 40(2):267.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 132 Tuesday, January 2, 2007 7:36 PM

132

FLOTATION FUNDAMENTALS

Wada, M. 1960. The wetting of solid in solutions of surface-active substances as a function of solute concentration. Page 287 in Proceedings of the 6th International Mineral Processing Congress. London: Institution of Mining and Metallurgy. Wade, W.H., and N. Hackerman. 1964. Thermodynamics of wetting of solid oxides. Adv. Chem. Ser. 43:222. Wakamatsu, T., and D.W. Fuerstenau. 1968. Effect of hydrocarbon chain length on the adsorption of sulfonates at the solid/water interface. Adv. Chem. Ser. 79:161. ———. 1973. Effect of alkyl sulfonates on the wettability of alumina. Trans. AIME 254(2):123. Yoon, R.-H. 2000. The role of hydrodynamic and surfac3 forces in bubble–particle interactions. Int. J. Miner. Process. 58:129. Yopps, J.A., and D.W. Fuerstenau. 1964. The zero point of charge of α-alumina. J. Colloid Sci. 19(1):61. Yousef, A.A., M.A. Arafa, and M.A. Malati. 1971. Adsorption of sulfite, oleate, and manganese(II) ions by β-manganese dioxide and its activation in flotation. J. Appl. Chem. Biotechnol. 21(7):200. Zettlemoyer, A.C. 1968. Hydrophobic surfaces. J. Colloid Interface Sci. 28(3–4):343.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 133 Tuesday, January 2, 2007 7:36 PM

The Nature of Hydrophobic Attraction Forces Jan Christer Eriksson and Roe-Hoan Yoon

INTRODUCTION

To gain an improved understanding of the molecular basis of hydrophobicity and hydrophobic attraction forces is of crucial importance for several scientific and technological fields, such as surface and colloid science, biochemistry, and mineral processing. The formation of surfactant micelles and biomembranes, for instance, can be traced to the tendency of hydrocarbon chains to associate when present in a water medium at a high enough (though still very low) concentration. In addition, many enzymatic reactions take place while the substrate is locked in a hydrophobic pocket. In the field of flotation, on the other hand, long-range hydrophobic attraction forces contribute to establishing the mineral–particle/ air–bubble attachment necessary to achieve selective ore enrichment. Early experiments indicating the existence of long-range hydrophobic attraction forces were carried out by Blake and Kitchener (1979) on thin water films sandwiched between hydrophobic surfaces. The instability observed was due to the attractive forces operating across the film that arise because of the contact of the water with the hydrophobic surfaces. Derjaguin and Churaev (1974) and Derjaguin, Churaev, and Muller (1987), referring to a large body of experimental results, discussed the same effect in terms of “the structural component of the disjoining pressure.” Most of the hydrophobic force measurements conducted during the last several decades have been based on employing the surface force apparatus (SFA), developed by Tabor and Winterton (1969) and Israelachvili and Tabor (1972) (Figure 1); the atomic force microscope (AFM) (Figure 2); or some similar direct-surface force measurement device such as the measurement and analysis of surface interaction forces (MASIF) of Parker (1994) and the interfacial gauge of Yaminsky, Ninham, and Stewart (1996). This particular research area, which focuses on thin, aqueous films between hydrophobic surfaces, was inaugurated in 1982 by Israelachvili and Pashley by using mica surfaces and successively adding the cationic surfactant hexadecyltrimethylammonium bromide (CTAB) to the water medium (Israelachvili and Pashley 1982, 1984). They established that in addition to the normal DLVO (Derjaguin–Landau–Verwey–Overbeek) forces due to dispersion and electrostatic interactions, a long-range attractive interaction force operates as a consequence of the hydrocarbon–water contact. Hence, they identified this surface force as the “hydrophobic force,” a term originally coined by Blake and Kitchener (1979). Several years later, Claesson and Christenson (1988) and Christenson and Claesson (1988) experimented with mica surfaces that had been modified by Langmuir–Blodgett (LB) deposition of dioctadecyldimethylammonium (DODA) bromide or, alternatively, a double-chain cationic fluorocarbon surfactant. For these cases, the hydrophobic attraction was considerably stronger and of longer range than observed by Israelachvili and Pashley 133

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 134 Tuesday, January 2, 2007 7:36 PM

134

FLOTATION FUNDAMENTALS

Fringes

Coarse Separation Distance Control

Piezoelectric Tube

Double-Beam Cantilever

Curved Mica Surfaces

Clamp to Adjust Spring Stiffness White Light

NOTE: The interacting (mica) surfaces have cylindrical shape, and their separation is determined optically by means of an interferometric technique. The lower surface is attached to a cantilever spring, the deflection of which serves to determine the interaction force (Flinn 1997).

FIGURE 1 SFA based on the original design by Tabor and Winterton (1969) and Israelachvili and Tabor (1972)

Laser Beam Cantilever Deflection

Flat Plate Sample Microsphere

Piezoelectric Tube

NOTE: The top portion shows a drawing of a triangular cantilever with an attached glass sphere. The distance between the sphere and flat plate (sample) is measured by monitoring the deflection of the cantilever spring using a laser beam. The deflection of the cantilever of a known spring constant gives the force at a given separation distance (Flinn 1997).

FIGURE 2

Schematic representation of AFM

(1982, 1984)—about 100 times stronger than the van der Waals attraction; decay length about 15 nm—and measurable up to about 80 nm. Independently, measurements were also conducted in Moscow by Rabinovich and Derjaguin (1988) using silanated silica filaments, which yielded mostly equivalent results (Figure 3). Subsequent control experiments employing a variety of prepared hydrophobic surfaces have only partially confirmed the previous findings and perceptions. An abundance of biased and bewildering results have been presented by many investigators, making it difficult © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 135 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

135

h, nm 0

10

20

30

40

50

60

70

80

–10

Rabinovich and Derjaguin (1988)

F / R , μN m–1

–102

–103

–104

NOTE: The points demark experimental values measured by Claesson and Christenson (1988). The solid line represents Equation 56, obtained by inserting B = 0.6 mJ/m2 and b–1 = 15.8 nm from Equation 31. For comparison, the dotted line shows the surface force function determined by Rabinovich and Derjaguin (1988) for silica filaments immersed in 0.1 mM KCl (potassium chloride) solution. The corresponding experimental points are scattered around the dotted line within a factor of about 3.

FIGURE 3 Plot of the attractive surface force, F/R, vs. the surface separation, h, for DODAcovered mica surfaces in contact with pure water

to summarize the current status of research. Nevertheless, Christenson and Claesson (2001) presented a detailed account of the scientific state of the art from an experimental perspective. They classified the non-DLVO attractive forces observed between hydrophobic surfaces into the following classes: 1. A fairly short-range but strongly attractive force, much stronger than the van der Waals force, between stable hydrophobic surfaces 2. An attraction of variable strength and range caused by the presence of small bubbles sporadically adhering to hydrophobic surfaces 3. A very long-range attractive force with exponential decay operating between a variety of hydrophobic surfaces, in particular those hydrophobized by means of Langmuir– Blodgett deposition on mica or silica, or adsorption from solution of an ionic amphiphile

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 136 Tuesday, January 2, 2007 7:36 PM

136

FLOTATION FUNDAMENTALS

Spalla (2000) presented a complementary paper, primarily dealing with the theoretical understanding of the hydrophobic force. Furthermore, papers by Tsao, Evans, and Wennerström (1993) and Craig, Ninham, and Pashley (1998) incorporate illuminating discussions about the possible origins of the hydrophobic force. More recently, Israelachvili and co-workers have begun to critically re-examine the entire issue, and extensive reference lists are included in their papers (Meyer, Lin, and Israelachvili 2005; Lin et al. 2005). At this point, the chapter focus is on some typical results obtained for the main hydrophobic surfaces investigated so far, and in particular, for hydrophobic surfaces that are of superior quality. The ultimate purpose is to scrutinize whether there is enough evidence today in support of the existence of a “true” hydrophobic force (i.e., a surface force of long range that is related to the structural response of a thin water film between hydrophobic surfaces). First, some recent findings as to the molecular organization of water interfaces and hydrophobic interactions in general are discussed. After dwelling on the thermodynamics of the surface force experiment, the concept of an ideal hydrophobic surface is introduced, thus providing a frame of reference for judging experiments with real hydrophobic surfaces that usually exhibit many deficiencies. One section is devoted to the possible formation of bridging air bubbles (and cavities) for hydrophobic surfaces with (equilibrium) contact angles against water in excess of 90°. M O L E C U L A R O R G A N I Z AT I O N O F WAT E R AT I N T E R FA C E S

Because of the advent of novel spectroscopic and computational methods in recent years, considerable progress has been made in the probing and modeling of the molecular state of bulk as well as of surface water. These novel methods include sum-frequency generation (SFG) spectroscopy, X-ray absorption spectroscopy (XAS), and X-ray Raman scattering (XRS) (Miranda and Shen 1999; Wernet et al. 2004). Hence, it has been established that about one-fourth to one-fifth of the water molecules in the top monolayer of the water–air interface have a non-hydrogen-bonded OH group as a unique feature. The dangling –OH group spectral peak for a hydrocarbon–water interface remains essentially the same as for the free water surface toward air, and can be rationalized on the notion of oriented oxygen double layers similar to those present in the ice Ih lattice (Figure 4). Evidently, a large part of the comparatively high surface energy of water is connected with the excess of broken hydrogen bonds in the surface, compared to the situation in bulk water. Rather surprisingly, however, Cavalleri (2004) has recently found that for bulk water there is also a tendency to form linear aggregates of water molecules (based on two strong hydrogen bonds rather than four of medium strength), resulting in rings and chains of water molecules. On this point, one may note that according to Hill’s small-system thermodynamics, a distribution of linear aggregates may arise insofar as the free energy cost of introducing one additional molecule in the central part of such an aggregate is small enough to become compensated by the fluctuation entropy contribution that is inherent in the length distribution. In the absence of any long-range forces, a mechanism of this thermodynamic nature is decisive for the formation of elongated, rod-shaped surfactant micelles. A similar scheme can be applicable even for chains of water molecules and might be at the root of water effects of long range. It remains to be assessed, however, as to what extent the novel findings and ideas about the structural aspects of liquid water will necessitate revising the conventional molecular picture based on four-coordinated water molecules in small clusters having external surfaces with less-well-bonded molecules (Ludwig 2001). © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 137 Tuesday, January 2, 2007 7:36 PM

137

712

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

275

275

450

FIGURE 4 The ideal ice Ih structure, characterized by stacked oxygen double layers. The numbers represent dimensions in picometers.

Thus, when considering hydrophobicity and hydrophobic surface forces, bear in mind the ongoing endeavors to probe and to simulate interfacial and bulk water by means of molecular dynamics, which may result in a substantially revised basis for a comprehensive theoretical treatment of these phenomena (Cavalleri 2004). The classical simulations of Lee, McCammon, and Rossky (1984), which are often referred to in this context, may have been based on a simplistic water potential (of ST2 type). This has already been indicated by the fact that the predictions of the melting point of ice using water models of this kind are, for the most part, grossly in error. Moreover, to only explore density changes in a thin water film might not be sufficient because minor, although thermodynamically significant, structural rearrangements can occur without sensible volume changes. Speaking in broad terms, an atomically smooth, chemically inert solid surface that is subject to thermal motions of small amplitudes may be expected to • Force the water molecules to pile up against the surface, as would actually be the case for any liquid, and • Force the hydrogen bond network to rearrange in its vicinity so as to limit the number of broken hydrogen bonds, thus forming a clathrate-resembling contact monolayer of water molecules. On average, this would result in a roughly tangential alignment of the water dipoles. In a cooperative fashion, such a contact monolayer, in turn, imposes bonding constraints on the successive layers of water molecules beneath it so as to generate a surface-induced network. A crucial and much-debated question arises: How deep toward the bulk may such a network prevail? Interestingly, the SFG spectra recorded by Miranda and Shen (1999) indicate that a well-ordered, ice-like water structure predominates at a solid hydrophobic surface, whereas a more disordered water structure prevails for the water–air and water–hexane interfaces. The previously mentioned simulations by Lee, McCammon, and Rossky (1984) and similar ones by other researchers (Forsman, Jönsson, and Woodward 1996) indicate that a significant surface effect is noticeable only for short distances in liquid water, on the order of © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 138 Tuesday, January 2, 2007 7:36 PM

138

FLOTATION FUNDAMENTALS

1 nm. In contrast, many surface force measurements (detection limit ≈ 10 μNm–1) are indicative of non-DLVO attractions extending as far as 50 nm and, occasionally, even longer. Eventually, the reasons for these widely contrasting results will have to be determined before a convincing theoretical description can be achieved for why and how the hydrophobic attraction arises. H Y D R O P H O B I C E F F E C T S I N S O L U T I O N S A N D AT T H E WAT E R – A I R I N T E R FA C E

The concept of hydrophobicity stems from the common observation that many substances, such as oil and hydrocarbons, do not mix with water. Yet, for entropic reasons, there is always a slight solubility, even for hydrocarbons, that reflects the local energetic conditions of a single hydrophobic solute molecule surrounded by water molecules. By analyzing the (mole–fraction-based) solubility in water of a number of straight-chain hydrocarbons, Tanford (1980) was able to quantify the Gibbs (free) energy change associated with transferring a –CH2– or a –CH3 group from bulk hydrocarbon to the dilute water state. These thermodynamic quantities are necessarily positive and of the following magnitude at room temperature (per methyl or methylene group): 0 Δμ CH = 3.548k B T 3 0 Δμ CH = 1.492k B T 2

(EQ 1)

where kB = R/NAvogadro and is the Boltzmann constant and T is the absolute temperature. 0 0 With sign reversed, Δμ CH and Δμ CH determine the driving force for aggregation of 3 2 hydrocarbon chains present in water, provided that the concentration of the chains exceeds the normal water solubility of the bulk hydrocarbon, a condition that can easily be realized by attaching a polar group to the hydrocarbon chain, thus forming an amphiphilic molecule. Generally, the solubility of a hydrocarbon in water is nearly temperature-independent at about room temperature. Referring to the thermodynamic Gibbs–Helmholtz equation for the case under discussion, 0 d ln x hc Δh hc ----------------- = ---------d( 1 ⁄ T ) kB

(EQ 2)

where xhc denotes the mole fraction of dissolved hydrocarbon. As xhc is approximately independent of temperature within the room-temperature range, this relation implies that the enthalpy change for the process hc (bulk hydrocarbon) → hc (water)

(EQ 3)

is small, which in turn means that the difference between the standard state chemical potentials is given by 0 = Δh 0 – T Δ s 0 ≈ – T Δ s 0 Δμ hc hc hc hc

(EQ 4)

0 is largely determined by the (negative) entropy change, Δs 0 , In other words, Δμ hc hc accompanying the hydrocarbon dissolution in water.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 139 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

139

According to Frank and Evans (1945), and Shinoda (1978), these thermodynamic results can be understood in the following way. To insert a hydrocarbon molecule into water, it takes a certain amount of energy to form a nano-cavity where the hydrocarbon molecule can reside. At room temperature, however, this energy expense is almost exactly counterbalanced by the energy released upon structuring the surrounding water to create a more 0 is ordered hydration shell, Δh structure . What is left, then, to obtain an estimate of Δμ hc largely the concomitant entropy change: 0 ≈ Δs Δs hc structure ≈ Δh structure ⁄ T

(EQ 5)

that pertains to a process similar in nature to a freezing phase transition and is, hence, a negative quantity. The last approximate relation in Equation 5 follows because for such a phase transition, there is almost no net change in Gibbs energy. Conversely, T Δs structure may be identified as being the main driving force for hydrocarbon association in a water environment at ordinary temperatures. In other words, the overall excess free energy due to the water structuring can be reduced by hydrocarbon association. In this way, one can readily rationalize why, for example, ordinary surfactant micelles start to form in solution (and hemimicelles on surfaces) at some fairly low “critical” surfactant concentration (i.e., the critical micelle concentration). A few decades ago, the negative thermodynamic quantity Δs structure discussed above was often attributed to “iceberg” formation that was considered to be the general cause of hydrocarbon association in water, the prevailing belief being that release of structurally constrained water molecules furnishes the free energy required to accomplish the association. The similarities with clathrate formation for inert gases in water were also frequently mentioned as supporting evidence. The full experimental picture is, however, somewhat more complex; compare Figure 5, which shows that water exhibits “normal” liquid properties at high temperatures where the solubility of a hydrocarbon increases with temperature but special, structurally related properties at lower temperatures. This behavior can, thus, be traced back to the formation of hydrogen-bonded clusters and networks in water. Basically, much the same explanation accounts for the relatively low surface entropy of liquid water. On a molar basis, this entropy amounts to about 10 J/K·mole, whereas for normal, non-hydrogen-bonded liquids, a value of about 25 J/K·mole (approximately one-fourth of the entropy of vaporization) is typical (Eriksson 1966). Moreover, at ordinary temperatures, the surface entropy of water increases with temperature, as appears from the following experimentally determined surface tension function for pure water (Cini, Loglio, and Ficalbi 1972), recalling that the surface entropy is determined by (the negative of ) the temperature derivative of the following function: γ w, air = 75.653 – 0.1379t – 0.2717 × 10 –3 t 2

( mNm –1 )

(EQ 6)

where t denotes temperature in centigrade. Because the surface entropy is an excess property relative to the entropy of water in the bulk state, for an ordinary liquid the surface entropy is expected to be practically T-independent. The fact that the surface entropy of water increases with temperature is thus indicative of a gradual breakdown of an interfacial hydrogen-bond-dependent structure. This conclusion is supported by the temperature dependence observed for the SFG spectrum of the pure water–vapor interface that indicates a relative loss of ordered water as the temperature is raised (Miranda and Shen 1999). © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 140 Tuesday, January 2, 2007 7:36 PM

140

FLOTATION FUNDAMENTALS

ºC 200 180 160 140

120

100

80

60

40

20

2

3

log X 2

Benzene

Toluene 4

Ethylbenzene

Extrapolated Solubility Curve

Experimental Solubility Curve

5 2.0

2.2

2.4

2.6

2.8

3.0

3.2

3.4

3.5

1/T × 103

NOTE: Typically, the solubility passes through a flat minimum at about room temperature. The effect of water structuring is related with the vertical distances to the corresponding extrapolated lines. Source: Data from Shinoda 1978.

FIGURE 5 Diagram displaying how the solubilities of alkylbenzenes in water vary with temperature

For the case of a planar geometry, however, the enthalpy change associated with the structure formation is insufficient to counterbalance the energy expenses because of rupturing hydrogen bonds toward air and the concomitant loss of dispersion interactions. Therefore, it can be assumed that the hydrophobic free energy is a curvature-dependent quantity, being less for a strongly curved droplet of a hydrocarbon fluid than for a planar hydrocarbon– water interface. Although there is presently consensus among theoretical physicists and chemists about the ability of liquid water to respond structurally to the formation of “internal” microscopic as well as “external” macroscopic interfaces, obviously there is an unresolved issue about the range that is affected. Calculations and molecular dynamics simulations using various water interaction potentials made thus far show that the range having a molecular organization different from that of bulk water may extend merely about 1 nm, whereas some surface force © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 141 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

141

measurements indicate attractive hydrophobic forces for water films as thick as 100 nm. Hence, in an attempt to bridge the gap between theory and experiment, pertinent questions to consider are as follows: 1. How realistic are the molecular model calculations/simulations for thin water films carried out thus far? Do they comply with the notion of restructuring at a minor free-energy expense? Would they be capable of predicting film tensions with the necessary precision? 2. To what extent do the experimental hydrophobic surfaces employed actually mimic an ideal hydrophobic surface? The questions presented in no. 1 above are currently being pursued by means of novel spectroscopic and computational methods. Certainly, the difficulties encountered along this fundamentally oriented route are rather formidable. The latter question is more practically oriented and leads to the almost arcane art of measuring contact angles in order to assess the hydrophobicity/polarity of solid surfaces. C O N TA C T A N G L E M E A S U R E M E N T S T O A S S E S S HYDROPHOBICITY

It is common practice to quantify the hydrophobicity of a solid surface 1 by means of making contact-angle measurements, usually employing pure water 3 as the contacting liquid. Assuming the solid–liquid interfacial tension (γ13), or more precisely, the reversible cleavage work in water (compare Eriksson 1969) to be γ13 = 51 mJm–2 (i.e., the same as for an entirely fluid hydrocarbon–water interface); taking the hydrophobic surface against air 2 to have a corresponding interfacial tension value of γ12 = 21 mJm–2; and invoking γ32 = 72 mJm–2, followed by using the Young equation, one can readily predict the contact angle of the hydrophobic surface to be –1

–1

θ = cos [ ( γ 12 – γ 13 ) ⁄ γ 32 ] = cos ( – 30 ⁄ 72 ) = 114.6°

(EQ 7)

where the term γ 12 – γ 13 (= 30 mJ m–2) is referred to as the “superficial tension” of water in contact with the hydrophobic surface. This particular tension, introduced by Gibbs, accounts for the tendency of water to contract on a solid surface. A prerequisite for the above identification is that no, or just minute amounts of, water vapor adsorbs at the hydrocarbon–air interface. According to Equation 7, a contact angle of 90° is reached when the superficial tension equals zero, whereas a completely water-indifferent, polar solid surface resembling ice, with γ13 = 0 mJm–2 and γ32 = 72 mJm–2, would exhibit a contact angle of 0°. Based on discussions in the previous paragraph, γ1 and γ13 have been plotted as functions of contact angle (cosθ), as shown in Figure 6. It is expected that γ13 = 36.0 mJm–2 at θ = 90° and 18.0 mJm–2 at θ = 60°. Thus, one realizes that even surfaces exhibiting a contact angle substantially less than 90° are presumably capable of exerting a structural hydrophobic effect on water. The interfacial tension, γ13, between solid 1 and water 3 can be considered to consist of apolar and polar components. Thus, LW + γ AB γ 13 = γ 13 13

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 8)

flotation0.book Page 142 Tuesday, January 2, 2007 7:36 PM

142

FLOTATION FUNDAMENTALS

60

γ13

50

γ, mJm–2

40

30

γ1

20

10 120°C

θ = 90°C

Hydrophobicity

Polarity

60°C

0 –0.5

–0.4

–0.3

–0.2

–0.1

0.0

0.1

0.2

0.3

0.4

0.5

cos (θ)

FIGURE 6 Diagram showing how, in theory, the cosine of the contact angle of water (cosθ ) and the solid–water (γγ13) and solid–air (γγ1) interfacial tensions depend on the polarity/ hydrophobicity of a solid surface

where LW refers to Lifshitz–van der Waals (i.e., apolar) interactions, and AB refers to acid– base (polar) interactions. Good and Girifalco (1960), and Fowkes (1963) showed that 2

LW = ⎛ LW LW ⎞ γ 13 ⎝ γ1 – γ3 ⎠

(EQ 9)

where γ 1LW is the LW component of the surface free energy of solid, and γ 3LW is the LW component of water. For the acid–base interactions, van Oss, Chaudhury, and Good (1987) showed that AB = 2 ⎛ + +⎞ ⎛ – –⎞ γ 13 ⎝ γ1 – γ3 ⎠ ⎝ γ1 – γ3 ⎠

(EQ 10)

where γ 1+ and γ 3+ are the acidic components of the surface free energy of the solid and water, respectively; and γ 1– and γ 3– are the basic components of the same. Inserting Equations 9 and 10 into Equation 8, one obtains γ 13 = ⎛⎝ γ 1LW – γ 3LW ⎞⎠ + 2 ⎛⎝ γ 1+ – γ 3+ ⎞⎠ ⎛⎝ γ 1– – γ 3– ⎞⎠

(EQ 11)

From the following fundamental relationships: γ i = γ iLW + γ iAB

(EQ 12)

where the subscript i refers to phases of interest, and γ iAB = 2 γ i+ γ i–

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

(EQ 13)

flotation0.book Page 143 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

143

one can rewrite Equation 11 to yield the following expression: γ 13 = γ 1 + γ 3 – 2 γ 1LW γ 3LW – 2 γ 1+ γ 3– – 2 γ 1– γ 3+

(EQ 14)

The values of γ 1+ and γ 1– of many hydrophobic substances (e.g., Teflon, polypropylene, and dodecane) are zero and therefore have high interfacial tensions. Surfactants are used to render various solids hydrophobic. Because the acidic ( γ 3+ = 25.5 mJ/m2) and basic components ( γ 3– = 25.5 mJ/m2) of water are fixed under most experimental conditions, the major role of a surfactant would be to decrease γ 1+ and γ 1– by blocking the hydrogen donor or acceptor sites of the solid surface. Laskowski and Kitchener (1969) noted that hydrophobicity arises when water molecules are prevented from forming hydrogen bonds with the polar sites on the surface of a solid. Pazhianur and Yoon (2003) conducted surface force measurements between silica surfaces treated with octadecyltrichlorosilane (OTS) using an AFM and compared the results with the changes in the surface free energies of the treated surfaces. The results are given in Figure 7, in which K (of Equation 35) represents the magnitude of a hydrophobic force measured. As shown, K increases with increasing advancing contact angle (θa) of the silica surfaces. The sharp increase in K at θa close to 90° may represent a change in orientation of the OTS molecule from flat to vertical orientation (Flinn, Guzonas, and Yoon 1994). As the contact angle increases in excess of 90°, cavitation may occur and cause a strong, additional non-DLVO attraction, providing an explanation for the change in slope of the K versus θa plot. It is also possible that the second inflection point of the plot is caused by a flat (or flipflop) orientation of the additional OTS molecules adsorbing at high surface coverages. It is important to note here that K increases with decreasing γ 1+ , γ 1– , and γ 1LW , as suggested by Equation 14. These findings are consistent with the work of Ederth (1999), who showed that both advancing and receding water contact angles on thiol-coated gold increase with increasing fraction of CH3 groups relative to that of OH groups (Figure 8). Note also that the surface free-energy data given in Figure 7 agree well with the approximate relationship between surface and interfacial tensions (γ1 and γ13) and cos θ shown in Figure 6. The interfacial tension (γ13) can be used to obtain the free energy of hydrophobic interaction (ΔG131) between two solid surfaces 1 in water 3 as follows: ΔG 131 = – 2γ 13

(EQ 15)

Hence, from Equation 11, one obtains the following expression: 2 ΔG 131 = – 2 ⎛⎝ γ 1LW – γ 3LW ⎞⎠ – 4 ⎛⎝ γ 1+ γ 1– + γ 3+ γ 3– – γ 1+ γ 3– – γ 1– γ 3+ ⎞⎠

(EQ 16)

in which the first term is small for hydrophobic solids (e.g., hydrocarbons or alkanes adsorbing on a solid surface), because the values of γ 1LW and γ 3LW are close to each other. If γ 1+ and γ 1– are small (i.e., a solid is apolar), the free energy of the hydrophobic interaction arises predominantly from the 4 γ 3+ γ 3– (= 102 mJm–2) term. On the other hand, if the value(s) of γ 1+ and/or γ 1– is large, ΔG131 becomes positive (i.e., the hydrophobic interaction vanishes), which means that water molecules form hydrogen bonds with the surface hydroxyl groups and render the surface hydrophobic. If a solid is basic (i.e., γ 1+ = 0, and γ 1LW = 40 mJm–2, the free energy of hydrophobic interaction vanishes at γ 1– > 28.3 mJm–2 (van Oss 1994). © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 144 Tuesday, January 2, 2007 7:36 PM

144

FLOTATION FUNDAMENTALS

45

40 10–16

γ1LW 35

10–17 25

K, Joules

Surface Free Energy, mN/m

30

γ1–

20

10–18 15

γ1AB 10 10–19 5

γ1+ 0 10–20 50

60

70

80

90

100

110

120

θ2 (Water)

NOTE: K is a parameter of a power law (Equation 35). γ1LW and γ1AB are the nonpolar and polar components of the surface tension of the solid surface, γ1, whereas γ1+ and γ1– denote acidic and basic components, respectively, of γ1AB.

FIGURE 7 Changes in surface free energy of OTS-treated silica plate in air contact as a function of the advancing water contact angle, and the changes in hydrophobic force constant, K, as a function of the water contact angle

100

θ

90

80

70 55

60

65

70

75

80

85

Solution Fraction C16, %

Source: Ederth 1999.

FIGURE 8 Advancing ( ) and receding ( ) contact angles of water with surfaces prepared from mixtures of C16 and C16OH-thiols of various mixing ratios

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 145 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

145

Thus, the hydrophobic interaction is seen to be related to the strong cohesive energy of liquid water. The free energy gain due to the hydrogen bond interactions among water molecules is large as compared to the interaction with the hydrophobic surfaces, eventually causing the water film tension to diminish as the thickness shrinks, which in turn is manifested as a hydrophobic attraction or a negative disjoining pressure. A question remains then as to how the hydrophobic force decays with the distance separating two macroscopic particles. Laskowski and Kitchener (1969) suggested that the multimolecular water layer on the surface of a hydrophobized silica is unstable, which is ascribed to a less favorable state of molecular association at a certain distance from the surface than in ordinary (bulk) water. These investigators were the first to recognize the existence of a long-range, non-DLVO hydrophobic force and to suggest that the long-range character arises from the structural properties of water. T H E R M O DY N A M I C A S P E C T S O F S U R FA C E F O R C E MEASUREMENTS

Consider an idealized thin film/surface force experiment using two plane-parallel, atomically smooth, laterally homogeneous hydrophobic surfaces with a thin water solution film between them (Figure 9). The thermodynamic variables of prime interest as determined in the adjacent bulk water solution are the temperature T and the solute chemical potential μS(or the concentration cS of solute). To establish equality of the chemical potentials everywhere, a lateral tension γf develops in the film which at large thicknesses of h will be equal to twice the interfacial tension of the hydrophobic surface–water solution interface (i.e., about 100 mNm–1). In order to keep the film at a certain thickness h, an extra pressure, positive (repulsive) or negative (attractive), the so-called (Derjaguin) disjoining pressure has to be applied. This pressure, πD, has alternatively been called surface “interaction pressure.”

πD

Water Solution

γf

πD

NOTE: The excess (interaction) pressure operating perpendicular to the film is, using Derjaguin’s terminology, the disjoining pressure, πD. In lateral directions, a film tension, γf, is acting. The surface force is defined as 2π times the difference in film tension between a thin film with interacting film faces and an infinitely thick film.

FIGURE 9

Schematic of a thin liquid film between two planar, solid surfaces

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 146 Tuesday, January 2, 2007 7:36 PM

146

FLOTATION FUNDAMENTALS

At constant temperature and pressure, the thin film system obeys the following thermodynamic fundamental equation to a very good approximation: – d γ f = Γ sf,ex d μ s + π D dh

(EQ 17)

which constitutes an extension of the classical Gibbs surface tension equation to cover the case of interacting solid surfaces separated by the film thickness h (compare to the appendix at the end of this chapter). The film thickness may most conveniently be defined as the separation between the hydrophobic surfaces devoid of any loosely adsorbed solute. This means that the Gibbs equimolecular dividing surfaces of the (hydrocarbon-covered) hydrophobic surfaces are being employed to delimit the water film in the direction perpendicular to the film faces. Furthermore, n sf,ex ≡ A Γ sf,ex (where A denotes the area of the film) is an excess quantity in the following sense: n sf,ex = n sf – n wf c s ⁄ c w

(EQ 18)

where n sf and n wf denote the actual mole numbers of surfactant and water in the film, respectively, and cs and cw are the corresponding bulk phase concentrations. In other words, n sf,ex represents the film content of the solute component in excess of what would be a corresponding slab of bulk solution containing the same amount of water as is present in the film. From Equation 17 it can be seen that the disjoining pressure πD is generally related to the film tension by the derivative γ-f⎞ ⎛ ∂-----= –πD ⎝ ∂h ⎠ T, p, μs

(EQ 19)

Hence, an attractive surface force, characterized by πD < 0, will result when the film tension γf decreases as the thickness diminishes, whereas πD > 0 means repulsion and a film tension that increases with decreasing h. In addition, Equation 17 includes an interesting Maxwell relation, f,ex

Γs ⎞ ∂π D⎞ ⎛ ∂-----------= ⎛ --------⎝ ∂h ⎠ T, p, μs ⎝ ∂μ s ⎠ T, p, h

(EQ 20)

stating that the amount of (surfactant) solute in the film will become less with decreasing thickness if the disjoining pressure tends to increase when raising the chemical potential of the solute. The film tension γf is formally an Ω-potential per unit area,

γ f = Ω ⁄ A ≡ G f,ex ⁄ A = G f ⁄ A – Γ wf μ w – Γ sf μ s

(EQ 21)

Thus, when equilibrium prevails at constant T, p, and, μs, the film tension γf is necessarily at a minimum. To derive the film tension change, Δγf, arising because of diminishing the thickness from h = ∞ where πD = 0 (i.e., beyond the range of the surface forces), Equation 19 can be integrated as follows while assuming that the bulk phase state will remain the same:

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 147 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

147

h

Δγ f = γ f ( h ) – γ f ( ∞ ) = –

π D dh

∫

(EQ 22)

h=∞

In practice, it is not feasible to employ a plane-parallel experimental setup as that indicated in Figure 9. Instead, one has to resort to sphere–sphere, sphere–plane, or crossed cylinder– cylinder configurations. At this point, the amazingly powerful Derjaguin approximation (1934) takes place that, in effect, constitutes a geometrical integration of Equation 19 (compare Figure 10), yielding an additional geometry-dependent factor such that the measured surface force F becomes equal to F = 2πRΔγf or sphere–plane and crossed cylinder–cylinder geometry and F = πRΔγf for sphere–sphere geometry. These simple relations effectuate the transformation from a curved to a planar geometry. They are valid for various surface interaction forces insofar as the ranges of these forces are much less than the inverse curvature of the surfaces involved. Although they are actually based on the very circumstance that (equilibrium) surface forces represent little else but film tension changes that arise when the interacting surfaces are brought close to contact, it has become common practice to report surface forces “normalized” in the form F/R, corresponding to 2πΔγf and πΔγf for these two standard geometries. Next, consider a pure water film between hydrophobic surfaces. Because of the unfavorable water–hydrocarbon contact combined with the necessity of attaining a uniform value of the water chemical potential everywhere, the local tangential pressure pT will have to

h1 = h + r 2/2R

R Sphere

r

h

Ring of Radius r Volume 2π drdh

dr dh

FIGURE 10 Diagram of the geometrical disjoining pressure integration involving the Derjaguin (1934) approximation

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 148 Tuesday, January 2, 2007 7:36 PM

148

FLOTATION FUNDAMENTALS

assume a variable negative value, corresponding to tension for each water layer located at x outside the solid surface. Hence p – p T ( x ) = [ p – p T ( 0 ) ] × exp ( – x ⁄ D )

(EQ 23)

Here, pT(0) denotes the tangential pressure next to the hydrophobic solid surface, whereas p represents the atmospheric pressure of the surroundings. Equation 23 assumes an exponential decay (decay length = D) of the thermodynamic effect caused by the presence of the surface as a function of the distance x from the surface. Equation 23 might constitute a reasonable first approximation, at least when the outer part of the surface zones begin to overlap (compare Marcˇelja and Radic 1976). Moreover, making the reasonable assumption that the surface force arising as the two hydrophobic surfaces are brought closer is entirely due to the elimination of surface-affected water molecules, integrate Equation 23 to yield (crossed cylinders): F ⁄ ( 2 π R ) = Δγ f = – const. × exp ( – h ⁄ 2D )

(EQ 24)

Hence, for the large separation part of a surface force versus thickness isotherm on this admittedly oversimplified basis, an exponential behavior of the (attractive) surface force with respect to the film thickness h can be predicted. Physically, this means that upon diminishing the film thickness, water molecules with slightly higher (Helmholtz) free energy than in the bulk are being released from the thin water film and transferred to the adjacent bulk water, thereby lowering the overall free energy of the film. Essentially the same mechanism (i.e., replacement of surface-located water) is responsible for the adsorption of surfactant and polymers at the air–water interface in the dilute Henry’s law region (Kumpulainen et al. 2005). In the following paragraphs, a more elaborate and more realistic treatment of the structural effect arising for the water in a thin film between ideal hydrophobic surfaces is considered. Diminishing film thickness from infinity to h = 0 will yield the adhesion force, as given by the Derjaguin–Muller–Toporov relation (for nondeformed interacting cylindrical surfaces, compare Israelachvili 1991): F adhesion ⁄ R = 2 πΔγ f ( h = 0 ) ≅ 4 πγ hc,w

(EQ 25)

The last of these relations assumes that there is no residual excess free energy associated with the touching surfaces in direct adhesive contact. Accordingly, the maximal adhesion force between fully hydrophobized solid surfaces in water is estimated to be 640 mN m–1. Evidently, 2γhc,w represents all the free energy per square meter that is gained in the planar case upon eliminating the hydrocarbon–water contact area. The additional 2πR-factor in Equation 25 originates from the Derjaguin approximation. However, when surface deformations are significant, the JKR ( Johnson, Kendal, Roberts) theory instead applies, which predicts a pull-off force equal to F adhesion ⁄ R ≅ 3 πγ hc,w

(EQ 26)

that is, a maximal pull-off force of about 480 mN m–1 for two crossed-cylindrical, hydrocarbon-covered surfaces (compare Israelachvili 1991). Pull-off forces refer to the forces measured when two surfaces in contact with each other are separated.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 149 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

149

The Ideal Hydrophobic Surface Versus Real Hydrophobic Surfaces

For experimental purposes, an almost-perfect hydrophobic surface and, hence, an “ideal” hydrophobic surface in the theoretical, philosophical sense is often realized. The question immediately arises about the most desirable properties of such a surface as to smoothness, hydrophobicity, molecular constitution, motional state, fluidity, stability toward water solutions and other media, and so forth. Regarding smoothness, the answer seems simple: a minimal roughness at the atomic level is desired, on average approaching 0.1 nm in peak-to-valley distance. First of all, this requires a smooth enough substrate; mica, silica, or glass is usually preferred. In addition, however, the hydrocarbon chains used as modifying agents must be attached in such a manner that the smoothness is maintained. Chemical reaction and polymerization schemes for the bonding of alkyl or silyl chains can be risky in this regard as they may involve strong mechanical pressures/tensions to operate in the surface during the reactions. Smoothness is likely to be a crucial factor for the occurrence of a long-ranged hydrophobic attraction force, and for this reason, it should be fully verified. The main experimental difficulty encountered in this context involves obtaining a sufficient degree of stability for the hydrophobic layer attached, and at the same time preserving surface smoothness at the atomic level. Concerning hydrophobicity, the answer may seem almost self-evident at first: maximal hydrophobicity calling for fluorocarbon rather than hydrocarbon chains. By making such an extreme choice, however, one may easily end up with studying the effects of vapor/air cavities and bridging bubbles rather than the hydrophobic attraction per se. For this reason, one might well prefer surfaces exhibiting a contact angle against water slightly less than 90° for which capillarity phenomena of this extraneous nature, in principle at least, should not occur. As to the molecular constitution, one would preferably desire the hydrocarbon chains employed to be sufficiently long (≥ C16 ), and have them fairly densely packed on the surface, to make up a certain minimum thickness of the resulting surface-modifying layer. Yet, forcing them to adopt a crystalline state with tilted chains may cause complicating domain structures and grain boundaries to arise. Concerning the motional state, one might prefer the hydrocarbon chains to be in a “semifrozen” solid (rather than close-packed crystalline) state, having a packing density between 0.22 and 0.25 nm2 per single chain. For a less crowded amphiphile monolayer in a liquid-expanded state, one runs the risk of the thermal motions making the water–hydrocarbon phase boundary fuzzy. Clearly, when ionic head groups are attached to the bonding sites of opposite charge on a solid surface for which the thermal amplitudes are small, the chains may end up in some “frozen” gel state, hampering the equilibration of the adsorbate, or at least making equilibration times exceedingly long. For mica and silica substrates, there seems to be a big difference between C16 and C18 cationic surfactants in this regard, the latter approaching adsorption equilibrium at an amazingly slow rate (Zhang et al. 2005). In summary, an ideal, stable, hydrophobic surface would exhibit an advancing contact angle against water of about 110°, and would in some solid state maintain a certain degree of molecular mobility but nevertheless be smooth on the atomic scale. The fairly high mechanical tension acting in the water adjacent to the hydrocarbon film will further tend to dampen the thermal amplitudes in the interface. If properly balanced in terms of its mechanical properties to withstand mechanical loads, and if smooth enough from the © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 150 Tuesday, January 2, 2007 7:36 PM

150

FLOTATION FUNDAMENTALS

outset for such a surface contact angle, hysteresis might be limited and the risk minimized for having bubbles attaching and bridging between approaching surfaces. Real surfaces that exhibit pronounced hydrophobic properties to a variable degree are listed as follows: 1. Cleaved mica with a deposited LB layer of a double-chain cationic surfactant (i.e., the hydrophobic surfaces employed by Claesson and Christenson [1988] and, more recently, by Meyer, Lin, and Israelachvili [2005] and Lin et al. [2005]) fulfils the quality criteria with respect to smoothness and motional state but shows lack of stability toward salt solutions and even toward passage through the water–air interface. One must realize that the chemical potential of the attached amphiphile is raised by compressing the monolayer spread on the Langmuir trough to the desired packing density of about 0.25 nm2 per single chain, which, by the way, matches almost exactly the anionic site density of the mica surface (≈0.50 nm2). Consequently, upon screening (by adding salt), there is a tendency for the surfactant to pass over to some less strained (bilayer) state. Further, the whole ion exchange process is prone to be facilitated kinetically by the presence of small ions. 2. A surface that is similar to point 1 but instead relies on adsorption from a cyclohexane solution to attach the double-chain cationic surfactant onto the mica. This scheme was successfully applied by Tsao, Evans, and Wennerström (1993) and Tsao et al. (1991), who obtained hydrophobized mica surfaces that were quite similar to those made according to point 1. In particular, by means of AFM they convincingly demonstrated that their surfaces were laterally homogeneous and free of defects over great distances, on the order of microns. Interestingly, a substantially smaller attraction force was recorded for C16 than for C18 chains, and significant temperature effects were noted, the attraction becoming weaker and less long-ranged at higher temperature. However, stability toward salt solutions was less satisfactory than for the corresponding LB monolayers prepared according to point 1. 3. Self-assembling alkylthiols dissolved in, for example, alcohol are known to bind strongly to gold surfaces, a circumstance that was utilized by Ederth, Claesson, and Liedberg (1998) to prepare hydrophobic surfaces on borosilicate glass substrates, which started by forming (by electron beam evaporation) a 1-nm titanium layer, followed by a 10-nm gold layer. The gold surfaces made in this manner are slightly rough with peak-to-through values of ~1.5 nm. The hydrocarbon chains become tilted and are tightly packed in a crystalline state. These surfaces exhibit a hydrophobic attraction of medium range but have been plagued by sporadic bridging bubble steps in the surface force curves. Stability is not a problem. 4. Glass, silica, or mica surfaces rendered hydrophobic by reaction with silanation agents (e.g., (tridecafluoro-1,1,2,2-tetrahydrooctyl)dimethylchlorosilane; Parker and Claesson 1994) and capable of reacting with surface OH groups. For mica surfaces, a water plasma pretreatment is necessary to introduce surface OH groups (Parker, Cho, and Claesson 1989). As a rule, stability toward salt solutions is obtained for this type of hydrophobic surface. Yet, to quote Christenson and Claesson (2001): “Experimental results with silylated surfaces have shown great variability depending on exact preparation conditions, and further underscored the critical connection between details of surface chemistry, surface morphology and the

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 151 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

151

measured forces.” Hence, it seems hardly feasible to fully control a silanation reaction scheme so as to generate a sufficiently smooth, charge-free, and laterally homogenous hydrophobic surface. 5. Wood and Sharma (1994) believed that one might succeed in making a more stable and better characterized hydrophobic surface by first polymerizing OTS at the air– water interface (pH = 2) on a Langmuir trough, followed by LB deposition and chemical grafting to a mica surface. Certainly, the surfaces obtained proved stable toward salt solutions but laterally inhomogeneous on the micrometer-length scale. No long-range hydrophobic attraction was observed, presumably because of the lack of molecular smoothness required. 6. Adsorption from water solution of a cationic surfactant such as CTAB or C18TAC (octadecyl-trimethylammonium chloride) onto mica, glass, or silica. Though easily prepared, these types of surfaces have a major drawback: they are uncharged (or carry a minimal surface charge) only at one particular surfactant concentration, often denoted as the charge neutralization concentration (CNC). Furthermore, for the most desirable chain length, C18, adsorption equilibrium is only approached very slowly at room temperature. Hence, in spite of the circumstance that the phenomenon of the long-range hydrophobic attraction is readily observed provided that the right surfactant concentration is chosen, fundamental studies are hampered by the variable state of the hydrophobic surfactant layer. Presumably, the best option is to concentrate on those surface force curves that show the strongest hydrophobic attraction rather than to try to sort out the exact electrostatic mechanisms that operate at concentrations different from the CNC. The temperature dependence of the hydrophobic force, for instance, could perhaps be investigated in this manner by changing the temperature, followed by adjustment of the surfactant concentration to recover a minimal surface charge. This listing, though certainly not complete, may serve as a ranking list of hydrophobic surfaces based on experiences from several laboratories during the past several decades. Preparing hydrophobic surfaces of sufficient quality is of crucial importance for making reliable conclusions about the strength, range, and nature of the hydrophobic force. Although hydrophobic surfaces prepared by adsorption of surfactants from solution are at the bottom of the list, they are probably the most important ones in terms of applications, particularly in flotation. Hydrophobic Attraction Forces Under Ideal Conditions

For large separations between two interacting hydrophobic surfaces, an exponentially decaying, negative tangential pressure component, pT, gives rise to a likewise exponentially decaying surface force. However, for closer distances between the two hydrophobic surfaces, a more elaborate model is necessary—in principle, similar to one by Cevs, Podgornik, and Zeks (1982)—to account for repulsive hydration forces. In this type of model, changes of state arising in the residual thin water film itself are also considered. A quasi-thermodynamic/structural model for hydrophobic attraction forces was presented in 1989 by Eriksson, Ljunggren, and Claesson (1989). In contrast to most of the other tentative explanations of hydrophobic force, this model has not been properly falsified in the Popper sense. Rather, it has been refuted by the vague claim that long-ranged structural effects are virtually excluded for (normal) liquids. However, with its ability to form a © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 152 Tuesday, January 2, 2007 7:36 PM

152

FLOTATION FUNDAMENTALS

variety of hydrogen-bonding patterns, water is still far from being a well-understood liquid at the molecular level. Hence, it is worthwhile to again scrutinize the basis of the waterstructure-based model and assess its predictive power by comparing it with the wider range of experimental results that are available today. After all, this model might serve as a valid point of departure for a deeper understanding of the hydrophobic force. Alternative attempts to fully understand the hydrophobic attraction will also be addressed. The premises are as follows: 1. The hydrophobic solid surfaces that interact in a liquid water medium are ideal: they are geometrically smooth to within about 0.1 nm at 0 K, they exhibit small thermal amplitudes at room temperature, and they are laterally close to being homogeneous in every respect. 2. Relatively speaking, there is a free-energy decrease associated with the molecular reorganization of the first monolayer of water molecules next to the hydrophobic surfaces because of minimizing the number of hydrogen bonds broken when contacting water with the hydrophobic surfaces. 3. The cooperatively enhanced tendency to avoid rupturing hydrogen bonds causes the surface-induced structure to be propagated (with a certain decay rate) toward the center of the thin film, resulting in a somewhat larger average number of hydrogen bonds per water molecule in the film than in the bulk. Hence, assuming only short-range interaction forces, there is a free-energy increase arising throughout the core of the thin film, owing to the imposed hydrogen-bond network formation. The final (inhomogeneous) water state in the film reflects a balance between the favorable molecular reorganization occurring in the first (contact) layer of water molecules and the induced, unfavorable restructuring of the remainder of the film. In the following discussion, these free-energy changes are accounted for by making use of a dimensionless order parameter s(x), which is a measure of the local increase of the free-energy density in the thin water film as compared with a corresponding slab of bulk water (x denotes the coordinate perpendicular to the thin film). Alternatively, it may be assumed that s(x) mirrors the local increase of the average number of hydrogen bonds per water molecule in the film, or the associated decrease in local density. Now imagine that the final equilibrium state of the thin water film sandwiched between two hydrophobic surfaces is reached in a stepwise fashion. Starting from a thin film cut out of the bulk state, the first step involves establishing the proper molecular interactions at hydrocarbon–water interfaces while retaining the average spherical-symmetric orientation of all the water molecules in the film. The second step implies a change of the packing and of the average orientation of the water molecules in each of the first molecular layers next to the hydrophobic surfaces to yield a less dense and more ordered molecular state with an increased number of hydrogen bonds and a preference for tangential alignment of the H–O–H bisectors of the water molecules. The film tension attained after these first two equilibration steps are denoted by γ 0f . The third step comprises a reorganization of the hydrogen-bond network throughout the film, whereby the parameter s becomes a function of the x coordinate and the final equilibrium value of γf is reached.

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 153 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

153

In an approach similar to the so-called square gradient approximation, frequently used in the past to model liquid–vapor interfaces, the model features invoked previously can be expressed in the following manner:

γ f = γ 0f – as 0 +

h⁄2

∫

[ c 2 s 2 + c 3 ( ds ⁄ dx ) 2 ] dx

(EQ 27)

–h ⁄ 2

The second term on the right-hand side of this expression (where a is a constant) accounts for an assumed linear free-energy reduction because of changing the order parameter s0 (and, hence, the packing density) for the water layers in direct contact with the hydrophobic surfaces, whereas the integral accounts for the free energy expense associated with structuring the core of the thin water film. Inclusion of the squared gradient term is essential because it furnishes a mechanism of energetic coupling between successive layers of water molecules. Hence, the constant c3 reflects the (average) tendency to cooperative structure generation. Upon minimizing the film tension γf while taking into account the proper boundary conditions, the following is readily derived cosh ( bx ) a s ( x ) = ⎛ ----------⎞ ---------------------------⎝ 2c 3 b⎠ sinh ( bh ⁄ 2 )

(EQ 28)

and

γ f ( h ) = γ 0f – ( a 2 ⁄ 4c 3 b ) coth ( bh ⁄ 2 ) = γ 0f – as 0 ⁄ 2 = γ 0f – ( B ⁄ 2 π ) coth ( bh ⁄ 2 )

(EQ 29)

The constant b stands for the quotient, whereas the interaction constant B introduced in Equation 29 can also be written in the form B = π a 2 ⁄ 8c 2 c 3 . For infinitely large film thicknesses, a lowering of the film tension caused by the imposed structuring due to the hydrophobic surfaces is obtained:

γ f ( h = ∞ ) – γ 0f = – B ⁄ 2 π

(EQ 30)

implying that B/4π is the corresponding reduction of the interfacial tension between water and a hydrophobic surface. It can be estimated to be ≈ 50 μNm–1, that is, to just about 0.1% of the total interfacial tension value of approximately 50 mNm–1. By taking the difference between Equation 29 and Equation 30 and making use of the Derjaguin approximation, the hydrophobic attraction force as measured by means of an SFA with cylindrically shaped hydrophobic surfaces having radii equal to R is given by the following expression: F ⁄ R = 2 πΔγ f = – B [ coth ( bh ⁄ 2 ) – 1 ]

(EQ 31)

For sufficiently large film thicknesses, the right-hand side of this equation becomes equal to –2Bexp(–bh), that is, in the (weak overlap) regime, ln (–F/R) is predicted to be a linear function of h, in line with Equation 23. Moreover, in this range, b–1 has the nature of a decay length. The assumption invoked to derive Equation 23 was that the removal of surface-perturbed water (due to the overlap of surface zones) is predominantly free energy-wise. On the other hand, Equation 31 is more general because it also accounts for the free energy changes arising © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 154 Tuesday, January 2, 2007 7:36 PM

154

FLOTATION FUNDAMENTALS

in the residual thin film itself due to the elimination of the outer parts of the water surface zones that are energetically perturbed by the hydrophobic surfaces. The major contribution to this part of the lowering of the film tension is an additional reduction of the number of broken hydrogen bonds for the contact layers of water molecules. As a consequence, the hydrophobic surface force increases in magnitude at an accelerating rate when the film thickness is diminished (compare Figure 17). Equation 31 can be placed in an approximate double-exponential form, namely F ⁄ R = – 2B ( e –bh + e –2bh )

(EQ 32)

where the second exponent accounts for the amplification of the surface force usually seen at medium separations. However, unlike Equation 31, Equation 32 does not extrapolate properly down to small h-values. Hence, the common practice of making use of a double-exponential expression entailing four parameters instead of just two is understandable, namely – F ⁄ R = C 1 e –h ⁄ λ1 + C 2 e –h ⁄ λ2

(EQ 33)

which, in many cases, has been shown to represent experimental surface force data satisfactorily in a wide separation range. The first exponent yields the largest contribution for small separations, whereas the second one dominates at large separations. Typical values derived are C1 = 0.2 Nm–1, λ1 = 2 nm, C2 = 1 mNm–1, and λ2 = 20 nm. Another convenient, though somewhat less accurate, manner to represent experimental surface force data is to invoke an expression of the same mathematical form as for the van der Waals attraction, that is, Δγ f = ( – K ⁄ 12 π )h –2

(EQ 34)

resulting in the one-parameter surface force expression: – F ⁄ R = ( K ⁄ 6 )h –2

(EQ 35)

Obviously, the strength of the hydrophobic attraction can easily be judged by comparing the value of the constant K with the corresponding value of the Hamaker constant, which is usually on the order of joules. Typical K values range between 10–18 and 10–19 J. The disjoining pressure expression, which can be derived from Equation 31 by differentiation with respect to h (compare Equation 19), is as follows: 2

π D = – ( bB ⁄ 2 π ) [ coth ( bh ⁄ 2 ) – 1 ]

(EQ 36)

Especially when displayed in this derivative mode, experimental hydrophobic interaction curves may appear to belong to two distinct regimes: below and above 10–20 nm, respectively (Figure 11). The same holds true for the order parameter s in the middle of the thin film (where x = 0), as derived theoretically. Below about 15 nm, s(x = 0) rapidly diminishes with the film thickness h, whereas above about 15 nm, s(x = 0) decreases very slowly in an almost linear fashion with h (Figure 12). The hydrophobic surface forces arising within the large separation range are generally rather weak, at most about 1 mNm–1, as compared with ≈500–600 mNm–1 for the maximum adhesion force at h = 0, and can thus easily be concealed (e.g., by a surface force of electrostatic origin). © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 155 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

155

106 DHDA DODA DEDA

–(2π)XπD, Nm –2

105

10

4

103 0

100

200

300

400

500

600

h, Å

FIGURE 11 Surface interaction pressures between mica surfaces modified with dihexadecyldimethylammonium (DHDA), DODA, and dieicosyldimethylammonium (DEDA) monolayers in water at 25°C as measured by Tsao et al. (1991). Note that the surface force curves are indistinguishable for DODA (C18 chains) and DEDA (C20 chains). 0.10

0.08

s(x=0)

0.06

0.04

0.02

0.00 0

5

10

15

20

25

30

35

h, nm

FIGURE 12 Order parameter s (x ) in the mid-plane of a thin water film sandwiched between hydrophobic surfaces (Eriksson, Ljunggren, and Claesson 1989)

Further, on the basis of Equation 36, it is estimated the excess free energy per water molecule in the middle of the water film where the pressure tensor is likely to be approximately isotropic. Typically, in the medium separation range, h ≈ 10 nm, this excess free energy amounts to about 4 × 10–4 kBT per molecule, as compared with the energy of a hydrogen bond at room temperature, ≈ 7 kBT, again recognizing the very minute thermodynamic effects by means of a sensitive SFA or AFM setup. According to the theory summarized, the strength of the hydrophobic attraction force is determined by the interaction constant B = πa 2 ⁄ 8c 2 c 3 , which in turn is strongly © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 156 Tuesday, January 2, 2007 7:36 PM

156

FLOTATION FUNDAMENTALS

dependent on the constant a that is related to the free-energy change associated with the reorganization of the contact layers of water molecules. Thus, B is anticipated to vary strongly with the degree of hydrophobicity as well as the smoothness on the molecular scale of the solid (or liquid) surface. On the other hand, the decay length b –1 = c 3 2c 2 should depend primarily on the properties of the water in the thin film, becoming large if structuring occurs readily (as for bulk water at the freezing point) or when there is a strong tendency to avoid rapid changes of the order parameter s(x), that is, when cooperativeness plays a significant role. However, if the hydrophobic surfaces are also charged and (overlapping) electrostatic double layers are present, one must anticipate competition between the (tangential) polarization of the water molecules due to the hydrophobic surfaces and the (perpendicular) dipole alignment in the electrostatic (mean) field. Such a coupling might result in a larger constant c2 and, hence, a shorter decay length. Salt effects for uncharged or nearly uncharged surfaces are expected to be rather minor, provided that the hydrophobic surfaces themselves are stable in contact with salt solutions. Generally, these features have been experimentally documented. In particular, note that the strength of the hydrophobic attraction scales semi-quantitatively with the contact angle for water on hydrophobic surfaces (Figure 6), and that Angarska et al. (2004) have shown that Equation 31 applies even for thin foam films at high salt concentrations. Furthermore, generalization of the water-structure-based theory to the case of an unsymmetric aqueous thin film between two different hydrophobic surfaces has been accomplished recently, an interesting case that has been studied experimentally by Yoon, Flinn, and Rabinovich (1997). In conclusion, it is seen that the quasi-thermodynamic theory due to Eriksson, Ljunggren, and Claesson (1989), which focuses on the rather minor free-energy effects that are associated with restructuring of water in contact with hydrocarbon surfaces, is capable of systematizing several experimental findings concerning the attractive hydrophobic surface force. Yet a fundamental problem related to this approach is that it does not provide an understanding as to why the effect, in some instances, can be of such an amazingly long range that it can be detected even for separations beyond 100 nm. One must bear in mind, however, that the long-ranged hydrophobic surface forces are extremely weak and represent very minute thermodynamic effects and that hydrogen-bonded networks and chains of water molecules are known to be cooperatively stabilized, that is, larger clusters are inherently more stable than smaller ones. This may set the stage for extended water clusters of various shapes to occur, provided that the associated free-energy expenses are small enough to be counterbalanced by corresponding size-fluctuation entropies, a thermodynamic scenario that is familiar from the field of surfactant aggregation. Nevertheless, there is definitely a need for more sophisticated models based on the concept of structure generation in water due to contact with a hydrophobic surface, as well as for novel experimental methods by which one can investigate the detailed state of thin water films. B U B B L E AT TA C H M E N T A N D C AV I T Y F O R M AT I O N AT H Y D R O P H O B H I C S U R FA C E S

Upon observing more or less distinct steps in the surface force curves for hydrophobic surfaces (plasma-treated mica silylated with (tridecafluoro-1,1,2,2-tetraoctyl)dimethyldichlorosilane) submerged in water at surface separations in the range of 100 nm, Parker Claesson, and Attard (1994) suggested that these steps might actually demark the onset of the hydrophobic

© 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 157 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

157

attraction, albeit an attraction that is totally different from the water-structure-based one discussed thus far. Hence, they attributed the attraction to tiny adhering air bubbles that, upon the approach of the hydrophobic surfaces, suddenly form bridging (quasi-cylindrical) bubbles. Although the concept of bridging bubbles is, in many ways, appealing, there are many difficulties associated with the claim that a capillary mechanism of this kind is the chief reason for the long-range effects observed. A primary difficulty is that very small air bubbles nucleated in a water phase are surprisingly short-lived. Because of the rather sizeable Laplace excess pressure, the air dissolves and the bubbles diminish in size at an accelerating rate. Straightforward calculations show that the lifetime of a small air bubble in water scales with the bubble radius squared. It is about 10 ms for a radius of 1 μm, and about 1 μsec for a 10-nm bubble, whereas a bubble that is millimeter-sized may subsist for days or even months (Epstein and Plesset 1950; Ljunggren and Eriksson 1997). Although air bubbles, as a rule, are generated upon contacting two hydrophobic surfaces in an SFA device filled with water and then followed by pulling them apart before starting the measurements, adhesion of small air bubbles is nonetheless a rare event because of the limited life span of these bubbles. Moreover, unlike the situation for larger air bubbles, which (because of the small excess pressure) can be treated as if they were thermodynamically closed, very small open air bubbles are not expected to adhere to an ideal hydrophobic surface in a stable manner (Kralchevsky 1996; Eriksson and Ljunggren 1999; Ryan and Hemmingsen 1993). Most real hydrophobic surfaces, however, contain defects that presumably play a crucial role in promoting bubble adhesion to occur to some minor extent. What remains, following the “bubble-mechanism” line of reasoning, is the option that a limited number of small bubbles, which have survived (because they started out as relatively large bubbles), happen to adhere in an irreversible, defect-dependent manner to the hydrophobic surfaces, and that a few of these adhering bubbles rather quickly form approximately cylindrical air bridges to the approaching hydrophobic surface. In this way, the excess (Laplace) pressure will be efficiently cancelled, and a bridging bubble can consequently subsist for a long time. To assume irreversibility for the wetting behavior is essential here. Otherwise, the formation of one large, air-filled cavity, eventually giving rise to a huge surface force, would have to be inferred. Simple calculations show that in order to account for the surface force steps observed by Ederth (1999), which amount to about 2 × 10–8 N in terms of force (rather than surface force), one must invoke an adhering (spherical cap) bubble that fulfils the contact angle condition, having a radius of curvature of about 70 nm. The surface area covered by such a bubble would be ≈ 1.3 × 104 nm2. Supposing the three-phase contact line to be pinned in a fixed position on some surface defects, the curvature is likely to decrease as some air dissolves, thus lowering the Laplace pressure and extending the expected lifetime of the bubble. Gas supersaturation in the surrounding water phase may likewise prolong the lifetime of an adhering bubble. As is easily confirmed, the formation of a bridging bubble out of an adhering bubble is likely to be advantageous free-energy-wise. Nevertheless, it is significant as well as understandable, in view of the short lifetime and submicroscopic size of such a bubble, that it has met with experimental difficulties to positively verify the presence of adhering air bubbles on hydrophobic surfaces (Lin et al. 2005; Mao et al. 2004). To judge from the evidence available today, the occurrence of adhering air bubbles depends strongly on the nature of the hydrophobic surface. This might be anticipated, of course, because non-ideal features in the hydrophobic surfaces themselves are likely to play a decisive role. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 158 Tuesday, January 2, 2007 7:36 PM

158

FLOTATION FUNDAMENTALS

R rc

FIGURE 13

h

Two approaching spheres with a bridging (quasi-cylindrical) air bubble

For a few years, it was a widely held belief that the long-ranged part of the hydrophobic attraction is an artifact that can be attributed to a capillary bridging mechanism of the previously mentioned character. More recently, however, several investigators have emphasized that the long-range hydrophobic attraction is actually present for several hydrophobic surfaces for which there are no signs of bubbles whatsoever. Hence, the supposition that small adhering bubbles which collapse into bridging bubbles is the major cause of the “true” longrange hydrophobic attraction, must be rejected. Strong additional support for this conclusion has recently been obtained from careful degassing experiments (Meyer, Lin, and Israelachvili 2005; Zhang et al. 2005). Conversely, large-scale cavity formation is a capillary phenomenon that is fairly well understood. It may occur when the surface tension of a solid surface in contact with air (or vapor) is less than the surface tension of the solid surface in contact with a liquid so as to make the cavity state the thermodynamically favored state. Expressed otherwise, it is a prerequisite that the (equilibrium) contact angle exceeds 90°. A somewhat simplified, though still sufficiently accurate version of the classical treatment presented by Yushenko, Yaminsky, and Shchukin (1983) is discussed in the following paragraphs. Consider the case of two approaching hydrophobic spheres (radii = R) submerged in water (Figure 13). Suppose a quasi-cylindrical (minimal-surface) cavity is formed that has almost no excess air pressure associated with it. The free-energy cost of forming such a cavity stems from forming the air–water interface, whereas the gain in free energy is due to replacing the hydrocarbon–water interface by hydrocarbon–air interface. For the free energy of cavity formation, the following approximate expression is obtained: ΔG ------- = r c ( h + r c2 ⁄ R ) γ air,w – r c2 ( γ hc,w – γ hc,air ) = γ air,w [ r c ( h + r c2 ⁄ R ) + r c2 cos θ eq ] 2π (EQ 37) where rc denotes the cylinder radius, and the r c2 ⁄ R term develops because the curvature of the spherical hydrophobic surfaces is taken into account to the first order. The expression in Equation 37 is apparently a third-order function of the cylinder radius rc with a minimum for a rather large value (where the contact angle condition is, in principle, fulfilled) and a maximum for some smaller rc value. The barrier associated with the maximum is present, however, only when the distance of approach, h, is greater than zero. (compare Figure 14). Provided that the equilibrium contact angle is large enough, the free-energy value at the minimum (representing the formation of a fully equilibrated air-filled cavity) will have a negative value, implying that the cavity state constitutes the thermodynamically stable state of the system. However, the intervening barrier will usually prevent the cavity from forming © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 159 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

159

1E – 15 1.0E – 09

θ = 105° 0

θ = 115°

∆G/2π, J

–1E – 15 0.0000

0.0002

0.0004

0.0006

0.0E + 00

θ = 105°

θ = 115° –1.0E – 09 0.0

0.1

0.2

0.3

0.4

0.5

rc , mm

FIGURE 14 Interfacial free-energy function according to Yushenko, Yaminsky, and Shchukin (1983), which accounts for the thermodynamically stable cavity formation between spherical, hydrophobic surfaces submerged in water. The inset shows the barrier at short separations that prevents cavity formation when the surfaces are brought close, though not touching, from a great distance.

when the thin film state is approached by gradually diminishing the surface separation. In other words, the thin film state of water between hydrophobic surfaces is actually a metastable state, as was noted earlier by Blake and Kitchener (1979). The cavity state can easily be realized, however, by first bringing the hydrophobic surfaces into contact, thus reducing the barrier to zero, then slowly pulling them apart. The attractive surface forces arising because of a large equilibrium cavity with an air pressure somewhat less than atmospheric pressure (undersaturation) may approach as much as about 1 N when the cavity is formed in an SFA between crossed cylinders (R = 1 cm). The disappearance of the cavity for some rather large h value is likewise readily understood on the basis of the previously mentioned simple equation. By increasing the surface separation h, the entire ΔG function is rotated counterclockwise. Eventually, the second (negative) term can no longer outweigh the first term, resulting in a minimum with a positive ΔG value, thus providing the impetus for cavity annihilation. In general, the experimental experience is in line with the thermodynamic description of cavity formation as outlined here; however, lack of complete equilibration for the air dissolved as well as complications arising because of contact angle hysteresis must be considered. Electrostatic Interaction Forces

To estimate repulsive electrostatic double-layer forces, the nonlinearized, mean-field Poisson– Boltzmann (PB) scheme is usually applied in some numerical version. The theory behind it assumes an evenly-smeared-out surface charge and a laterally homogeneous counterion distribution outside the charged (geometrical) surface. These assumptions seem more and more questionable when the surface charge density is smaller. At the same time, however, the magnitude of the repulsion becomes less, making the problems that may arise redundant in any case. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 160 Tuesday, January 2, 2007 7:36 PM

160

FLOTATION FUNDAMENTALS

The special case of one single (noninteracting) double layer is known as the Gouy– Chapman approach, according to which can be obtained the electrostatic free energy per charge, gel, from the following set of equations: g el = e φ 0 + γ el a charge

(EQ 38)

where γel is the electrostatic contribution to the surface tension, and acharge is the area per surface charge. This particular equation is, in fact, generally valid within the realm of the PB theory ( Jönsson 1981; Evans and Ninham 1983). For the Gouy–Chapman case, the surface potential is given by the following expression:

φ 0 = ( 2k B T ⁄ e ) [ S + ( S 2 + 1 ) 1 ⁄ 2 ]

(EQ 39)

and the electrostatic contribution to the surface tension is given by

γ el = ( – 2k B T ⁄ a charge ) [ ( S 2 + 1 ) 1 ⁄ 2 – 1 ] ⁄ S

(EQ 40)

where the negative sign indicates that this contribution is actually a surface pressure. Further, the dimensionless, reduced-charge parameter S introduced previously is defined by S = ( σ 2 ⁄ 8RT ε 0 ε r c t ) 1 ⁄ 2

(EQ 41)

where σ is the surface charge density, e the proton charge, εr is the relative dielectric number for water, and ct is the total (1:1) electrolyte concentration. In this context, the Debye length is introduced by means of the following expression (F denotes the Faraday constant): κ –1 = F –1 ε 0 ε r RT ⁄ 2c t

(EQ 42)

Hence, the Debye length, which characterizes the extension of the diffuse part of the double layer, according to the Debye–Hückel scheme, depends on the electrolyte concentration but not on the surface charge density. From this background, briefly consider the case of an (1:1) ionic surfactant adsorbed at an air–water interface in the dilute Henry’s-law regime. In this range, where γel constitutes the chief contribution to the reduction of the overall surface tension, the surface tension drops linearly with the surfactant concentration. Taking this circumstance into account and making use of the Gibbs surface tension equation, one can readily show that the gas law–like relation – γ el a charge = 2k B T

(EQ 43)

must hold exactly for thermodynamic reasons. Evidently, twice the –kBT/acharge can be attributed to the pressure effects of both anions and cations in the surface. On the other hand, in Equation 40, considering S to be >>1, about 1 kBT stems from the osmotic effect of the counterions, whereas ≈ 1 kBT stems from the polarization of the water in the electrostatic mean field (Ljunggren and Eriksson 1988). This discrepancy immediately conveys that the current theoretical description of electrostatic interactions has its weak points. Furthermore, from Equation 40 it appears that the Gouy–Chapman theory incorporates an additional factor [(S2 + 1)1/2 – 1]/S, which deviates substantially from unity, especially when S becomes less than about 10, which is the case for dilute ionic surfactant © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 161 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

161

layers. In other words, in spite of its many virtues, the Poisson–Boltzmann description has a serious flaw, as it does not show a correct limiting behavior for charged equilibrium (Gibbs) monolayers. This is in sharp contrast with, for example, the Flory–Huggins theory of polymer solutions. The main reason behind these deficiencies is probably related to both the mean-field concept as such and, in addition, to the particular model feature of a geometrical (i.e., volumeless), evenly charged surface devoid of entropy of mixing for the charged groups with water. This latter shortcoming also shows up in the fairly low surface potentials and charge densities that are commonly derived from surface force data. Nevertheless, in the PB scheme, the electrostatic disjoining pressure is primarily given by the following expression (for simplicity, no added salt):

π Del = c ion RT ( e e φm ⁄ kB T + e –e φm ⁄ kB T – 2 )

(EQ 44)

where φm represents the mid-plane potential and cion for the bulk concentration (moles per cubic meter) of positive and negative ions. In particular, in the much-employed weak-overlap approximation that is applicable insofar as the surface potential is small (< 25 mV),

π Del ≈ ( ε r ε 0 κ 2 ⁄ 2 ) φ m2 ≈ 64k B Tc s Γ 02 e –κ h

(EQ 45)

Γ 0 = tanh ( eφ 0 ⁄ 4k B T )

(EQ 46)

where

By integrating Equation 45, the corresponding contribution to the surface force is easily found: ( F ⁄ 2 π R ) = 64k B Tc s Γ 02 e –κ h ⁄ κ

(EQ 47)

implying an exponential asymptotic behavior determined by the (salt-dependent) Debye length. Thus, for large separations, there is a formal similarity (though the sign is opposite) between the electrostatic repulsion and the water-structure-dependent hydrophobic attraction treated previously. Moreover, the electrostatic surface force contribution can readily match the hydrophobic contribution in strength, provided that there is a significant surface charge. Still, a major difference from the theoretical standpoint is that the electrostatic doublelayer repulsion can be rationalized by referring to the well-established osmotic effect of dissolved (point) ions in a structureless solvent medium, whereas up to this point, a firm fundamental basis is lacking for an explanation of the hydrophobic attraction in terms of the molecular properties of water. In the 1970s, it was realized that correlation effects may become significant, especially for interacting double layers encompassing divalent counterions (Guldbrand et al. 1984; Kjellander and Marcˇelja 1984). An attraction of this origin is conceptually similar to the dispersion interaction because it relies on instantaneous, laterally inhomogeneous counterion distributions outside the charged surfaces. On average, the fluctuations cause more weight to be given to attractive rather than repulsive configurations, resulting in an overall attraction. On this basis, the large deviations from PB theory for charged systems with divalent ions have been accounted for while at the same time verifying that the same PB theory as a rule is reasonably accurate for systems containing monovalent ions. © 2007 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2009.

flotation0.book Page 162 Tuesday, January 2, 2007 7:36 PM

162

FLOTATION FUNDAMENTALS

For the correlation interaction between surfaces carrying mobile charges, the Ninham– Parsegian expression holds true for the disjoining pressure. At large separations, it takes on the following limiting form: 3

κ π D = – kT ------------- e –2 κ h κh

h » κ –1

(EQ 48)

The magnitude of the attractive interaction pressure predicted by means of this equation is, however, on the order of the van der Waals attraction. Hence, it is evident that charge correlation attraction arising at the molecular (ion) length scale is not sufficiently strong to rationalize the experimentally measured hydrophobic attraction. This was realized by Forsman, Jönsson, and Åkesson (1998), Miklavic et al. (1994), and Kekicheff and Spalla (1995), who argued that more sizeable contributions of this nature can be obtained for charged surfaces that are heterogeneous, containing (mobile) charged surfactant aggregates of some size, such as, for example, hemimicelles. In particular, Miklavic et al. (1994) were able to demonstrate that at relatively high salt concentration, the correlation attraction should decay with a decay length equal to κ–1/2(i.e., half of the Debye length) upon increasing the surface separation. However, their calculations presuppose that the surface density of adsorbed micellar aggregates is basically independent of the concentration of added salt, a condition that for the most part is unlikely to be fulfilled. Although these concepts about ionic correlation attractions may be of some relevance for hydrophobic surfaces formed by adsorbing (e.g., a water-soluble cationic surfactant onto mica, glass, or silica), it seems unacceptable to employ them to account for the archetypal long-ranged hydrophobic attraction documented for the LB-monolayer-modified mica surfaces prepared by Claesson and Christenson (1988) or the corresponding mica surfaces prepared by adsorption from cyclohexane solution by Tsao et al. (1991), at least insofar as no salt is added when making the surface force runs. The surface characterizations carried out as well as the preparation protocols employed seem to leave little room for speculation in this direction. In a similar vein, it was suggested by Tsao, Evans, and Wennerström (1993), and Yoon and Ravishankar (1996) that correlations among colloidal-grained, ordered dipole domains across the thin film may give rise to sizable attractions, far stronger than the van der Waals interaction. Although these matters, perhaps, are not yet resolved, it seems implausible that all observations regarding the hydrophobic attraction can be accounted for on the basis of correlation attractions alone. Returning to the subject of hemimicelle formation, an aggregation of this kind is anticipated for hydrophobic surfaces prepared by surfactant adsorption from water solution, starting at the critical hemimicelle concentration where the hydrophobic attraction is already high. Herder (1990) demonstrated that upon adsorbing a cationic surfactant onto Langmuir–Blodgett–DODA-modified mica, the hydrophobic attraction rapidly vanishes and is replaced by electrostatic repulsion. Furthermore, Craig, Ninham, and Pashley (1998) thoroughly investigated the interaction between silica surfaces in dilute CTAB and cetylpyridinium chloride (CPC) solutions in the presence of electrolytes, and firmly concluded that a direct electrostatic mechanism for the hydrophobic attraction is hardly an option. Parker and Claesson (1994) arrived at the same conclusion using the MASIF setup and silanated glass spheres.

flotation0.book Page 163 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

163

S U R FA C E F O R C E D ATA S U P P O R T I N G T H E WAT E R STRUCTURE MECHANISM

After about two decades of research efforts devoted to searching for the mechanism that gives rise to the hydrophobic attraction, it is probably time to turn back to where the surface–force-based development phase started—the measurements by Israelachvili and Pashley (1982). These authors made use of mica surfaces submerged in dilute CTAB solution and interpreted the results rather straightforwardly in terms of the hydrophobic effect. In the 1960s, the “mysteries” of water were still fashionable topics for study. Much inspiration came from Pauling’s (1960) investigations of the hydrogen bond and from the fascinating field of clathrate physical chemistry (Franks 1973). Later, the tendency to restrict the discussion to structural aspects only—instead of employing the full statisticalmechanical machinery—and the polywater affair made this research direction fall in disrepute. More recently, however, it has become abundantly clear that the molecular understanding of liquid water is incomplete to an extent that renders it virtually excluded for making reliable theoretical predictions about (transient) H-bond-generated structures. In order to cope with the long-ranged hydrophobic attraction, a few different approaches than the water-structure method have been rather thoroughly scrutinized. In particular, electrostatics beyond the standard PB mean-field description have been reexamined, and more trivial, though quite cumbersome, capillary effects have been investigated in detail. The latter phenomena include the formation of cavities and bridging bubbles between (hysteretic) hydrophobic surfaces submerged in water. In the end, however, both of these major alternative approaches have turned out to be untenable. To further reveal the extent to which the water–structure-based theory of the hydrophobic force compares with experimental data, perhaps the most illuminating and most well-documented experimental surface force study carried out so far will be examined—one that clarifies the hydrophobic attraction (presented by Tsao et al. [1991]). These investigators made use of hydrophobized mica surfaces (mounted in an SFA), prepared by means of adsorption from cyclohexane solutions of cationic surfactants of different chain lengths: DHDA, DODA, and DEDA, with bromide or acetate as the counterions. The resulting adsorption layers were characterized by AFM. Hence, it was verified that the double-chain surfactant cations employed are electrostatically bonded, one to each anionic mica site, resulting in a packing density close to 0.50 nm2 (0.25 nm2 per single chain), and that they remain quantitatively bonded to their sites (though in a metastable state) even after contacting the surface with water and raising the temperature to 50°C. Full stability toward salt solutions was, however, not achieved using this surface preparation method. Moreover, the aforementioned authors demonstrated that while the DEDA monolayer preserves a frozen chain state at 50°C, this is not the case for the DHDA and DODA monolayers. In fact, DHDA appears to be present in a melted chain state already at 25°C, whereas DODA melts in the range between 40° and 50°C. Correlating changes were observed in the surface force curves, and it was concluded that a well-organized, smooth hydrocarbon monolayer of frozen hydrocarbon chains causes the strongest hydrophobic attraction. This finding was later corroborated by Rabinovich, Guzonas, and Yoon (1993) by means of Fourier transform infrared (FTIR) measurements. The chain length primarily matters in the sense that it determines where on the temperature scale chain melting occurs. Moreover, for one and the same temperature, the decay length at large separations was found to be nearly the same.

flotation0.book Page 164 Tuesday, January 2, 2007 7:36 PM

164

FLOTATION FUNDAMENTALS

These observations fit well into the theoretical model description presented previously that presumes structural water effects. Accordingly, the strength of the hydrophobic force is determined primarily by the constant a of the (linear) response function –as0 , which participates in the expression for the constant B in Equation 31:

πa2 B = ---------------8c 2 c 3

(EQ 49)

In view of the cooperative nature of water structure generation, it seems very likely that a stronger response (larger a) in terms of the lowering of free energy should result for the contact monolayer insofar as it is fairly unperturbed by the thermal motions of the hydrocarbon surface. This dependence may suffice to account for the distinctly different surface force behaviors recorded by Tsao et al. (1991) for hydrocarbon surfaces in a frozen (larger B) or melted (smaller B) state. Similarly, raising the temperature should tend to make a smaller, and, in addition, make the structuring of the film core more costly in terms of free energy, yielding a higher c2, that should likewise tend to diminish B. Also, the decay length b–1, which is given by c 3 ⁄ 2c 2 , depends inversely on c2 and should thus diminish with temperature, which is for the most part observed. Concerning c3, just a modest dependence on temperature is anticipated. On the other hand, the (unintentional) presence of a polarizing electrostatic field would tend to increase c2 and hence shorten the decay length, whereas the presence of inert gas molecules in the form of clathrate guest molecules might conceivably make it less costly to restructure the water film core, implying a smaller c2 and a longer decay length. More importantly, however, a virtually model-independent, surface-thermodynamic analysis of surface force data as complete as those of Tsao et al. (1991) can be carried through. Revisit the thermodynamic fundamental equation governing the present film case (see appendix to this chapter). For a pure water film in contact with water, d γ f = ( V f,ex ⁄ A )dp – ( S f,ex ⁄ A )dT – π D dh

(EQ 50)

where Vf,ex represents the volume in excess of the volume of a slab of bulk water that contains the same number of water molecules as the film of a certain thickness. The excess entropy Sf,ex is analogously defined. For an infinitely thick film, accordingly, obtain the following: d γ f ( ∞ ) = ( V ∞f,ex ⁄ A )dp – ( S ∞f,ex ⁄ A )dT

(EQ 51)

Upon deducting Equation 51 from Equation 50, next obtain d ( Δγ f ) = ( Δ V f,ex ⁄ A )dp – ( Δ S f,ex ⁄ A )dT – π D dh

(EQ 52)

This equation is the appropriate thermodynamic relation for dealing with surface force data for thin films consisting of pure water. It obviously includes the partial derivative ( Δγ f )-⎞ Δ S f,ex ⎛ ∂--------------= – ------------⎝ ∂T ⎠ p, h A

(EQ 53)

flotation0.book Page 165 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

165

6 DEDA (s) and DODA (s) at 25°C DEDA (s) at 50°C DODA (s) at 40°C DHDA (l) at 25°C DHDA (l) at 40°C

5

–F/2π R, mJm –2

4

3

2

1

0 0

50

100

150

200

250

300

350

400

h, Å

FIGURE 15 Film tension changes for thin water films between hydrophobized mica surfaces at different temperatures derived from the disjoining pressure data presented by Tsao et al. (1991)

Thus, if the surface force increases with temperature, which in the present case means that the attractive hydrophobic surface force becomes less negative when the temperature is raised, it necessarily results in ΔSf,ex as a negative quantity. In turn, this evaluation result would imply that the thin water film is actually more ordered than an imaginary water film of the same thickness encompassing two noninteracting surface zones of water next to the hydrophobic surfaces. Provided that the hydrophobic surfaces themselves do not suffer any changes in their intrinsic thermodynamic properties as the film thickness varies (which should not be the case, especially if the surfaces are solid), there will be no contribution to ΔSf,ex other than the one arising in the water film. By integrating the disjoining pressure functions derived experimentally by Tsao et al. (1991), the corresponding relative film tensions can be obtained (compare Figure 15). These curves display the crucial feature discussed previously, that is, upon diminishing the surface separation, the film tension reduction is larger at room temperature than it would be if the room temperature were greater. Consequently, as soon as the surfaces interact, ΔSf,ex is always a negative quantity. Also, Equation 31 with B-, b- values generated from the measurements of Claesson and Christenson (1988; who used similarly prepared hydrophobic surfaces), yields a reasonably good fit for the film tension curve obtained by Tsao and colleagues for DODA at room temperature. The relative excess film entropies per unit area, ΔSf,ex/A, can be quantified using Equation 53, resulting in the curves shown in Figure 16. Evidently, the excess entropy becomes increasingly more negative when the water film gets thinner, especially for the frozen hydrocarbon surfaces, DODA and DEDA. Regarding the numerical values, it is illuminating to compare with the entropy reduction one would get for a corresponding slab of bulk water that undergoes freezing to ice. By invoking the entropy of fusion of ice (≈ 22 J mol–1 K–1) for a 5-nm water film, estimate f ΔS freezing = – 6 mJm –2 K –1 , which is about 30 times more than the entropy reduction quantified from the surface force measurements. In other words, the enhanced molecular order in the thin film is equivalent to introducing about 3% of ice ordering.

flotation0.book Page 166 Tuesday, January 2, 2007 7:36 PM

166

FLOTATION FUNDAMENTALS

250

DODA (s)

–∆S f,ex/A , μJm–2K–1

200

150 DEDA (s) 100

50

DHDA (l) 0 0

50

100

150

200

250

300

h, Å

FIGURE 16 Excess film entropies as functions of the separation between mica surfaces hydrophobized with DHDA, DODA, and DEDA. The difference between the DODA and DEDA curves is presumably due to experimental uncertainty.

Furthermore, comparing with the relative film tensions shown in Figure 15, observe that for every thickness h, it holds true that

Δγ f 0; that is, the air gases are enriched in the thin water film between hydrophobic surfaces in comparison with an infinitely thick film that was scaled down to the same thickness. By applying Equation 61 to films of 10 to 20 nm thickness, as examples, the film excesses of N2 and O2 are about the same as the amount of dissolved gases in corresponding (imaginary) thin films made up of bulk solution. This observation is in fair agreement with the idea of an enhanced water structure in thin water films delimited by hydrophobic surfaces. The solvent properties of the water change as additional water cage volume becomes available to the (clathrate-forming) gas molecules, similar to when the temperature is reduced to below room temperature. Relatively speaking, the higher content of the gas solutes for thin films will cause the film tension to become more negative in accordance with Equation 61. Conversely, deaeration should tend to diminish the magnitude of the attractive hydrophobic force, as observed by Meyer, Lin, and Israelachvili (2005). Moreover, it seems most likely that the effects of deaeration documented by Pashley (2003) on the stability of colloids and emulsions can also be understood, at least in a preliminary way, on this basis. Referring to Equation 58, an analogous equation for ΔΓ sf,ex is ∂ ln B Bbh ∂ ln b ⁄ d μ s ⎞ ΔΓ sf,ex = – ------------ Δγ f – --------- ⎛⎜-----------------------------⎟ 4π ⎝ ∂μ s 2 sinh ( bh ⁄ 2 ) ⎠

(EQ 62)

which should be relevant insofar as the solute concentration is kept low and the extent of solute adsorption is limited on the hydrophobic surfaces. From this expression, an increase in ΔΓ sf,ex , such as found in the case under discussion, is strongly coupled with the constant B becoming larger as the chemical potential μs is raised: the H-bond network in the film is reinforced by introducing inert gas molecules. Concerning the addition of alcohols and surfactants that readily adsorb on hydrophobic surfaces, in the dilute regime the situation is principally opposite: The surface force rises with μs, yielding a negative ΔΓ sf,ex that, according to the quasi-thermodynamic relation (Equation 62), can be interpreted in terms of a smaller constant B and a diminishing decay length b–1. This is true for alcohols as well as nonionic surfactants. For a contact monolayer mixed with such surface-active species, there is simply less free energy to be gained by

flotation0.book Page 171 Tuesday, January 2, 2007 7:36 PM

THE NATURE OF HYDROPHOBIC ATTRACTION FORCES

171

0.10

0.05

F/R, mN/m

0.0

–0.05 Water –0.10

12.5% Ethyl Alcohol 20% Ethyl Alcohol

–0.15

–0.20 0

10

20

30

40

50

60

Separation, nm

FIGURE 19 Surface force isotherms presented by Ederth (1999) for water and ethanol–water mixtures between hexadecanethiol surfaces, showing the pronounced effect on the hydrophobic attraction of adding alcohol. The advancing contact angles are 94° (12.5%) and 88° (20%), respectively.

restructuring (smaller a). Evidently, for ionic surfactants, the electrostatic (double layer) repulsion must also be invoked. Surface force curves for water and ethanol–water mixtures recorded by Ederth (1999) using C16-alkylthiolated gold surfaces are reproduced in Figure 19. It appears that the surface force is substantially reduced in magnitude by adding alcohol. Using Equation 61, estimate ΔΓ sf,ex to –4 × 1016 molecules m–2 for a 10-nm-thick film in contact with ≈16% alcohol solution to be compared with the alcohol content of a corresponding bulk solution film: ≈ 2 × 1018 m–2. The reason why the surface force diminishes as alcohol is added is that the lowering of the film tension γf is larger for an infinitely thick film than for a thinner film. In turn, this is because adding alcohol to the thin film counteracts the favorable structure formation. On the other hand, adding salts should not change the situation very much with respect to the hydrophobic force (provided that the metastable attachment of hydrocarbon chains remains unaffected), the chief reason being that small ions do not mix with the water in the contact monolayers adjacent to the hydrophobic surfaces. Hence, the B constant should be left much the same, and likewise b–1, insofar as the ions do not interfere significantly with the H-bond network formation. This agrees with a multitude of observations using hydrophobic surfaces that are sufficiently stable. Hydrophobic Forces in Flotation

In flotation, hydrophobic particles are selectively collected on bubble surfaces and separated from the hydrophilic particles suspended in aqueous slurry. Thermodynamically, the bubble– particle adhesion occurs when contact angle of the particle is larger than zero. In view of their high interfacial tensions in water, air bubbles should be considered hydrophobic. Thus, the bubble–particle interaction occurring during flotation may be seen as a hydrophobic interaction, the kinetics of which is controlled by the surface forces involved, namely electrostatic, van der Waals, and hydrophobic forces. The electrostatic forces are repulsive when

flotation0.book Page 172 Tuesday, January 2, 2007 7:36 PM

172

FLOTATION FUNDAMENTALS

both particles and bubbles are negatively charged. The van der Waals forces operating in wetting films are also repulsive according to the Lifshitz theory. Therefore, it is difficult to explain flotation without considering the existence of the attractive hydrophobic force discussed in this chapter. In flotation, the role of the hydrophobic force is to reduce the energy barrier so that the bubble–particle interaction becomes a “fast heterocoagulation” process. The process can be “slow” if the hydrophobic force is too small to counterbalance the repulsive electrostatic force. It is well known that flotation rate increases with increasing particle hydrophobicity and decreasing double-layer potentials. In general, surface forces are weaker than the hydrodynamic forces operating in a flotation cell by orders of magnitude. As a particle approaches a bubble in close proximity, however, the hydrodynamic force is much reduced because of the hydrodynamic resistance against film thinning. The hydrodynamic forces become comparable to the surface forces as the separation distance between bubble and particle reaches the critical rupture thickness of the wetting film between the two surfaces. By assuming that the probability of the bubble– particle adhesion is determined by the hydrodynamic and surface forces, a flotation rate equation has been developed both under quiescent and turbulent flow conditions (Schimoller, Luttrell, and Yoon 1994; Yoon and Mao 1996; Mao and Yoon 1997; Do and Yoon 2005). This approach made it possible to develop a model that can predict flotation rates using both hydrodynamic and surface chemistry parameters. Hydrophobic forces also play a role in particle–particle interactions. Hydrophobic particles coagulate at a pH well above the isoelectric point, which cannot be explained without assuming the existence of a hydrophobic force (Xu and Yoon 1989, 1990). The hydrophobic coagulation, which is driven by the hydrophobic force, also plays an important role in flotation. In general, flotation rate decreases with decreasing particle size. Gaudin (1957) showed, however, that the flotation rate of galena particles stayed constant at particle sizes 160 Molecules per Aggregate

Source: Chandar, Somasundaran, and Turro 1987.

FIGURE 31 Schematic representation of the correlation of surface charge and the growth of aggregates for various regions of the adsorption isotherm depicted in Figure 29

likely to occur through the growth of existing aggregates rather than the formation of new ones. Here, the adsorption goes up by a factor of at least 5, while the aggregation number goes up by about 2, so there must be about 2 1/2 times as many aggregates. The factor of 2 in aggregate size indicates that adsorption possibly goes from a patchy monolayer (head facing alumina) to a patchy bi-layer (one head facing solution, the other facing alumina). This is possibly due to better gain in energy by its hydrophobic effect between the hydrophobic tails of the already adsorbed surfactant molecule and the unadsorbed ones. Such a situation can be expected to result in a reverse orientation of the surfactant molecules as illustrated in Figure 31, where the whole process of adsorption has been schematically portrayed. These studies on the adsorbed layer of SDS on alumina further confirm the earlier concepts of hemimicellization. Surfactant aggregation occurs above a critical concentration referred to as “hemimicellar concentration,” which is marked by a sharp increase in the adsorption isotherm, eventually leading to the formation of highly organized and finite size assemblies even at relatively low surface coverages (Chandar, Somasundaran, and Turro 1987). Electron Spin Resonance

The ESR spectroscopic technique deals with transitions induced between Zeeman levels of a paramagnetic system situated in a static magnetic field. Only species with a magnetic moment are capable of interacting with the magnetic field. Three types of ESR studies may be applied to the micellar systems. They are spin-probing, spin-labeling, and spin-trapping techniques. In the spin-probing technique, a molecule with a spin is externally added to the system, whereas in the spin-labeling technique, a spin-bearing moiety through covalent bonding forms a part of the molecule. The spin-trapping technique is mainly applied in the

flotation0.book Page 209 Tuesday, January 2, 2007 7:36 PM

ADSORPTION OF SURFACTANTS

209

identification of radicals produced thermally, photochemically, or radiolytically by trapping the radical through chemical reactions with a spin-trap and converting the radical into a stable free radical to be examined by ESR. In the study of the alumina–SDS system, stable free-radical nitroxide spin labels were chosen for use of ESR spectroscopy. These spin labels (in micromolar levels) were co-adsorbed individually on alumina along with the SDS (Waterman et al. 1986). As shown in Figure 32, the shape of the adsorption isotherm of SDS on alumina with probe is similar to that in the absence of the probe, except at low SDS concentrations where an enhancement in SDS adsorption is observed because of the synergistic co-adsorption of the surfactant with the probe. The ESR spectrum of 16-doxylstearic acid probe in aqueous solution shows the typical isotropic three-line spectrum characteristic of the nitroxide. The spectra obtained from the alumina–water interface, on the other hand, are distinctly different from the solution spectrum, with three types of ESR spectra. At low SDS concentrations ( covellite > chalcopyrite, which indicates that the separation by flotation of the arsenic minerals of tennantite and enargite from the other copper minerals based on differences in surface oxidation is only possible if chalcocite is absent from the mineral mixture. Other Surface-Sensitive Analytical Techniques to Measure the Oxidation of Sulfide Minerals

Oxidation and reduction reactions occurring on the surface of sulfide minerals may be identified by cyclic voltammetry (Gardner and Woods 1979; Hamilton and Woods 1984). Figure 10 shows a typical voltammogram of a chalcocite electrode conditioned with oxygen at pH 8.5 and 10.8. At pH 8.5, the first major peak at +0.08 V (vs. the standard calomel electrode, or SCE) in the anodic scan has been assigned to the oxidation of the chalcocite mineral to form covellite and also copper hydroxide (Roos, Celis, and Sudarsono 1990). Cu 2 S + 2H 2 O ↔ CuS + Cu ( OH ) 2 + 2H + + 2e –

E h = +0.06 V (vs. SCE) (EQ 5)

The second major peak in the anodic scan is present at +0.21 V (vs. SCE) and has been identified as the oxidation of covellite (Reaction 3). The covellite (CuS) is present in the chalcocite sample but is also formed during the previous oxidation reaction (Equation 5) (Roos, Celis, and Sudarsono 1990). CuS + 2 H 2 O ↔ Cu ( OH ) 2 + S + 2H + + 2e –

E h = +0.12 V (vs. SCE)

(EQ 6)

flotation0.book Page 242 Tuesday, January 2, 2007 7:36 PM

242

FLOTATION FUNDAMENTALS

Chalcopyrite

40

Bornite

20

0

–20 H2O2 Zeta Potential, mV

O2 –40

O2 N2

N2 Chalcocite

40

Covellite

20

0 O2 –20 O2

–40 N2

N2

–60 4

6

8

10

4

6

8

10

12

pH NOTE: Filled and empty symbols refer to a pH change from high to low pH values and from low to high pH values, respectively (the arrows show the direction of pH change).

FIGURE 9 Zeta potential versus pH curves of copper sulfide minerals conditioned at pH 11.0 for 20 minutes in nitrogen (N2), for 60 minutes in oxygen (O2), and for 60 minutes with hydrogen peroxide (H2O2)

On the cathodic scan, the double peaks at –0.10 and –0.16 V correspond to the reduction of the oxidation products formed at 0.21 and 0.08 V, respectively (Reactions 5 and 6). A similar voltammogram is observed at pH 10.8. The only difference is the shift of all the peaks by approximately –0.16 V compared with those at pH 8.5. Also, the –0.08 V anodic peak is very broad, especially on its more positive side, and hides the smaller peaks observed at +0.08 and +0.21 V at pH 8.5. A good understanding of sulfide mineral oxidation has been obtained by combining several analytical techniques such as cyclic voltammetry and XPS (Buckley, Hamilton, and Woods 1985). XPS is, by far, the most used surface analytical technique for the study of sulfide mineral oxidation, because it can more directly measure the type and proportion of surface species. It has been able to confirm the mechanism of oxidation through Reactions 1 to 4, as a function of pulp conditions such as pH and dissolved oxygen (DO) content (Buckley and

flotation0.book Page 243 Tuesday, January 2, 2007 7:36 PM

PULP AND SOLUTION CHEMISTRY

243

1

3

0

Current, mA

pH = 10.8

3

2 1 0

1 1

pH = 8.5 3

–0.85

–0.65

–0.45

–0.25

–0.05

0.15

0.35

Potential, V (vs. SCE)

FIGURE 10 Voltammograms of chalcocite conditioned in oxygen at pH values of 8.5 and 10.8 (1, 2, and 3 refer to Reactions 3, 4, and 5, respectively)

Woods 1983, 1984; Buckley and Walker 1988; Buckley, Woods, and Wouterlood 1989; Buckley and Riley 1991; Fornasiero et al. 1994, Fairthorne, Fornasiero, and Ralston 1997). As an example, Figure 11 shows the XPS spectra of chalcopyrite conditioned at pH 5.0 or 9.5 and with nitrogen or oxygen gas. With this analytical technique, all the major species involved in the sulfide mineral oxidation mechanism (Reactions 1 to 4) were identified, and their proportions on the chalcopyrite surface were monitored with changes in pulp conditions. It was found that more iron dissolves from the chalcopyrite lattice than copper, and, therefore, more ferric hydroxide was formed than copper hydroxide. In particular, the ratio of ferric hydroxide (Fe(OH)3) to ferrous sulfide (FeS) doubled with oxygen purging, whereas the ratio of copper hydroxide (Cu(OH)2) to CuS remained constant at pH 9.5 (Figure 11). U S E O F D I AG N O S T I C P U L P A N D S O L U T I O N C H E M I S T RY I N S E L E C T I V E F L O TAT I O N

This section discusses some practical case studies where pulp and solution chemical measurements are used as a first step toward process optimization. First, a typical plant survey and key sample points for pulp chemical measurements are discussed. Next, the importance of correlating laboratory conditions with plant performance are outlined. Finally, examples of measurements taken from a plant and their interpretation are discussed in two studies. Pulp Chemical Measurements

It is very important to measure both the primary ball mill discharge Eh and pH during a typical plant survey because this is usually the point in the process where the Eh is at its lowest value ( Johnson 1988) and corresponds to the state of least oxidation (Figure 12). This sample point is also important because it indicates the extent of media oxidation (Grano et al. 1994), and the possibility of maintaining low Eh values during collector conditioning and subsequent flotation ( Johnson, Jowett, and Heyes 1982). Measurements conducted at the primary ball mill discharge point can also be very useful in demonstrating

flotation0.book Page 244 Tuesday, January 2, 2007 7:36 PM

244

FLOTATION FUNDAMENTALS

5.0/N2

Fe(OH)3

Intensity, arbitrary units

9.5/N2

9.5/O2

FeS

704

708

712

716 704

708

712

716 704

708

712

716

Binding Energy, eV

Intensity, arbitrary units

S2p

Cu2p(3/2)

Sn2– CuS

S2– S0

158

162

Cu(OH)2

166

930

934

938

Binding Energy, eV

Source: Fairthorne, Fornasiero, and Ralston 1997.

FIGURE 11 Top: Fe2p(3/2) XPS spectra of chalcopyrite as a function of conditioning pH and type of gas; Bottom: S2p and Cu 2p(3/2) XPS spectra of chalcopyrite conditioned at pH 9.5 in oxygen (lines represent the calculated XPS spectra with their individual components)

Reagents Primary Cyclone Overflow Process Water

Conditioned Rougher Feed

Rougher Tailing X Rougher Block

X

X

Process Water

X SAG Feed

SAG Mill X

Primary X Ball Mill

X Rougher Concentrate

X

Reagents Regrind Ball Mill

Reagents Plant Condtioning

Includes Regrind Pump Box and Cyclone

Cleaner Block X = Sampling Point

Reagents

X Regrind Cyclone Overflow Reagents Process Water

X Cleaner Tailing

X Final Concentrate

FIGURE 12 Typical points (x) for measuring pulp chemistry values in a flotation plant, showing typical reagent and process water addition points

flotation0.book Page 245 Tuesday, January 2, 2007 7:36 PM

PULP AND SOLUTION CHEMISTRY

245

the difference between the semiautogenous grinding (SAG) and primary ball mill chemical environments (Grano et al. 1994). This is critical for subsequent unit flotation stages in the primary grinding circuit that may be adversely affected by low Eh values. The pulp chemical change across the primary ball mill can be assessed by comparing the primary cyclone underflow with the primary ball mill discharge. The cyclone overflow should be measured separately from the “conditioned rougher feed” because of the reagents, as well as aeration in cycloning and pumping from the primary ball mill circuit to the flotation feed. The “conditioned rougher feed,” of course, is the feed presented to rougher flotation with all reagents added. The regrind ball mill discharge and regrind cyclone overflow should also be separately measured. If there are any conditioning stages, such as pulp heating, pH adjustment, aeration, or reagent additions (e.g., CuSO4, SO2), the change in pulp chemistry across the conditioning stage should also be assessed. An excellent example of the use of diagnostic Eh and pH values to solve plant problems is discussed by Johnson and Munroe (1988). The utility of pulp and solution chemistry measurements in solving processing problems is outlined in the following section. These measurements involve Eh and pH inorganic and organic composition of the circuit water, extractable metal ions from mineral surfaces, and temperature. The circuit water should be assessed, including other key streams in the process water circuit (e.g., tailings return water, makeup water, mine water, tailings thickener overflow water, concentrate thickener overflow water), as required (Levay, Smart, and Skinner 2001). The importance of process water chemistry is illustrated by an example where copper (II) in solution, emanating from a thickener overflow stream (which contained 10 ppm Cu concentration at pH 7), inadvertently activated sphalerite in the lead circuit of a lead/zinc flotation plant. This caused sphalerite to report to the lead concentrate and, thus, dilute the latter. With this analysis information available, steps were taken to minimize this inadvertent copper activation. Tailings return water, which contained much lower levels of copper ( PPG 400 > PPG 192. Considering the difference in molecular weight, 75 PPG 192 PPG 400 PPG 725 PPG 1000 PPG 2000 MIBC

Equilibrium Surface Tension, mN/m

70 65 60 55 50 45 40 35 30 0.0001

0.01

1

100

10,000

Concentration, mM A. Equilibrium surface tension vs. concentration of PPG and MIBC. The arrows indicate where droplets were observed in the solution (phase separation). 75 PPG 192 PPG 400 PPG 725 PPG 1000 PPG 2000

Equilibrium Surface Tension, mN/m

70 65 60 55 50

Droplet Formation

45 40 35 30 1E-06

0.0001

0.01

1

100

10,000

C/C* B. Equilibrium surface tension vs. concentration of PPG relative to the phase separation concentration.

Source: Tan et al. 2005.

FIGURE 2

Surface tension vs. concentration

flotation0.book Page 263 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

263

MIBC was slightly more surface active than PPG 192. For the PPG series, this increase in surface activity follows the increasing number of hydrophobic groups on the polymers. In the plots for Figure 2a, there was no evidence of a critical micelle concentration (CMC), which would have resulted in a breaking point in the curve and a plateau. However, in all solutions, phase separation occurred in the higher concentration ranges, causing the solution to become turbid. The earliest stage of this phenomenon resulted in droplet formation, which was detected by dynamic light scattering on the prefiltered solution as previously reported (Tan et al. 2004). The critical concentration of the phase separation (CPS, C*) is indicated by arrows in Figure 2a. To enable a comparison between the different CPS values of the various polymers, the results were normalized, and the surface tension versus concentration data were plotted in the form of surface tension versus C/C* (Figure 2b). This plot shows that the C/C* values coincide, indicating that phase separation occurs at the same relative concentration for the different PPG polymers. This value corresponds to a solution surface tension value in the range of 42 to 48 mN/m. The average HLB and the range of HLB values are shown in Figure 3 for the PPG polymers together with the value of MIBC for comparison purposes. The results show a decrease in the solubility limit and HLB value with an increase in molecular weight, with the average HLB value falling from 10.5 to about 6.5 for the PPG polymers. The HLB number of MIBC is indicated. The frothing performance of the chemicals was also evaluated at a 20-ppm concentration level, and the intermediate molecular weight frother (PPG 400) with HLB performed better. 1E+0 PPG MIBC 1E–1

C*(M)

1E–2

1E–3

1E–4

1E–5 2

4

6

8

10

12

Hydrophilic/Lyphophilic Balance

Source: Tan et al. 2005.

FIGURE 3 The phase separation concentration, C*, versus the average HLB and range of HLB for the PPG. The HLB of MIBC is indicated.

flotation0.book Page 264 Tuesday, January 2, 2007 7:36 PM

264

FLOTATION FUNDAMENTALS

R E L AT I O N S H I P B E T W E E N F R O T H I N G A N D P H A S E S E PA R AT I O N O F F R O T H E R S

Additional studies of foamability measurements of the PPGs were conducted by Tan et al. (2004, 2005) over a wide range of concentrations and gas flow rates. They reported that the foam height (or foamability) followed a characteristic plot (Figure 4), which could be separated into three distinct regions. At low concentrations, foamability increases and reaches a critical foam height at concentration C1. At this point, the foamability reaches a maximum value and remains constant until another point, C2, is reached. At concentrations greater than C2, the foamability decreases. The values of C1 and C2 decreased with increasing PPG molecular weight. Clearly, these results demonstrate that each foamer has a well-defined concentration range for foaming and defoaming. The results for the PPG series of frothers are shown in Figure 5 and show characteristic plots over different concentration ranges. In the low-concentration range (where foaming increased following an increase with concentration), the amount of surfactant adsorbed at the air–solution interface increased because of an increase in surface pressure. In the plateau concentration region, although the concentration of surfactant in solution is increased, the foaming characteristics of the system remain constant. Finally, a critical point is reached at a higher concentration where the foamability begins to decrease. The plateau region is likely to be related to the influence of bulk surfactant on the Marangoni effect. Surface tension gradients build up at the air–solution interface, resulting from thinning of the film. This causes the transport of bulk liquid into the thin film, restoring the thickness and preventing rupture from occurring. Generally, the plateau concentration C1 is found to decrease for molecules with increasing surface activity, because a more-surface-active molecule adsorbs more readily at a gas–liquid interface. At concentrations beyond a critical limit, C2, the surfactant exceeds its solubility limit and

Constant

Decreasing

Foamability

Increasing

C2

C1 Concentration

FIGURE 4

Foamability versus concentration profile of PPG frothers

flotation0.book Page 265 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

265

phase separation occurs as droplet formation. In this high-concentration region, the droplets act as an antifoamer. Many years ago, this foam depression effect caused by excessive frother was recognized as “over oiling.” In fact, increasing the frother dosage caused a rise, then a leveling-off, and finally, a drop in recovery of minerals. Curiously, the effect seems to be almost forgotten among researchers today, although Klimpel (1987) studied over-oiling by frothers on an industrial scale. As demonstrated in the previous section, the wide variation in concentration, which results in this depressant effect depending on the molecular weight of the PPG, emphasizes the importance of frother dosage changes as a key factor in flotation operations. Gas Flow Rates 33 mL/min 54 mL/min 43 mL/min 65 mL/min Parts per Mllion 1.9 25

19

192

1,920

Parts per Million 19,200 A

20

15

10

5 C1 0 0.01

0.1

1

42

10

5

C2 10

100

0 0.01

1,000

0.1

Steady-State Foam Height, mm

Steady-State Foam Height, mm

1.9 25

15

10

5

Concentration, mM

19.4 PPG 2000

20

0.1

1.0

194 D

20

15

10

5

0 0.001

0.01 Concentration, mM

Source: Tan et al. 2004.

FIGURE 5

1,000

Parts per Million 940 C

0.01

10

Concentration, mM

94

PPG 1000

0 0.001

B

15

Parts per Million 9.4

420,000

20

Concentration, mM

0.94 25

4,200

PPG 400

Steady-State Foam Height, mm

Steady-State Foam Height, mm

PPG 192

0.42 25

Foamability versus concentration profile for PPG frothers

0.1

flotation0.book Page 266 Tuesday, January 2, 2007 7:36 PM

266

FLOTATION FUNDAMENTALS

A D S O R P T I O N O F F R O T H E R S AT T H E M I N E R A L – S O L U T I O N A N D A I R – S O L U T I O N I N T E R FA C E S

Ideally, in flotation circuits the frother preferentially adsorbs only at the air–solution interface. However, in the case of excessive overdosing of frother where phase separation occurs, the droplets can adsorb at the mineral–solution interface. Interestingly, the adsorption of MIBC and kerosene-type frothers on hydrophobic solids such as coal has been reported in the literature (Aktas and Woodburn 1994), but this only appears to occur at high concentrations (greater than C2). It was suggested that the adsorption is influenced by the internal pore structure of the mineral. In fact, the rival effects of pore penetration and surface spreading of the frother droplets on the surface of the hydrophobic solid have been a controversial topic of debate. For coal particles, the flotation yield using short-chain hydrocarbon frothers was retarded at high concentrations by the penetration of the surfactants, such as diacetone alcohol and 2-ethyl hexanol, into the pores of coal. Early studies by Leja and Nixon (1957) also suggested that the interaction between collector and frother could lead to a complex, which, upon adsorption onto the mineral, could remove frother from solution. Some form of collector-frother adduct was suggested by Crozier and Klimpel (1989) for the xanthate-alcoholic frother on the surface of sulfide minerals. However, to date, there has been no follow-up, and this work has yet to be verified. S U R FA C E T E N S I O N A N D B U B B L E S I Z E

Typically, frothers are used in the very low dosage range, and at these concentrations they cause very limited reduction in the surface tension of the solution. Generally, bubble size is much more sensitive to very low frother concentrations than the surface tension, but few study results have been reported in these low frother concentration ranges. The air-bubble size dependence on the surface tension of the liquid has been studied by Rao and Stenius (1998) for a series of frothers. They measured the population size and distribution of bubbles (using a laser light-scattering technique) formed in solutions of nonionic alkyl PEO frothers and a long-chain anionic surfactant. These co-workers showed that as the surface tension of the liquid decreased, the average maximum size of the bubble population decreased, and the distribution became narrower. In addition, larger numbers of smaller bubbles were produced at increased surfactant concentrations. The experimental data reported from these studies are shown in Figures 6 and 7. In Figure 8, the surface tension of the solution was related directly to the bubble size in agreement with the Laplace equation for the nonionic frother systems. The coalescence of bubbles is also highly dependent on the electrolytes in the solution. The flotation efficiency increases as bubble size decreases. Because of streamlines around a rising air bubble, small particles frequently do not collide with the bubble, which is much larger than the particles. In addition, the stability of bubbles and foaming characteristics are dependent on the combination of both equilibrium and dynamic effects, with the dynamic effects dominating. For strong foaming, the literature concludes that the surfactant must be capable of rapidly lowering the surface tension. However, a relatively slow rate process is also required by which a freshly created liquid surface retains high, non-equilibrium surface tension long enough for surface flow to occur to stabilize the film. Mere rapid reduction in surface tension does not lead to the stabilization of the foam. What is necessary is the slow attainment of equilibrium after a fresh surface is produced. Surface elasticity arises from the variation of the surface tension during deformation of a liquid film. This can be manifested under equilibrium conditions (when the surface layer

flotation0.book Page 267 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

267

300 Surface Tensions, mN/m

C11E7 C11E10 C12SO3 Water

28.44

250

Number of Bubbles

32.28 200 35.08 150

100 71.81 50

0 15

25

35

45

55

65

Diameter, μm

Source: Rao and Stenius 1998.

FIGURE 6

Bubble size distribution in water and nonionic surfactant solution close to CMC 20 Surface Tensions, mN/m 30.6

55.5

71.8

Bubble Population, %

16

12

8

4

0 20

30

40

50

60

70

Bubble Diameter, μm

Source: Rao and Stenius 1998.

FIGURE 7

Bubble size distribution in water containing C11E10 at different concentrations 50

Maximum Bubble Diameter, μm

46

42

38

34

30 30

40

50

60

70

80

Surface Tensions, mN/m

Source: Rao and Stenius 1998.

FIGURE 8

Relationship between bubble size and surface tension for C11E10

flotation0.book Page 268 Tuesday, January 2, 2007 7:36 PM

268

FLOTATION FUNDAMENTALS

is under tension during equilibrium with the bulk phase), and this is defined by the Gibbs elasticity. In non-equilibrium conditions, it is defined by the Marangoni elasticity. The Marangoni elasticity of the monolayer can be determined from the dynamic (non-equilibrium) surface tension as the surface is abruptly extended or pulsated. The Marangoni elasticity is usually larger than the Gibbs elasticity for the same system and is a more important parameter in foam stability. The Marangoni elasticity occurs at a range of frequencies, but it can be evaluated from the dynamic surface tension data, usually at low frequency of dilation (compression). In practical foam systems, this could be as low as 1 cycle/min or as high as 900 cycles/sec. Essentially, it depends on the extension–contraction cycle in the foam. In fact, measurements (using a suitable experimental technique) ideally need to be made to correspond with measurements of the change in surface area of the actual bubbles during foaming. T H E D R A I N AG E O F F ROT H E R S O L U T I O N

Generally, it has become clear from model studies that the same factors which play a role in foam stability (film thickness, elasticity, etc.) also have a decisive influence on the stability of the isolated thin films. Hence, model film studies are very important, and the drainage of thin foam films will be considered in the next section. Experimental thinning studies have been reported with a microfilm formed from a concave liquid drop suspended in a short, vertical capillary tube. The apparatus was originally developed by Schedludko (1966) and has been well documented in the literature. Direct measurement of the thickness of the aqueous film versus time (the drainage process) can be determined by micro-reflectance methods. The drainage time of a flat film is determined from the Reynolds equation: ho

T =

∫ dh ⁄ VRe

(EQ 2)

ht

where T = the time of drainage of the film, which has an initial thickness of ho and drains tothe thickness ht h = the distance between surfaces VRe = the Reynolds thinning velocity t = time For the case of horizontal, fairly thick films (>100 nm), an expression has been derived for the thinning between two disc surfaces under the influence of a uniform external pressure. The change in film thickness with drainage time can be expressed as the Reynolds drainage: dh2h 3 ΔP V Re = –-------= ---------------dt 3ηR 2

(EQ 3)

where R = the radius of the disc η = the viscosity of the liquid ΔP = the pressure difference between the film and the bulk solution and is taken to be equal the capillary suction Pc in the surrounding “plateau” border

flotation0.book Page 269 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

269

Figure 9 shows the drainage patterns of films produced with three different frothers: ethoxylated nonionic—C10(EO)5 and C12(EO)5—and Dowfroth 200, which is a PPG with a molecular weight of about 200. In all cases, the results—expressed as the experiment-measured drainage rates (V) as compared to VRe versus bulk concentration (see Figure 9)—indicate that the films thin faster (i.e., drainage times are shorter) than those calculated from the Reynolds equation (Equation 3). The drainage is of the same order of magnitude as in previous results and follows the order: Dowfroth 200 > C10(EO)5 > C12(EO)5. The increase in frother concentration has a pronounced effect on drainage but only at higher concentrations where drainage is considerably reduced. Polyethylene glycols at low concentrations are well-known drag-reducing agents, and high drainage velocities are expected. 2.0 2.0

2.0 2.0

V/VRe V/VRe 1.5 1.5

1.5 1.5

Dowfroth Dowfroth C10(EO)5 C10(EO) 5 C12(EO)5 C (EO) 12

5

1.0 1.0

1.0 1.0

0.5 0.5

0.5 0.5

τ/ττ/τ Re

0.0 0.0 10–610

–6

10–410

–4

10–210

–2

Re

0.0 0.0

Bulk Concentration, Bulk Concentration, M M Source: Manev and Pugh 1992.

FIGURE 9 The relative thinning ratios of C10(EO)5 and C12(EO)5 and the ratio of measured drainage times (V to the calculated VRe) are compared to Dowfroth 200. Below the CMC the results are similar, but above the CMC the ethoxylated frothers drain slower than the Dowfroth 200.

flotation0.book Page 270 Tuesday, January 2, 2007 7:36 PM

270

FLOTATION FUNDAMENTALS

25

25

B 20 Foam Height, mm

20 Foam Height, mm

A

PPG 1000 MIBC

15

10

15

10

5

5

0

0

Mixture PPG 1000/MIBC 0

10

20

30

40

50

0

Concentration, ppm

10

20

30

40

50

PPG 1000 Concentration in Mixture, ppm

Source: Tan et al. 2005.

FIGURE 10 Foam behavior of (a) PPG 1000 and MIBC from 0 to 5 ppm, and (b) PPG 1000/MIBC mixture at a total concentration of 50 ppm

M I X E D C H E M I C A L F ROT H E R S Y S T E M S

In the flotation of minerals, a frother or combination of frothers is chosen depending on the requirements of the mechanical and interfacial rheology characteristics of the froth. In addition to technical advantages, there are also economical advantages in using blends. Often in sulfide flotation, two or more frothers will be blended to improve flotation efficiency. Frequently, a short-chain alcohol such as MIBC is chosen as a versatile frother together with a higher-molecular-weight frother, such as pine oil or a PPG type, which leads to improvements of the froth or bubble size, especially when coarser particles must be floated. Figure 10a shows the foam height versus concentration at a constant specific gas volume for a series of individual and mixtures of frothers. In this figure, the froth heights for the individual frothers, PPG 1000 (HLB 8.4) and MIBC (HLB 6.1), are shown for a concentration level from 0 to 50 ppm. These plots clearly show that for the single-frother systems, PPG 1000 performs better than MIBC with the foam height increasing with frother concentration, whereas with MIBC the foam height remains relatively low throughout the concentration range. In Figure 10b, the foam height for a mixture of PPG and MIBC is shown with a total frother concentration of 50 ppm. These results show that the mixed system with lower amounts of PPG 400 (5 ppm) but higher amounts of MIBC (45 ppm) gives a better foaming performance than that of MIBC or PPG 1000 alone at 50 ppm. Additional experiments have shown that a mixed surfactant film, consisting of a lowmolecular-weight/high-HLB frother mixed with a high-molecular-weight/low-HLB frother, results in improved foaming compared to either frother alone. In Figure 11, configurations are suggested for the low-, high-, and mixed-PPG polymers at the air–solution interface. In the mixed systems, these structures suggest that a closed, packed, cohesive film at the air–solution interface can be achieved, which could explain the higher foaming. S TA B I L I Z AT I O N O F B U B B L E A N D F OA M B Y PA R T I C L E S

The terms foaming and frothing are used interchangeably, but it is more usual in mineral processing to refer to the gas–water macro-cluster systems (two-phase) where the broken structure leaves a homogeneous aqueous phase as a foam. In mineral processing, the froth contains dispersed solid particles and is a three-phase system, so that the broken structure

flotation0.book Page 271 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

A

271

B Air

Air

Water

Water

Low-Molecular-Weight/ High-HLB Polymer

C

Medium-Molecular-Weight/ Medium-HLB Polymer

D Air

Air

Water

Water

High-Molecular-Weight/ Low-HLB Polymer

Mixed Film Consisting of Low-Molecular-Weight (High-HLB) and High-Molecular-Weight (LowHLB) Polymers

Source: Tan et al. 2005.

FIGURE 11

Suggested structures of PPG polymers and mixtures at the air–solution interface

gives a two-phase system (aqueous solution and finely divided particles). In mineralized froths, the important parameter would appear to be the wettability of the particles, but size and shape also have some influence. Although it has been clearly established that foam stability can be increased or decreased by many types of particles, to some extent the mechanism is complex because, frequently, there are several mechanisms operating in the same system. Usually, the particles have some critical degree of hydrophobicity, which plays a critical role in the dynamics of rupturing the thin foam films. Both the particle size and shape have been shown to play an important role, and systems have been studied using particles within a wide size range ( Johansson and Pugh 1992; Dippenaar 1982). It must be also stressed that throughout frothing technology, the main problems are frequently related to both stabilization and destabilization of the system, and the role of the particles is crucial in both functions. In fact, the use of particles as foam breakers is well known throughout industry, and hydrophobic particles are important ingredients in many foam-breaking formulations (Pugh 2002a). F R O T H I N T H E F L O TAT I O N P R O C E S S

Overall, the dynamic behavior of particles and the interaction with frothers are critical, but steps in industrial froth flotation that are poorly understood usually lead to loss of recovery during the froth phase. Although foams are stabilized by adsorbed surfactant molecules (Figure 12a), froths are usually stabilized by small particles with a critical degree of wetting at the gas–liquid interface, which causes the bubbles to become “armored” (Figure 12b). Several early ideas were based on the premise that coalescence is prevented because of a steric interaction. As reported by many flotation researchers, the stability and drainage of a three-phase froth depends on hydrophobicity of the mineral particles present in the froth. Froth becomes stabilized and draining of the liquid from the thin layer is restricted by hydrophobic solid particles. The more strongly the particles adhere to the bubble, the more stable the froth becomes; thus, an increase in contact angle of particles to a certain critical value benefits froth stability. However, there is considerable evidence that the highly hydrophobic particles (e.g., with contact angle greater than 90°) will destabilize froth (Garrett 1993).

flotation0.book Page 272 Tuesday, January 2, 2007 7:36 PM

272

FLOTATION FUNDAMENTALS

A

FIGURE 12

B

Different stabilizing actions of (a) frother molecules and (b) solid particles

A. Nonwetting particles cause receding of the liquid and premature rupture of the film at liquid–solid interface, as indicated by arrows.

B. Wetting liquid allows particles to retain liquid within the film, delaying rupture.

Source: Ip, Wang, and Toguri 1999.

FIGURE 13 The influence of wettability on film stability as the film approaches the particle size thickness

C A P I L L A R Y E F F E C T S O N F R O T H S TA B I L I T Y

As the liquid film drains down to a critical thickness, the nonwetted particles can reduce froth stability by inducing the liquid to de-wet around the particle, causing the liquid to recede from the particle at areas indicated by the arrows in Figure 13a. This leads to rapid rupture, but in the case of partial wetting, the particles trap the liquid, making the film more stable (Figure 13b). However, the influence of particle concentration, density, and shape needs to be taken into consideration. As shown in Figure 14, the effect of low and high concentrations of nonwetting particles and the effect of plate-like particles (for example, clay particles) on thin film stability are illustrated. Two mechanisms have been suggested to explain the froth stabilization effects caused by hydrophobic particles adsorbed at the interface. The first effect results from a change in capillary pressure. This is caused by the presence of adsorbed particles modifying the curvature of the gas–liquid interface, which reduces the pressure difference between the plateau borders and the three films associated with it. The situation is illustrated in Figure 15.

flotation0.book Page 273 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

273

Air Water Air B. High Mineralization of Bubbles

A. Poor Mineralization of Bubbles

C. Effect of Plate-like Particles

Source: Warren 1981.

FIGURE 14

Effect of hydrophobic particles on bubble stability

Without Particles A

With Particles D

Film

Profile View

Plateau Border

B

E Pgas

Pgas

Pfilm

Pfilm

Pgas

Pgas

Ppb

Ppb

Pgas = Pfilm > Ppb

Pgas > Pfilm ≈ Ppb

Near Particles

Film

C

F

Plateau Border

Pressure

Source: Ip, Wang, and Toguri 1999.

FIGURE 15 The effect of hydrophobic particles on the pressure difference between the foam film and the plateau borders

In the case of the foam with no particles (Figure 15a), the liquid can flow from the film into the plateau border and then flow through the structure by gravity. The flow rate would be proportional to the pressure difference, ΔP, expressed by ΔP = P film – P pb = γ ⁄ R pb

(EQ 4)

where Pfilm is the pressure in the films, Ppb is the pressure in the plateau border, γ is the surface tension of the liquid, and Rpb is the radius of curvature of the gas–liquid film interface (Figures 15b and 15c). Therefore, when ΔP is high, the flow rate is increased, which causes faster drainage, and the foam becomes less stable. If many hydrophobic particles are attached to the gas–liquid interface (Figure 15d), the radius of curvature of the gas–film

flotation0.book Page 274 Tuesday, January 2, 2007 7:36 PM

274

FLOTATION FUNDAMENTALS

interface would change and become almost equal to that of the gas–plateau border interface (Figures 15e and 15f ). This will cause the pressure difference to decrease, resulting in a more stable froth. According to Lucassen (1992), capillary effects are also very important for small particles where gravitational forces are negligible. This results from small floating solid particles at the fluid interface that can interact with each neighbor because any solid particle will virtually always cause deformation of interfaces. The extent of deformation should be altered by the approach of the neighboring particle. This especially occurs in cases where particles have an irregular wetting perimeter, which disturbs the smoothness of the interface. Calculations made for a model particle with a sinusoidal edge indicate that the disturbances can become significant when induced by capillary interaction with neighboring particles. Restricted Drainage Mechanism

It has also been suggested that attached particles can cause the overall drainage within the thin film to be hindered as the film and the liquid passages become constricted and tortuous. To some extent, the volume of the particles stabilized by froth is, therefore, roughly proportional to the amount of hydrophobic solids present, but there is an upper limit. It appears that if the size of the hydrophobic particle is small enough compared with the film thickness, as previously discussed, then they can arrange themselves at the liquid–gas interface and stabilize the films by the capillary mechanism described. If the particles are large (i.e., the diameter is larger than the film thickness), the particles can bridge and rupture the foam film. Frothing experiments by Garrett (1993) using fairly large particles (1–100 microns) showed that these particles could easily destroy the foam. This effect has been observed in the flotation process. For instance, it has been reported that 0.1-mm galena particles can prolong the life of froths of isoamyl alcohol aqueous solutions from 17 seconds to several hours; whereas 0.3-mm galena particles can only increase it to 60 seconds (Lu, Pugh, and Forssberg 2005). Frothing Studies with Model Quartz Particles of Well-defined Size and Hydrophobicity

This section studies the influence of size and hydrophobicity of particles on the stability of froths using both the modified Bikermann column and the thin film balance ( Johansson and Pugh 1992). In these experiments, surface-modified quartz particles were used as models. The hydrophobicity of the quartz surface was controlled by reacting the dry surface with trimethylchlorosilane in cyclohexane under a dry environment according to standard procedures. In order to characterize the surfaces, a quartz plate was also placed in the reaction vessel together with the particles. After the reaction, the plate and particles were removed and rinsed with cyclohexane and washed with water. Finally, the contact angle of a drop of water on the plate was determined using the Ramé-Hart goniometer. The hydrophobicity of the quartz particles was quantified from small-scale flotation experiments using a Hallimond tube apparatus. The froths (containing quartz particles) were characterized by both dynamic and static frothing tests. The dynamic test (carried out during froth generation) quantified the equilibrium state of the froth, whereas the static tests (after the gas flow had ceased) determined the rate of collapse of the froth. In this study, a modified Bikermann test was used consisting of a glass column where the maximum equilibrium volume of the froth (Hmax) was determined at a standard flow rate. A typical set of data is shown in Figure 16 for dynamic frothing with

flotation0.book Page 275 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

275

Size Fraction (Frother Concentration) 26–44 μm (20 mg/L) 74–106 μm (20 mg/L) 26–44 μm (50 mg/L) 74–106 μm (50 mg/L) 12 PPGMME, n = 3

A

PPGMME, n = 4

B

12

Hmax, cm

Hmax, cm

8 8

4 4

0

0 0

20

40

60

80

0

100

20

Flotation Yield, %

40

60

80

100

Flotation Yield, %

10

α -Terpineol

MIBC

C

D

10

8

8 Hmax, cm

Hmax, cm

6

4

6

4 2 2

0

0 0

20

40

60

Flotation Yield, %

80

100

0

20

40

60

80

100

Flotation Yield, %

Source: Johansson and Pugh 1992.

FIGURE 16 Relationship between the dynamic stability expressed as the maximum froth height (Hmax) at the flow rate of 60 L/hr vs. the hydrophobicity of the quartz particles (expressed as the flotation yield)

four commercial mineral processing frothers: polypropylene glycol monomethyl ether PPGMME with n = 3, PPGMME with n = 4, ∝-terpineol, and MIBC. These results express the dynamic froth characteristics in terms of Hmax versus the hydrophobicity of the particles expressed in terms of the flotation yield. From the results obtained with the small particle fraction (26–44 microns), there appears to be trend for a distinct maximum corresponding to a flotation yield of about 70%. This corresponds to a critical degree of hydrophobicity of 60° for the small particle fraction. This value seems to be reproducible in all the frother systems. Also, at higher flotation yield, the froth collapsed, indicating that the particles were acting as foam breakers; therefore, it could be concluded that below this critical degree of hydrophobicity, the particles appear to have less effect on the system. These trends are observed at both high and low frother concentrations. However,

flotation0.book Page 276 Tuesday, January 2, 2007 7:36 PM

276

FLOTATION FUNDAMENTALS

these effects were not observed with the larger size fraction. In fact, the particles do not appear to influence the stability of the system. Similar trends have been observed at both low (20 mg/L) and high (50 mg/L) frother concentrations. Additional experiments were carried out at a range of frother concentrations where a similar trend was observed for Hmax values versus concentration plots. E N T R A I N M E N T O F S U B M I C R O N - S I Z E PA R T I C L E S I N T H E F R O T H P H A S E A N D T R U E F L O TAT I O N

The process leading to the formation of three-phase froth layers involves not only the hydrophobic particles carried over in the froth but also the hydrophilic entrained gangue particles. These particles are only feebly adhered to the gas bubbles or are situated in the water film in the froth and are usually washed out during drainage (Figure 17). At the same time, the ascending air bubbles carry the hydrophobic particles to the top of the froth. Thus, the top of three-phase froth contains the more hydrophobic particles and has the higher grade; the grades of the floated material decrease from top to bottom of the froth. This phenomenon is referred to as a secondary concentration effect and is useful for upgrading the concentrate quality and, therefore, has found application in column flotation. In most cases, complete separation between gangue and valuable mineral in the pulp zone is difficult to achieve and, nearly always, some gangue minerals are transported into the froth with the entrained liquid. As the froth ages, some of the hydrophilic gangue returns to the pulp because of slurry drainage, while the remainder is carried up with the concentrate, reducing the quality of the product. The drainage of hydrophilic particles (or recovery of gangue in the concentrate) is largely affected by the proportion of the gangue minerals, such as density and particle size. To date, the effect of the gangue characteristics on the drainage has been well established, but the influence of the froth characteristics on the gangue recovery has not been fully investigated. This is due to many factors that influence froth stability, such as particle size, hydrophobicity, and reagents. Recently, however, some interesting experiments have been carried out by Ata, Ahmed, and Jameson (2004) that have helped to reveal some new aspects of this complex problem. A specially designed flow cell was used in which the froth could be isolated from the pulp zone; this enabled the collection of particles that had dropped out of the froth. In these studies, hematite particles (82 microns in diameter) were floated forming the froth phase in Air

Air Air

Air

Source: Kelly and Spottiswood 1982.

FIGURE 17

Schematic illustration of froth drainage

flotation0.book Page 277 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

277

the conventional way, but glass particles of about the same diameter were added to the wash water and fed to the froth to study the influence of hydrophobicity of the particles on the recovery of gangue. The froth-fed glass particles had a range of contact angles of 0°, 50°, 66°, and 82°. In other tests, hydrophilic and hydrophobic glass particles were mixed in various ratios, and tests were conducted with the blended feed. This enabled studies to be conducted on the effect of the hydrophobicity of the particles attached to the bubbles and on the drainage of gangue minerals in the froth phase. From these studies, it was concluded that the hydrophobic particles had a strong effect on the structure of the froth, and this had an influence on the drainage of the hydrophilic particles. The hydrophobic particles significantly affected the drainage rate of hydrophilic particles in the froth. The recovery of hydrophilic particles in the flotation product increased as the hydrophobicity of the floatable particles increased, and the water recovery rate decreased with increasing contact angle of the particles. It was concluded that the hydrophobic particles had an effect on the structure with higher hydrophobicity, causing increased drainage rate of water leading to drier froths in which the hydrophilic particles become more easily entrapped. Gangue recovery in the concentrate also increased with the concentration of the gangue mineral presented to the froth, even in the presence of wash water, presumably as a result of the increased viscosity of the liquid in the froth. C O L L E C T O R L E S S F L O TAT I O N

The frothing and flotation of hydrophilic metal hydroxides in the absence of frother and collector, also known as collectorless flotation, could be considered as an area of special interest and is sometimes referred to as contactless flotation. There are several reviews on the microflotation of solids that occurs in the presence of hydrolyzing ions and in dissolved air flotation circuits. The region of floatability corresponds closely with the regions where precipitation of the metal ion occurs. In fact, maximum coagulation corresponds to maximum flotation. Hydrophilic solids, such as quartz, are usually readily coagulated by iron or aluminum, and various degrees of flotation occur with microbubbles. In these systems, it has been suggested that entrapment of bubbles occurs in aggregates, and it has been reported that for efficient frothing and flotation, small bubbles are required (Solari and Gochin 1992). One theory suggests that naturally occurring organic compounds may be responsible for particle hydrophobicity, which causes frothing in some industrial dissolved air flotation operations. It has been well documented that natural water frequently contains biologically derived surfactants which could stabilize microbubbles. In clean water systems, the flotation of nonfloating suspension of ferric hydroxide flocs can occur fairly readily in the presence of a few parts per million of collector. Also, coagulation with hydroxyl ions increases the effective particle size and decreases the number of particles to be collected. General particle aggregation is needed, but in many cases, a collector may be also needed to improve the process efficiency. Experiments of Kitchener and Gochin (1981), have confirmed that the floatability of metal hydroxides is very sensitive to the presence of organic impurities in the system. They suggested natural water contains surface-active impurities that are adsorbed onto metal precipitated hydroxides, thereby forming insoluble hydrophobic soaps that provide sites for bubble adhesion. Because the flocs have low density, microbubble attachment to a few hydrophobic spots on the flocs would be sufficient to ensure flotation.

flotation0.book Page 278 Tuesday, January 2, 2007 7:36 PM

278

FLOTATION FUNDAMENTALS

S T R U C T U R E O F F R O T H S A N D F L O TAT I O N I N A Q U E O U S S O L U T I O N S O F I N O R G A N I C E L E C T R O LY T E S

The frothing and flotation of naturally hydrophobic particles in inorganic electrolyte solutions was first documented in the former U.S.S.R. during the 1930s. This work was mostly related to the flotation of coals in saline waters. Electrolytes also played a role in stabilizing the froth. Following these early studies, many other cases have been reported in the general area of mineral flotation. Wellham, Elber, and Yan (1992); Yoon (1982); and Yarar (1988) have reviewed this specific area. Several surface chemical mechanisms have been proposed to explain the flotation process. These include the action of the electrolytes in disruption of hydration layers surrounding the particles and enhancing bubble–particle capture, reduction of the electrostatic interactions, and an increasing charge on the surface of the bubbles to prevent primary bubble coalescence. However, none of these appear to satisfactorily explain the experimental observations. Yoon (1982) studied flotation coal cleaning using inorganic electrolyte, and it was concluded that the flotation efficiency increased with additional salt concentration. Flotation rate and cleaning efficiency were also improved. More recently, Craig, Ninham, and Pashley (1993) assessed the inhibition of bubbles to coalescence in electrolytes by the application of a combining rule based on the nature of the cationic/anionic pair. This rule enables one to predict whether or not the electrolyte would inhibit coalescence of gas bubbles in the electrolyte solutions. Viscosity and electrostatic repulsion were ruled out as possible explanations. In fact, following conventional electrostatic double-layer theory, an increase in salt concentration would reduce the double-layer repulsion and should induce inhibition. It was also suggested by Craig, Ninham, and Pashley (1993) that the coalescence in pure water was caused by a strong hydrophobic attractive force, which opposed the hydrodynamic repulsion existing between the colliding bubbles. Paulson and Pugh (1996) carried out flotation experiments with graphite (≈20 μm in diameter) with a series of different inorganic electrolytes at a range of concentrations (in the absence of an organic frother) in a small glass, cylindrical column cell. The flotation recovery of graphite as a function of the electrolyte concentration is shown in Figure 18. The plot in Figure 18 shows that recovery generally increases with concentration and varies according to the cationic/anionic pair. In fact, it is possible to classify the electrolytes into three groups according to their flotation performance. Group A, salts with divalent and trivalent cations or anions, include MgCl2, CaCl2, Na2SO4, MgSO4, and LaCl3, which give high flotation response. In this group, flotation begins at about 0.02 M and reaches maximum recovery at about 0.06 to 0.1 M concentrations. Group B includes NaCl, LiCl, KCl, CsCl, and NH4Cl, which give medium flotation response with flotation beginning at about 0.05 to 0.1 M electrolyte. Finally, group C includes NaAc, NaClO4, HClO4, HCl, H2SO4, LiClO4, and H3PO4, which give a very low flotation response, even up to concentrations as high as 0.3 M electrolyte. A plot of double-layer thickness versus flotation recovery (Figure 19) showed that for group A and B electrolytes, a correlation exists between the flotation performance and the Debye length (1/κ), which suggests that the electrostatic interaction plays a role in the process. In this study, a relationship was also found between flotation recoveries and the surface tension concentration gradient of the electrolyte solution. Additionally, a correlation showing a decrease in gas solubility occurred with increasing electrolyte concentration. Thus, the increased flotation performance of the hydrophobic graphite appears to be linked with the increase in stability of the gas bubbles and froth caused by a decrease in dissolved gas concentration in the electrolyte solutions.

flotation0.book Page 279 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

279

100 90 80 A

B

Recovery, %

70 60 50 40

C

30 20 10 0 0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

Salt Concentration, M H 2O MgCl2 NaCl

CaCl2 MgSO4 Na2SO4

LaCl3 LiCl KCl

CsCl NH4Cl NaAc

LiClO4 NaClO4 HClO4

HCl H2SO4 H3PO4

Source: Paulson and Pugh 1996.

FIGURE 18 The flotation recovery of graphite particles vs. the electrolyte concentration for a series of different types of electrolytes. Group A = high flotation performance; Group B = intermediate flotation performance; Group C = low flotation performance.

100 90 80

Recovery, %

70 60 50 40 30 20 10 0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Debye Length, mm CaCl2 NaClO4 MgSO4

MgCl2 Na2SO4 NH4Cl

LaCl3 LiClO4 CsCl

H3PO4 KCl H2SO4

LiCl HCl NaCl

HClO4 HaAc

Source: Paulson and Pugh 1996.

κ) and the flotation performance of FIGURE 19 The relationship between the Debye length (1/κ graphite in different inorganic electrolytes

flotation0.book Page 280 Tuesday, January 2, 2007 7:36 PM

280

FLOTATION FUNDAMENTALS

BIBLIOGRAPHY

Aktas, Z., and E.T. Woodburn. 1994. The adsorption behavior of nonionic reagents on two low rank British coals. Miner. Eng. 7:1115–1126. Aston, J.R., C.J. Drummond, F.J. Scales, and T.W. Healy. 1983. Frother chemistry in fine coal processing. Pages 148–160 in Proceedings of the 2nd Australian Coal Preparation Congress. Edited by R.I. Whitmore. Brisbane, Australia: Westminster Press. Ata, S., N. Ahmed, and G.J. Jameson. 2004. The effect of hydrophobicity on the drainage of gangue minerals in flotation froths. Miner. Eng. 17:897–901. Becher, P. 1984. Hydrophile-lipophile balance. J. Dispersion Sci. Technol. 5(1):81–96. Craig, V.S.J., B. Ninham, and R.M. Pashley. 1993. The effect of electrolyte on bubble coalescence. J. Phys. Chem. 97:10192. Crozier, R.D., and R.R. Klimpel. 1989. Frothers: Plant practice. Pages 257–279 in Frothing in Flotation. Edited by J.S. Laskowski. New York: Gordon and Breach. Dippenaar, A. 1982. The stabilization of froth by solids. Int. J. Miner. Process. 9:1–14. Garrett, P.R. 1993. Defoaming. Surfactant Science Series, Volume 45. New York: Marcel Dekker. Ip, S.W., Y. Wang, and J.M. Toguri. 1999. Aluminum foam stabilization solid particles. Can. Metall. Q. 38(1):84. Johansson, G., and R.J. Pugh. 1992. The influence of particle size and hydrophobicity on the stability of mineralized froths. Int. J. Miner. Process. 34:1–22. Kelly, E.G., and D.J. Spottiswood. 1982. Introduction to Mineral Processing. New York: WileyInterscience. Kitchener, J., and R.J. Gochin. 1981. The mechanism of dissolved air flotation for potable water— basic analysis and proposal. Water Res. 15:585–590. Klimpel, R.R. 1987. The industrial practice of sulphide mineral collectors. In Reagents in the Mineral Industry. Edited by P. Somasundaran and B. Moudgil. New York: Marcel Dekker. Klimpel, R.R., and R.D. Hansen. 1988. Frothers. Pages 385–409 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Klimpel, R.R., and S. Isherwood. 1991. Some industrial implications of changing frother chemical structure. Int. J. Miner. Process. 33:369–381. Leja, J., and J.C. Nixon. 1957. Ethylene oxide and propylene oxide compounds as flotation reagents. Pages 297–307 in 2nd Congress of Surfaces Activity. Volume 3. Edited by J.H. Schulman. London: Butterworths. Lu, S., R.J. Pugh, and E. Forssberg. 2005. Interfacial Separation Processes. Studies in Interfacial Science, Volume 20. Amsterdam: Elsevier. Lucassen, J. 1992. Capillary forces between solid particles in fluid interfaces. Colloids Surf. 65(2–3):131. Manev, E., and R.J. Pugh. 1992. Study of drainage and equilibrium thickness of single foam films containing non-ionic frothers and a short chain xanthate. J. Colloid Interface Sci. 151:505–516. Moxton, N.T., C.N. Bensley, R. Keast Jones, and S.K. Nicol. 1987. Insoluble oils in coal flotation: The effect of surface spreading and pore penetration. Int. J. Miner. Process. 21:261–274. Paulson, O., and R.J. Pugh. 1996. Flotation of inherently hydrophobic particles in aqueous solutions of inorganic electrolytes. Langmuir 12:4808–4813. Pugh, R.J. 1996. Foaming, foam films, antifoaming and defoaming. Adv. Colloid Interface Sci. 64:67–142. ———. 2000. Flotation of graphite with polyglycol frothers. Miner. Eng., 13(2):151–162. ———. 2002a. Foam Breaking in Aqueous Solution. In Handbook of Applied Colloid and Surface Chemistry. Edited by K. Holmberg. New York: Wiley & Sons. ———. 2002b. Foams and Foaming. In Handbook of Applied Colloid and Surface Chemistry. Edited by K. Holmberg. New York: Wiley & Sons. Rao, R., and P. Stenius. 1998. The effect of flotation de-inking chemicals on bubble formation. J. Pulp Paper Sci. 24(5):156–160. Schedludko, A. 1966. Thin films. In Colloid Science. New York: Elsevier. Solari, A., and R.J. Gochin. 1992. Fundamental aspects of microbubble flotation process. In Colloid Chemistry in Mineral Processing. Edited by J.S. Laskowski and J. Ralston. New York: Elsevier.

flotation0.book Page 281 Tuesday, January 2, 2007 7:36 PM

THE PHYSICS AND CHEMISTRY OF FROTHERS

281

Tan, S.N., D. Fornasiero, R. Sedev, and J. Ralston. 2004. Polypropylene glycol frothers. Colloids Surf. A 250:370–315. Tan, S.N., R.J. Pugh, R. Sedev, and J. Ralston. 2005. Foaming of polypropylene glycols and glycol/ MIBC mixtures. Miner. Eng. 18:179–188. Warren, L.J. 1981. Shear flocculation. Chemtec 11:180–185. Wellham, E.J., L. Elber, and D.S. Yan. 1992. Coal flotation. Miner. Eng. 5(3–5):381. Yarar, B. 1988. Gamma flotation: A new approach to flotation, using liquid vapour surface tension control. Page 41 in Froth Flotation. Edited by S.H. Castro and J. Alvarez. New York: Elsevier. Yoon, R.H. 1982. Coal flotation. Min. Congr. J. 68(12):76.

flotation0.book Page 282 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

flotation0.book Page 283 Tuesday, January 2, 2007 7:36 PM

Surface Characterization and New Tools for Research R.St.C. Smart, W.M. Skinner, A.R. Gerson, J. Mielczarski, S. Chryssoulis, A.R. Pratt, R. Lastra, G.A. Hope, X. Wang, K. Fa, and J.D. Miller

INTRODUCTION

In the selective separation of mineral phases by flotation, surface chemistry is the principal determinant of the average contact angle for a specific mineral phase in a flotation pulp. The average contact angle is, in turn, the principal determinant of the bubble–particle attachment efficiency (Ea) in the overall collection efficiency (Ec) from which the flotation rate constant can be determined (Ralston 1994a). The recovery and selectivity in sulfide flotation is ultimately dependent on the relative rate constants of the different mineral phases. The average contact angle is not only mineral-specific, based on a statistical average of the mineral particles in that phase, but also the contact angle for each particle is an average of hydrophobic and hydrophilic areas across the particle surface. Determination of this hydrophobic/hydrophilic balance by particle therefore requires selection of the particular mineral phase. Obtaining this information is not necessarily a simple task in a flotation pulp containing many different mineral phases, different particle sizes of individual phases, adsorbed and precipitated species (often colloidal), and oxidized products. The hydrophobic/hydrophilic balance by particle and its statistical average by mineral phase require identification of the major species contributing to each category in surface layers (Smart, Jasieniak et al. 2003). In addition to adsorbed collector molecules and their oxidized products (e.g., dimers), hydrophobicity can be imparted to sulfide mineral surfaces by oxidation to produce polysulfide Sn2– species resulting from loss of metal ions (usually Fe2+) from surface layers. In acid solution, hydrophobic elemental sulfur can also be formed and is usually imaged in patches on the sulfide mineral surface (Smart, Amarantidis et al. 2003). Almost all other species found on sulfide mineral surfaces, such as oxide/oxyhydroxide/hydroxides, oxy-sulfur (e.g., sulfate), carbonate, hydrous silica, and fine gangue particles, are essentially hydrophilic but may be in the form of localized particles, colloids, and precipitates or continuous, reacted, or precipitated surface layers (Smart, Amarantidis et al. 2003). The action of collector molecules in inducing hydrophobicity can be assisted by activating species such as copper and lead ions that complex the collector on the surface. Previous research has shown that this activation can be inadvertently produced by dissolution and transfer via solution of these ions to the surfaces of mineral phases not intended to float (e.g., Smart 1991; Lascelles and Finch 2002; Finkelstein 1997). Other complex mechanisms can affect both hydrophilic and hydrophobic contributions. They include, for instance a. The extent of liberation of individual mineral phases by grinding or, conversely, the extent of remaining composite particles b. Chemical alteration of the surface layers of sulfide minerals induced by oxidation reactions in the pulp solution 283

flotation0.book Page 284 Tuesday, January 2, 2007 7:36 PM

284

FLOTATION FUNDAMENTALS

c. The presence of a wide range of particle sizes in the ground sulfide ores d. Galvanic interactions between different sulfide minerals to produce different reaction products on the mineral surfaces e. Chemical interaction between particles in the form of aggregates and flocs f. The presence of colloidal precipitates arising from dissolution of the minerals, particularly sulfides, and grinding media g. The mechanism of adsorption of reagents on specific surface sites h. Competitive adsorption between oxidation products, conditioning reagents, and collector reagents The primary purpose of surface characterization is, therefore, to identify these mechanisms with secondary application to the understanding and control of these mechanisms in plant practice. This chapter summarizes some of the advances in surface analytical techniques and knowledge gained from surface characterization applied to flotation systems in the last two decades. There have been major additions to both the armory of instruments and the methodologies available for these studies. The selection of examples from both techniques and methods has been made as objectively as possible but necessarily has resulted in many excellent reports of research and plant surveys (or diagnoses) that have been missed or inadequately described. An attempt was not made to describe the theoretical or practical basis for each technique but, rather, the type of new information that it can provide. In compensation, reference is made to technique descriptions, reviews, and more complete reports in the references list. Some of the techniques surveyed in this chapter are now used in ore and plant surveys to provide part of the metallurgical information on which plant control is based but, more often, to diagnose reasons for losses in recovery or selectivity. Other techniques have been applied to basic studies of processes and mechanisms controlling flotation. Some case studies, combining information from different techniques, have been included to illustrate the now-established place of surface analysis and characterization in flotation research and practice. X - R AY P H O T O E L E C T R O N A N D A U G E R E L E C T R O N SPECTROSCOPIES

X-ray photoelectron and Auger electron spectroscopy techniques (Briggs and Seah 1992; O’Connor, Sexton, and Smart 2003), both analyzing the first 2–5 nm of the surface layers used in static (spot) and imaging modes, have produced primary information on processes b, c, e, f, and g previously described. Sulfide Oxidation: Polysulfides

In sulfide flotation, the mechanisms of surface oxidation and the consequent physical and chemical forms of oxidation products on the surface, which are derived from studies using these surface analytical techniques, can be summarized as • Metal-deficient (sulfur-rich) surfaces, polysulfides, and elemental sulfur • Oxidized fine particles attached to larger sulfide particle surfaces • Colloidal precipitates of metal hydroxide particles and flocs • Continuous oxidized surface layers (e.g., oxide/hydroxide) of varying depth

flotation0.book Page 285 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

285

Adsorbed sulfate and carbonate species Nonuniform spatial distribution with different oxidation rates (e.g., isolated, patchwise oxidation sites; face specificity) Scanning Auger microscopy (SAM) can provide both electron (secondary and backscattered) microscopic imaging for topography and phase identification together with surface elemental mapping, line scans, spot (submicron) analysis, and depth profiles. This type of information is illustrated in Figure 1 where a 50 wt % pyrite/50 wt % chalcopyrite mineral mixture was ground and conditioned in pH 9 nitrogen purged solution for >14 days. A very wide range of particle sizes is evident. Oxidized fine particles 0.1–10 μm are imaged, as well as flocs comprising clumps of loose aggregates with dimensions 1–3 μm, consisting of smaller spheroidal particles each with approximate diameters 0.1–0.5 μm. A similar range of particles is usually seen in samples taken from operating flotation plants. The flocs are not necessarily observed on similar surfaces that have been laboratory-conditioned for shorter periods (i.e., 300-nm signal is from a ferric hydroxide floc. All of these surface species, which are typically in islands or reacted patches of the particle surface, contribute to hydrophobicity. • •

1 MS + xH 2 O + --xO 2 ↔ M 1 – x S + xM ( OH ) 2 2

(EQ 1)

It is now well established that iron-containing sulfide minerals (e.g., pyrite, pyrrhotite, chalcopyrite, pentlandite, arsenopyrite) essentially follow a reaction mechanism similar to that in Equation 1, in that iron hydroxide products and an underlying metal-deficient or sulfur-rich sulfide surface are formed. The seminal work of Buckley, Woods, and their colleagues, using a combination of X-ray photoelectron spectroscopy (XPS) and electrochemical techniques (reviewed in Smart, Amarantidis et al. 2003), has clearly demonstrated this mechanism in single mineral studies. In their work, oxidation of abraded pyrite surfaces exposed to air for a few minutes produced a high binding energy (BE) doublet component of the S 2p spectrum in addition to ferric oxide/hydroxide reaction products (Buckley and Woods 1987). The sulfur product was attributed to an iron-deficient Fe1–xS2 surface layer with the later proposition that polysulfide-like species Sn2– are formed. Specifically, Mycroft et al. (1990) have correlated XPS and Raman spectra of electrochemically-oxidized pyrite surfaces with polysulfide model compounds but only at Eh > 600 mV, pH 5. Recently, monolayer-sensitive synchrotron radiation XPS (SRXPS) and time-of-flight secondary ion mass spectrometry (TOF-SIMS) (discussed in the following sections) have been used to verify the polysulfide formation in surface oxidation under plant conditions. The importance of the Sn2– n>2 polysulfides is that they contribute to hydrophobicity independently of collector addition. Collectorless flotation of pyrite in alkaline solution, correlated to electrochemical oxidation, can be explained by the production of a hydrophobic sulfur-rich surface together with hydrophilic iron hydroxide species (Ahlberg, Forssberg,

flotation0.book Page 286 Tuesday, January 2, 2007 7:36 PM

286

FLOTATION FUNDAMENTALS

A

B

C

Source: Adapted from Smart, Amarantidis et al. 2003.

FIGURE 1 SAM images from pyrite and chalcopyrite particles at successively higher magnifications (i.e., white bar = (a) 100 μm, (b) 10 μm, (c) 5 μm). Oxygen depth profiles for Auger analysis gave approximately 5, 30, >300 (pyrite particle), 80, and 10 nm (chalcopyrite particle), respectively, for each of the five points in (c).

and Wang 1990). After grinding, the surface becomes substantially covered by the hydrophilic species, and no significant flotation is observed without addition of collector. Collectorless flotation can, however, be easily obtained after complexing the iron with ethylenediamine tetraacetic acid (EDTA) in solution, indicating that the underlying hydrophobic sulfur-rich layer is responsible for pyrite flotation under these conditions. Elemental sulfur was not evident at pyrite surfaces exposed to air or neutral to alkaline solutions (Buckley and Woods 1987; Buckley, Hamilton, and Woods 1985). Thin layers of elemental sulfur were, however, observed on pyrite surfaces exposed to aerated, dilute sodium sulfide solutions (Buckley and Woods 1987; McCarron, Walter, and Buckley 1990). The collectorless flotation of chalcopyrite after air exposure or solution oxidation has been directly correlated with the surface composition determined by XPS (Zachwieja et al. 1989). Removal of iron hydroxide species during conditioning in alkaline solution to leave the hydrophobic sulfurrich sulfide surface showed strong flotation. Conversely, oxidized chalcopyrite surfaces reduced in situ became copper deficient and were unfloatable. A specific illustration of the XPS observation of polysulfide formation is found in galena oxidation in pH 9 solution from fresh fracture to 30-minutes aeration (Figure 2) (Smart et al. 2000). The growth of the high BE components of the sulfur S 2p spectra, due to Sn2– species (B n=2; C n>3), is correlated with increasing contact angle and flotation recovery (Prestidge and Ralston 1995). The evidence for assignment of the high BE components of S 2p XPS spectra to metal-deficient, polysulfide defect sites and elemental sulfur has been reviewed (Smart, Skinner, and Gerson 1999).

flotation0.book Page 287 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

B. 10 Minutes Oxidation in pH 9 Solution

A 160.8 eV 100% A

170

168

166

164

162

160

158

Intensity, arbitrary units

Intensity, arbitrary units

A. As Fractured

287

A 160.8 eV 77% B 162.0 eV 12% C 163.1 eV 11%

A C

170

168

Binding Energy, eV

166

164

B 162

160

158

Binding Energy, eV

Intensity, arbitrary units

C. 30 Minutes Oxidation in pH 9 Solution

A 160.8 eV 69% B 162.0 eV 16% C 163.1 eV 15%

A C

170

168

166

164

B 162

160

158

Binding Energy, eV Source: Adapted from Smart et al. 2000.

FIGURE 2

XPS S 2p spectra from galena

Collector Actions

XPS and SAM have also done much to increase understanding of the actions of collector molecules in flotation, which are considerably more complex than the earliest simplistic model of adsorption of the head group and dangling hydrophobic tail. Their actions in several different modes have been studied using XPS and SAM in recent research (Smart, Amarantidis et al. 2003), namely • Adsorption to specific surface sites • Colloidal precipitation of metal–collector species from solution • Detachment of oxidized fine sulfide particles from larger particle surfaces • Detachment of colloidal metal oxide/hydroxide particles and flocs • Removal of adsorbed, oxidized surface layers • Inhibition of oxidation • Aggregation and disaggregation of particles • Patchwise or face-specific coverage The presence of adsorbed xanthate on freshly fractured galena surfaces has been confirmed from both S 2p spectra and the more surface-sensitive X-ray induced Auger spectra (i.e., S LMM and Pb NOO) signals. The work of Buckley and Woods (1991) has correlated xanthate coverage (using voltammetry) with XPS spectra and flotation recovery showing that only a fraction of the monolayer is adsorbed at maximum recovery. Sub-monolayer, perpendicularly-oriented, adsorbed lead ethyl xanthate was confirmed in combined XPS, Fourier transform infrared (FTIR) spectroscopy, and controlled potential studies (Suoninen and Laajalehto 1993). There are now many examples of studies in the literature in which uptake of the collector molecules on the sulfide mineral surface occurs through the formation of colloidal (precipitated) metal-xanthate or metal-hydroxyxanthate species in solution

flotation0.book Page 288 Tuesday, January 2, 2007 7:36 PM

288

FLOTATION FUNDAMENTALS

TABLE 1 Survey of the atomic concentrations (%) of elements for undeslimed pyrite particles conditioned at pH 4, N2 purged Initial Element Carbon (C) Oxygen (O) Iron (Fe) Sulfur (S) Calcium (Ca) Phosphorus (P)

Etch

[DCDTP] = 0 M 28 7.8 28 24 10 27 27 33 3.6 4.6 0 0

Initial

Etch

[DCDTP] = 7.5 × 10–3 M 26 8.3 11 5.2 21 39 40 47 0.45 95%), whereas at pH 8.5, only weak, larger aggregates appear to form, and the recovery does not increase to the same extent (i.e., 60%–80%). In the presence of coarser particles (38–75 μm) at pH 5.5 with Cu(II) addition alone, a very high percentage of the fine particles were recovered and a fine/coarse particle aggregation (i.e., “piggybacking”) mechanism was confirmed by on-line particle size analysis and optical microscopy. At the higher pH 8.5, the interaction between fine and coarse sphalerite is slower and much less complete with correspondingly poorer flotation of both the fine and coarse fractions. XPS surface analysis confirms that this is due to partial coverage of the surface by colloidal hydroxides and overall hydrophilicity inhibiting strong hydrophobic interactions between the particles (Lange, Skinner, and Smart 1997). There are now many case studies in which XPS surface analytical studies, combined with flotation metallurgy and solution chemistry, have directly contributed to improvements in flotation recovery and grade. For example, naturally floatable iron sulfides with graphitic surface layers have been identified and separately removed in copper flotation at Mount Isa, Australia (Grano, Ralston, and Smart 1990). This study also identified the selective removal of ferric hydroxides and carbonates by collector addition. The effects of fine grinding on flotation performance have been surveyed in a correlated XPS-flotation study (Frew et al. 1994) and specific application to the zinc regrind at Cominco Alaska’s Red Dog mine has been reported (Frew, Smart, and Manlapig 1994). The action of an extended period of aerated conditioning before copper activation and collector conditioning in increasing sphalerite flotation at the Murchison Zinc (Australia) concentrator (Kristall et al. 1994) was explained, using XPS, by the removal of zinc hydroxides from the sphalerite surface and the concomitant appearance of a metal-deficient sphalerite surface. XPS also demonstrated an increase in the oxidation state of pyrite after aerated preconditioning. The presence of excessive surface oxidation in copper reflotation at Western Mining Corporation’s Olympic Dam operation (Australia) (Smart and Judd 1994), identified by XPS analysis, led to improved operation of Lasta filters. A low flotation rate of galena in lead roughers at the Hilton Concentrator of Mount Isa Mines was analyzed (Grano et al. 1993, 1996) using XPS. The presence of precipitated species and their removal by a change of conditioning reagents (i.e., lime to soda ash) and collector reagent (i.e., ethyl xanthate to DCDTP, a collector that is stable in the presence of sulfite species over a wide pH range) has been used to address this problem.

flotation0.book Page 290 Tuesday, January 2, 2007 7:36 PM

290

FLOTATION FUNDAMENTALS

S Y N C H R O T R O N R A D I AT I O N X P S Spectroscopic Advantages

Conventionally, obtaining a profile of composition as a function of depth requires either sequential ion beam sputtering and analysis, or the use of angle-resolved photoemission. Sputter profiling is limited because the disruption of chemical bonds beyond the top few monolayers in the surface detracts from any chemical environment information. Differential sputtering can also occur in compounds, resulting in a progressive enrichment of one constituent over another. Angle-resolved XPS can yield nondestructive, surface-sensitive information but requires flat surfaces and depth accuracy requires knowledge of photoelectron inelastic mean free paths in overlying reaction products. The majority of exposed sulfide mineral surfaces are not flat. Indeed, only a few sulfide minerals exhibit good cleavage planes (e.g., galena, sphalerite). It is also rare for the thickness distribution of reaction products to be uniform. The XPS analysis depth can be modulated by controlling the kinetic energy and, hence, the escape depth of the emitted photoelectron of interest. This capability is afforded through the use synchrotron radiation. Current soft X-ray synchrotron beamlines offer high resolution monochromation in the 10–2,000 eV energy range with resolving powers in excess of 104 (E/ΔE), and photon flux that is several orders of magnitude higher than conventional sources. Energy resolution and photon flux allow for full capability of the electron analyzer to be realized, giving superior photoelectron spectral line widths and excellent signal-to-noise ratio in short analysis times. The universal curve of photoelectron escape depth as a function of kinetic energy is reproduced in Figure 3. The maximum surface sensitivity is attained when the kinetic

hν = BE + KE + Φ

1,000

S 2p BE ~ 160 eV

λ, monolayers

100

hν = 1,487 eV

10

1 hν = 210 eV 1

10

100

1,000

Electron Energy, eV

Source: Adapted from Carlson 1975.

FIGURE 3 Universal curve of photoelectron escape depth as a function of kinetic energy. Photoemission energy conservation equation shows the relationship between incident photon ν), photoelectron BE, kinetic energy (KE), and instrumental work function (φ φ). energy (hν

flotation0.book Page 291 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

291

energy of the emitted photoelectron is in the range of 40–50 eV. To achieve this for the S 2p core-line (~160 eV BE), for example, an X-ray energy of approximately hν = 210 eV should be used. These conditions yield an almost fivefold enhancement in surface sensitivity over a conventional Al Kα source of hν = 1,487 eV for the S 2p signal. Surface Sensitivity and Fracture Surfaces

The enhanced surface sensitivity and spectral resolution afforded by SRXPS is exemplified by the analysis of the fracture surface of pyrite (FeS2). Figure 4a and 4b reproduce the S 2p core-line measured using a conventional monochromated Al Kα source (1,487 eV) and a synchrotron source (206 eV), respectively. The narrower line widths are immediately apparent in Figure 4(b). Two new S 2p doublet contributions (peaks a and b) appear to be located at lower BE relative to the main peak. These two new peaks in the pyrite S 2p spectrum are confirmed to derive from surface contributions as their intensities relative to the main peak (bulk peak) decrease with increasing excitation energy (Nesbitt et al. 1998). The interpretation of these surface core-level shifts requires consideration of the molecular orbitals involved in bonding, the mineral structure (e.g., metal and ligand coordination), and the fracture mechanism. A detailed description is not possible within the constraints of this review and the reader is directed to work by Nesbitt et al. (1998) for a comprehensive explanation. Put simply, during pyrite fracture, both Fe–S and S–S bonds may be ruptured. This results in a reduction in the coordination of metal and ligand (in this case, sulfur) sites at the surface and changes in electron density associated with these sites. Electrons involved in bonding are usually retained at the ligand (sulfur) site. The rupture of Fe–S bonds leaves the

A

(i)

S 2p

B

S 2p

(iii)

hν = 206 eV hν = 206 eV

2–

S2O3

hν = 1,487 eV

hν = 1487 eV

S2x(2x–n)– SO32–

(ii)

C

(iv)

D

hν = 210 eV

c

hν = 206 eV

b

hν = 206 eV

III

II

hν = 210 eV a

170

168

166

164

162

Binding Energy, eV

160

170

168

166

164

162

160

Binding Energy, eV

FIGURE 4 Sulfur 2p core-line spectra of fractured pyrite surfaces (a) using a conventional α source, and (b) a 206-eV synchrotron source (Schaufuss et al. 1998). monochromated Al Kα Subsequently reacted pyrite surfaces are shown for (c) air oxidation for 14 hours (Schaufuss et al. 1998) and (d) adsorption of mercaptobenzothiazole from solution (Szargan, Schaufuss, and Rossbach 1999). The spectral contributions from bulk sulfur dimers (S22–) are indicated by the solid circles ( ) above the primary S 2p3/2 component.

flotation0.book Page 292 Tuesday, January 2, 2007 7:36 PM

292

FLOTATION FUNDAMENTALS

surface sulfur dimer site with increased electron density and, hence, will emit photoelectrons of lower BE than the fully coordinated bulk sulfur dimer (peak b in Figure 4b). The formal oxidation state of the monomeric S1– produced by S–S bond rupture reduces to S2– through the associated oxidation of Fe2+ to Fe3+, giving rise to even greater electron density associated with this site and hence an even lower BE (peak a) (Nesbitt et al. 1998). This level of detail is not possible using conventional, laboratory-based XPS instrumentation and the level of surface sensitivity afforded by SRXPS approaches that of static secondary ion mass spectrometry (static SIMS), as well as atomic force and scanning tunneling microscopies (AFM and STM) and low-energy electron diffraction. Moreover, nondestructive and quantitative surface- and bulk-sensitive measurements may be made in the same experiment, simply by varying the incident photon energy. A range of important metal sulfide minerals has been examined by SRXPS over the last decade, looking at surface states and initial reactions. These include, for example, galena, PbS (Paulucci and Prince 1990; Leiro et al. 1998; Kartio et al. 1998); pyrite, FeS2 (Bronold, Tomm, and Jaegermann 1994; Nesbitt et al. 1998; Schaufuss et al. 1998); marcasite, FeS2 (Uhlig et al. 2001); pyrrhotite, Fe1–xS; millerite, NiS (Nesbitt et al. 2001); arsenopyrite, FeAsS (Schaufuss et al. 2000); loellingite, FeAs2 (Nesbitt, Uhlig, and Szargan 2002); gersdorfite, NiAsS (Nesbitt, Schaufuss et al. 2003); covellite, CuS; chalcocite, Cu2S (Laajalehto et al. 1996); and chalcopyrite, CuFeS2 (Harmer et al. 2004). This is by no means an exhaustive list; however, Harmer and Nesbitt (2004) have published a treatment of SRXPS interpretation showing how the sulfide mineral structure and composition determine whether surface reconstruction occurs as a result of fracture and the type of species exposed at the surface (e.g., metal oxidation, ligand polymerization). Recent work, combining ab initio calculations with spectroscopic analysis (von Oertzen, Skinner, and Nesbitt 2005), has confirmed this interpretation for pyrite, chalcopyrite, and molybdenite, MoS2 (von Oertzen, Harmer, and Skinner, in press). This provides a high level of support for the synchrotron XPS interpretation of other important sulfide mineral fracture surfaces. Surface Reaction

The implications of using these kinds of measurements for minerals processing are manifold as they enable the first surface exposed to solution to be probed (i.e., the fracture surface immediately exposed during grinding). The subsequent initial reactions at this surface may also be studied in depth using SRXPS in similar detail. Figures 4c and 4d illustrate further SRXPS examples of oxidation and collector adsorption at pyrite fracture surfaces. From these types of studies, it is possible to follow the relative reactivity of the various surface states exposed on fracture and, in turn, relate this reactivity back to the structure and bonding within the mineral. Other Aspects of SRXPS

In the soft X-ray region, photoionization cross-sections can vary strongly with photon energy. This can be used to great advantage in SRXPS, particularly in valence band studies. By collecting the valence band spectrum of a mineral at several photon excitation energies, it is possible to enhance or diminish the spectral intensity contributions from metal and ligand bonding and nonbonding orbital, thereby identifying them and monitoring their involvement in surface reaction (Nesbitt et al. 2002; Nesbitt, Uhlig et al. 2003). Where possible within the constraints of surface roughness, angle-resolved measurements may also be performed to further enhance surface sensitivity, particularly for adsorbate studies.

flotation0.book Page 293 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

293

SRXPS in conjunction with near-edge X-ray absorption fine structure (NEXAFS) spectroscopy provides an extremely powerful combination for the study of sulfide minerals (Goh, Buckley, Lamb, Skinner et al. 2006; Goh, Buckley, Lamb, Roseberg et al. 2006). NEXAFS yields information on metal and ligand coordination, oxidation state and relative location. It is particularly useful in monitoring organic reagent bonding mechanisms and molecular orientation at the mineral surface. The Future

Much of the methodology for the study of fracture surfaces has now been established, despite the past scarcity of appropriate beamline/end-station combinations for these studies. Recent developments such as the soft X-ray facilities at the Canadian Light Source and the new Australian XPS/NEXAFS end station (initially located at the National Synchrotron Radiation Research Center in Taiwan) have expanded the available capabilities for such investigations. The investigation of the mechanisms of subsequent surface reaction will necessarily provide impetus into the future, together with further instrumental developments (e.g., imaging photoemission, effective charge neutralization for insulating minerals). T I M E O F F L I G H T S E C O N DA RY I O N M A S S S P E C T RO M E T RY

Diagnosis of the surface chemical factors playing a part in flotation separation of a value mineral phase requires measurement of the species that are statistically different between the concentrate and tail streams, together with an estimate (if possible) of the magnitude of the differences. The recently developed statistical methods, based on TOF-SIMS, have moved toward this ultimate aim. This technique used in static mode involves a very low flux of heavy ions impacting surface layers with mass spectrometric analysis of the secondary ions emitted from the surface. In the time of routine measurement, only 1–2 surface atoms in 1,000 are impacted. The secondary elemental and molecular fragment ions come from the first two molecular layers of the surface and provide a very detailed set of positive and negative mass fragments from simple ions (e.g., Na+, OH–) through to molecular ions of specific reagents (e.g., isobutyl xanthate (CH3)2CHOCS2–). Identification of molecular mass peaks for collectors, activators, depressants, precipitates, and adsorbed species is possible with comparative surface concentrations by particle and by phase between feed, concentrate, and tail streams. The pioneering work of Nagaraj and Brinen (Brinen et al. 1993) with TOF-SIMS and the initial statistical analysis of air-dried particles using laser ionization mass spectrometry by Chryssoulis and colleagues (Chryssoulis, Reich, and Stowe 1992) have greatly contributed to the approaches described here. The improvements in the methodology include introduction of the mineral particles without exposure to air (Smart 1991), analysis of surface monolayer (or two), and full statistical analysis of all surface species. These analyses provide a statistical basis for assessment of surface chemical factors that have differentiated particles of a particular mineral phase that have reported to a concentrate from those that have reported to the tail. Validation: Statistical Analysis

The use of TOF-SIMS to quantify changes in surface chemistry has been extensively validated in several ways. The amount of collector (e.g., xanthate, dithiophosphate) adsorbed from solution and monitored by UV adsorption was calibrated against the normalized

flotation0.book Page 294 Tuesday, January 2, 2007 7:36 PM

294

FLOTATION FUNDAMENTALS

TOF-SIMS intensity for the molecular parent ion. A linear relationship up to monolayer coverage was found with a transition to a plateau for multilayer coverage due to the extreme surface sensitivity of the TOF-SIMS analysis (Piantadosi 2001). This is not a serious limitation because most minerals processing plant dosages are sub-monolayer. A second validation of the TOF-SIMS representation of the surface chemistry was conducted in comparison of spectra from polished surfaces of stoichiometric troilite, FeS; iron-deficient pyrrhotite, Fe1–xS; and stoichiometric pyrite, FeS2. The products of air oxidation of troilite and pyrrhotite have previously been shown to include disulfides and polysulfides (Smart, Amarantidis et al. 2003). The experiment was designed to test the representation of this surface chemistry in TOF-SIMS spectra. For the iron-sulfur fragments, the FeS2/FeS ratios for troilite, pyrrhotite and pyrite were 0.59, 1.2, and 32, respectively. A similar sequence of Sn/S atomic fragment distributions confirmed the presence of polysulfides in these slightly reacted surface layers (Smart et al. 2000). Correlation of TOF-SIMS with XPS spectra (Figure 2) for freshly-cleaved galena (PbS) surfaces reacted in pH 9 solution for increasing periods of time has also shown a systematic increase in Sn/S ratios with increasing components of S 2p XPS spectra corresponding to polysulfide formation (Smart et al. 2000). The basis for the methodology of sample preparation, mineral phase recognition, TOF-SIMS analysis, and statistical evaluation has been described in the paper by Piantadosi and Smart (2002). TOF-SIMS images of the particles in total ion yield mode are similar to those shown in Figure 5. Scanning for specific signals (e.g., Pb, Zn, Cu, Fe) can then be used to identify the particles of a specific mineral phase (e.g., galena, sphalerite, covellite, pyrite, chalcopyrite) for specific analysis. The region-of-interest (ROI) facility in the software allows definition of selected particles, as in Figure 5, corresponding to a specific mineral phase with the boundary for analysis set at a fixed position inside the contrast edge. When sufficient particles of the mineral phase have been identified for reliable statistics, a mass spectrum from each particle is recorded and stored. The statistical analysis (Piantadosi et al. Pulses/Pixel: 1

10 μm In

G

In

Sp Sp

In In

In

G

Sp Py

Sp

Sp Ch Py

Ch Py

Py

Sp

Ch Ch Py

Py

Sp

Ch

Ch

Ch Ch

Sp

G

Sp

Sp

Ch

Py

Py

Sp Ch

Sp Sp

Ch

Sp

Sp

Sp

sum of rest 2589522 285

Source: Hart et al. 2004.

FIGURE 5 Principal component analysis identification of mineral phases: pyrite (Py), sphalerite (Sp), chalcopyrite (Ch), gangue materials (G), indium mounting material (In)

flotation0.book Page 295 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

295

2000) then determines a mean value for each atomic and molecular species with 95% confidence intervals for each signal. This analysis was first applied to the effects of calcium ion depression on galena flotation (Piantadosi et al. 2000). Correlation for 26 particle sets of Ca/Pb ratios for concentrate and tail streams gave 0.022 and 0.048, respectively, indicating that there is statistically more calcium (~2x) on galena particles surfaces in the tails compared with the concentrate. A second study (Piantadosi and Smart 2002) of the effect of iron hydroxides and collector, isobutyl xanthate (IBX), on galena flotation compared normalized intensities of IBX in feed, concentrate, and tail streams. There is a clear separation between particles of galena reporting to concentrate (0.005) and tail (0.001). The IBX concentrations on particles in the tail are not statistically different from those in the feed, due to increased hydrophilicity of galena particles in the tail rather than to reduced hydrophobicity. An early attempt to derive a hydrophobicity/hydrophilicity index, based on a ratio of signals from the (hydrophobic) collector (IBX) to (hydrophilic) oxy-sulfur products (SO3) and iron hydroxide (FeOH), gave a value for the concentrate of 44.7 ± 13.7 compared with 7.1 ± 2.4 for the tail, but it is recognized that the index does not include all hydrophobic or all hydrophilic species contributing to the separation. Further comparisons for laboratory separation of galena and pyrite using di-isobutyl dithiophosphate (DIBDTP) collector have also been reported showing statistically ~12 times more collector on galena compared to pyrite. Galena particles reporting to the concentrate show statistically less calcium, lead hydroxide, and oxy-sulfur species on their surfaces compared to tail particles. The early flotation of galena was also considerably assisted by the presence of colloidal as well as adsorbed Pb DIBDTP. Plant Diagnosis

This methodology has now been applied to full ore samples from plant operations including, as examples, Mount Isa Mines (MIM), Ok Tedi Mining Ltd. (OTML, Papua New Guinea), Falconbridge (Strathcona, Canada), Anglo Platinum (South Africa), Mineracao Caraiba (Brazil), and Inco Matte Concentrator (Sudbury, Ontario, Canada)—a total of 18 full statistical analyses to date (Smart, Jasieniak et al. 2003). Examples of results from samples supplied by the client from rougher and rougher scavenger flotation are shown in Figure 6. Poor flotation kinetics were exhibited by fine sphalerite (–10 μm) copper-activated down the rougher-scavenger banks. The study was designed to determine whether the poor flotation response to fine sphalerite was due to differences in mineral surface chemistry rather than hydrodynamic collision frequency factors alone. The process characteristics included pH 10.5 adjusted with lime, collector addition of IBX, and copper sulfate activation. Sphalerite particles in the –10-μm size range were selected using the ROI methodology so that the surface chemistry of this mineral phase was examined selectively. The bars in Figure 6 show the median value of each positive and negative ion signal with the 95% confidence intervals indicated by the smaller intervals at the top of each bar. Comparison between the rougher concentrate, scavenger concentrate, and scavenger tail eliminates all species for which confidence intervals overlap as not statistically significantly different (at least to the first level of statistical analysis). Other signals are clearly statistically different with dissimilar magnitudes in this comparison. In the selected positive ion species, discrimination into the concentrate streams is indicated for Zn, Cu, and Na with discrimination into the tail for increasing Fe, K, Si, Al, and particularly Mg. Low surface concentrations of Ca appear to favor the rougher concentrate but are apparently depressant into the scavenger concentrate and tail. In negative ion SIMS, the concentrates are statistically favored by high exposure of

flotation0.book Page 296 Tuesday, January 2, 2007 7:36 PM

296

FLOTATION FUNDAMENTALS

(+) SIMS, Sphalerite Particles (Slimes) 0.40 Scavenger Tail Scavenger Concentrate Rougher Concentrate

0.35

Normalized Intensity

0.30 0.25 0.20 0.15 0.10 0.05 0.00 Na

Mg

Al

Si

K

Ca

Fe

Cu

Zn

FeOH

(–) SIMS, Sphalerite Particles (Slimes) 0.50 Scavenger Tail Scavenger Concentrate Rougher Concentrate

0.45

Normalized Intensity

0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 C

CH

O

OH

F

S

SO3

IBX

Source: Smart, Jasieniak et al. 2003.

FIGURE 6 Statistical TOF-SIMS spectra for sphalerite particles in the scavenger tail, scavenger concentrate, and rougher concentrate (bars = 95% confidence intervals)

S, CH, and by low surface exposure of F, OH, and O. More detailed examination of the collector IBX signal shows close correlation with the Cu signals. But the oxy-sulfur SO3 signals are not statistically different between the three streams. Comparisons show that the surface chemistry of the fine sphalerite is significantly different between concentrates and tails and even, for some species, between the rougher and scavenger concentrates. The most important difference is the absence of copper exposure and associated IBX on fine sphalerite in the tail stream, indicating low hydrophobicity of these particles. This difference is exaggerated by the presence of high concentrations of Mg, Ca, Al, OH, and F ions, apparently in the form of hydroxides and (alumino)silicates obscuring copper activation. Oxidation to oxy-sulfur species is not a major factor in the depression of fine sphalerite. Calcium ions, in particular, appear to have a depressant role between the rougher and scavenger flotation stages. A second example of this statistical analysis for chalcopyrite in the OTML system has recently been published (Piantadosi, Pyke, and Smart 2001). It has been possible to estimate an average contact angle for this mineral phase by comparison with single mineral studies using the same collector.

flotation0.book Page 297 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

297

Scores on PC# 2 250

0.4 0.3 0.2 0.1 0.0 –0.1 –0.2 –0.3

Scaled 95% Limits are 149.93 and 70.4232

Scores on PC# 2

Zn

PC2

Cu 0

20

40

60

80

Fe

100 120 140 160 180 200 220 240 m/z

2-PC2 2 1

0

1 5 Scaled 95% Limits are 162.531 and 86.1104

Fe

–1 0

20

40

60

80

100 120 140 160 180 200 220 240 m/z

Source: Hart, Biesinger, and Smart 2006.

FIGURE 7 Second PC image scores and factor loadings for positive ion TOF-SIMS image data from the chalcopyrite/pyrite/sphalerite mixture. Top: autoscaled data; bottom: mean centered data. Images are 100 × 100 microns. Positive factor loadings appear bright in image; negative loadings appear dark.

Principal Component Analysis: Phase Recognition and Statistics

A recent improvement in the statistical analysis has been the introduction of principal component analysis (PCA) applied to the mass spectra. Reliable identification of specific mineral particles is central to this statistical analysis. A chalcopyrite/pyrite/sphalerite mineral mixture conditioned at pH 9 for 20 minutes to study transfer of Cu from chalcopyrite via solution to the other two mineral surfaces—because this mechanism can be responsible for their inadvertent flotation in copper recovery—showed no statistical difference in the copper intensities on pyrite and sphalerite (selected from Fe and Zn images) after this conditioning. PCA identifies combinations of factors strongly correlated (positively or negatively) in images or spectra from sets of data. In images, PCA selects these correlations from the mass spectra recorded at each of 256 × 256 pixels in a selected area of particles. In the image mode, PCA has proved to be a much better method of selecting particles by mineral phase with clearer definition of particle boundaries because of multivariable recognition. The first principal component, with factor loadings that are positive in weighting for all masses, is representative of the largest variance in the data set: topography and matrix (ion yield intensity) fluctuations. The second and subsequent principal components (PCs) will then have this variance removed and, as such, are topography- and matrix-corrected. Figure 7 illustrates the selection of sphalerite, chalcopyrite, and pyrite phases in the second PC from each of autoscaled and mean centered calculation modes. The transfer of copper ions from chalcopyrite dissolution to both pyrite (Smart 1991; Hart et al. 2004; Hart, Biesinger, and Smart 2006) and sphalerite surfaces (Finkelstein 1997) is confirmed by the surface analysis but it has also clearly separated a statistical difference in copper intensities between the sphalerite and pyrite phases in favor of sphalerite. The PCA method has been applied to concentrate and tail samples collected from the Inco Matte Concentrator demonstrating extensive CuOH and NiOH transfer between the

flotation0.book Page 298 Tuesday, January 2, 2007 7:36 PM

298

FLOTATION FUNDAMENTALS

chalcocite (Cc) and heazlewoodite (Hz) minerals (Hart et al. 2004). The PCs gave excellent recognition of the two mineral phases with reliable statistics on the regions selected. The close correspondence in these correlations between the isotopes of Cu (63, 65) and Ni (58, 60) gave further confidence in the interpretation. Statistical differences in normalized intensities (values are given in parentheses) illustrate the important discriminating depressant action of NiOH in flotation despite the activation of Hz by Cu transfer. The inadvertent flotation of Hz in the concentrate appears to be a result of Cu activation (0.16). There is also abundant Cu on Hz particles in the tails (0.08), but this is about half that in the concentrate. The Cu distribution between Cc and Hz particles in both concentrate and tails is the same within statistical 95% confidence intervals. The large statistical difference is in the Ni distribution where there is much more (~5×) hydrophilic Ni ions on Cc particles in the tail compared with the concentrate. Hence, Cc in tails appears to be the result of high depressant hydrophilic loadings rather than absence of hydrophobic Cu-collector surface species. The exposure of Cu on Cc particles in the tails as compared with concentrate is ~0.5, corresponding to an increase in Ni exposure of ~7.5. Both Cu activation of Hz and Ni depression of Cc are clearly operating in this system. There is also considerably more collector (>4×) on Cc particles in the concentrate than in the tails. In the tail samples, there is no statistical difference in intensity of the collector signals between Hz and Cc. The possible depressant action of Ca ions is not found to be selective in this surface analysis. Ca is found on both Cc and Hz surfaces in the concentrate and tails in statistically inseparable signals. The reduced chalcocite hydrophobic/hydrophilic ratio is, therefore, related to the presence of Ni ions on the surface, with a consequent reduction in bubble attachment efficiency. Hence, the statistical analysis can be used to confirm some mechanisms and deny others proposed to control recovery and selectivity, giving more focus on the control factors. T I M E O F F L I G H T L A S E R I O N I Z AT I O N M A S S S P E C T R O M E T R Y

Laser probe microanalysis combining laser excitation of small samples with a time-of-flight mass spectrometer dates back to the mid-1960s (Ruckman 1986). The application of the time-of-flight laser ionization mass spectrometry (TOF-LIMS) to analyze solid surfaces was introduced in 1986 by Clarke, Ruckman, and Davey. In 1988, while analyzing pyrite from the Brunswick Cu-Pb-Zn concentrator, it was accidentally discovered that free pyrite particles, devoid of galena inclusions, had significant levels of lead on their surfaces. The analyses showed that lead was confined to the surface and that it was more abundant on floated, as compared to rejected, free pyrites. Thus, the potential of TOF-LIMS to analyze the surface of mineral particles from plant samples in order to explain phenomena such as loss in selectivity during differential flotation, concentrate dilution, and rejection of free valuable minerals, was recognized. Comparative surface microanalysis by TOF-LIMS did become an integral part of several plant surveys but in most cases, this type of work is used for troubleshooting in (Cu)Pb-Zn, (Au)Cu, and platinum-group element (PGE) flotation plants (54 in total during the last 15 years). Sample Preparation

Sample preparation for surface microanalysis with the laser microprobe was kept as simple as possible, on the premise that surface compositional differences and, in particular, their magnitude as opposed to absolute values, dictate the distinct response to flotation of otherwise similar particles. The only requirement is that all samples are treated in exactly the same way.

flotation0.book Page 299 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

299

The sample preparation protocol requires cutting representative samples, filtering, followed by a brief rinse with deionized water to displace mill water, and then air drying. Cold storage in vials under a nitrogen atmosphere is encouraged for samples with readily oxidizable minerals (e.g., galena, pyrrhotite). Gravity separation by panning is used only when required to preconcentrate free particles of rare minerals, thereby facilitating picking under the stereoscope (e.g., gold, PGE minerals). Typically, 20 to 30 particles of the mineral of interest are picked from each sample or from two size fractions of the same sample when compositional differences between coarse and fine particles are being investigated. Analysis

The commercially available laser microprobe instruments, the LIMA-2A, and its successor the PRISM, use Cassegrain optics, which permit specimen illumination and viewing, laser irradiation, and ion extraction—all normal to the sample surface (Figure 8). This unique feature is ideal for rough surfaces such as mineral particles because it minimizes topographic effects and increases elemental sensitivity. The use of a two-laser system (Figure 9) for the nonresonant multiphoton ionization (NR-MPI) of neutrals ablated by the laser microprobe technique (Schueler, Odom, and Evans 1986) allows for the decoupling of the sampling from the ionization step. This, in turn, enables an increase in surface sensitivity to the point where only copper is detected from a copper-activated sphalerite particle (i.e., monolayer detection), with a concurrent increase in elemental sensitivity (yielding minimum detection limits in the 1–50-ppm range (Figure 10). The analytical spot size (1 to 30 μm) and surface sensitivity (0.001 to 0.025) are inversely related to the degree of focusing and the power of the ablator laser. Laser probe microanalysis is very fast, with more than 100 analyses performed and processed per hour. Thus, large data sets, typically 200–400 strong, are collected and form the basis of t-test comparative statistical analysis, which is used to identify and rank activators and depressants (Bolin, Chryssoulis, and Martin 1997). Rotational factor analysis, a multivariate statistical analysis program, is used to validate t-test findings by analyzing data groupings based on factor loadings. The TOF-LIMS technique in the NR-MPI mode is ideally suited for elemental surface microanalysis. In the single-laser negative ion mode, simple radicals (OH, CO3, SO3, SO4, AsO4, FeOHx) associated with surface oxidation can be easily detected. Collector identification and loading measurements, although possible by TOF-LIMS (Chryssoulis et al. 1995), is preferably done by the complementary TOF-SIMS and vacuum UV surface analysis by laser ionization (VUV-SALI) techniques, as they are more sensitive for organic surface microanalysis because of better preservation of molecular ions (Chryssoulis, Weisener, and Dimov 1995). The use of TOF-LIMS to quantify changes in surface composition has been extensively validated using several elements (e.g., Cu, Pb, Au, Ag) and minerals (e.g., pyrite, sphalerite, carbonaceous matter). Two examples are the linear relationship of TOF-LIMS data on surface Cu on sphalerite with the milligrams of Cu consumed, determined from solution assays (Chryssoulis, Kim, and Stowe 1994); and the surface concentration values of copper (in atomic %) measured by XPS (Stowe et al. 1993). Quantification of TOF-LIMS surface data is possible using minerals loaded to different degrees with the element of interest or with the help of relative sensitivity factors determined under standardized conditions (Dimov and Chryssoulis 1997). Surface characterization by TOF-LIMS and, in fact, any other surface microanalytical technique, is usually not a standalone investigation. In most cases, it is a followup investigation on a liberation study, which identified liberated minerals of readily

flotation0.book Page 300 Tuesday, January 2, 2007 7:36 PM

300

FLOTATION FUNDAMENTALS

Ionizing Laser

Ablator Laser and Light Path

L1 L2 L3 Target

Ions to TOF

x y z

Cassegrain Optics

FIGURE 8

Ion Lens

Schematic of TOF-LIMS sample analysis region Ablator Laser Ablated Ions and Neutrons Particle

Ion Lens

Ion Deflector Ion Trajectory

Sample Holder

Focusing Optics Ion Reflectron Ion Detector

Ionizing Laser

FIGURE 9

TOF-LIMS principle of operation

floatable size classes to be a significant fraction of losses to tails or a concentrate grade loss or misplacement to the wrong concentrate. Several case studies are discussed in the follow sections to illustrate this point. Brunswick Cu-Pb-Zn Concentrator

In the Brunswick Cu-Pb-Zn concentrator (British Columbia, Canada), two long-standing issues have been the premature flotation of sphalerite in the primary Cu-Pb circuit, culminating in the production of a bulk Zn-Pb concentrate (Figure 11) and the dilution of the final Zn concentrate by pyrite. Several detailed liberation studies on extensive plant surveys documented that floated free particles are a significant part of both issues. TOF-LIMS studies did consistently show, firstly, significant amounts of Pb on the surfaces of the sphalerite and pyrite particles and, secondly, that this Pb had a major effect on the flotation of both minerals (Kim, Chryssoulis, and Stowe 1995). The surfaces of sphalerite and pyrite from competent run-of-mine (lump-size ore from the semiautogenous-grinding-mill feed) are clean of surface contaminants in contrast to conveyor belt fines (20 μm) and fine ( n3

θ

n3

Evanescent Field

Incident Phase, n1, k1

n1

d

Film, n 2, k 2

θ > θc θc = sin–1 (n 3/n1)

n1 > n3

θ

Third Phase, n 3, k 3 θ

FIGURE 18

309

θ

Typical FTIR/IRS experimental setup

Kellar, Cross, and Miller 1989; Simon-Kutscher, Gericke, and Huhnerfuss 1996; Jang and Miller 1995; Tejedor-Tejedor and Anderson 1990; Buffeteau et al. 1999; Lee and Sung 2001; Ren and Kato 2002; Buffeteau et al. 2001; Tickanen, Tejedor-Tejedor, and Anderson 1997) have been quantitatively measured from the FTIR/IRS spectra. One advantage of FTIR/IRS over other sampling techniques is that surfactant adsorption at the IRE surface can be studied by in-situ experiments. Generally, the advantage of in-situ investigation of surfactant adsorption is that it offers both the adsorption kinetics and chemical information regarding the adsorption state. The adsorption density equation was originally derived by Sperline, Muralidharan, and Freiser (1987) and later modified and applied to flotation chemistry research by Miller and co-workers ( Jang and Miller 1993; Kellar, Cross, and Miller 1989). The adsorption density equation in one form is A – NC b ε d e Γ = 10 7 ----------------------------------------------------N ε ( 2d e ⁄ d p + p ⁄ cos φ )

(EQ 2)

where Γ= A= N= Cb = ε= de = dp = p=

adsorption density, μmol/m2 integrated absorbance, cm–1 Avogadro’s number bulk solution concentration, mol/L absorptivity, L/mol·cm2 effective thickness of the sample, cm penetration depth, cm number of adsorbate-coated IRE entrance and exit faces

Adsorption density measurements using the FTIR/IRS technique (Figure 19) have shown that at a fluorite surface, monolayer coverage can be achieved at very low initial oleate concentrations (below 1 × 10–6 M). Such behavior is characterized by a monolayer plateau in the adsorption isotherm but is not the case for calcite. For fluorite, chemisorption is clearly the dominant adsorption mechanism at low oleate concentrations with oleate bonding directly with calcium sites at the fluorite surface. Chemisorption also occurs to a more limited extent (incomplete monolayer) at the surface of other calcium minerals, such as calcite and apatite for low oleate concentration. On the other hand, at higher concentrations of oleate and with calcium ion, the growth of multilayers and/or the physical adsorption of calcium dioleate collector colloids are observed from the adsorption isotherm measured using FTIR/IRS spectra.

flotation0.book Page 310 Tuesday, January 2, 2007 7:36 PM

310

FLOTATION FUNDAMENTALS

Adsorption Density, μmol/m 2

1,000

100

10

1 Calcite Fluorite 0.1 10–6

10–5

10

–4

10–3

10–2

Equilibrium Oleate Concentration, M

FIGURE 19 Adsorption isotherms of oleate at semisoluble mineral surfaces as determined by direct FTIR/IRS analysis of the surface

Of course, in-situ FTIR/IRS has also been used extensively for qualitative studies of surfactant adsorption at mineral surfaces (Young and Miller 2000; Free and Miller 1996; Lu and Miller 2002). In these studies, high S/N ratios were obtained. These studies have also provided considerable information regarding surfactant adsorption thermodynamics and kinetics. However, it is difficult to obtain information regarding interfacial water structure from in-situ FTIR/IRS experiments (Hancer, Sperline, and Miller 2000). O T H E R V I B R AT I O N A L S P E C T R O S C O P Y T E C H N I Q U E S

Vibrational spectra, as well as infrared spectroscopy, can be obtained by other spectroscopic techniques such as Raman and sum-frequency generation (SFG). Although it is difficult to obtain detailed information about natural mineral surface composition at monolayer coverage by these experimental techniques, they present other advantages, which are described in the following sections. Raman Spectroelectrochemistry

Raman spectroscopy is an in-situ technique that is well suited to the investigation of a mineral surface. In Raman spectroscopy, the spectrum of the collected light comprises a series of discrete bands at frequencies higher to (anti-Stokes) or lower than (Stokes) the elastically scattered (Rayleigh) laser line. These lines are said to be Raman shifted from the Rayleigh line, and the shift is equal to vibrational energy of the transition. Minerals may exhibit vibrational spectra because of the structure of the molecular units that they contain as in the case of S–S in pyrite, or the spectra may be derived from the vibrational structure of the mineral crystal (sphalerite) (Hope, Woods, and Munce 2001). Many common materials (e.g., glass and water) are transparent to visible and near-infrared radiation, and they are weak Raman scatterers in the wavelengths where the majority of laser excitation sources operate. The dependence of the Raman shift on the vibrational and rotational transitions of the scattering molecules means that some minerals may not exhibit a significant Raman spectrum, galena and chalcocite being Raman-inactive sulfide minerals. Raman scattering is an inefficient process with less than 1 in 106 photons undergoing a Raman interaction. Typically, this requires a sample thickness of 5 to 50 nm for the spectra

flotation0.book Page 311 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

311

to be detectable with the most sensitive modern commercial instruments. In spectroelectrochemical studies of collector adsorption, electrodes of gold, silver, or copper have been used for surface-enhanced Raman scattering (SERS). This technique enables the formation of a monolayer of adsorbed collector molecules to be characterized. SERS has been used to investigate the adsorption of a range of sulfide mineral flotation collectors: ethyl xanthate (Woods and Hope 1998; Woods, Hope, and Brown 1998a, b); isopropyl, isobutyl, and isoamyl xanthate (Hope, Watling, and Woods 2001b); O-isopropyl-N-ethylthionocarbamate (Woods and Hope 1999); 2-mercaptobenzothiazole (Woods, Hope, and Watling 2000; Hope, Watling, and Woods 2001a); di-isobutyldithiophosphinate (Hope et al. 2003; Hope, Woods and Watling 2003); and butyl ethoxycarbonylthiourea (Hope , Woods and Watling 2001; Hope et al. 2004). In the case of the xanthates, the collector adsorption can be controlled using the electrode potential. Spectra can be obtained from surfaces prior to collector adsorption, with partial monolayer coverage and for multilayer coatings. This behavior can also be observed with thiocarbamates. The Raman spectra of these collectors were consistent with adsorption of the collector molecules on the surface and intact, through a collector sulfur group. Decomposition of a collector was only observed under applied potentials greater than the solution potentials encountered in typical sulfide mineral flotation. Mercaptobenzothiazole, di-isobutyldithiophosphinate and butyl ethoxycarbonylthiourea flotation reagents adsorbed onto the metallic electrode surface throughout the accessible potential range. The Raman spectra of the deposits formed on the electrode surface were very close to the spectra of the relevant metal compounds prepared in bulk. Metal-collector compounds could be extracted from the electrode surface after extended reaction times and, when characterized by Raman, FTIR, and nuclear magnetic resonance, were found to be the same as the bulk compounds. Results indicate that sulfide mineral flotation reagents can act either through the chemisorption of collectors (xanthates and thiocarbamates), or through the formation of a metal compound on the mineral surface (mercaptobenzothiazole, di-isobutyldithiophosphinate and butyl ethoxycarbonylthiourea). Sum-Frequency Generation

SFG was first introduced in 1987 by Shen’s research group (Zhu, Suhr, and Shen 1987). Since then, SFG has been further developed into a surface-specific vibrational spectroscopy and used to study the vibrational modes and orientations of molecules and monitor reactions at interfaces. SFG is a nonlinear optical process, where the signal is generated at a frequency that is the sum of the frequencies of two incident optical fields due to the nonlinear interaction of infrared and visible lasers as shown in Figure 20 (Shen 1989). SFG is forbidden in a medium with inversion symmetry, but this nonlinear optical process occurs at surfaces where the inversion symmetry is broken. Most bulk materials have inversion symmetry; thus, they do not generate SFG signals. This unique feature makes SFG a sensitive and powerful tool for the study of various interfaces and surfaces (Nickolov, Wang, and Miller 2004). The SFG spectrum of molecules with long hydrocarbon chains is very sensitive to hydrocarbon chain order—loosely packed chains and disordered monolayers will, in general, have more random orientations of the methyl and methylene groups, and the intensity of the SFG signal will be much smaller than in the case of an all-trans state. When an alkyl chain is in an all-trans conformation, it is locally centrosymmetric around the C–C bond, and the CH2 symmetric stretching mode at ~2,850 cm–1 is SFG inactive and does not

flotation0.book Page 312 Tuesday, January 2, 2007 7:36 PM

312

FLOTATION FUNDAMENTALS

ωvis

ωsum

ωIR

Sample Surface

FIGURE 20

SFG generation

appear in the spectra. When the chains are disordered because of gauche defects, this local inversion symmetry is lifted, and a peak at 2,850 cm–1 will appear in the spectra to an extent that depends on the degree of disorder. The ratio of intensity of the CH3 symmetric stretching to CH2 symmetric stretching can be used to provide a relative measurement of hydrocarbon chain ordering (Conboy et al. 1997). The peak ratio method, which directly depends on intermolecular and intramolecular symmetries, therefore provides the most quantitative measure of chain conformation in the monolayer (Smiley and Richmond 1999). In a closely packed surfactant monolayer in all-trans conformation, there are only two dominant bands at 2,875 cm–1 and ~2,940 cm–1 that correspond to the symmetric stretching vibration and the symmetric stretching Fermi resonance with a bending overtone of the CH3 group, respectively. In the following discussion, the acronyms SSP and SPS (where S and P represent polarization that is perpendicular and parallel, respectively, to the incident plane) designate the state of polarization of the beams in the following order: output sum frequency beam, input visible beam, and input IR signal beam. For example, Figure 21 shows the spectrum for an oleic acid Langmuir–Blodgett (LB) film at the surface of fused silica (Wang 2004). The spectrum suggests that the oleic acid forms a closely packed monolayer in all-trans conformation because there are only two dominant bands at 2,878 cm–1 and 2,940 cm–1 which correspond to the symmetric stretching vibration and the symmetric stretching Fermi resonance with a bending overtone of the CH3 group, respectively, in the SSP polarization spectrum. Shown in Figure 22 are the spectra taken from the interface of CaF2 and a hydroxamic acid D2O (10–3 M hydroxamic acid) solution (Wang 2004). The spectrum for the SSP polarization state is similar to that for a monolayer of hydroxamic acid at a fused silica surface (Wang 2004). The CH2 symmetric stretching is observed at 2,850 cm–1. The peak at 2,877 cm–1 is due to the CH3 symmetric stretching. The Fermi resonance of CH3 symmetric stretching appears at 2,940 cm–1. A small peak at 2,920 cm–1 is assigned to CH2 asymmetric stretching. For the SPS polarization state, the major peak is due to CH3 asymmetric stretching at 2,961 cm–1. The peak from the CH3 symmetric stretching appearing at 2,880 cm–1 for the SPS polarization state indicates that the CH3 symmetric axis has a certain angle with respect to the surface normal. The SFG intensity from the CH2 gauche structure is greater than that for the LB-transferred monolayer. The strong CH3 symmetric stretching peak that appears in the SPS spectrum means there is a significant tilt angle between the C– CH3 axis and the surface normal.

flotation0.book Page 313 Tuesday, January 2, 2007 7:36 PM

4.5

9

4.0

8

SFG Intensity, arbitrary units

SFG Intensity, arbitrary units

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 2,800

2,850

2,900

2,950

3,000

Wavenumber, cm –1

FIGURE 21 SFG spectra for oleic acid LB film at the surface of fused silica under SSP polarization conditions

313

7 6 SSP

5 4 3 2

SPS

1 0 2,750

2,850

2,950

3,050

Wavenumber, cm –1

FIGURE 22 SFG spectra taken from hydroxamic acid at D2O solution/CaF2 interface SSP and SPS polarization conditions

M I C R O S C O P Y A N D M I N E R A L O G Y : PA S T, P R E S E N T, AND FUTURE

For many years, traditional descriptive mineralogical examination has been performed using polished sections of samples. Determination of the mineral quantities (modal analysis) and liberation analysis in ores and mineral processing samples has been a basic requirement. The era of quantitative mineralogy (quantitative modal analysis and quantitative liberation analysis) began in 1848 when Delesse published a method to determine the mineral quantities based on the areas of the minerals of interest in a polished section. The method was very labor intensive: it involved tracing the outlines of the mineral grains onto a cloth, sticking this cloth onto a tin foil, cutting the tin foil, sorting according to the different minerals, and then measuring the relative weighting of each group of foil cutouts. The ratio of the weights of the tin foil cutout minerals to the weight of the original tin foil was proportional to the area of that mineral in the sample surface and was correlated to the volume percent and the weight percent. In 1898, Rosiwal published an improved method to determine the mineral quantities in polished sections. Rosiwal’s method, which became known simply as “cord analysis,” involved a grid of parallel lines covering the image of the specimen surface that were used to determine the linear length of intercepts onto each mineral of interest. From the addition of the linear intercept in the mineral of interest and the total length of the lines, the linear percent of the mineral was calculated, which was correlated to the volume percent and the weight percent. In 1934, Glagolev published an improved method for determining minerals quantities. Glagolev’s method, which became known simply as “point counting,” involved manual counting of the points in a grid that coincided with a mineral of interest in a polished section observed with an optical microscope. The ratio of the points on a mineral of interest to all points on the examination grid was equivalent to the volume fraction of that mineral, which was correlated to the weight percent. The advent of affordable computers in the 1970s allowed the development of systems that improved the speed for determining the mineral quantities. Basically, the methods developed by Delesse, Rosival, and Glagolev (area grade, linear grade, and point counting) are behind the computerized systems used to determine the mineral quantities. The advent

flotation0.book Page 314 Tuesday, January 2, 2007 7:36 PM

314

FLOTATION FUNDAMENTALS

of computers also provided the required speed to quantitatively determine the mineral liberation, which basically involves the measurement of the particle mineral composition. A review of the capabilities of early image analysis is given in works by Martens, Morton, and McCarthy (1978) and Taylor et al. (1978). Many of the developments of early image analysis were spurred by the needs of metallography. In 1974, the CANMET Mining and Mineral Sciences Laboratories (CANMETMMSL) in Canada adapted a Quantimet system to perform quantitative mineralogy. Petruk (1976) described this system, which was interfaced to an optical reflected light microscope with a black-and-white video camera, and performed measurements of areas of mineral grains. Other systems interfaced to optical microscopes were used in other parts of the world. However, system limitations were soon recognized in that many minerals could not be properly discriminated. Very early in the development of systems for quantitative mineralogy, it was realized ( Jones 1984) that minerals could be better identified automatically by using scanning electron microscopes (SEMs) or electron microprobes (EMPs) with X-ray detectors for elemental analysis. Si(Li) energy-dispersive spectroscopy (EDS) detectors or wavelength-dispersive spectroscopy (WDS) detectors were used. The system developed by Jones at Imperial College, London, United Kingdom ( Jones 1984), was based on an electron microprobe (Camebax) with WDS detectors. The system identified the minerals using linear scans at a pixel size of ~2 μm and the elemental X-ray signals from the WDS detectors. Measurement time was ~10 msec per pixel if the mineral was identified from the presence or absence of an element, and 100 msec if more complete elemental information was required. The identification of a single mineral grain requires acquiring and processing information from10 to 30 pixel spots. The QEM*SEM system was developed by the Commonwealth Scientific and Industrial Research Organization (CSIRO) in Australia. It was initially presented by Grant and colleagues in 1976. The present configuration, commercially available, is called QEM*Scan. The system is based on a SEM and identifies minerals mainly based on their elemental X-ray counts obtained from EDS detectors. EDS detectors have a dead time that is a function of the count rate. The first EDS detectors had analog pulse processors and could process a maximum of 25,000 counts per second (cps). Although it was possible to purchase QEM*SEM systems with one EDS detector, to increase its speed the first full configuration of the QEM*SEM had four EDS detectors with analog pulse processing. In the 1990s, EDS detectors with digital pulse processors became available and could process a maximum of ~50,000 cps (the actual count rate of an EDS with analog pulse processors is ~2,000 cps, whereas the actual count rate for an EDS detector with digital pulse processor is ~20,000 cps). The present configuration of the QEM*Scan has four EDS detectors with a digital pulse processor. In general, the system uses a grid of points that are superimposed on the image of the particles. From each of these stop points, the system acquires elemental X-ray information from the group of EDS detectors. The EDS information is compared with a database of minerals, and the mineral at the point is identified. The acquisition time per stop point is fast, ~10 msec. However, because a mineral grain may comprise 10–30 stop points, the overall speed of the system to process several thousands of mineral grains is slow. To partially address speed concerns, the system has different operational modes to perform bulk mineralogical analysis or to perform a more detailed mineralogical analysis including the liberation analysis. In general, the main difference is the spacing of the stop points to acquire EDS X-ray data.

flotation0.book Page 315 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

315

In 1984, a system was integrated at the CANMET-MMSL (Petruk 1988) to determine mineral quantities and liberation analysis by measurements of areas of mineral grains and particles. The system was based on a programmable generic image analyzer interfaced with an electron microprobe ( JEOL 733). The highly stable electron beam current in the microprobe allows the system to identify the minerals using mainly the backscattered electron (BSE) image. Acquisition of BSE images is much faster than acquisition of X-ray elemental information. The first generation of the system (1984) was based on an IBAS image analyzer (Kontron) hosted in an MCP computer (Motorola); this system allowed identification and measurement of ~5,000 particles per hour. The last generation of the system is based on a KS400 image analyzer (Zeiss Vision) hosted in a computer that allows identification and measurement of ~100,000 particles per hour. There are ~5,000 known minerals; among those are many minerals with similar average atomic numbers (Zav) and, thus, similar BSE gray levels. However, a given sample does not have 5,000 minerals but probably only ~20 minerals and of those, commonly, only five minerals are of interest. Thus, for many samples, it is possible to use only the information from the BSE image to perform the image analysis. For example, BSE images allow discrimination between pyrrhotite (Fe1–xS, x~0.2 with Zav = 22.1) and troililite (FeS, with Zav = 22.4), and discrimination between magnetite (Zav = 21.0) and hematite (Zav = 20.1). There are some cases where there are minerals of interest with overlapping BSE gray level. For these cases, the system at CANMET-MMSL can obtain elemental X-ray information from any of its X-ray detectors. The speed of the system is reduced when the image analysis requires the use of X-ray data. The first generation of the system had two WDS detectors and one EDS detector. The image analyzer automatically controls the electron beam and obtains the elemental X-ray information from any of the X-ray detectors. The system can acquire X-ray information from the whole field, or from rectangular windows around the mineral grains or from stop points similar to the QEM*Scan. Thus, the system can identify mineral grains by a combination of BSE imaging and elemental X-ray data. The system could also be operated in a mode fully similar to the QEM*Scan, although it would rarely be necessary (i.e., identifying all minerals based on elemental X-ray information from stop points). In this latter mode, the first generation of image analyzer at CANMET-MMSL was slower than the QEM*Scan, because it used X-ray data from a single X-ray detector (either EDS or WDS). WDS detectors have better energy resolution than EDS detectors (~5 eV vs. ~140 eV, respectively; e.g., Goldstein et al. 1994). Thus, considering the whole population of ~5,000 existent minerals, there will be minerals whose elements overlap in the EDS spectra but can be well resolved using WDS detectors. However, WDS detectors have focusing limitations (i.e., the signal intensity is not homogeneous for the entire field of view at magnifications lower than ~300×). X-ray information can be acquired very quickly by deflecting the electron beam to a required position at the field of view. This can be done for any magnification using an EDS detector. However, this cannot be done for magnifications lower than ~300× using the WDS detectors, because the count rate for a given mineral will be different at the extreme corners of the field of view. It is possible to automatically control the stage by moving it to a given point to the center of the viewing field and obtaining the X-ray data from a WDS detector. This is known as stage-scanning and is a common feature of modern controllers of electron microprobes. The image analyzer at CANMET-MMSL can be programmed for stage-scanning and using WDS data at low magnifications for minerals that cannot be discriminated by EDS information. However, the speed of the system is further reduced. Another possibility is to acquire BSE images and WDS information at magnifications of

flotation0.book Page 316 Tuesday, January 2, 2007 7:36 PM

316

FLOTATION FUNDAMENTALS

300× or higher and then zoom down the images to a lower magnification that would be better suited to the size of the particles in the sample (Matos, Lastra, and Petruk 1996). Since 1984, the operational philosophy with the system at CANMET-MMSL has been to discriminate as many minerals of interest as possible using only BSE information and combine the information from EDS or WDS only when needed. Since 1984, having analyzed several thousand samples, the overall experience at CANMET-MMSL has been that ~80% of the sample cases can be analyzed using BSE information only, ~10% can be analyzed by adding single elemental information from the EDS detector, ~5% can be analyzed by adding multiple elemental information from the EDS detector, and another 5% of the sample cases require adding information from the WDS detectors. There is another advantage of performing quantitative mineralogy based on BSE imaging. This is related to the excitation volume than can be demonstrated from first physical principles. X-ray signals from electron-based instruments do not come from an area on the sample surface; rather, they come from an excitation volume. This excitation volume is a function of the accelerating voltage of the electron beam and the density of the sample point. In general terms, the diameter of the excitation volume for X-ray signals from minerals is ~3 μm at 20 kV of accelerating voltage. On the other hand, the BSE signal provides a spatial resolution of ~0.1–0.2 μm. Thus, for grains smaller than ~3–5 μm, the X-ray signals will give information that may not be appropriate to properly identify the corresponding mineral. Similarly, because the spatial resolution of X-ray signals is ~3 μm for minerals, borders of mineral particles and borders of minerals within particles may not be properly identified. In the late 1990s, QEM*SEM became QEM*Scan based on a LEO (Zeiss) SEM. At about the same time, Kontron imaging was acquired by Zeiss, whereby Kontron interfaces for SEM instruments and electron microprobes became unavailable, and the Julius Kruttschnitt Mineral Research Centre ( JKMRC, Australia) developed the Mineral Liberation Analyser (MLA) (Gu 2003). The MLA is based on a SEM (FEI, formerly Philips), and is commercially available. The MLA is similar to the system at CANMET-MMSL. The MLA discriminates many minerals of interest using only BSE information and combines with information from EDS only when needed to discriminate certain specific minerals. When analyzing only BSE images, the system is faster than the QEM*Scan. The MLA has up to two EDS detectors; therefore, when identifying minerals based solely on X-ray information, it is slower than the QEM*Scan. In 1997, a new type of X-ray energy-dispersive detector for SEM instruments and electron microprobes was introduced. This new technology is based on the silicon drift chamber (SDC) detector (Figure 23). The SDC detector does not require cooling by liquid nitrogen, and its count rate capability is extremely high at ~400,000 to 1,000,000 cps. Thus, one SDC detector at a count rate of 400,000 is equivalent to ~8 conventional EDS detectors with digital pulse processor (Figure 24). Initially, the SDC detectors had a beryllium window and could not detect elements lighter than ~Na. In addition, the energy resolution was not as good as that obtained by conventional EDS detectors. Thus, the electron microscopy community was not very enthusiastic in adopting SDC detectors. Presently, SDC detectors are available with a polymer window and can detect elements lighter than Na. In addition, by compromising on the count rate, SDC detectors can achieve better energy resolution than conventional EDS detectors. It is also possible to obtain counts from a combination of four SDC detectors to acquire even higher count rates. The present configuration of the image analyzer at CANMET-MMSL is still based on an electron microprobe with two WDS detectors and one EDS detector, but it has an

flotation0.book Page 317 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

317

450

p+

400 Last Ring

n– S i

Back

Count Rate Capabilities, kcps

Clear Anode Field Effect Transistor Ring #1

350 300 250 200 150 100 50 0 Analog

FIGURE 23 detector

General scheme of the SDC

Digital

SDC

FIGURE 24 Differences in count rate capabilities of different EDS detectors at similar energy resolutions (analog = conventional EDS detector with analog pulse processor; digital = conventional EDS detector with digital pulse processor; SDC = silicon drift chamber detector)

additional SDC detector (Röntec). The SDC detector allows acquisition times of 10 msec per pixel and shorter times if electron beam currents of at least 15 nA are used. In full BSE mode, the system at CANMET-MMSL processes ~100,000 particles per hour. In full X-ray mode, it processes ~20,000 particles per hour; it is presently faster than any commercially available systems. Speed is important, especially when considering the analysis of tailing samples that contain very few grains of the mineral of interest, a common situation when investigating the losses of valuable minerals. Electron microprobes are instruments that are designed to provide very stable electron beam currents. A stable electron beam current is a basic requirement for EMP analysis. The electron beam current in an EMP is stable over a period of weeks and can be automatically monitored and maintained constant every second. On the other hand, the SEM is basically an imaging instrument and typically provides less-stable electron beam currents. In many SEM instruments, the electron beam current varies over periods of 1 hour (Gu 2003). To fully exploit the speed of quantitative mineralogy based on BSE imaging, the instrument must provide a stable electron beam current over the full time of sample analysis. To analyze a set of samples or to search and analyze precious minerals, the electron beam current must be stable during a period of up to 24 hours. In addition, the EMP can yield higher beam currents than the SEM. Higher beam currents yield better contrast between minerals in the BSE image. Also, EMP instruments commonly have four large ports and two additional auxiliary ports. These six ports can be used to connect a range of X-ray detectors. Thus, EMP instruments are better suited for quantitative mineralogy studies. Without budget constrictions, the future instruments for quantitative mineralogy studies will be EMPs with two WDS detectors, each with four analyzing crystals and four SDC detectors. The cost of these instruments could be similar to the cost of the present commercially available systems for quantitative mineralogy studies. Of course, there will be some simpler instruments for basic requirements and limited budgets.

flotation0.book Page 318 Tuesday, January 2, 2007 7:36 PM

318

FLOTATION FUNDAMENTALS

X-RAY ABSORPTION SPECTROSCOPY

Practically, X-ray absorption spectroscopy (XAS) must be carried out on a synchrotron radiation source to provide the wide X-ray wavelength range and high intensity required. XAS is primarily of use in mineral flotation studies where information about the form of trace elements is required; that is, the form of the element that is of interest is not obscured by the signal from the same element within the bulk sample. However, in certain modes, XAS can also be used to probe surface phenomena where the element of interest is also present in the underlying specimen. Surprisingly, XAS has been used relatively rarely for mineral flotation studies, although there are a number of advantages to its use as compared to techniques that require a vacuum environment. There is generally little sample preparation required, and the sample turnaround time is rapid as most measurements can be carried out in air. XAS measurements lend themselves to the use of environmental cells and provide both spatial and electronic structure information. XAS can be divided into several subtechniques, which are more commonly referred to in the literature (for a general review of XAS and other synchrotron techniques, see Gerson, Halfpenny et al. 1999). In general, XAS is divided into two categories: XANES (X-ray absorption near-edge structure), and EXAFS (extended X-ray absorption fine structure). The former occurs up to approximately 40 eV above the absorption edge and is the result of excitation from the valence to conduction bands. XANES is sensitive to the coordination geometry and oxidation state. EXAFS results from scattering of excited photoelectrons off neighboring atoms and occurs at higher incident energies than XANES. EXAFS is sensitive to local structure out to approximately 5 Å. XAS can be measured in several modes. Most commonly, the X-ray fluorescence (measured at 90º to the incident X-ray beam) or transmitted intensities are measured. Transmission measurements enable the bulk sample to be probed, whereas fluorescence measurements are more surface sensitive with a measurement depth dependent on the energy of the fluorescence yield (see, e.g., Kasrai et al. 1996). However, both measurement modes can be used effectively to study surface-related phenomena where the element under investigation is surface specific. More surface-sensitive measurements can be obtained using total electron yield (TEY), partial electron yield (PEY) or reflection EXAFS (REFLEXAFS). TEY is measured via the drain current experienced on electron excitation to the continuum. The depth of analysis of this form of XAS is dependent on the incident X-ray energy. Thus, for the Si L-edge (95–120 eV) the depth of measurement is about 50 Å, whereas for the Si K-edge this would be approximately 700 Å (1,830–1,900 eV) (Kasrai et al. 1996). PEY is obtained via the measurement of electron current induced on a conducting grid a short distance from the sample surface. A voltage can be applied to the grid to repel low-energy electrons (i.e., those resulting from deeper within the sample). REFLEXAFS requires the incident beam to intercept the surface at a very low angle, thus enabling reflection rather than absorption. An angle of 100 millidegrees, half of the critical angle for chalcopyrite, has been adopted (England et al. 1999b). In this instance, the depth of penetration was less than 50 Å. The data can be collected via measurement of the intensity of the reflected beam or the fluorescence yield at 90° to the sample surface. This enables surface-sensitive measurements to be obtained at monolayer or even sub-monolayer coverage (Greaves 1991). However, REFLEXAFS requires a flat, polished specimen, whereas the other modes of measurement can be carried out on powders, slurries, or liquids.

flotation0.book Page 319 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

319

XAS has been used for the study of Zn and Pb sorption onto chalcopyrite (CuFeS2) (England et al. 1999a, b and references therein). Chalcopyrite tends to have natural hydrophobicity and thus self-floatability. This characteristic is attributable to the formation of a sulfur-rich surface layer due to metal loss under conditions of low pH. The adsorption of metal cations, as hydroxides, can result in a reduction in the flotation response. REFLEXAFS data revealed that Zn adsorbed at pH 5.5 is bonded to O (at 2.00 Å) only. There was no evidence of bonding to S. On addition of xanthate (at pH 10.2), the Zn coordination changed radically so that the nearest neighbor coordination sphere contained both O and S (at 1.97 Å and 2.33 Å, respectively). For Pb adsorption, both O and S coordination is present prior to the addition of xanthate. After xanthate addition, only S coordination is observed. It appears, therefore, that the initial adsorption site of the Zn species is onto the oxidized Fe(OH)x species present on the surface, whereas the Pb is adsorbed onto the sulfurrich surface regions. This interpretation provides an explanation for the observation that only small amounts of Pb in solution are required to remove chalcopyrite self-flotation. Adsorption of Pb (probably as a hydroxide) would render the hydrophobic sulfur-rich regions hydrophilic. The loss of self-floatability would be much less affected by adsorption of Zn species onto the already hydrophilic, oxidized Fe surface regions. Another study, originating from the same research team, examined the adsorption of Cu and Pb onto sphalerite (ZnS) (England et al. 1999a; Pattrick et al. 1998, 1999). Cu adsorption activates the sphalerite surface for enhanced collector adsorption. ZnSe was also examined, as an isostructural analogue to sphalerite so that the sulfur XAS data could be obtained from the adsorbed xanthate without the spectra being swamped by bulk S contributions. On the basis of analysis of the REFLEXAFS data, it was proposed that 3 S atoms and 1 O atom are bonded to the adsorbed Cu atoms (at 2.25 Å and 2.07 Å, respectively). On addition of xanthate, the O were replaced by an S atom. The fourth S atom has a considerably longer bond distance to the Cu atom as compared to the three initial S atoms, 2.75 Å as compared to 2.22 Å. Pb adsorbed onto sphalerite was proposed to be coordinated solely to O atoms. On addition of xanthate, the Pb coordination consisted of two bonds to S and two to O. It would appear, therefore, that, at least under the conditions used for this study, Pb does not specifically adsorb onto sphalerite as does Cu. A similar study of the adsorption of Cu onto sphalerite has also been carried out by Gerson, Lange et al. (1999). Analysis of the EXAFS data indicated Cu coordination to S (2.26–2.30 Å) but no evidence of Cu coordination to O. A data set plus Fourier transform for Cu adsorbed onto sphalerite sample are shown in Figures 25a and 25b. The Fourier transform is similar to a one-dimensional electron density distribution surrounding but offset by a phase shift. The large central amplitude is the result of the surrounding S atoms. The other two maxima centered at approximately 1 Å and 2.7 Å are the result of the truncation of the Fourier transform series from the experimental data. The exact location of the maxima will depend on the data range that is fitted by the simulated model data. In order to fit, the analytical program must truncate the model data in a manner similar to the experimental data. It is unclear in the studies by Pattrick et al. (1999) whether the interpretation of O coordination is, in fact, the result of this artefact induced by Fourier series truncation. On the basis of the interpretation of the EXAFS and XANES data, together with other experimental evidence and previous knowledge, a mechanism for the Cu activation of sphalerite was presented by Gerson, Lange et al. (1999). This mechanism proposes the replacement of Zn by Cu on the sphalerite surface to form a distorted trigonal planar structure, similar to the Cu structure found for half the Cu atoms within covellite (CuS). In-situ

flotation0.book Page 320 Tuesday, January 2, 2007 7:36 PM

320

FLOTATION FUNDAMENTALS

A

B

0.01

3.5 –0.01

4.5

5.5

6.5

–0.03 k, Å –1

7.5

8.5

Transform Amplitude, arbitrary units

k3Xobs(k), arbitrary units

0.03 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.5

1.0

1.5

2.0

2.5

3.5

r, Å

C NOTE: (a) EXAFS data collected for Cu adsorbed –4 onto sphalerite conditioned in 10 M Cu(NO3)2 for 15 minutes at pH 5.5. The regular sinusoidal nature of the data is indicative of a single coordination shell. (b) The Fourier transform of the model and experimental data sets. Although three maxima are shown, only one atomic shell has been used to fit this data (Cu–S 2.26 Å). (c) The proposed structure of Cu adsorbed onto a sphalerite (110) surface. For both (a) and (b), (x) shows the experimental data, and (–) shows the data simulated from the model.

FIGURE 25

Experimental EXAFS data sets with correspondence to proposed copper structure

reduction of Cu(II) to Cu(I) occurs, resulting in the oxidation of the surrounding S atoms. This smearing out of the surface charge results in the increased degree of hydrophobicity for Cu-activated sphalerite as compared to unactivated sphalerite that has been observed and, hence, results in increased floatability even without collector addition. The conditions under which Cu activations were carried out during this study were carefully chosen to inhibit copper(II) hydroxide (Cu(OH)2) formation. The Cu activation of pyrite (FeS2) has also been investigated by Weisener and Gerson (2000). On the basis of EXAFS data, a similar structure for adsorbed Cu to that projected for sphalerite was proposed (i.e., distorted trigonal planar) and near-identical Cu–S bond lengths were determined. Again, there was no evidence of Cu–O coordination except in the instance where Cu(OH)2 was purposefully precipitated within the activating solution (pH 8.5, 2.84 × 10–4 mol m–2 Cu). In this instance, Cu–S bond lengths of 2.29 Å and Cu–O bond lengths of 2.00 Å were derived from the EXAFS data. This also provides a possible explanation for the observations by Pattrick et al. (1999) as to the coordination of adsorbed Cu to O. Todd and Sherman (2003) used XAS to probe the surface oxidation mechanism of chalcocite (Cu2S). This leads to reduced flotation of chalcocite in minerals processing circuits. In this instance, TEY collection mode was used. For the Cu L-edge measurements undertaken, there is likely to be a contribution from the bulk mineral. However, the O K-edge measurements will be surface specific only because of the absence of O in the bulk. Oxidation of chalcocite in different pH solutions indicated that under acidic pH, cuprous oxide (Cu2O) dominates the oxidation products, whereas under alkaline pH, cupric oxide (CuO) is the major oxidation product. XANES has been employed in a study of selected elements (Ti, V, Cr, Mn, and As) in deep-cleaned Kentucky No. 9 coal (Huggins et al. 1997). Tail and float samples were prepared

flotation0.book Page 321 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

321

using both a Denver cell and column flotation. The tails and float fractions were examined in order to gain a better understanding of the distribution and form of elements that may be released during combustion and therefore form a pollutant risk and, hence, to identify appropriate coal-cleaning strategies. XANES data has been interpreted through comparison with data obtained from standards. V, Ti, Cr, and Mn were all found to be different in form in the tail and float fractions. The tailings fraction was interpreted as containing Ti and V in interlayer positions within illite, whereas the Ti and V present in the float fraction was proposed to be in an organic association. The pre-edge features from the V spectra indicate the oxidation state of V to be higher in the float fraction than in the tails fractions. The difference between the data collected for Cr in the float and tailings fraction was found to be subtler with Cr present in the tails as illite and in float fractions as a hydroxide species. Mn in the tails was associated with calcite (CaCO3) and illite, whereas in the float fraction, two different organic forms were identified. The XANES data for the float fractions for these elements suggest the principal coordination is to O anions. In contrast, As was primarily associated with pyrite and oxidation products thereof, and the relative proportions of these forms was not float or tail dependent. Additional opportunities exist to further utilize XAS measurements to advance the understanding of mineral flotation studies. In particular, an opportunity exists to exploit energy-dispersive XAS for mineral processing applications. This type of measurement can be carried out rapidly as it uses a broadband incident X-ray beam rather than the monochromatic beam traditionally employed. The latter requires a monochromator sweep in order to scan the energy range required for the spectra. Energy-dispersive XAS can be carried out on in-situ materials and in real time. Data acquisition can be carried out in the order of 10 seconds. SCANNED PROBE MICROSCOPIES

Though the scanned probe microscopies—scanning tunneling microscopy (STM), and atomic force microscopy (AFM)—are essentially research tools applied to model systems, they have added much in the last two decades to the understanding of surface reactions and adsorption mechanisms of minerals under flotation-related conditions. Both techniques can image surfaces close to the atomic level. STM requires reasonably conductive samples (which has made galena a mineral of choice) with some chemical identification in scanning tunneling spectroscopic mode. AFM can image insulating surfaces, cannot chemically identify atoms, but can give astonishing information on particle–particle and even particle–bubble interactions and changes with reaction and adsorption in force–distance (approach–retract) mode. Hochella (1995) has reviewed STM and AFM studies of mineral surfaces and their oxidation. Smart, Amarantidis et al. (2003) have reviewed applications to oxidation and collector additions in flotation. Some examples of particular insights, related to surface processes described previously, will illustrate the unique types of information that these techniques can provide. The development of isolated, patchwise oxidation in air and solution has been very well illustrated by STM studies of galena surfaces. Eggleston and Hochella (1990, 1991, 1992) have imaged (001) surfaces of galena at atomic scale after exposure to water for 1 minute. Apparent vacancies at the sulfur sites are correlated with oxidation in their model of this process. The oxidized regions do not initiate randomly, but after oxidation has begun at a site, these regions tend to nucleate and grow without initiation of new sites. The boundaries of the oxidized regions tend to lie along the [110] directions, apparently due to S atoms

flotation0.book Page 322 Tuesday, January 2, 2007 7:36 PM

322

FLOTATION FUNDAMENTALS

across this front having only one nearest-neighbor-oxidized sulfur whereas an unoxidized S across a [100] boundary would have two nearest-neighbor-oxidized sulfurs. As with crystallization processes, [100] fronts move fast and disappear, leaving the slow-moving [110] dominant. At lower magnification, the process of galena oxidation in air has also demonstrated random sites of oxidation and growth on (001) galena surfaces with no clear preference for initiation at step edges or corners (Laajalehto et al. 1993). This process is illustrated in Figure 26 from that work and correlated with XPS spectra, showing that the initial oxidation products are peroxide, hydroxide, and carbonate species successively. With time up to 270 minutes in air, the oxidation products grow from surface features with lateral dimensions 9 nm diameter with “holes” in the overlayer that still allow access to the underlying sulfide surface. Further studies of galena oxidation in air (Kim et al. 1994), comparing synthetic and natural galena samples, confirmed that the growth mechanism on natural galena with the oxidation initiation sites correlated with impurity atoms in the surface layer. The very much slower oxidation of synthetic galena did occur preferentially on edges, dislocations, and lattice defect sites on the (001) faces of the galena crystal. The XPS spectra in this case show predominantly lead hydroxide and sulfate with a smaller contribution from carbonate in the oxidation products. In solution, STM (and AFM) imaging showed the development of subnanometer pits with increasing reaction time in air-purged water at pH 7 (Kim et al. 1995). The boundaries of the pits lie in the (100) and (010) directions in the galena surface with depths corresponding to unit cell dimensions of galena (i.e., 0.3 and 0.6 nm). The process occurring in solution is congruent dissolution, confirmed by XPS spectra showing unaltered Pb4f and S 2p signals. The x- and y-dimensions of the pits and their rates of formation depend strongly on the pH and purging gas (i.e., O2, air, N2) used. Dissolution rates, determined directly from STM images of monolayers removed, decrease with increasing pH in agreement with the reported dissolution studies on galena (Fornasiero, Ralston, and Smart 1994;

A

B

C

FIGURE 26 STM images from a 70 × 70 nm area of (a) freshly cleaned galena surface; (b) the same surface after 70 minutes standing in air; (c) after 270 minutes in air. The upper row are grayscale images; the lower row are 3-D (rotated) images with the vertical scales 1.8, 2.0, and 3.8 nm, respectively. Constant current mode (0.2–0.25 nA), bias ~0.35 V.

flotation0.book Page 323 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

323

Hsieh and Huang 1989). For all pH values studied, the growth of the dissolution pits is significantly greater in the x- and y-directions than in the z-direction, suggesting that the edges are more active toward dissolution than the faces (Kim et al. 1995). At the relatively low surface-area-to volume ratios used in the STM studies, there is no evidence for the growth of surface oxidation products similar to those observed in air or to adsorption/precipitation of lead hydroxide colloids from solution. Increasing the lead ion concentration to 10–3 M in solution resulted in surface product formation as elongated, oval colloidal projections with dimensions of ~50 nm × 20 nm and average heights of 14 nm, with a distinct [110] directionality. XPS analysis has confirmed that these species are predominantly lead hydroxide, presumably formed from the hydrolysis of lead ions followed by surface attachment. It is not yet clear whether the mechanism of surface attachment involves the formation of lead hydroxide colloids in solution and their precipitation onto the galena surface or adsorption of Pb2+/Pb(OH)2 molecular species at specific sites on the galena surface before in-situ growth. However, the formation of these patchy surface layers shows that the galena surface is heterogeneous and that its overall hydrophobicity and flotation response will be controlled not only by the surface chemistry but by the surface arrangement of hydrophilic and hydrophobic patches. Atomic level imaging of the (001) surface of galena, reviewed by Hochella (1995), has been achieved, including observation of the oxidation of a single S site at which the tunneling current has been effectively extinguished. In-situ STM images of a freshly cleaved galena crystal in contact with an air-equilibrated 10–4 M ethyl xanthate solution show colloidal particles of lead ethyl xanthate (as confirmed by XPS and FTIR) formed at the surface corresponding to multilayer surface coverage. In-situ STM studies of ethyl xanthate treated preoxidized galena surfaces have also shown the removal of oxidized lead species and the formation of colloidal lead ethyl xanthate particles as flattened spheres with diameters of 10–20 nm and average heights of 6 nm (Ralston 1994b; Kim et al. 1995). Combined XPS and AFM studies of galena oxidation in acetate buffer (pH 4.9) by Wittstock et al. (1996) produced dramatic imaging of elemental sulfur protrusions 10– 200 nm after initial roughening of the galena surface. These protrusions are separated by several hundred nanometers and appear to result from a process of diffusion in the aqueous phase. XPS shows the formation of elemental sulfur starting at potentials more anodic than 160 mV SHE (saturated hydrogen electrode). AFM imaging first detects the protrusions at +236 mV SHE. The authors, therefore, propose that the process causing surface roughening is dissolution of PbS to Pb(II) ions and HS– ions, whereas the deposition reaction is the electrochemical oxidation of HS– ions to elemental sulfur. It seems likely that sulfur formation starts at impurity locations leading to different rates and sizes of protrusion development. In nickel sulfide processing, magnesium silicate (MgO) gangue minerals often report to the concentrate, causing downstream processing problems as well as increased smelting costs. In addition, these hydrophilic MgO minerals may interfere with the flotation of valuable sulfide minerals such as pentlandite [(Fe, Ni)9S8]. Flotation of the MgO particles may be via composite particles or through attachment to the valuable minerals as slime coatings. A coating of hydrophilic slime particles will decrease the hydrophobicity of the sulfide particle and may also reduce collector adsorption (Learmont and Iwasaki 1984). Either of these flotation mechanisms will reduce both the flotation rate and recovery, and will, therefore, result in lower recoveries of the valuable sulfide minerals (Trahar 1981; Senior and Trahar 1991; Wellham, Elber, and Yan 1992). Slime coatings of lizardite and chrysotile have been found to adhere to the surface of pentlandite, reducing its flotation rate (Edwards, Kipkie,

flotation0.book Page 324 Tuesday, January 2, 2007 7:36 PM

324

FLOTATION FUNDAMENTALS

0.04

Force–Distance, mN/m

0.02

0.0 0

20

40

60

70

100

–0.02

–0.04

–0.06 Apparent Separation, nm

FIGURE 27 Direct interaction forces as a function of apparent separation measured on approach between a pentlandite surface ( ) and lizardite particle ( ) in 10–3 M KNO3 (potassium nitrate) at pH 9.4 and in the presence of 20 ppm CMC

and Agar 1980; Li 1993; Chen at al. 1999a, b; McQuie 1999). The interaction between lizardite and pentlandite has been directly investigated using the atomic force microscope in force–distance mode and electrokinetic zeta potential determinations as a function of pH and in the presence of the polymeric dispersant, carboxymethyl cellulose (CMC) (Bremmell, Fornasiero, and Ralston 2005). The lizardite mineral was positively charged with the zeta potential independent of pH. The magnitude and sign of the pentlandite particles were pH dependent and were negative for pH values above 4.5. At pH values greater than 9, where flotation of nickel sulfide ores is routinely performed, the two minerals are oppositely charged and, therefore, attract through an electrostatic mechanism. Direct interaction force measurements between pentlandite and lizardite surfaces as a function of pH demonstrate this attractive interaction. Adsorption of CMC at the lizardite–solution interface overcompensates the positive charge on the lizardite particle, and its zeta potential is rendered negative. In the presence of CMC, a repulsive interaction force between lizardite and pentlandite, which was concluded to be of electrosteric origin, was measured in the AFM (Figure 27). The results explain the flotation behavior of the minerals performed in this and previous studies. P L A N T C A S E S T U DY : U S I N G S U R FA C E A N A LY S I S T O EXAMINE ZINC CIRCUIT MINERAL LOSSES Introduction

The Matagami mine sulfide flotation plant in Northern Quebec is the focus operation for this study. The Matagami plant produces copper and zinc concentrates. This study investigated zinc losses in the zinc circuit using surface analysis. For this work, the rougher feed, rougher concentrate, rougher tails, cleaner tails, and zinc concentrate streams were sampled and examined in the CANMET-MMSL surface science laboratories located in Ottawa, Canada. Pulp samples were collected using established and modified protocols designed to preserve the surface chemistry on particles from the time of sampling until the time of analysis. The surface chemistry of particles in each pulp was examined using the surface analysis methods of X-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy

flotation0.book Page 325 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

325

(AES). The bulk mineralogy of each pulp sample was determined from X-ray diffraction (XRD) powder patterns. XRD was used because the materials are crystalline, and the method has a detection range that is readily comparable to that of XPS. Standard XPS provides excellent chemical state information, but the spatial resolution of the data is limited to hundreds of micrometers. For the acquisition of surface chemical information from individual particles, AES was used. The spatial resolution of the AES surface chemical data collected using a scanning Auger microprobe is limited essentially by the diameter of the primary electron beam; therefore, AES spectra can be collected from exceedingly small particles. Methods

To prevent the exposure of individual particle surfaces to air and preserve the chemistry of the solid–liquid interface present at the time of collection, the sampling steps outlined in the next section were followed. These procedures are in a continual state of development and are the product of discussion, consultation, and experimentation with surface science groups in academia, government, and industry in Africa, Australia, Europe, and North America. The protocols are based on the premise that each particle in a pulp will have a hydrated boundary layer that will prevent direct contact between the particle surface and the atmosphere. When frozen, it is a passivating and protecting layer of ice that isolates individual particles from the air (Smart 1991; Love, Cayless, and Hazell 1993; Pratt, Nesbitt, and Muir 1994). Pulp samples were prepared for XRD analysis by grinding 2 g of material to less than 10-μm particle diameters using a custom micromill. XRD analyses were conducted on the samples using an automated Rigaku diffractometer equipped with a rotating copper anode X-ray source. XRD powder patterns were collected using monochromatic radiation; they were then processed, and the mineral phases were identified using the Materials Data Inc. powder diffraction pattern analysis program JADE (Release 6.1) and the ICDD Powder Diffraction Database (Release 2001). XPS spectra were collected using a PerkinElmer Corporation (Physical Electronics Division) PHI-5600 spectrometer equipped with an OMNI V lens system. XPS data were collected using 400-W achromatic Mg X-rays (Ex-ray = 1,353.6 eV) and Al X-rays (Ex-ray = 1,486.6 eV). The two X-ray sources were utilized as a means of resolving contributions from coincident photoelectron and X-ray-induced Auger electron emissions. The energy scale of the spectrometer was calibrated to the metallic Au(4f7/2) line at 84.0 eV and was to give an energy dispersion of 857.8 eV between the metallic copper 2p3/2 and 3p lines. The analyzer pass energy was 187.0 eV for broad-energy-range “survey” scans and 29.35 eV for narrowenergy-range “multiplex” scans. The binding energy scale of the spectra is referenced to the C1s peak from adventitious hydrocarbon (static charge referencing) fixed at 285.0 eV (Swift 1982). XPS information was collected from spots measuring nearly 400 μm in diameter, and the vacuum in the analytical chamber was approximately 8.0 × 10–10 Torr during analysis. Details on the analysis of XPS spectra are provided in work by Pratt, Nesbitt, and Muir (1994). AES spectra were collected using a PerkinElmer Corporation (Physical Electronics Division) PHI-600 scanning Auger microprobe equipped with a LaB6 thermionic electron emitter. Analyses were obtained using an electron beam accelerated to a potential of 3.0 kV at a current of 90 nA and an analyzer energy resolution set to 0.6%. Under these instrument conditions, exceedingly small particles could be examined without any interferences from the sample mount matrix. This is because the volume of analysis is defined by the primary

flotation0.book Page 326 Tuesday, January 2, 2007 7:36 PM

326

FLOTATION FUNDAMENTALS

beam diameter (~300 nm) and the escape depth (λ) of Auger electrons, which is approximately 1 to 3 nm beneath a particle surface. Lambda (λ) is dependent on the kinetic energy of the Auger electron and the solid’s density. Semiquantitative surface compositions were calculated using peak-to-peak heights and manufacturer-supplied, empirically-derived sensitivity factors. The vacuum in the analytical chamber was about 6.0 × 10–10 Torr during analysis. Metallurgical Processing Surface Science Protocols for Sampling and Analysis

The experimental protocol for preparation of surface samples is as follows: 1. Sample the pulp stream using a beaker. 2. Place several milliliters of pulp into a vial. 3. Purge with N2(g). 4. Cap the vial and seal the cap with silicon. 5. Freeze the sample as quickly as possible, and maintain in a frozen state until time for analysis. 6. Thaw the sample within 30 minutes of the scheduled analysis. 7. Using a pipette, remove 1.5 mL of the process water and deposit it into a micro test tube. 8. Using a micro spatula, remove a minute amount of pulp and place it into the micro test tube. For XPS-destined samples, the solution should be slightly clouded; this amount provides about a monolayer of particle coverage on the filter. For AESdestined samples, the solution should remain clear; this amount provides the dispersed particle coverage needed for Auger analysis. 9. Separate the solids from the solution onto a 2.5-cm-diameter 0.45-μm nitrocellulose membrane filter using a vacuum filtration system. (NOTE : For these experiments, the membrane filter was sputter-coated with gold to increase conductivity and fixed to the appropriate sample platen.) 10. Prepare the damp samples for surface analysis using an ultrahigh vacuum (UHV) conditioning chamber that is attached to the XPS instrument. (NOTE : The conditioning chamber used in these experiments was designed and built by one of the authors. One function of the conditioning chamber is to prepare damp samples for the UHV conditions required for spectroscopic surface analysis. The chamber design is attached to the XPS via a series of UHV gate valves and stainless-steel conduits to the analytical chamber of the XPS. This permits all manipulations to be undertaken at high to ultrahigh vacuum conditions. Conditioning takes up to 15 hours; therefore, UHV conditions should be maintained throughout the experiment.) 11. For collection of AES data, transfer the conditioned samples to the scanning Auger microscope instrument from the XPS using a high-vacuum transfer vessel. (NOTE : The high-vacuum transfer vessel used in these experiments was designed and built at CANMET by Dr. Jim Brown.) Results and Discussion

The XRD powder pattern and XPS survey scan data collected from the selected zinc circuit feeds and the concentrates show that there is a progressive lessening in complexity with the evolution of the pulps within the circuit. This discussion focuses on the results obtained from the sphalerite component in the pulps. Contributions from sphalerite can be identified

flotation0.book Page 327 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

327

in XRD patterns collected from each of the pulps examined. XPS survey scans collected from each of the pulps show contributions that can be attributed to sphalerite as well. They also show distinct Cu contributions. Qualitative evaluation of the XRD data shows that only minor to trace amounts of chalcopyrite (CuFeS2) can be found in the pulps. The source for the Cu detected by XPS is most likely the Cu added for activation of sphalerite. Because the vast majority of the particles in the Zn concentrate is sphalerite, the approximate surface Zn/Cu ratio of 3:1, obtained from this pulp, can be used as a model ratio for the qualitative evaluation of sphalerite activation in each of the pulps examined. The approximate Zn/Cu ratios for the pulps examined are as follows: rougher feed >30:1, rougher concentrate 4:1, rougher tails 10:1, and cleaner scavenger tails 6:1. As inadvertent activation of the sphalerite in the feed should not be very high, a high Zn/Cu ratio is expected. Activation appears to have occurred within the rougher, and the rougher concentrate has the appropriate Cu activation ratio. For the rougher tails, the high Zn/Cu ratio obtained may indicate that the small amount of sphalerite present in the pulp has not been sufficiently activated. For the cleaner scavenger tails, the Zn/Cu ratio is about 50% greater than the model ratio. These results indicate that there is a problem with the activation of the sphalerite in the circuit. Information into Cu chemistry at particle surfaces can be obtained through evaluation of Cu 2p spectra collected from the rougher concentrate, rougher tails, cleaner scavenger tails, and Zn concentrate (Figure 28). The spectra have been normalized such that the intensities near 932 eV are nearly coincident. The overlain spectra are characterized by Cu 2p3/2 and 2p1/2 peaks, respectively, at 931.8 eV and 951.7 eV. The mineralogy of the Zn concentrate is mainly sphalerite, and the Cu 2p collected from this pulp is interpreted to originate from activated sphalerite surfaces. The shapes and positions of the Cu 2p peaks are similar to those reported for Cu(I) (Chawala, Sankarraman, and Payer 1992), and, in agreement, the signals are interpreted to originate from Cu(I) ions. The pulp with activated surfaces most closely resembling those of the Zn concentrate data is the rougher concentrate data. The Cu 2p spectra collected from the rougher tails and cleaner scavenger tails have an additional contribution near 942 eV. The position and shape resemble those reported for Cu(II) ions (Chawala, Sankarraman, and Payer 1992) and are interpreted to be from Cu(II) species. These results show that a portion of the Cu on the sphalerite surfaces in the tails is found as Cu(II). AES Analyses of Individual Sphalerite Particles

AES spectra were collected from individual sphalerite particles in the process stream pulps. A minimum of 10 particles in each pulp was analyzed. The AES spectra showed contributions from S, C, O, and the transition metals Fe, Cu, and Zn. On many of the sphalerite particles examined, contributions from Ca were detected. Two spectra representative of the AES data collected in the study are shown in Figure 29. The Ca detected is interpreted to be associated with a precipitated Ca sulfate species, possibly gypsum. Although the sampling size is small, the AES results show that Ca concentrations are consistently the lowest on sphalerite particles in concentrates. Conversely, Ca concentrations are consistently higher on sphalerite particles in tails. Using O concentrations as a guide to the degree of surface oxidation, the AES data shows that the tails have sphalerite particles that are the more oxidized (Figure 29a), and the two concentrate pulps have sphalerite particles that are the least oxidized (Figure 29b). These trends appear to apply to both the coarse and fine particles examined within the pulps.

flotation0.book Page 328 Tuesday, January 2, 2007 7:36 PM

328

FLOTATION FUNDAMENTALS

Cu 2p3/2

Cu 2p1/2

Zn Rougher Concentrate Zn Rougher Tails Zn Cleaner Scavenger Tails Zn Concentrate 1 Cu(I)

Normalized Intensity

2 Cu(II)

2

1

965

960

955

1

950

945

940

935

930

925

Binding Energy, eV

FIGURE 28 Overlain narrow region Cu 2p spectra collected from four zinc circuit pulps. The peak intensities shown have been normalized. The dotted traces are the concentrates and the solid traces are the tails.

dN(E)

A

Cu 0.3% Ca 3.4%

Si 3.0%

Fe 9.2% Zn 7.3%

O 10.5% C 46.7%

S 17.7%

dN(E)

B

O 1.3% C 18.5%

Fe 7.9%

Cu 9.1% Zn 25.4%

S 37.8%

40

180

320

460

600

740

880 1,020 1,160 1,300 1,440

Kinetic Energy, eV

FIGURE 29 Representative AES scans collected from sphalerite particles in the (a) cleaner scavenger tails and (b) zinc concentrate. Values are atomic %.

flotation0.book Page 329 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

329

XRD shows that an appreciable amount of sphalerite is found in tailings pulps within the circuit, and the XPS results show that the sphalerite is found in the tails because of inadequate Cu activation. Examination of the Auger spectra collected from individual sphalerite particles in the circuit pulps corroborates the XPS interpretation. The sphalerite particles examined in the two concentrate pulps show well-defined Cu peaks (Figure 29b). Those in the tails have no clear and unambiguous contributions from Cu (Figure 29a). The number of particles examined is small and a larger sample population would greatly increase the confidence in the interpretations put forward for the Auger data. S U M M A RY

AES examination of sphalerite in the two concentrates shows that these particles are less oxidized and cleaner (less calcium species) than those in the two tails. The XPS and AES results show that sphalerite component in the tails has not been sufficiently activated, and the XPS results show oxidized copper is found on the surface of the particles. REFERENCES

Ahlberg, E., K.S.E. Forssberg, and X. Wang. 1990. The surface oxidation of pyrite in alkaline solution. J. Appl. Electrochem. 20:1033–1039. Bolin, N.J., S.L. Chryssoulis, and C.J. Martin. 1997. A surface study of Boliden ore by TOF-LIMS. Int. J. Miner. Process. 51:27–37. Bremmell, K.E., D. Fornasiero, and J. Ralston. 2005. Pentlandite-lizardite interactions and implications for their separation by flotation. Colloids Surfaces A 252:207–212. Briggs, D., and M.C. Seah, editors. 1992. Practical Surface Analysis. 2nd edition. Volume 1: Auger and X-ray Photoelectron Spectroscopy. United Kingdom: John Wiley & Sons. Brinen, J.S., S. Greenhouse, D.R. Nagaraj, and J. Lee. 1993. SIMS and SIMS imaging studies of adsorbed dialkyl dithiophosphinates on PbS crystal surfaces. Int. J. Miner. Process. 38:93–109. Bronold, M., Y. Tomm, and W. Jaegermann. 1994. Surface states of cubic d-band semiconductor pyrite FeS2. Surf. Sci. 314:L931–L936. Buckley, A.N., I.C. Hamilton, and R. Woods. 1985. Investigation of the surface oxidation of sulfide minerals by linear potential sweep voltammetry and X-ray photoelectron spectroscopy. Pages 41–60 in Flotation of Sulfide Minerals. Edited by K.S.E. Forssberg. Amsterdam: Elsevier. Buckley, A.N., and R. Woods. 1987. The surface oxidation of pyrite. Appl. Surf. Sci. 27:347–452. ———. 1991. Adsorption of ethyl xanthate on freshly exposed galena surfaces. Colloids Surf. 53:33–45. Buffeteau, T., D. Blaudez, E. Pere, and B.B. Desbat. 1999. Optical constant determination in the infrared of uniaxially oriented monolayers from transmittance and reflectance measurements. J. Phys. Chem. B 103:5020. Buffeteau, T., E. Le Calvez, B. Desbat, I. Pelletier, and M. Pezolet. 2001. Quantitative orientation of α-helical polypeptides by attenuated total reflection infrared spectroscopy. J. Phys. Chem. B 105:1464. Carlson, T.A. 1975. Photoelectron and Auger Spectroscopy. New York: Plenum Press. Chawala, S.K., N. Sankarraman, and J.H. Payer. 1992. Diagnostic spectra for XPS analysis of Cu-O-SH compounds. J. Electron Spectrosc. 61:1–18. Chen, G., S. Grano, S. Sobieraj, and J. Ralston. 1999a. The effect of high intensity conditioning on the flotation of a nickel ore. Part 1: Size-by-size analysis. Miner. Eng. 12:1185–1200. ———. 1999b. The effect of high intensity conditioning on the flotation of a nickel ore. Part 2: Mechanisms. Miner. Eng. 12:1359–1373. Chryssoulis, S.L. 2001. Using mineralogy to optimize gold recovery by flotation. JOM 53:48–50. Chryssoulis, S.L., L.J. Cabri, J.L. Campbell, and W.J. Teesdale. 1991. Comparison of in-situ gold analyses in arsenian pyrite. Appl. Geochem. 6:225–230.

flotation0.book Page 330 Tuesday, January 2, 2007 7:36 PM

330

FLOTATION FUNDAMENTALS

Chryssoulis, S.L., R. Dunne, and A. Coetzee. 2004. Diagnostic microbeam technology in gold ore processing. JOM 56(7):53–57. Chryssoulis, S.L, J.Y. Kim, and K.G. Stowe. 1994. LIMS study of variables affecting sphalerite flotation. Proceedings of the 26th Annual Meeting of the Canadian Mineral Processors. Paper no. 28. Office of the Secretary of the Canadian Mineral Processors. Chryssoulis, S.L., F. Reich, and K.G. Stowe. 1992. Characterization of mineral surface composition by laser probe microanalysis. Trans. Inst. Min. Metall., Sect. C 100:1–6. Chryssoulis, S.L., K.G. Stowe, E. Niehuis, H.G. Cramer, C. Bendel, and J.Y. Kim. 1995. Detection of collectors on concentrator mineral grains by time of flight secondary ion mass spectrometry TOF-SIMS. Trans. Inst. Min. Metall., Sect. C 104:141–150. Chryssoulis, S.L., C.G. Weisener, and S. Dimov. 1995. Detection of mineral collectors by TOFLIMS. Pages 899–902 in Proceedings of the Secondary Ion Mass Spectrometry SIMS X. Edited by A. Benninghoven, B. Hagenhoff, and H.W. Werner. New York: John Wiley & Sons. Clarke, N.S., J.C. Ruckman, and A.R. Davey. 1986. Laser microprobe mass spectrometry of solid surfaces. In Proceedings of the Third International Laser Microprobe Mass Spectrometry Workshop. Edited by F. Adams and L. van Vaeck. Belgium: University of Antwerp. Conboy, J.C., M.C. Messmer, R. Walker, and G.L. Richmond. 1997. An investigation of surfactant behavior at the liquid/liquid interface with sum frequency vibrational spectroscopy. Amphiphiles at Interfaces. Prog. Colloid Polym. Sci. 103:10. Delesse, A. 1848. Procédé mechanique pour determiner la composition des roches. 4th series. Annales des Mines 13:379–388. Dimov, S., and S.L. Chryssoulis. 1997. Relative sensitivity factors for quantitative TOF-LIMS analysis of mineral surfaces. Pages 815–818 in Proceedings of the 11th International Conference on Secondary Ion Mass Spectrometry–SIMS XI. Edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie. New York: John Wiley & Sons. Edwards, G.R., W.B. Kipkie, and G.E. Agar. 1980. The effect of slime coatings of the serpentine minerals, chrysotile and lizardite on pentlandite flotation. Int. J. Miner. Process. 7:33–24. Eggleston, C.M., and M.F. Hochella Jr. 1990. Scanning tunneling microscopy of sulfide surfaces. Geochim. Cosmochim. Acta 54:1511–1517. ———. 1991. Scanning tunneling microscopy of galena 100 surface oxidation and sorption of aqueous gold. Science 254:983–986. ———. 1992. Tunneling spectroscopy applied to PbS001 surfaces: Fresh surfaces, oxidation and sorption of aqueous Au. Am. Mineral. 78:877–883. England, K.E.R., R.A.D. Pattrick, J.M. Charnock, and J.F.W. Mosselmans. 1999a. Floating sulfide: Activating and poisoning surfaces. J. Synchrotron Rad. 6:664–666. ———. 1999b. Zinc and lead sorption on the surface of CuFeS2: A fluorescence REFLEXAFS study. Int. J. Miner. Process. 57:59–71. Fa, K., and J.D. Miller. 2003. Surfactant adsorption density calculation from Fourier transform infrared external reflection spectroscopy FTIR/ERS. J. Chem. Phys. 119(24):13068. Finkelstein, N.P. 1997. The activation of sulfide minerals for flotation: A review. Int. J. Miner. Process. 52:81–120. Fornasiero, D., F. Li, J. Ralston, and R.St.C. Smart. 1994. Oxidation of galena surfaces, I. X-ray photoelectron spectroscopic and dissolution kinetics studies. J. Colloid Interface Sci. 164:333– 344. Free, M.L., and J.D. Miller. 1996. The significance of collector colloid adsorption phenomena in the fluorite/oleate flotation system as revealed by FTIR/IRS and solution chemistry analysis. Int. J. Miner. Process. 48:197. Frew, J.A. K.J. Davey, R.M. Glen, and R.St.C. Smart. 1994. Effects of fine grinding on flotation performance: Zinc regrind at Cominco Alaska’s Red Dog mine. Pages 287–288 in Proceedings of the 5th Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Frew, J.A., R.St.C. Smart, and E.V. Manlapig. 1994. Effects of fine grinding on flotation performance: Generic statements. Pages 245–250 in Proceedings of the 5th Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy.

flotation0.book Page 331 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

331

Gerson, A.R., P.J. Halfpenny, S. Pizzini, R. Risti´c, K.H. Roberts, D.B. Sheen, and J.N. Sherwood. 1999. Applications of synchrotron X-radiation to problems in materials science. Pages 105–169 in X-ray Characterization of Materials. Edited by E. Lifshin. New York: Wiley-VCH. Gerson, A.R., A.G. Lange, K.P. Prince, and R.St.C. Smart. 1999. The mechanism of copper activation of sphalerite. Appl. Surface Sci. 137:207–223. Glagolev, A.A. 1934. Quantitative analysis with the microscope with the microscope by the point method. Eng. Min. J. 135:399–400. Goh, S.W., A.N. Buckley, R.N. Lamb, R.A. Rosenberg and D. Moran. 2006. The oxidation states of copper and iron in mineral sulfides, and the oxides formed on initial exposure of chalcopyrite and bornite to air. Geochim. et Cosmochim. Acta 70(9):2210–2228. Goh, S.W., A.N. Buckley, R.N. Lamb, W.M. Skinner, A. Pring, H. Wang, L-J. Fan, L-Y. Jang, L-J. Lai and Y-W. Yang. 2006. Sulfur electronic environments in a-NiS and b-NiS: Examination of the relationship between coordination number and core electron binding energies. Phys. Chem. Minerals 33:98–105. Goldstein, J.I., D.A. Newbury, P. Echlin, D.C. Joy, A.D. Romig, C.E. Lyman, C. Fiori, and E. Lifshin. 1994. Comparison of WDS and EDS. Pages 331–336 in Scanning Electron Microscopy and X-ray Microanalysis. 2nd edition. New York: Plenum Press. Grano, S., J. Ralston, and R.St.C. Smart. 1990. Influence of electrochemical environment on the flotation behavior of Mt. Isa copper and lead-zinc ore. Int. J. Miner. Process. 30:69–97. Grano, S.R., D.W. Lauder, N.W. Johnson, S. Sobieraj, R.St.C. Smart, and J. Ralston. 1993. Surface analysis as a tool for problem solving: A case study of the Hilton concentrator at Mt. Isa Mines Ltd. Pages 1–15 in Proceedings of the Symposium Polymetallic Sulfides Iberian Pyrite Belt (Evora, Portugal). Lisbon: Portuguese Mining Industry Association. Grano, S.R., P.L. Wong, W. Skinner, N.W. Johnson, and J. Ralston. 1996. Detection and control of calcium sulfate precipitation in the Hilton concentrator of Mt. Isa Mines, Ltd. Pages 171–179 in Proceedings of the XIX International Journal of Mineral Processing Congress, Australia. Volume 3. Melbourne: Australasian Institute of Mining and Metallurgy. Grant, G., J.S. Hall, A.F. Reaid, and M.A. Zuiderwyk. 1976. Multi-compositional particle characterization using a SEM microprobe. Pages 401–408 in Scanning Electron Microscopy. Volume III. Chicago: ITT Research Institute. Greaves, G.N. 1991. Glancing angle X-ray absorption spectroscopy. Adv. X-ray Anal. 34:13–22. Gu, Y. 2003. Automated scanning electron microscope based mineral liberation analysis. J. Miner. Mater. Charact. Eng. 2(1):33–41. Hancer, M., R.P. Sperline, and J.D. Miller. 2000. Anomalous dispersion effects in the IR-ATR spectroscopy of water. Appl. Spectrosc. 18(1):138. Harmer, S.L., and H.W. Nesbitt. 2004. Stabilization of pyrite FeS2, marcasite FeS2, arsenopyrite FeAsS and loellingite FeAs2 surfaces by polymerization and auto-redox reactions. Surf. Sci. 564:38–52. Harmer, S.L., A.R. Pratt, H.W. Nesbitt, and M.E. Fleet. 2004. Sulfur species at chalcopyrite CuFeS2 fracture surfaces. Am. Mineral. 89:1026–1032. Hart, B., M. Biesinger, J.D. Miller, J. Francis, and R.St.C. Smart. 2004. Principal component analysis applied to surface chemistry in minerals flotation. Pages 73–88 in Particle Size Enlargement in Mineral Processing, Proceedings of the Fifth UBC McGill International Conference on Fundamentals of Mineral Processing. Edited by J.S. Laskowski. Montreal: Canadian Institute of Mining, Metallurgy and Petroleum. Hart, B., M.C. Biesinger, and R.St.C. Smart. 2006. Improved statistical methods applied to surface chemistry in minerals flotation. Miner. Eng. 19:790–798. Hochella, M.F., Jr. 1995. Mineral surfaces: Their characterisation and their chemical, physical and reactive natures. Pages 17–60 in Mineral Surfaces. Edited by D.J. Vaughan and R.A.D. Pattrick. London: Chapman and Hall. Hope, G.A., K. Watling, and R. Woods. 2001a. An electrochemical investigation of the suppression of silver dissolution in aqueous cyanide by 2 mercaptobenzothiazole. J. Appl. Electrochem. 31:703– 709. ———. 2001b. A SERS spectroelectrochemical investigation of the interaction of isopropyl, isobutyl and isoamyl xanthates with silver. Colloids Surf. A 178:157–166.

flotation0.book Page 332 Tuesday, January 2, 2007 7:36 PM

332

FLOTATION FUNDAMENTALS

Hope, G.A., R. Woods, S. Boyd, and K. Watling. 2003. A spectroelectrochemical investigation of the interaction of diisobutyldithiophosphinate with copper, silver and gold surfaces: I. Raman and NMR spectra of diisobutyldithiophosphinate compounds. Colloids Surf. 214:77–85. ———. 2004. A SERS spectroelectrochemical investigation of the interaction of butylethoxycarbonylthiourea with copper surfaces. Colloids Surf. A 232:129–137. Hope, G.A., R. Woods, and C.G. Munce. 2001. Raman microprobe mineral identification. Miner. Eng. 14:1565–1577. Hope, G.A., R. Woods, and K. Watling. 2001. A spectroelectrochemical investigation of the influence of sodium diisobutyldithiophosphinate on silver dissolution in aqueous cyanide. J. Appl. Electrochem. 31:1285–1291. ———. 2003. A spectroelectrochemical investigation of the interaction of diisobutyldithiophosphinate with copper, silver and gold surfaces: II. Electrochemistry and Raman spectroscopy. Colloids Surf. 214:87–97. Hsieh, Y.H., and C.P. Huang. 1989. The dissolution of PbS in dilute aqueous solutions. J. Colloid Interface Sci. 131:537–549. Huggins, F.E., S. Srikantapura, B.K. Parekh, L. Blanchard, and J.D. Robinson. 1997. XANES spectroscopic characterization of deep-cleaned fraction of Kentucky No. 9 coal. Energy Fuels 11:691–701. Jang, W.H., and J.D. Miller. 1993. Verification of the internal reflection spectroscopy adsorption density equation by Fourier transform infrared spectroscopy analysis of transferred LangmuirBlodgett films. Langmuir 9:3159. ———. 1995. Molecular orientation of Langmuir-Blodgett and self-assembled monolayers of stearate species at a fluorite surface as described by linear dichroism theory. J. Phys. Chem. 99:10272. Jones, M.P. 1984. Recent advances in the rapid collection of quantitative mineralogical data. Pages 141–155 in Applied Mineralogy ICAM 84. Edited by W.C. Park, D.M. Hansen, and R.D. Hagni. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Kartio, I.J., K. Laajalehto, E. Suoninen, A.N. Buckley, and R. Woods. 1998. The initial products of the anodic oxidation of galena in acidic solution and the influence of mineral stoichiometry. Colloids Surf. A 133:303–311. Kasrai, M., W.N. Lennard, R.W. Brunner, G.M. Bancroft, J.A. Badwell, and K.H. Tan. 1996. Sampling depth of total electron and fluorescence measurements in Si L- and K-edge absorption spectroscopy. Appl. Surf. Sci. 99:303–312. Kellar, J.J., W.M. Cross, and J.D. Miller. 1989. Adsorption density calculations from in situ FT-IR/ IRS data at dilute surfactant concentrations. Appl. Spectrosc. 43(8):1456. Kim, B.S., R.A. Hayes, C.A. Prestidge, J. Ralston, and R.St.C. Smart. 1994. Scanning tunneling microscopy studies of galena: The mechanism of oxidation in air. Appl. Surf. Sci. 78:385–397. ———. 1995. Scanning tunneling microscopy studies of galena: The mechanisms of oxidation in aqueous solution. Langmuir 11:2554–2562. Kim, J.Y., S.L. Chryssoulis, and K.G. Stowe. 1995. Effects of lead ions in sulphide flotation. Paper 9 in Proceedings of the 27th Annual Meeting of the Canadian Mineral Processors. Office of the Secretary of the Canadian Mineral Processors. Kim, J.Y., J.E. Nesset, S. Parker, and S.L. Chryssoulis. 1997. TOF-LIMS studies of pH modifiers in sulphide flotation. Pages 807–810 in Proceedings of the 11th International Conference on Secondary Ion Mass Spectrometry, SIMS XI. Edited by G. Gillen, R. Lareau, J. Bennett, and F. Stevie. New York: John Wiley & Sons. Kristall, Z., S.R. Grano, K. Reynolds, R.St.C. Smart, and J. Ralston. 1994. An investigation of sphalerite flotation in the Murchison Zinc concentrator. Pages 171–180 in Proceedings of the 5th Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Laajalehto, K., I. Kartio, T.L. Karila, T.L. Laiho, and E. Suoninen. 1996. Investigation of copper sulfide mineral surfaces using synchrotron radiation excited photoemission spectroscopy. Pages 717–720 in Proceedings of the European Conference on Applications of Surface and Interface Analysis ECASIA ’95. Edited by H.J. Mathieu, N. Reihl, and D. Briggs. New York: Wiley. Laajalehto, K., R.St.C. Smart, J. Ralston, and E. Suoninen. 1993. STM and XPS investigation of reaction of galena in air. Appl. Surf. Sci. 64:29–39.

flotation0.book Page 333 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

333

Lange, A.G., W.M. Skinner, and R.St.C. Smart. 1997. Fine:coarse particle interactions and aggregation in sphalerite flotation. Miner. Eng. 10(7):681–693. Lascelles, D., and J.A. Finch. 2002. Quantifying accidental activation I. Cu ion production. Miner. Eng. 15:567–571. Learmont, M.E., and I. Iwasaki. 1984. Effect of grinding media on galena flotation. Miner. Metall. Process. 1:136–143. Lee, S., and C.S.P. Sung. 2001. Surface chemical composition analysis in polystyrene/poly(vinyl methyl ether) blend films by UV reflection spectroscopy. Macromolecules 34:599. Leiro, J.A., K. Laajalehto, I. Kartio, and M.H. Heinonen. 1998. Surface core-level shift and phonon broadening in PbS 100. Surf. Sci. 412–413:L918–L923. Leppinen, J., and J.A. Mielczarski. 1986. Spectroscopic study of the adsorption of thiol collectors on lead sulphide in the presence of sodium sulphide. Int. J. Miner. Process. 18:3–20. Li, Z. 1993. Effect of gangue minerals containing magnesium on pentlandite flotation. Kuangye Xueyuan Xuebao 24:36–44. Love, C.J., R.A. Cayless, and L.B. Hazell. 1993. The SPLINT technique for sample preservation using liquid nitrogen transfer. Surf. Interface Anal. 20:565–568. Lu, Y., and J.D. Miller. 2002. Carboxyl stretching vibrations of spontaneously adsorbed and LB-transferred calcium carboxylates as determined by FTIR internal reflection spectroscopy. J. Colloid Interface Sci. 256:41. Martens, A.E., R.R.A. Morton, and J.C. McCarthy. 1978. The application of advanced image analysis techniques. Pages 426–432 in Proceedings of the Symposium on Quantitative Analysis of Microstructures in Materials Science, Biology and Medicine. Edited by J.L. Chermant. Stuggart, Germany: Rieder-Verlag GmbH. Matos, M.J., R. Lastra, and W. Petruk. 1996. Characterization of feldspar texture and liberation by automated image analysis. Trans. Inst. Min. Metall., Sect. C 105:133–140. McCarron, J.J., G.W. Walker, and A.N. Buckley. 1990. An X-ray photoelectron spectroscopic investigation of chalcopyrite and pyrite surfaces after conditioning in sodium sulfide solutions. Int. J. Miner. Process. 30:1–16. McQuie, J.D. 1999. Influence of particle aggregation and pulp chemistry on the flotation of pentlandite fines in the slimes stream at Mt. Keith. MappSc thesis, University of South Australia, Adelaide. Mielczarski, E., Y. Duval, and J.A. Mielczarski. 2002. Spectroscopic characterization of the nature and structure of adsorbed organic monolayers on quartz in the region of very strong absorption of substrate. Reverse surface selection rule. J. Phys. Chem. B 106(46):11985–11992. Mielczarski, E., and J.A. Mielczarski. 2003a. Influence of galvanic effect on adsorption of xanthate on pyrite, galena and chalcopyrite. Pages 866–873 in Proceedings of the XXII International Mineral Processing Congress, Cape Town, South Africa, September 29–October 3. Volume 2. Edited by L. Lorenzen, et al. Johannesburg: South African Institute of Mining and Metallurgy. ———. 2003b. Surface modification at molecular level in mineral beneficiation. Int. J. Min. Proc. 72(1–4):129–140. ———. 2005. Infrared spectroscopic studies of galvanic effect influence on surface modification of sulfide minerals by surfactant adsorption. Environ. Sci. Technol. 39:6117–6122. Mielczarski, E., J.A. Mielczarski, and J.M. Cases. 1998. Molecular recognition effect in monolayer formation of oleate on fluorite. Langmuir 14:1739–47. Mielczarski, E., J.A. Mielczarski, J.M. Cases, B. Rai, and Pradip. 2002. Influence of solution conditions and mineral surface structure on the formation of oleate adsorption layers on fluorite. Colloids Surf. 205:73–84 Mielczarski, E., J.A. Mielczarski, Z. Lei, and P. Somasundaran. 2004. Structure of adsorbed n-dodecylβ-D-maltoside layers on hematite. J. Colloid Interface Sci. 275:403–409. Mielczarski, J.A. 1986. In situ ATR IR spectroscopic study of xanthate adsorption on marcasite. Colloids Surf. 17:251–271. ———. 1993. External reflection infrared spectroscopy at metallic, semiconductor and nonmetallic substrates. I. Monolayers films. J. Phys. Chem. 97:2649–2663.

flotation0.book Page 334 Tuesday, January 2, 2007 7:36 PM

334

FLOTATION FUNDAMENTALS

Mielczarski, J.A., J. Cases, E. Mielczarski, P. Marion, M. Fordham, and C. Valente. 1999. Production of clean copper sulphide concentrate. In Proceedings of the Rewas’99 Global Symposium on Recycling Waste Treatment and Clean Technologies, San Sebastian, Spain, September 5–9. Warrendale, PA: The Minerals, Metals and Materials Society. Mielczarski, J.A., and E. Mielczarski. 1995. Determination of molecular orientation and thickness of self-assembled monolayers of oleate on apatite by FTIR reflection spectroscopy. J. Phys. Chem. 99:3206–3217. ———. 1999. Infrared external reflection spectroscopy of adsorbed monolayers in a region of strong absorption of substrate. J. Phys. Chem. 103:5852–5859. Mielczarski, J.A., E. Mielczarski, and J.M. Cases. 1996. Interaction of amyl xanthate with chalcopyrite, tetrahedrite and tennantite at controlled potentials. Simulation and spectroelectrochemical results for two component adsorption layers. Langmuir 12:6521–6529. ———. 1997. Infrared evaluation of composition and structure of ethyl xanthate monolayers produced on chalcopyrite, tetrahedrite and tennantite at controlled potentials. J. Colloid. Interface Sci. 188:150–161. ———. 1998. Influence of chain length on adsorption of xanthates on chalcopyrite. Int. J. Miner. Process. 52:215–231. ———. 1999. Dynamic of fluorite-oleate interactions. Langmuir 15:500–508. Mielczarski, J.A., E. Mielczarski, J. Zachwieja, and J.M. Cases. 1995. In situ and ex situ infrared studies of nature and structure of thiol monolayers adsorbed on cuprous sulfide at controlled potential. Langmuir 11:2787–2799. Mielczarski, J.A., P. Nowak, and J.W. Strojek. 1980. Spectrophotometric investigation on products of ethyl xanthate sorption on lead sulphide and galena surfaces. Part I. Oxidized samples. Pol. J. Chem. 54:279–91. ———. 1983. Correlation between the adsorption of sodium dodecyl sulphate on calcium fluorine fluorite and its floatability—an infrared internal reflection spectroscopy study. Int. J. Miner. Process. 11:303–317. Mielczarski, J.A., Z. Xu, and J.M. Cases. 1996. Qualitative and quantitative evaluation of heterogeneous adsorbed monolayers on mineral electrodes by infrared reflection spectroscopy. J. Phys. Chem. 100:7181–7184. Mielczarski, J.A., and R.-H. Yoon. 1989. FTIR external reflection study of molecular orientation in spontaneously adsorbed layers on low absorption substrates. J. Phys. Chem. 93:2034–2038. Mycroft, J.R., G.M. Bancroft, N.S. McIntyre, J.W. Lorimer, and I.R. Hill. 1990. Detection of sulfur and polysulfide on electrochemically oxidized pyrite surfaces by XPS and Raman spectroscopy. J. Electroanal. Chem. 292:139–152. Nesbitt, H.W., G.M. Bancroft, A.R. Pratt, and M.J. Scaini. 1998. Sulfur and iron surface states on fractured pyrite surfaces. Am. Mineral. 83:1067–1076. Nesbitt, H.W., A.G. Schaufuss, G.M. Bancroft, and R. Szargan. 2002. Crystal orbital contributions to the pyrrhotite valence band with XPS evidence for weak Fe-Fe p bond formation. Phys. Chem. Miner. 29(1):72–77. Nesbitt, H.W., A.G. Schaufuss, M. Scaini, G.M. Bancroft, and R. Szargan. 2001. XPS measurement of fivefold and sixfold co-ordinated sulfur in pyrrhotites and evidence for millerite and pyrrhotite surface species. Am. Mineral. 86:318–326. Nesbitt, H.W., A.G. Schaufuss, M.J. Scaini, H. Hochst, G.M. Bancroft, and R. Szargan. 2003. Monitoring fundamental reactions at NiAsS surfaces by synchrotron radiation X-ray photoelectron spectroscopy: As and S air oxidation by consecutive reaction schemes. Geochim. Cosmochim. Acta 67:845–858. Nesbitt, H.W., I. Uhlig, G.M. Bancroft, and R. Szargan. 2003. Resonant XPS study of the pyrite valence band with implications for molecular orbital contributions. Am. Mineral. 88:1279– 1286. Nesbitt, H.W., I. Uhlig, and R. Szargan. 2002. Surface reconstruction and As-polymerization at fractured loellingite FeAs2 surfaces. Am. Mineral. 87:1000–1004. Nesset, J.E., C. Sui, J.Y. Kim, M. Cooper, M. Li, and S.L. Chryssoulis. 2001. The effect of soda ash and lime as pH modifiers in sphalerite flotation. Annual Meeting of the Canadian Mineral Processors. Office of the Secretary of the Canadian Mineral Processors.

flotation0.book Page 335 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

335

Nickolov, Z.S., X. Wang, and J.D. Miller. 2004. Liquid-air interfacial structure of alcohol-octyl hydroxamic acid mixtures: A study by sum-frequency spectroscopy. Spectrochim. Acta A 60A(12):2711. O’Connor, D.J., B.A. Sexton, and R.St.C. Smart, editors. 2003. Surface Analysis Methods in Materials Science. 2nd edition. Springer Series in Surface Science. Berlin: Springer-Verlag. Pattrick, R.A.D., J.M. Charnock, K.E.R. England, J.F.W. Mosselmans, and K. Wright. 1998. Lead sorption on the surface of ZnS with relevance to flotation: A fluorescence REFLEXAFS study. Miner. Eng. 11:1025–1033. Pattrick, R.A.D., K.E.R. England, J.M. Charnock, and J.F.W. Mosselmans. 1999. Copper activation of sphalerite and its relation to flotation: An X-ray absorption spectroscopy reflection extended X-ray absorption fine structure investigation. Int. J. Miner. Process. 55:247–265. Paulucci, G., and K.C. Prince. 1990. Surface core-level shift of lead sulfide. Phys. Rev. B 41:3851– 3853. Petruk, W. 1976. The application of quantitative mineralogical analyses of ores to ore dressing. CIM Bull. 69(767):146–153. ———. 1988. Capabilities of the microprobe Kontron image analysis system: Application to mineral beneficiation. Scanning Microsc. 2(3):1247–1256. Piantadosi, C. 2001. Competitive collector adsorption in the selective flotation of galena and chalcopyrite from iron sulphide minerals. Ph.D. thesis, University of South Australia, Adelaide. Piantadosi, C., M. Jasieniak, W.M. Skinner, and R.St.C. Smart. 2000. Statistical comparison of surface species in flotation concentrates and tails from ToF-SIMS evidence. Miner. Eng. 13:1377–1394. Piantadosi, C., B.L. Pyke, and R.St.C. Smart. 2001. TOF-SIMS statistical analysis of surface hydrophobic-hydrophilic species ratios and contact angle estimation in real ore systems. Pages 305–323 in Interactions in Mineral Processing. Edited by J.A. Finch, S.R. Rao, and L. Huang. Toronto: Metallurgical Society. Piantadosi, C., and R.St.C. Smart. 2002. Statistical comparison of hydrophobic and hydrophilic species on galena and pyrite particles in flotation concentrates and tails from TOF-SIMS evidence. Int. J. Miner. Process. 64:43–54. Pratt, A.R., H.W. Nesbitt, and I.J. Muir. 1994. Generation of acids in mine waste: Oxidative leaching of pyrrhotite in dilute H2SO4 solutions pH 3.0. Geochim. Cosmochim. Acta 58:5147–5159. Prestidge, C.A., and J. Ralston. 1995. Contact angle studies of galena particles. J. Colloid Interface Sci. 172:302–310. Prestidge, C.A., A.G. Thiel, J. Ralston, and R.St.C. Smart. 1994. The interaction of ethyl xanthate with copper II-activated zinc sulfide: Kinetic effects. Colloid Surf. A 85:51–68. Ralston, J. 1994a. Bubble-particle capture. Page 1464 in Flotation II. Volume 2. Edited by S. Castro and J. Alvarez. Chile: Andros. ———. 1994b. The chemistry of galena flotation: Principles and practice. Miner. Eng. 7:715–735. Ren, Y., and T. Kato. 2002. Polarized infrared external reflection spectroscopy of a uniaxial monolayer at the air-water interface. Langmuir 18:6699. Rosiwal, A. 1898. Über Geometrische Gesteinsanalysen. Ein einfacher Weg zür ziffermassigen Feststellung des Quantitätsverhältnisses der Mineral-Bestandteile gemengter Gesteine, Verh, Kaiserlich Koeniglichen Geologischen Reichsanstaet, Vienna, 5/6. (Translated by H.C. Ranson. On Geometric Rock Analysis. Pages 143–75 in A Simple Method for the Numerical Determination of the Quantitative Ratios of the Mineral Fractions of Mixed Rocks. Farnborough, UK: Royal Aircraft Establ. Lib. Trans. No. 871, 1960.) Ruckman, J.C. 1986. Laser microprobe mass analysis, Part II. In Proceedings of the Third International Laser Microprobe Mass Spectrometry Workshop. Edited by F. Adams and L. van Vaeck. Belgium: University of Antwerp. Schaufuss, A.G., H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M. Bancroft, and R. Szargan. 1998. Incipient oxidation of fractured pyrite surfaces in air. J. Electron. Spectrosc. Relat. Phenom. 96:69–82. Schaufuss, A.G., H.W. Nesbitt, M.J. Scaini, H. Hochst, G.M. Bancroft, and R. Szargan. 2000. Reactivity of surface sites on fractured arsenopyrite FeAsS toward oxygen. Am. Mineral. 85:1754–1766.

flotation0.book Page 336 Tuesday, January 2, 2007 7:36 PM

336

FLOTATION FUNDAMENTALS

Schueler, B.W., R.W. Odom, and C.A. Evans. 1986. Non-resonant multiphoton ionization of neutrals ablated by the laser microprobe technique. In Proceedings of the Third International Laser Microprobe Mass Spectrometry Workshop. Edited by F. Adams and L. van Vaeck. Belgium: University of Antwerp. Senior, G.D., and W.J. Trahar. 1991. The influence of metal hydroxide and collector on the flotation of chalcopyrite. Int. J. Miner. Process. 33:321–341. Shannon, L.K., and W.J. Trahar. 1986. The role of collector in sulfide ore flotation. Pages 408–425 in Advances in Mineral Processing. Edited by P. Somasundaran. Littleton, CO: SME. Shen, Y.R. 1989. Surface properties probed by second-harmonic and sum-frequency generation. Nature 337(6207):519. Simon-Kutscher, J., A. Gericke, and H. Huhnerfuss. 1996. Effect of bivalent Ba, Cu, Ni, and Zn cations on the structure of octadecanoic acid monolayers at the air-water interface as determined by external infrared reflection-absorption spectroscopy. Langmuir 12:1027. Smart, R.St.C. 1991. Surface layers in base metal sulphide flotation. Miner. Eng. 4:891–909. Smart, R.St.C., J. Amarantidis, W.M. Skinner, C.A. Prestidge, L. LaVanier, and S.G. Grano. 2003. Surface analytical studies of oxidation and collector adsorption in sulfide mineral flotation. Pages 3–60 in Topics in Applied Physics. Volume 85. Solid-Liquid Interfaces. Edited by K. Wandelt and S. Thurgate. Berlin: Springer-Verlag. Smart, R.St.C., M. Jasieniak, C. Piantadosi, and W.M. Skinner. 2003. Diagnostic surface analysis in sulfide flotation. Pages 241–248 in Proceedings, Flotation and Flocculation: From Fundamentals to Applications, Hawaii, July 28–August 2, 2002. Edited by J. Ralston, J.D. Miller, and J. Rubio. Adelaide: Ian Wark Research Institute, University of South Australia. Smart, R.St.C., M. Jasieniak, K.E. Prince, and W.M. Skinner. 2000. SIMS studies of oxidation mechanisms and polysulfide formation in reacted sulfide surfaces. Miner. Eng. 13:857–870. Smart, R.St.C., and B. Judd. 1994. Improved Lasta filter and copper reflotation performance through surface analysis surveys at WMC’s Olympic Dam operation. Pages 1–4 in Proceedings of the 5th Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Smart, R.St.C., W.M. Skinner, and A.R. Gerson. 1999. XPS of sulfide mineral surfaces: Metaldeficient, polysulfides, defects and elemental sulfur. Surf. Interface Anal. 28:101–105. Smiley, B.L., and G.L. Richmond. 1999. Alkyl chain ordering of asymmetric phosphatidylcholines adsorbed at a liquid-liquid interface. J. Phys. Chem. B 103(4):653. Sperline, R.P., S. Muralidharan, and H. Freiser. 1987. In situ determination of species adsorbed at a solid-liquid interface by quantitative infrared attenuated total reflectance spectrophotometry. Langmuir 3:198. Stowe, K.G., S.L. Chryssoulis, J.Y. Kim, and C.G. Weisener. 1993. LIMS studies of mineral surfaces in sulphide flotation. Pages 457–460 in Secondary Ion Mass Spectrometry, SIMS IX. Edited by A. Benninghoven, Y. Nihei, R. Shimizu, and H.W. Werner. New York: John Wiley & Sons. Suoninen, E., and K. Laajalehto. 1993. Structure of thiol collector layers on sulfide surfaces. Pages 625–629 in Proceedings of the XVIII International Mineral Processing Congress, Sydney, Australia. Volume 3. Melbourne: Australian Institute of Mining and Metallurgy. Swift, P. 1982. Adventitious carbon—the panacea for energy referencing? Surf. Interface Anal. 4:47–51. Szargan, R., A. Schaufuss, and P. Rossbach. 1999. XPS investigation of chemical states in monolayers: Recent progress in adsorbate redox chemistry on sulphides. J. Electron. Spectrosc. Rel. Phenom. 100:357–377. Taylor, C.J., J.N. Brunt, N. Dixon, and P.J. Gregory. 1978. The application of advanced image analysis techniques. Pages 433–489 in Proceedings of the Symposium on Quantitative Analysis of Microstructures in Materials Science, Biology and Medicine. Edited by J.L. Chermant. Stuttgart, Germany: Rieder-Verlag GmbH. Tejedor-Tejedor, M.I., and M.A. Anderson. 1990. The protonation of phosphate on the surface of goethite as studied by CIR-FTIR and electrophoretic mobility. Langmuir 6:602. Tickanen, L.D., M.I. Tejedor-Tejedor, and M.A. Anderson. 1997. Quantitative characterization of aqueous suspensions using variable-angle ATR-FTIR spectroscopy: Determination of optical constants and absorption coefficient spectra. Langmuir 13:4829.

flotation0.book Page 337 Tuesday, January 2, 2007 7:36 PM

SURFACE CHARACTERIZATION AND NEW TOOLS FOR RESEARCH

337

Todd, E.C., and D.M. Sherman. 2003. Surface oxidation of chalcocite Cu2S under aqueous pH=2–11 and ambient atmospheric conditions: Mineralogy from Cu L-and O K-edge X-ray absorption spectroscopy. Am. Mineral. 88:1652–1656. Trahar, W.J. 1981. A rational interpretation of the role of particle size in flotation. Int. J. Miner. Process. 8:289–327. Uhlig, I., R. Szargan, H.W. Nesbitt, and K. Laajalehto. 2001. Surface states and reactivity of pyrite and marcasite. Appl. Surf. Sci. 179:222–229. van der Steldt, K., W. Skinner, and S. Grano. 1993. A study of the interaction of di-cresyl, dithiophosphate with galena and pyrite using micro flotation, z potential measurements and X-ray photoelectron spectroscopy. Report. Ian Wark Research Institute, University of South Australia. von Oertzen, G.U., S.L. Harmer, and W.M. Skinner. In press. XPS and ab initio calculation of surface states of sulfide minerals: Pyrite, chalcopyrite and molybdenite. Mol. Simul. von Oertzen, G.U., W.M. Skinner, and H.W. Nesbitt. 2005. Ab initio and X-ray photoemission spectroscopy study of the bulk and surface electronic structure of pyrite (100) with implications for reactivity. Phys. Rev. B 72(235427):1–10 Wang, X. 2004. Ph.D. thesis, Metallurgical Engineering Department, University of Utah, Salt Lake City. Weisener, C., and A.R. Gerson. 2000. Cu(II) adsorption mechanism on pyrite: An XAFS and XPS study. Surf. Interface Anal. 30:454–458. Wellham, E.J., L. Elber, and D.S. Yan. 1992. The role of carboxy methyl cellulose in the flotation of a nickel sulfide transition ore. Min. Eng. 5:381–395. Wittstock, G., I. Kartio, D. Hirsch, S. Kunze, and R. Szargan. 1996. Oxidation of galena in acetate buffer investigated by AFM and photoelectron spectroscopy. Langmuir 12:5709–5721. Woods, R., and G.A. Hope. 1998. Spectroelectrochemical investigations of the interaction of ethyl xanthate with copper, silver and gold: I. FT-Raman and NMR spectra of xanthate compounds. Colloids Surf. A 137:319–328. ———. 1999. A SERS spectroelectrochemical investigation of the interaction of O-isopropyl-Nethylthionocarbamate with copper surfaces. Colloids Surf. A 146:63–74. Woods, R., G.A. Hope, and G.M. Brown. 1998a. Spectroelectrochemical investigations of the interaction of ethyl xanthate with copper, silver and gold: II. SERS of xanthate adsorbed on silver and copper surfaces. Colloids Surf. A 137:329–337. ———. 1998b. Spectroelectrochemical investigations of the interaction of ethyl xanthate with copper, silver and gold: III. SERS of xanthate adsorbed on gold surfaces. Colloids Surf. A 137:339–344. Woods, R., G.A. Hope, and K. Watling. 2000. A SERS spectroelectrochemical investigation of the interaction of 2 mercaptobenzothiazole with copper, silver and gold surfaces. J. Appl. Electrochem. 30:1209–1222. Young, C.A., and J.D. Miller. 2000. Effect of temperature on oleate adsorption at a calcite surface: An FT-NIR/IRS study and review. Int. J. Miner. Process. 58:331. Zachwieja, J.B., J.J. McCarron, G.W. Walker, and A.N. Buckley. 1989. Correlation between the surface composition and collectorless flotation of chalcopyrite. J. Colloid Interface Sci. 132(2):462–468. Zhu, X.D., H. Suhr, and Y.R. Shen. 1987. Surface vibrational spectroscopy by infrared-visible sum frequency generation. Phys. Rev. B 35(6):3047.

flotation0.book Page 338 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

flotation0.book Page 339 Tuesday, January 2, 2007 7:36 PM

The Flotation of Fine and Coarse Particles Graeme J. Jameson, Anh V. Nguyen, and Seher Ata

INTRODUCTION

The flotation process is used to separate or recover particles over a very wide range of sizes. In the minerals industry, it is not unusual to grind an ore to a top size of 6–7 μm in order to liberate the valuable material, and the feed to flotation will presumably include a certain proportion of material that is below 1 μm. At the other end of the scale, the top size is limited by the performance of the flotation circuit; above a certain point, the recovery drops off as particle size increases. Thus, in mineral flotation, there is a range of particle size that is typically between 10 and 70 μm for base metals, where high recoveries can be obtained with acceptable cell residence times. Outside this range, however, the recoveries decrease for various reasons. This review will focus on the two regions where high recoveries are more difficult to achieve: the ultrafine (70 μm). The physical factors that affect each case and tend to reduce the recovery rate will be discussed. Special measures for increasing the recovery rates will be described. Emphasis will be placed on the physical effects that arise in the extremes of particle size, which helps to explain the difficulties encountered and paves the way for future action. In order to provide suitable bounds for this chapter, a number of relevant topics will be excluded, despite their obvious importance. It will be assumed that the particles to be recovered by flotation have been properly conditioned with appropriate reagents, so flotation chemistry will not be considered a factor in the flotation of fine and coarse particles, except where the hydrophobicity as reflected in the contact angle is specifically involved. Similarly, the general principles of particle capture, hydrodynamics of flotation cells, the bulk flow of froths, and the phenomena of liquid drainage within froths will not be discussed unless there is some relevance to the particular problems of the extremes of particle size in flotation. There are some otherwise interesting papers in which the Hallimond tube was used to study the effects of particle size on flotation, but these papers will not be discussed because of the difficulty in characterizing the hydrodynamics of this apparatus. In the same vein, although there have been numerous papers describing operational procedures in plants to improve the recovery of adventitious coarse particles, which are present only in small concentrations, such discussions will not be included. C A P T U R E O F PA R T I C L E S

The classical diagram showing the effect of particle size on recovery was given by Jowett (1980) using data from Trahar (Figure 1). When lead or zinc sulfides were floated for a fixed time in a batch cell, the recovery approached 100% for particles in the range of 20–70 μm. However, for particle sizes smaller than 20 μm, the recovery appears to diminish uniformly 339

flotation0.book Page 340 Tuesday, January 2, 2007 7:36 PM

340

FLOTATION FUNDAMENTALS

100

% Recovery from Size Fraction

90 80 70

Galena Sphalerite Pyrite Pentlandite

60 50 40 30 20 10 0 1

2

5

10

20

50

100

200

500

1,000

Average Particle Size, μm

Source: Jowett 1980 (using data from W.J. Trahar).

FIGURE 1

The recovery of sulfide mineral particles after 1 minute of flotation in a batch cell

as the particle size is reduced. Similarly, for particles larger than 70 μm, the recovery decreases as the size increases. This phenomenon is well known and is reproducible with many different systems (see, for example, Yianatos, Bergh, and Aguilar 2000). The size range of particles with maximum floatability depends on the density of the particles, so that for coal, high recoveries can be achieved at particle sizes up to 350 μm, and the recovery starts to decrease as the size drops to smaller than about 70 μm. Data, such as those shown in Figure 1, represent the gross recovery from a given cell, including material entrained in the froth. Special problems of fine particle flotation have been discussed in many publications, such as those by Trahar and Warren (1976), Fuerstenau (1980), Somasundaran (1984), Jameson (1984), and Sivarohan (1990). To seek the reasons for the influence of particle size, it will be assumed that the rate-determining step will be the consequence of a number of phenomena that take place in the liquid phase—the pulp—in the flotation cell. It is quite possible that the froth will play a role. It is well known that the froth can be manipulated to reject particles of low grade, which presumably will be attached less strongly to the bubble surfaces in the froth. However, in this chapter, it is assumed that behavior in the liquid phase dominates the kinetics of flotation, and that once a particle enters the froth, it is in effect removed from the cell. PA R T I C L E C A P T U R E I N Q U I E S C E N T L I Q U I D S Models for the Capture Efficiency

The overall process of particle capture in the liquid phase is a balance between two competing effects: those of particle collection or attachment, and those of detachment. Two distinct methods have been used to model the flotation process. In one, flotation is modeled as a sequence of events in which the probabilities of each event are calculated, and the overall probability of capture efficiency is the product of these probabilities. The attachment process is further broken down into a number of steps, such as induction and sliding, film thinning, and rupture, each with a probability. While this is a valid way of decomposing a very

flotation0.book Page 341 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

341

complex phenomenon, an alternative is to regard the capture process in holistic fashion, solving the equations of motion to find an overall capture efficiency and thereby subsuming the various mechanisms within an overall model. For convenience, the probability of collection for the various authors will be expressed in terms of a collection efficiency E, defined as the ratio of the area of the tube of radius R, that is coaxial with a bubble whose radius is rb, such that all particles of radius rp that lie within the collection circle will be captured by the bubble. Thus, 2

2

E = R ⁄ rb

(EQ 1)

Gaudin (1932, 1957) considered the shape of the streamlines around a moving sphere in the limits of creeping or viscous flow where the Reynolds number, Re, approaches zero, and what he called turbulent flow, where Re continues to infinity. For the latter case, Gaudin used the stream function for an inviscid fluid. For the probability of collection, he obtained Viscous fluid:

E = ( 3 ⁄ 2 ) ( rp ⁄ rb )

Inviscid fluid:

E = 3 ( rp ⁄ rb )

2

(EQ 2) (EQ 3)

In Sutherland’s (1948) model, a bubble is assumed to be rising in a liquid with a velocity U. The flow around the bubble is inviscid, so the streamlines can be calculated by irrotational flow theory, assuming that the particles had no effect on the flow field. Sutherland assumed that if a particle lay on a streamline that would bring it within one particle radius of the surface of the bubble, collision or capture would occur. He derived a theory for the “collision radius” for a particle of radius rp being captured by a bubble of radius rb. All particles lying within a circle of this radius, perpendicular to the axis of motion of the bubble, will collide with it. In the limit as rp Re > 0. Because of the asymmetry, the liquid streamlines and the particle trajectories are compressed toward the front surface of rising bubbles and more relaxed in the bubble rear. The asymmetry also has important effects on the attachment interactions. Dobby and Finch (1987) showed that the correction to Equation 7 gives 2 ⎧ sin ϕ a ⎫ 3 ⁄ 16 Re 3-- ⎛ r p⎞ 2 1 + ----------------------------------- ---------------E = 0.56 ⎨ sin ϕ ⎬ 2 ⎝ r b⎠ c⎭ ⎩ 1 + 0.249Re

(EQ 9)

flotation0.book Page 344 Tuesday, January 2, 2007 7:36 PM

344

FLOTATION FUNDAMENTALS

90 80

Collision Angle, degrees

70 60 50 40 Dukhin (1983) Exact Bubble Radius rb = 385 μm Bubble Velocity U = 0.196 m/sec Particle Density ρp = 2,500 kg/m3 Liquid Density ρL = 1,000 kg/m3

30 20 10 0 0

5

10

15

20

25

30

35

40

45

50

Particle Radius, μm

FIGURE 3 Comparison between results of Dukhin (1983) and the exact numerical solution (unpublished data) for the collision angle, ϕc, on the mobile surface of rising bubbles

The last term on the right-hand side of Equation 9 describes the attachment efficiency. The collision angle ϕc depends on the bubble Reynolds number. The attachment angle ϕa is a function of the induction time, similar to Equation 5. The first two terms on the righthand side of Equation 9 describe the collision efficiency and can be modified to include the particle inertia, gravitational forces, and other forces (Dobby and Finch 1987). Further analysis shows that the fore-and-aft asymmetry of the particle trajectories around air bubbles is also influenced by the surface mobility of rising bubbles and inertial forces. The surfactant molecules adsorbed at the surface of rising bubbles are swept to the bubble rear by the liquid, causing the front surface of rising air bubbles to become mobile while the bubble rear with the stagnation cap of adsorbed surfactants becomes immobile. The tangential component of the liquid flow on the mobile bubble surface is nonzero and significantly magnifies the effect of inertial forces governing the particle attachment and collection processes (Nguyen 1999; Dai et al. 1998). The effect of centrifugal force on the attachment in the limit of inviscid (potential) liquid flows was analyzed by Dai, Fornasiero, and Ralston (1999), who employed the approximate results of Dukhin’s analysis (Dukhin 1983). Note that Dukhin’s results for the collision angle (or the angle of tangency) follow the exact numerical solution of the particle motion equation only for very fine particles (Figure 3). For the particle size range (30–100 μm) often encountered in flotation, Dukhin’s results significantly deviate from the exact numerical solution. Further analysis of the particle attachment onto the mobile surface of rising bubbles in flotation is required. Collection Efficiency and Rate Constant

The collection efficiency, E, can be related to the rate constant, k, for batch flotation. Assuming first-order kinetics, the equation for the rate of removal of particles from the pulp ( Jameson, Nam, and Young 1977) can be written as dN p 3QEh --------- = – ⎛ --------------⎞ N p ⎝ 4d p ⎠ dt

(EQ 10)

flotation0.book Page 345 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

345

or dN p 1 6J g --------- = – -- E ⎛ -------⎞ N p 4 ⎝ db ⎠ dt

(EQ 11)

where Np is the number concentration of particles in the flotation cell (m–3); Q is the gas flow rate (m3/sec) into a batch flotation cell of height h (m); dp is the diameter of the particles being floated; and Jg is the superficial gas rate in the cell (m/sec). Jg = Q/A, where A is the area of cross section of the cell. It is easy to show that 6Jg/db is equal to the surface area flux, Sb, that is, the rate of flow of bubble surface area per unit area of cross section of the flotation cell. Thus, a rate constant, k, can be written as follows: 3EJ g ES k = ----------- = --------b2d b 4

(EQ 12)

Thus, if experiments are carried out in a batch flotation cell and the concentration of particles is followed with time, it is relatively simple to calculate the rate constant and, hence, the collection efficiency, E, and determine the effect of particle and bubble size. The bubble surface area flux in flotation cells depends on the bubble size, which is, of course, dependent on the way in which the bubbles are made and the surface chemistry of the flotation pulp, as well as the influence of the particles undergoing flotation. Equation 12 shows that it is pointless to perform batch flotation tests to obtain kinetic data for scale-up unless the gas superficial velocity, Jg , and the Sauter mean bubble diameter, d32, are measured. The importance of Equation 12 cannot be underestimated. The rate constant has dimensions of time–1. The relevant characteristic time for a continuous flotation process is the residence time, τ, while for a batch process, one could take the half-life, τ50, that is, the time taken for the concentration of floatable particles in the batch cell to drop to one-half of the initial value. Thus, an appropriate dimensionless rate constant can be written ES k continuous = --------b4τ

(EQ 13)

for a continuous flotation cell or bank, and ES k batch = ---------b4τ 50

(EQ 14)

for a batch flotation. Note that the two dimensionless rate constants, in general, will not be the same. However, batch tests on an ore will enable E to be determined for a given set of hydrodynamic and interfacial conditions. Scale-up to a larger cell will only be possible if the hydrodynamic conditions are the same, especially the power per unit volume dissipated in the impeller region, and if appropriate scaling is applied for differences in Sb and τ. Measurements of the bubble surface area flux in a wide range of industrial flotation cells have been reported by Jameson and Allum (1984).

flotation0.book Page 346 Tuesday, January 2, 2007 7:36 PM

346

FLOTATION FUNDAMENTALS

Experiments on Fine Particle Flotation

Reay and Ratcliff (1975) conducted an experimental investigation into the effect of particle size on the collection efficiency, using glass beads and latex particles, to test their finding that the rate of flotation should vary as dp2, at constant gas flow rate and bubble size. Using glass beads of diameters dp between 1 and 30 μm, and polystyrene latex particles of diameters between 4 and 10 μm. The results for the latex particles showed that E varied approximately as dp0.44, which completely varies from the expected outcome. It was suggested that the reason for the unexpected result could have been electrostatic double-layer effects, which may have been important with such small particles of almost neutral buoyancy. With the glass beads, however, E varied approximately as dp1.5, which is closer to the predicted value for the exponent. With respect to the dependence of E on bubble diameter, the experimental technique was inconclusive. Bubbles were generated from two frits, giving distributions of mean size 42 and 71 μm. The measured change in the flotation rate constants for these frits supported the prediction that collection efficiency should vary similar to db2.05. Collins and Jameson (1976) measured the flotation rate of fine polystyrene particles, whose diameters ranged from 4 to 20 μm. They studied the rate of flotation of particles in this range using a size-by-size analysis with a Coulter counter. The bubble size was kept constant at 53 μm with a standard deviation of 9 μm. They found that the collection efficiency varies as dp to the 1.5 power. Collins (1975) developed a theory based on a method suggested by the work of Spielman and Fitzpatrick (1973), and the results were in good agreement regarding the particle size dependence. For particles greater than about 50 μm, Collins’ theory predicted that E varies as dp2, essentially the same outcome as that of Reay and Ratcliff (1975). Anfruns and Kitchener (1977) generated single bubbles in a suspension of particles of quartz or glass beads. They measured the collection efficiency of the particles, which were made hydrophobic by surface methylation. For particles in the range of 10–50 μm and bubbles in the range of 500 μm–1 mm, their results indicate that E varied as dp2/db1.69. The flotation of silica, pyrite, and galena in a flotation column was studied by DiazPenafiel and Dobby (1994). With silica of d80 35 μm, they found that over the bubble size – 1.54 ) at constant gas range of 0.8–2.0 mm, the collection efficiency varied as db–0.54( k ∝ d b flow rate. The exponent is rather smaller than found by others, especially that of Anfruns and Kitchener (1977). Recalling that the rate constant is proportional to E/rb, the experimental results can be expressed as shown in Table 1. In the work of Diaz-Penafiel and Dobby (1994), the collection efficiency of bubbles in a flotation column was measured. The bubble sizes were similar to those used in industrial practice and were significantly larger than those used by previous workers. They argued that their experiments were conducted in columns with a high gas holdup, as compared with the single-bubble experiments of others, and that this was the reason for the differences in the observed collection efficiencies. They provided a graph that showed how their results compared with those of others. The effect of the gas holdup on the collection efficiency follows the prediction by Nguyen-Van and Kmet (1994). In Figure 4, data from various authors are plotted, together with the theoretical collection efficiencies predicted by Weber and Paddock (1983) and Yoon and Luttrell (1989). The bubble terminal velocities required for calculation purposes were taken from Motarjemi and Jameson (1978). The reasons for the small irregularities in the theoretical curves relate to the uneven behavior of the bubble rise velocity with bubble size, and the complicated

flotation0.book Page 347 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

347

TABLE 1 Flotation in a quiescent liquid: Experimental results for variation of rate constant with particle and bubble diameter Source

Particle/Bubble Size Range

Rate-Constant Dependence

Reay and Ratcliff (1975)

Latex particles, ρp 1,050 kg/m

3

db = 42 μm db = 71 μm

k independent of dp k ∝ dp0.14

42 < db < 71 μm

k ∝ 1/dp2.9

Glass beads, ρp = 2,500 kg/m3 db = 42 μm db = 71 μm

k ∝ db1.57 k ∝ db2.9

42 < db < 71 μm

k ∝ 1/db2.44

db = 42 μm

k independent of dp

Collins and Jameson (1976)

Latex particles, ρp = 1,050 kg/m3 4 < dp < 30 μm db < 100 μm

k ∝ dp1.65/db3

Anfruns and Kitchener (1977)

Quartz, ρp = 2,500 kg/m3 4 < dp < 50 μm 600 < db < 1,000 μm

k ∝ dp2/db2.69

Fine coal 11.4 dp < 40.1 μm 60 < db < 560 μm

Refer to Equation 8.

Quartz, d80 35 μm

k ∝ db–1.54

Yoon and Luttrell (1989)

Diaz-Penafiel and Dobby (1994)

0.1

Collection Efficiency

dp 11.4 μm Yoon and Luttrell 1989 (coal) dp 12.0 μm Anfruns and Kitchener 1977 (silica) dp 13.0 μm Diaz-Penafiel and Dobby 1994 (silica) 0.01 Weber and Paddock (1983)

0.001 Yoon and Luttrell (1989)

0.0001 100

1,000

10,000

Bubble Diameter, μm

FIGURE 4 Comparison of experimental results and predictions. Note that the predicted values do not include corrections for the induction time.

flotation0.book Page 348 Tuesday, January 2, 2007 7:36 PM

348

FLOTATION FUNDAMENTALS

dependence on the Reynolds number. The data appear to be consistent although there is some spread. The values predicted from theory tend to be too high, but the theoretical collection efficiency here does not include any correction for the induction time, so they are actually collision efficiencies, whereas the experimental values are true collection efficiencies. The data demonstrate the enormous influence of bubble size—there is an approximately eightyfold decrease in the collection efficiency when the bubble size increases from 100 to 2,000 μm. Taking into account the dependence shown in Equation 12, the rate constant will decrease by a factor of 1,600 over the same bubble size range. Thus, in flotation columns, it is essential to create bubbles that are as small as practicable so as to obtain high recoveries. The bubbles must not be too small, however, or they run the risk of passing out in the tailings stream from the column. In the design of the Jameson flotation cell, creating bubbles in the range of 250–300 μm was found to be useful ( Jameson 1988). A conclusion drawn from these studies is that for fine particles undergoing flotation in a quiescent liquid or a column, there is a strong influence of particle size and bubble size. At a constant gas rate and particle size, higher flotation recoveries are achieved with smaller bubbles. C O L L E C T I O N O F PA R T I C L E S I N S T I R R E D V E S S E L S Experiments on the Effect of Particle and Bubble Size

There are very few studies in which the particle size and the bubble size were varied under conditions of controlled agitation. Ahmed and Jameson (1985, 1989), using polymer latex, quartz, and zircon particles, investigated the effect of bubble size on flotation kinetics. They used a series of interchangeable glass frits to generate bubbles of known size: 75, 165, 360, and 655 μm in diameter. The frit was mounted in the bottom of a baffled tank and agitated by a Rushton turbine impeller. The particles used and their sizes were as follows: styrenedivinylbenzene copolymer latex, 4–26 μm; quartz, 5–42 μm; and zircon, 5–32 μm. The performance of the particles was followed on a size-by-size basis using a Coulter counter. There was a very strong effect of bubble size—the flotation rate increased up to one hundredfold when the bubble size was reduced from 655 μm to 75 μm. However, the results were much less regular than those in a quiescent liquid. With the polymer latex, it was possible to conduct experiments in a quiescent liquid, but agitation was required in the case of the quartz and zircon in order to keep the particles in suspension. The results of the flotation tests differed widely depending on the particle density. With the polymer latex, the rate constant was found to vary with the particle size with exponents in the range of 1.29 to 1.88, similar to those for the quiescent liquid shown in Table 1, and with the bubble size db to exponents to the negative powers 1.83 (for dp = 6.5 μm) and 1.13 (dp = 26 μm). If the collection had followed the same mechanism as that in a quiescent liquid, it would have been expected that k should vary approximately with db–3, but this was not observed. In the presence of agitation, the rate constants increased uniformly as the impeller speed increased, suggesting that the agitation led to higher rates of collision but that detachment forces were insignificant. At the highest speed, the dependence of the rate constant on the particle diameter had reduced to between dp0.54 and dp0.84. With the quartz particles, there was clear evidence of particle detachment as the impeller speed increased. The rate constants reached a maximum in the range 300–500 rpm, although with the largest bubbles, no maximum was seen. With zircon, the rate constants for the largest bubbles

flotation0.book Page 349 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

349

increased steadily with increasing speed, but for bubble sizes of 165–655 μm, the flotation rate reached a maximum at about 200–400 rpm, decreasing when the speed increased to 600 rpm. The authors concluded that, in the presence of agitation, the following were true: 1. Smaller bubbles are more efficient in the flotation of fine particles, but the rate constant is never as strongly dependent as the theory for quiescent liquids would suggest ( k ∝ d b–3 ). 2. At constant agitation, the dependence of the flotation rate on the bubble size is a complex function of the particle size and density. 3. The effect of the particle size on the rate constant diminishes as the agitation speed is increased, and the exponent in the relation k ∝ d pn in situations with quartz and zircon particles never exceeds unity, in agreement with results of Trahar (1981) in industrial flotation cells. 4. The effect of bubble size is strongly dependent on the density of the particles. Thus, for polystyrene latex (density 1,050 kg/m3), the exponent m in k ∝ 1 ⁄ d bm varied from 1.66 to 0.82 when the impeller was changed from low speed (200 rpm) to high speed (600 rpm). Over the same speed range, with quartz (density 2,650 kg/m3), m changed from 1.29 to 0.76, and with zircon (density 4,560 kg/m3), m changed from 1.86 to 0.90. In all cases, the rate constant increased with decreasing bubble size. 5. Impeller speed produces strong effects. When operated at zero or low rotational speed, the flotation cell functioned as a low-height column. With the latex, there was a tenfold increase in flotation rate when the speed increased to 600 rpm. With the denser minerals, at a constant particle size, there was clear evidence that particle detachment occurred because the rate constant reached a maximum and then dropped. Interestingly, this effect was most visible with the smallest bubbles used, 75 μm, which may be related to the concept that detachment occurs through the centrifugal force exerted when a bubble is rotating in a turbulent eddy, and that smaller bubbles rotate faster (see Equation 26). The results of Ahmed and Jameson (1985) suggest that the variation of rate constant with particle and bubble size in stirred vessels bears little relation to the results for the same system in the absence of agitation. The only other paper written about the effects of bubble size in mechanical flotation cells is that of Szatkowski and Freyberger (1988), who generated bubbles of various sizes by using different flotation machines and impellers and by changing the air rate and impeller speed. It is not possible to disengage, from their results, the individual effects of the impeller speed and gas rate on the flotation kinetics. Accordingly, their results will not be discussed here. Theoretical Modeling of Flotation in Stirred Vessels

There have been several papers written about the modeling of particle capture in agitated vessels, particularly those of Yoon (1991, 2000); Duan, Fornasiero, and Ralston (2003); and Pyke, Fornasiero, and Ralston (2003). These authors began with the basic equation: dN p --------- = – kN p = – Z pb E coll = – Z pb E c E a E s dt

(EQ 15)

flotation0.book Page 350 Tuesday, January 2, 2007 7:36 PM

350

FLOTATION FUNDAMENTALS

where Zpb is the collision frequency between particles and bubbles per unit volume of liquid; Ecoll is the overall collection efficiency of the bubble; and Ec, Ea, and Es are the efficiencies of the subprocesses of particle–bubble collision, attachment, and stability (related to the detachment efficiency). The collision frequency is found by considering the bubbles and particles in the agitated flotation cell to be suspended in a turbulent field of known energy dissipation rate per unit mass, ε. The collision mechanism is similar to that encountered in the formation of flocs or aggregates in the turbulent flow in a stirred vessel (Abrahamson 1975). Considering the work of Schubert (1999), the following expression for the collision frequency is derived: 4⁄9

d 2 0.33ε d b7 ⁄ 9 ⎛ Δρ⎞ 2 ⁄ 3 - -----Z pb = 5N p N b ⎛ ----b-⎞ -----------------------------⎝ ρ⎠ ⎝ 2⎠ 1⁄3 ν

(EQ 16)

Here, Np and Nb are the number concentrations of particles and bubbles (m–3); ε is the energy dissipation rate (W/kg); and Δρ is the density difference (kg/m3) between the particles and the liquid. The mean energy dissipation rate is simply the power input to the flotation cell divided by the mass of liquid in the cell. The equation appears to suggest that the collision rate increases indefinitely as the bubble diameter is increased. However, in stirred vessels, there is an upper limit on the bubble diameter because, for a given local energy dissipation rate in the vicinity of the impeller, there is a maximum in the size of bubbles that can survive intact (Parthasarathy, Ahmed, and Jameson 1992). In order to make use of Equation 16, an expression for the collection efficiency Ecoll must be determined. Duan, Fornasiero, and Ralston (2003) and Pyke, Fornasiero, and Ralston (2003) used a collection efficiency derived essentially from Sutherland’s equation for the capture of particles for a bubble rising in an inviscid stationary liquid, with modifications to account for the induction time. There was no reason given for their choice, which may be questionable. The mechanism by which the bubbles and particles collide in the turbulent field in the flotation cell is by turbulent transport through a spectrum of eddies of different sizes. The particles and bubbles are in relative motion because of the general motion of the bulk shear flow and the relative motions brought about by the turbulent fluctuations in the liquid in the near neighborhood of the particle and bubble when they are in close proximity. The bubble and the particle will almost certainly be rotating while translating, and this effect must be accounted for. It is this mechanism that allows the collision frequency Zpb to be calculated using Equation 16. Accordingly for Ecoll, it would be appropriate to use a collection efficiency derived for the same type of flow, but such a collection efficiency has not yet been developed. Consequently, it appears that a theoretical expression for the rate of capture of particles in a stirred vessel does not yet exist. Because of the complexity of the theoretical expressions for the effect of bubble size on the flotation rate constant, it is not possible to present a simple relationship. The theory of Duan, Fornasiero, and Ralston (2003) provides a first-order approximation, that k ∝ 1/db11/9. Experimental support for this dependence is found in the work of Ahmed and Jameson (1985, 1989) described earlier.

flotation0.book Page 351 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

351

F L O TAT I O N O F C O L L O I D S A N D N A N O PA R T I C L E S

Previously, particles of sufficient size had been treated so that the effects of electrostatic repulsion or attraction, and Brownian diffusion, were insignificant relative to the viscous and inertial forces in play. Because intermolecular forces are important, their effects were included in considerations of the induction time. This section focuses on capture by Brownian diffusion and the role of electrical double-layer force and hydrophobic surface forces. The capture of fine particles by diffusion was considered by Gaudin (1957) but was first semiquantitatively assessed by Reay and Ratcliff (1973). They wrote an equation of motion for a particle approaching a spherical bubble in a quiescent liquid, in creeping flow. The viscous force on the particle was calculated from Stokes’ law. The solution to their equation gave the collection efficiency for particles sufficiently large for diffusional effects to be significant. They also provided a model for particle capture by diffusion, based on existing equations for molecular mass transfer to a bubble. Thus, a particle was, in effect, treated initially as a molecule, whose diffusivity D was related to particle size by the Stokes–Einstein equation: D = k B T ⁄ 6πμr p

(EQ 17)

where kB is the Boltzmann constant, T is the absolute temperature, μ is the viscosity of the liquid, and rp is the particle radius. The particle capture was modeled using mass transfer theory in the high Peclet number regime (Pe = 2rbU/D), thus enabling a collection efficiency to be determined. Although the theory of Reay and Ratcliff (1973) considered the net transfer of particles by the law of diffusion, the particle collection efficiency was determined in a manner analogous to that found for particles captured by interception. The diffusive collection efficiency, E, was calculated by dividing the net flow of diffusive particles to the bubble surface per unit time by the number of particles swept out by the bubble per unit time, resulting in the following equation: 4k p ( c – c s ) –2 ⁄ 3 E = ------------------------ = 4f ( Sh ⁄ Pe ) = f ( Pe ) cU

(EQ 18)

where kp is the particle mass transfer coefficient; c and cs are the particle concentrations in the bulk and near the bubble surface, respectively; the Sherwood number Sh is defined by Sh = 2rbkp/D; rb is the bubble radius; and U is the bubble rise velocity. In the high Peclet number regime of mass transfer, one obtains Sh = Pe1/3 (Levich 1962, Davies 1972). In Equation 18, f = 1 – cs/c describes the dimensionless driving force for the diffusive mass transfer of particles. The function f has such properties that f → 1 for strong particle attachment (cs ≈ 0) to the bubble surface and that f → 0 for weak particle attachment (c ≈ cs). Reay and Ratcliff (1973) argued that f depends mainly on the interfacial chemistry of the bubble and particle surfaces and the liquid phase. Reay and Ratcliff (1973) treated particle capture by diffusion as quite separate from hydrodynamic capture. For pure diffusive capture, Equation 18 predicts that E ∝ 1/rp, which is opposite from results of the hydrodynamic regime of the particle capture by interception and inertia (Figure 5). The results are interesting in that they predict a minimum in the collection efficiency at a particle size that appears to be between 0.1 and 1 μm. Below 0.1 μm, the collection is enhanced by Brownian diffusion, and for larger particles, the hydrodynamic interactions are more favorable for collection (see Jameson, Nam, and Young 1977).

flotation0.book Page 352 Tuesday, January 2, 2007 7:36 PM

352

FLOTATION FUNDAMENTALS

100 Diffusion Regime

Hydrodynamic Regime

Collection Efficiency, %

10

Collins (1975) 1

Reay and Ratcliff (1973)

0.1

0.01 0.01

0.1

1

10

100

Particle Diameter, μm

Source: Jameson, Nam, and Young 1977.

FIGURE 5 diffusion

Collection efficiency governed by hydrodynamic forces and particle Brownian

Although Reay and Ratcliff ’s (1973) theory predicts the correct trend for the collection of diffusive particles versus the particle size, it remains rather semiquantitative because the theory contains a number of unknown parameters, including the particle subsurface concentration cs, the driving force f for the particle attachment, and the coefficient kp for the particle mass transfer by flotation collection. The intermolecular and surface forces governing the selective particle attachment onto the air bubbles are not included in the theory. The Levich (1962) theory of mass transfer employed in Reay and Ratcliff ’s (1973) model does not consider the microhydrodynamic interaction between a particle and a bubble surface at close approach, which changes the Stokes’ drag force. The microhydrodynamics were later considered by Collins (1975) and Jameson, Nam, and Young (1977) when they examined particle collection by air bubbles, including Brownian diffusion. Collins solved the equation of motion of a particle approaching a bubble in the Stokes’ flow with the inclusion of the van der Waals force and corrections to the drag force due to the microhydrodynamic interaction at short distances. The hydrodynamic corrections used to calculate the forces on the particle and the bubble at close approach distances followed the work of Brenner (1961); Goren and O’Neill (1971); and Goldman, Cox, and Brenner (1967). It has been shown (Collins 1975) that particle collection, which includes Brownian diffusion, requires the numerical solution of the mass conservation equation for simultaneous diffusion and convection, which are position-dependent. The numerical computation solution requires high computing power and sophisticated numerical computation methods, as demonstrated by Davis and colleagues (Loewenberg and Davis 1994; Ramirez et al. 1999; Ramirez, Davis, and Zinchenko 2000). These researchers conveniently formulated the problem in terms of the stochastic pair-distribution function, p ( r ) , which represents the probability density for finding a particle whose center is lying at r (measured from the bubble

flotation0.book Page 353 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

353

center). The pair-distribution function satisfies the quasi-steady Fokker–Planck equation described by Batchelor (1972): ∇ ⋅ [ p ( r )V ( r ) ] = 0

(EQ 19)

where ∇ · describes the divergence operator, and V is the relative velocity between the bubble and the particle. The solution of Equation 19 satisfies the boundary conditions that the function p is zero at the bubble–particle contact and is equal to unity if the particle is significantly far from the bubble surface. The relative velocity in Equation 19 can be resolved in terms of the two components of the relative (microhydrodynamic) mobility functions (of the position vector r ): one lying in the direction of the line of centers between bubble and particle, the other in a direction perpendicular to the line of centers. The numerical solution of the Fokker–Planck equation is then integrated to obtain the collection efficiency described by

∫ ( –p ( r )V ( r ) ( r ⁄ r ) ) dA E = ---------------------------------------------------------2 πU ( r p + r b )

(EQ 20)

The integration in Equation 20 is carried out over the bubble surface at the radial distance of Rp + Rb. Typical, exact numerical results obtained by Ramirez, and colleagues (1999) are shown in Figure 6. The unretarded component of the van der Waals force, as determined by the Hamaker macroscopic approach (and the combined rules), was taken into consideration by

Collection Efficiency, E

0.1

0.01

0.001

0.0001 0.01

0.1

1.0

Size Ratio, λ NOTE: The upper solid line represents a bubble with a free interface and a scaled Hamaker constant A = 0, and the lower solid line represents a bubble with a rigid interface. The dashed lines represent A = 5, 50, 500, 5,000, and ∝ (top to bottom). The dotted lines on the right-hand side represent the results of trajectory calculations in the absence of Brownian diffusion.

Source: Ramirez, Davis, and Zinchenko 2000.

FIGURE 6 Numerical collection efficiency vs. the particle-to-bubble size ratio at a fixed bubble Peclet number Pe = 4 × 104

flotation0.book Page 354 Tuesday, January 2, 2007 7:36 PM

354

FLOTATION FUNDAMENTALS

the authors. Unfortunately, for asymmetric bubble–particle systems in flotation, the unretarded van der Waals force is repulsive and cannot be considered as the driving force for the selective attachment of particles onto the bubble surface. The retarded van der Waals force is influenced by the electrolyte concentration and can be attractive at large separation distances. The prediction of this attractive van der Waals force between a bubble and a particle was not available until recently (Nguyen, Evans, and Schulze 2001). In works by Loewenberg and Davis (1994); Ramirez et al. (1999); and Ramirez, Davis, and Zinchenko (2000), no account was considered for the charge on the surfaces of the bubble and the particle. With colloids and nanoparticles, the charge as reflected in the zeta potential would be expected to have a first-order effect. This anomaly has recently been addressed by George, Nguyen, and Jameson (2004b) who included the effect of attractive van der Waals, electrical double layer, and hydrophobic forces on particle collection by diffusion. The modeling considers the particle trajectory approach and three predominant particle-transport phenomena, including interception, gravity, and Brownian diffusion in the mass balance equation mass, leading to ∂----c + ∇ ⋅ ( cV – D Δc ) = 0 ∂t

(EQ 21)

where c is the particle concentration, t is the reference time, V is the non-Brownian component of the particle velocity, and D is the tensor of the particle diffusivity, which is a function of the microhydrodynamic correction factors. The particle velocity V is determined from the force balance by considering the intermolecular and surface forces, the liquid resistance and gravitational forces, as well as the effect of the microhydrodynamics on the particle motion around the bubble. The numerical solution of Equation 21 was integrated to obtain the collection efficiency, which is described by π

2 E = -------------------------- ∫ – J r ( ϕ ) sin ϕ dϕ c∞ ( U + Vs ) 0

(EQ 22)

where c∞is the particle concentration in the bulk solution, U is the bubble slip velocity, Vs is the particle terminal settling velocity, Jr is the particle flux toward the bubble surface, and ϕ is the polar angle on the bubble surface (measured from the front stagnation point). Theoretical results from these calculations are shown in Figure 7. Figure 8 shows experimental data obtained with silica nanoparticles of different diameters. The flotation experiments were conducted in a 1-L batch column-type cell with an internal diameter of 70 mm and of a height variable of 305–405 mm. The silica particles were from commercial sources and ranged in size from 40 nm to 3.5 mm. The particle concentration in the pulp was about 1% by weight. Cetyltrimethylammonium bromide (CTAB) and Dowfroth 250 were used as the collector and frother, respectively. The bubbles were produced by introducing nitrogen gas through a 68-mmdiameter, heat-resistant glass, sintered disc to produce bubbles of an average diameter, determined photographically, of 150 μm. The experimentally measured zeta potential for particles was –35 mV. The zeta potential for nitrogen bubbles in the flotation collector solutions was measured using a method developed by Kubota and Jameson (1993). Entrainment was eliminated as a source of product recovery using a variety of correction techniques (George, Nguyen, and Jameson 2004a). The non-DLVO (Derjaguin–Landau–Verwey–Overbeek) colloidal

flotation0.book Page 355 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

355

10.0 0 mV

20 mV

50 mV

60 mV

115 mV

Collection Efficiency

1.0

0.1

0.01

0.001

0.0001

0.00001 10

100

1,000

Particle Diameter, nm NOTES: Numerical constants: bubble diameter (150 μm), particle refractive index (1.54), density (2,600 kg/m3), zeta potential (–35 mV), Debye constant (50 nm–1). Hydrophobic force parameters (for the double exponential approximation): K1 = –7 mN/m, K2 = –6 mN/m, l1 = 6 nm, and l2 = 20 nm.

Source: George, Nguyen, and Jameson 2004b.

FIGURE 7 Theoretical prediction of the effect of the bubble zeta potential (ZP) on the collection efficiency of silica nanoparticles at various particle diameters

Collection Efficiency

0.1

0.01

0.001

Nyasil 20 Snowtex ZL Snowtex 20L Model, Zp = 30 mV Model, Zp = 50 mV

0.0001 10

100

1,000

10,000

Particle Diameter, nm NOTES: The numerical constants used in this calculation were the same as those in Figure 7, except for K2 = –20 mN/m in the simulation with –30 mV for the bubble zeta potential.

Source: George, Nguyen, and Jameson 2004b.

FIGURE 8 Comparison between experimental data and the model for collection efficiency of silica particles as a function of particle diameter

forces between hydrophobic surfaces, referred to as hydrophobic forces, have not been precisely described and modeled and, therefore, were estimated in the modeling using the available experimental results obtained with a surface force apparatus and atomic force microscopy (Nguyen and Schulze 2004), resulting in the values shown in Figure 7. The hydrophobic forces are very long range and stronger than the van der Waals attractive forces by many orders of magnitude (Christenson and Claesson 2001). Direct

flotation0.book Page 356 Tuesday, January 2, 2007 7:36 PM

356

FLOTATION FUNDAMENTALS

measurements with the surface force apparatus and atomic force microscopy now reveal (e.g., Christenson and Claesson 1988; Ralston, Dukhin, and Mishchuk 2002; Nguyen et al. 2003) that very small (nanometer and submicron) gas bubbles in solution and at hydrophobic surfaces (Miller et al. 1999; Ishida et al. 2000; Yang et al. 2003) may influence the long range and strong attraction between hydrophobic surfaces. The prediction for the forces between hydrophobic surfaces in the presence of the submicroscopic gas bubbles has been a major challenge (Mishchuk, Ralston, and Fornasiero 2002). As a result, the inclusion of the hydrophobic forces in the modeling of particle collection by Brownian diffusion remains very clouded. The results in Figure 8 confirm the existence of a minimum in the collection efficiency at about 70 nm, which is not much different from the extrapolated minimum determined by Reay and Ratcliffe (1973) (refer to Figure 5). F L O TAT I O N L I M I T S F O R F I N E PA R T I C L E S

It has been hypothesized that there is a lower limit for the flotation of fine particles so that particles below a critical size will be incapable of attachment to a bubble. Scheludko, Toshev, and Bojadjiev (1976) predicted that the minimum diameter for flotation of a particle, dp min, is given by 3κ 2 d p min = 2 ⎛ ------------------------------------------⎞ ⎝ U 2 Δργ ( 1 – cos θ )⎠

1⁄3

(EQ 23)

where κ is the line tension (N), U is the approach velocity, Δρ is the difference in density between the particle and the liquid, and θ is the contact angle. George, Nguyen, and Jameson (2004b) can be used to calculate the minimum floatable size of the silica particles. Their bubble size was 150 μm with a rise velocity of 0.012 m/sec. Assuming a line tension of 2.1 × 10–10 N, following Chalyovska as reported by Schulze (1984), the result is a critical minimum floatable size of 4.6 μm, putting cos θ = 1 as the most favorable case. (The value of the line tension is in an acceptable range, as reported by Drelich [1996].) The critical diameter predicted by Equation 23 is in the range of 0.5–5 μm given by Schulze (1984). In their experiments, George, Nguyen, and Jameson were able to float silica particles down to 50 nm. Entrainment was eliminated as a source of product recovery using a variety of correction techniques (George, Nguyen, and Jameson 2004a). Their results do not support Schulze’s hypothesis. The results of Fukui and Yuu (1980), who floated polystyrene particles of mean diameter 0.6 μm using bubbles of mean diameter 20 μm without collector, can also be examined. Using the terminal rise velocity of 0.00079 m/sec for the bubbles and a particle density of 1,050 kg/m3, the result is a critical lower particle size of 78 μm, which is much greater than the size of the solids actually floated, which was 0.6 μm. Collins and Jameson (1976) had no difficulty in floating polystyrene lattices whose particle size was as low as 3 μm. If these particles followed the prediction of the equation of Scheludko, Toshev, and Bojadjiev (1976), the line tension would have to be at least four orders of magnitude greater than present information suggests. It appears from these comparisons that the hypothesis of Scheludko, Toshev, and Bojadjiev (1976) does not accurately predict the minimum diameter at which flotation can occur, if indeed such a minimum does exist. As Drelich (1996) pointed out, the experimental value of the line tension is so small that it challenges the ability of even the most sensitive

flotation0.book Page 357 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

357

instruments to measure it. He also remarked that “any experimental value of the line tension is a matter of controversy.” It remains to be seen whether the concept of line tension is of any relevance to flotation. F L O TAT I O N O F C O A R S E PA R T I C L E S

At the upper end of the flotation size range, coarse particles present a different set of problems than fine particles. The theoretical equations for the collision efficiency of particles suggest that the efficiency should continue to rise as the particle size increases, but the experimental evidence suggests otherwise, indicating that other subprocesses, particularly those that give rise to particle detachment, become rate-limiting. The main issues to consider are: • The need to provide sufficient buoyancy on the part of the bubbles to be able to lift large particles in the liquid and into the froth; • The detachment forces that come into play when large particles are rising in the liquid, and when they are in the vicinity of the impeller in a mechanical cell; and • The forces that are exerted on coarse particles in the froth layer in the flotation cell. Minimum Bubble Diameter

The minimum bubble diameter db,min necessary to lift a particle of diameter dp in the liquid in the flotation cell can easily be found from simple hydrostatics. Thus, for a bubble–particle combination to float, the mass of the combination must be less than the mass of liquid displaced, from which is found (Schulze 1984): ρp – ρL 1 ⁄ 3 d b,min ≥ d p ⎛ -----------------⎞ ⎝ ρL ⎠

(EQ 24)

where ρp and ρL are the densities of the solid and the liquid, respectively. In practice, the buoyancy of the bubble and attached particle is usually sufficient to lift the particle to the surface in a mechanical flotation cell. Thus, for a 1-mm coal particle of density 1,200 kg/m3 in a slurry of density 1,050 kg/m3, the minimum bubble size is 0.96 mm; and for a 1-mm particle of sulfide mineral of density 4,200 kg/m3 in a pulp of density 1,250 kg/m3, the minimum bubble size is 1.33 mm. Because there is a range of bubble sizes from approximately 200 μm to 3 mm in typical flotation devices, bubbles of the required minimum diameter will typically be present, and aggregates of positive buoyancy will usually be possible. Another factor that eliminates buoyancy as a problem in minerals flotation is the formation of bubble clusters or aggregates, held together by bridging particles that are attached simultaneously to two or more particles. Examples of these clusters will be found in Gaudin (1957); Klassen and Mokrousov (1963); Glembotskii, Klassen, and Plaksin (1963); and Schulze (1984). Recently, Ata and Jameson (2005) investigated, in a systematic way, the formation of clusters in the flotation of fine silica of average size 7 μm using dodecylamine as collector in a mechanical cell. As the collector concentration increased, there was a parallel increase in the fraction of bubbles rising out of the pulp that were engaged in clusters, reaching a maximum of 68% in their experiments. Clusters were also observed by Jameson and Allum (1984) in the second cell of a bank of four in a coal preparation plant. The formation of bubble clusters or aggregates seems to be favored by the presence of high concentrations of highly hydrophobic particles. It has long been claimed by reagent manufacturers for the

flotation0.book Page 358 Tuesday, January 2, 2007 7:36 PM

358

A

FLOTATION FUNDAMENTALS

B

FIGURE 9 Bubble clusters or aggregates formed in flotation cells: (a) flotation of 175-μm coal particles in a mechanical cell at an operating coal preparation plant (Jameson and Allum 1984); (b) flotation of 7-μm silica particles, using dodecylamine at 300 g/t as collector (Ata and Jameson 2005).

coal industry that the recovery of coarse coal particles by flotation can be improved by the use of special collectors. Recently, Yoon, Luttrell, and Asmatulu (2002) described the use of novel reagents that improved the recovery of coarser anthracite particles up to 4 mm in diameter. One of the ways these collectors could work would be by increasing the hydrophobicity of the particles, thereby increasing the probability of cluster formation, which presumably assists in the transport of large particles into the froth. Examples of clusters are shown in Figure 9. Upper Limits Imposed by Shear or Turbulence

The models for the collision efficiency of particles by single bubbles suggest that efficiency increases as the particle size increases. However, the experimental results show that above a certain size, recovery starts to decline, suggesting that detachment of particles must be important in such cases. An important reason for the observed decrease in recovery as particle size increases is the effect of stresses induced by shear or turbulence (Morris 1950; Jowett 1980; Schulze 1977, 1982, 1984; see also Mika and Fuerstenau 1968). It is argued that work must be done to separate bubble–particle aggregates against the restoring force of surface tension. The work of rupture can be found by solving the Young–Laplace equation with appropriate boundary conditions. It is then assumed that the energy for disruption comes from the turbulent field. An aggregate caught in a turbulent eddy will rotate with a frequency that varies with the size of the eddy. If the centrifugal force exerted on the particle exceeds the capillary force between the particle and the bubble (Figure 10), detachment will occur. In a mechanical flotation cell, the region surrounding the impeller and stator is highly turbulent, and it is here that most of the energy is imparted to the liquid slurry. Accordingly, a number of investigations have related the particle–bubble contact rates, the maximum size of bubbles produced, and the maximum floatable particle size to the properties of the turbulent field. Typically, the results are expressed in terms of a “machine acceleration” bm or the mean energy dissipation rate ε. Schulze assumed that bubbles in an agitated liquid behaved as if they were at the center of a vortex, so they rotated with the vortex. The rotational velocity was found using the isotropic turbulence theory. Any particle on the surface of the bubble would then experience a

flotation0.book Page 359 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

A

359

B Particle

Centrifugal Force

Liquid

Gas Bubble Surface Tension

Surface Tension

FIGURE 10 A particle attached to a bubble rotates in a turbulent eddy (a); the forces acting on the particle are depicted in (b)

centrifugal force, tending to move the particle away from the bubble. For the machine acceleration, Schulze (1982) used: ε2 ⁄ 3 b m = 1.9 ---------d b1 ⁄ 3

(EQ 25)

where the dissipation rate ε has units of watts per kilogram of liquid, and db is the diameter of the bubble. In the case of a particle that is much smaller than the relevant bubble, Schulze (1982) showed that the maximum particle size that could remain attached to a bubble of given diameter was – 6γ sin ( π – θ ⁄ 2 ) sin ( π + θ ⁄ 2 ) 1 ⁄ 2 d p,max = ⎛ --------------------------------------------------------------------------⎞ ⎝ ⎠ ( gΔρ + b m ρ p )

(EQ 26)

Here, dp,max is the maximum particle size; γ is the surface tension; θ is the contact angle; g is the acceleration due to gravity; and Δρ is the difference in density between the particle and the liquid. From Equations 25 and 26 it can be seen that when the machine acceleration bm >> g, the maximum particle diameter that can resist detachment varies as db1/6. Thus, these equations, on their own, suggest that a particle of a given size can always form a stable bubble–particle aggregate, provided that a bubble of sufficient size exists in the flotation cell. However, there is a limit to the size of bubble that can be produced in the cell, because the turbulence that gives rise to bubble detachment also causes large bubbles to be unstable and break up into smaller fragments. Maximum Stable Bubble Size and the Maximum Floatable Particle Size

When a bubble is in a shearing flow, it must respond to the velocity gradient in the flow, which causes differences in velocity between the two opposing apexes. The differences in velocity manifest themselves as differences in pressure, which are related to the properties of the turbulent flow field. The pressure differences tend to stretch the bubble against the restoring force of surface tension, and if the stretching forces are too large, the bubble will break into two segments. Levich (1962) and Hinze (1955) showed that for a given flow field, there is a maximum in the size of bubble that can be stable. Thus, in a flotation cell, although the theory suggests that there is no maximum in the size of particle that can be floated, there is, in fact, a maximum that is dictated by the maximum size of bubble that can

flotation0.book Page 360 Tuesday, January 2, 2007 7:36 PM

360

FLOTATION FUNDAMENTALS

exist in the turbulent field in the cell. Hinze (1955) suggested that the bubble would break up if the ratio of the inertial and surface tension forces, in the form of a critical Weber number, exceeded a critical value. The critical Weber number Wec is given as ρu 2 d max We c = -----------------γ

(EQ 27)

where u 2 is the mean square velocity difference between two points in the turbulent flow, a distance apart equal to the maximum bubble diameter, dmax. The mean square velocity difference is obtained from an expression due to Batchelor (1951): u 2 = C 1 ( εd max )

2⁄3

(EQ 28)

where C1 is an empirical constant, and ε is the energy dissipation rate per unit mass in the liquid (ε =, P/ρLV, where P is the power input, ρL is the density of the liquid, and V is the volume of liquid). Parthasarathy, Ahmed, and Jameson (1992) applied this theory to the prediction of the diameter of bubbles produced in a stirred vessel. They wrote an equation for the maximum bubble size in the following form: 3⁄5 ⎞ We 3 ⁄ 5 ⎛ γ D bmax = ⎛ ---------c⎞ ⎜ --------------------------------------------⎟ ⎝ C1 ⎠ ⎝ 3⁄5 2⁄5 ⎠ (P ⁄ ρ V) ρ L

(EQ 29)

L

The value of the critical Weber number has been variously reported as 1.2 (Hinze 1955) and 4.7 (Lewis and Davidson 1982). Parthasarathy, Ahmed, and Jameson (1992) compared predictions from Equation 26 with bubble size distributions measured in a cell stirred by a number of agitator types and found that the equation overpredicted the bubble diameters by a factor of 5 to 10. However, when the volume of liquid in the cell, V, was replaced by the swept volume of the impeller, Vi, excellent agreement was obtained. These authors also showed that the ratio of the Sauter mean diameter of the bubbles in a baffled stirred tank is 0.61dmax. The value of Wec/C1 was found to be 7.23, so that 3⁄5 ⎛ ⎞ ⎛ γ3 ⁄ 5 ⎞ γ d bmax = 3.27 ⎜ --------------------------------------------⎟ = 3.27 ⎜ ------------------------⎟ ⎝ ( P ⁄ ρ L V ) 2 ⁄ 5 ρ L 3 ⁄ 5⎠ ⎝ ε i2 ⁄ 5 ρ L 3 ⁄ 5⎠

(EQ 30)

where εi is the power per unit mass based on the mass of liquid in the volume swept by the impeller: εi = P/ρLVi. This point was investigated in general terms by Schubert (1999), who showed that subject to certain limits, the maximum stable bubble size db,max in the present nomenclature is given by 3⁄5 ⎛ ⎞ γ d b,max = We c0.6 ⎜ --------------------------------------------⎟ ⎝ ( P ⁄ ρ L V ) 2 ⁄ 5 ρ L 3 ⁄ 5⎠

(EQ 31)

flotation0.book Page 361 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

361

where the critical Weber number Wecrit was arbitrarily assumed to have a value of unity. Because the swept volume of the impeller is usually around one-tenth of the volume of the liquid in the cell, Schubert’s equation would overpredict the maximum bubble size by a factor of 102/5, or ~2.5. In Nguyen’s (2003) paper, the equation derived by Schubert (1999) was combined with the work of Schulze (1982) on the maximum size of particle that could remain attached to a bubble. Nguyen improved the accuracy of the solutions to the Young–Laplace equation and found that for most purposes, the following simple equation could be used to represent the maximum floatable particle size: 3γ ( 1 – cos θ ) 1 ⁄ 2 d pmax = ⎛ -------------------------------⎞ ⎝ Δρ ( g + b m ) ⎠

(EQ 32)

According to Equation 25 from Schulze (1977), the value of bm is unbounded as the bubble size increases, but, clearly, there has to be a limit to the size of the bubbles that can exist in the turbulent field in the cell. The turbulent shear field that is responsible for particle capture and detachment also limits the maximum size of the bubbles that can remain stable in the flow. Thus, it is expected that the turbulent shear environment will restrict the range of values of db that are relevant to particle collection. It is assumed that the factor which limits the maximum size of particle that can be recovered in a mechanical cell is the maximum size of bubble that is generated in the cell. Therefore, when the maximum stable bubble size calculated by Parthasarathy, Ahmed, and Jameson (1991) is inserted into Equation 25, the value of the machine acceleration appropriate to the maximum bubble size is b m ( dbmax ) = 1.9ε i2 ⁄ 3 ⁄ d bmax1 ⁄ 3 = 1.28ε i4 ⁄ 5 ρ L1 ⁄ 5 ⁄ γ 1 ⁄ 5

(EQ 33)

Substitution into Equation 32, with bm >> g, yields the following approximate equation for the maximum floatable particle size: γ 6 ⁄ 5 ( 1 – cos θ ) 1 ⁄ 2 d pmax = 1.53 ⎛ -----------------------------------⎞ ⎝ Δρε 4 ⁄ 5 ρ 1 ⁄ 5 ⎠ i

(EQ 34)

L

This equation can be tested by making some realistic assumptions. Use a mechanical flotation cell with power input of 3 kW/m3, and assume the swept volume of the impeller is onetenth of the volume of the liquid in the cell, so that the power consumption in the impeller region is 30 kW/m3. For the particles, chalcopyrite of density 4,200 kg/m3 in a pulp of density 1,280 kg/m3 was chosen. Assume that the surface tension is 0.060 N/m, and the contact angle is 60 degrees. Substitution into Equation 34 gives a maximum floatable particle size of 512 μm, which is quite within the range of expectations (Yianatos, Bergh, and Aguilar 2000). Knowing the maximum floatable particle size, the efficiency Es of the bubble–particle aggregate stability in the turbulent field, which is introduced in Equation 15, can be determined by 6⁄5 d pmax 2 ⎫ ⎧ ⎧ 2.34γ ( 1 – cos θ ) ⎫ E s = 1 – exp ⎨ 1 – ⎛ ------------⎞ ⎬ = 1 – exp ⎨ 1 – ---------------------------------------------- ⎬ ⎝ dp ⎠ d p2 Δρ ε i4 ⁄ 5 ρ L1 ⁄ 5 ⎭ ⎩ ⎭ ⎩

(EQ 35)

flotation0.book Page 362 Tuesday, January 2, 2007 7:36 PM

362

FLOTATION FUNDAMENTALS

This equation is valid when the hysteresis of contact angle— Δθ = θ A – θ R , where θA and θR are the (maximum) advancing and (minimum) receding contact angles—is smaller θR . If the contact angle hysteresis is significant (i.e., Δθ > θ R ), Equation 35 is replaced by 6⁄5

1.17γ sin θ R sin Δθ ⎫ ⎧ -⎬ E s = 1 – exp ⎨ 1 – -----------------------------------------------d p2 Δρ ε i4 ⁄ 5 ρ L1 ⁄ 5 ⎭ ⎩

(EQ 36)

Equations 35 and 36 are derived when the turbulent tensile stresses are dominant. If the turbulent shear stresses or the vibration of bubbles rising to the pulp–froth interface are the dominant forces causing the particle detachment, Equations 35 and 36 must be modified, as shown in works by Nguyen and Schulze (2004). C O A R S E PA R T I C L E F L O TAT I O N W I T H E L U T R I AT I O N

An interesting concept for the flotation of coarse particles was patented by Barbery, Bouajila, and Soto (1989). Experimental results were described in later papers (Soto and Barbery 1991; Oteyaka and Soto 1995). Essentially, the particles to be floated are suspended in a bed that is fluidized by an upflow of water, which passes over the overflow lip so that there is no froth layer at the top of the cell. Air bubbles are introduced into the fluidized bed and attach to coarse hydrophobic particles, lifting them to the overflow where they are removed from the cell as product. Hydrophilic particles sink to the bottom of the cell and pass out as tailings. The upflow of water counteracts the tendency of the particles to settle under gravity, and the additional buoyancy imparted to the hydrophobic particles by the air bubbles is sufficient to allow a separation. Soto and Barbery (1991) described the application of this principle to the separation of phosphate and quartz, and found that recoveries up to 100% could be achieved with particles as large as 0.5 mm. Entrainment of quartz gangue was observed only for particles up to 100 μm. The principles of the air-assisted elutriator used by these authors appear to be used in a proprietary device known as the HydroFloat Separator (Kohmuench, Luttrell, and Mankosa 2001). The HydroFloat is operated in such a way that the bed of particles behaves like a teeter bed (i.e. a dense-phase fluidized bed with a high solids holdup). Clearly, this device has application, especially in industries where coarse particles with a very low feed concentration of slimes are to be treated. However, for streams with a broad range of particle sizes, its application would probably be limited. P H E N O M E N A I N T H E F ROT H P H A S E

Discussion so far has centered on the collision, attachment, and detachment of particles in the pulp phase in a flotation cell. Given the general concern with the overall production of particles from the cell, it is also necessary to consider the froth phase. The first barrier to be overcome by particles in their journey to the overflow lip is to pass through the pulp–froth interface. Bubbles carrying particles will rise toward the interface with a net upward velocity of the same order as the terminal velocity of the bubble–particle aggregate. At the interface, the bubbles slow down in a short time while entering the froth. It is likely that some large particles do not survive the transition. Falutsu (1994) pointed out that when bubbles decelerate, they lose kinetic energy, and the energy released is of the same order as that required to dislodge a particle from the surface of the bubble. It is, therefore, possible that this is one mechanism by which coarse particles fail to be transferred from the pulp to the froth.

flotation0.book Page 363 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

363

It is now well established that selective detachment of particles takes place in the froth. The froth layer is not homogeneous, in that as the bubbles rise, they tend to coalesce, a process that intensifies toward the top layers. When coalescence occurs, the liquid that had been between small bubbles cannot be accommodated between the resulting large bubbles, and therefore tends to drain downward through the froth. Also, the specific bubble surface area is also diminished through coalescence, so if the bubbles are fully laden with particles in the lower levels in the froth, some of the particles will be released into the liquid lamellae after coalescence and will return to the pulp. Early studies of particle behavior in the froth were conducted by Watson and Grainger-Allen, Cutting and Devenish, Cutting et al., Hemmings, Moys, and Flynn and Woodburn and have been well described in papers by Ross (1991a, b, c). It has been observed that the detachment process is selective in that the particles that are only weakly held in the froth, especially those of lowest hydrophobicity, are rejected first, a result confirmed by Moys (1978); Yianatos, Finch, and Laplante (1988); and Ross (1997). The rejection of weakly held particles is desirable or undesirable, depending on the location of the flotation cell in the operating circuit. In rougher and scavenger cells, it is usually desirable to maximize the recovery of values no matter what their grade, so even poorly liberated composites should be recovered in the froth. In cleaner cells, however, it is preferable for the low-grade particles to be squeezed out of the froth by particles that are more strongly bound to the bubbles, so detachment here is desirable. When particles have been detached from bubbles near the top of the froth, they are not necessarily lost from the product. Cutting, Barber, and Newton (1986) and Yianatos, Finch, and Laplante (1987) found that froth upgrading occurs as the froth height increases, presumably because of the selective reattachment of particles, which can occur in lightly loaded froths with unoccupied sites on the surfaces of the bubbles at lower levels in the froth. However, when the availability of vacant sites diminishes, the particles that are streaming downward through the froth have to compete with particles that are already attached to the rising bubbles. The collection of particles in the froth phase has recently been studied by Ata, Ahmed, and Jameson (2002). They carried out experiments in a continuous flotation cell of special design, following Falutsu and Dobby (1989), in which the drop-back from the froth zone could be collected. They floated hematite in the customary way in the pulp zone of a mechanical flotation cell, and injected a continuous feed of small glass particles in suspension in water into the froth. The glass particles could be used in a hydrophilic state, in which case they acted as gangue; or they were hydrophobized to give contact angles of 50°, 66°, and 82°. Their results showed that recovery in the froth can be very rapid—for froth depths as low as 100 mm below the froth-feed entry point, recoveries as high as 87% were achieved where the froth was lightly loaded. However, the work also showed that particles that are already attached to the bubbles, in this case the hematite particles rising out of the flotation cell and into the froth, were difficult to dislodge. In fact, even when the froth was highly stressed, by the addition of high wash-water rates and high froth depths to induce froth drainage and coalescence, the hydrophobic particles added to the froth had little effect on the recovery of hematite. When the bubbles were lightly loaded with hematite by using a feed of low solids content in the flotation cell, it was possible for the feed to the froth to have relatively high solids contents. No practical difficulties were encountered in handling froth feeds solids up to at least 25% w/w. At a constant volumetric froth-feed flow rate, there was a linear relation between the solids content and the recovery. There are reports in Russian literature in which the feed to a flotation cell is spread over the froth (see, for example, Malinovskii et al. 1974). However, the work of Ata, Ahmed, and

flotation0.book Page 364 Tuesday, January 2, 2007 7:36 PM

364

FLOTATION FUNDAMENTALS

Jameson (2002) appears to be the first in which the recovery of solids deliberately fed to the froth phase has been studied in depth. The results suggest that there may be practical cases where it could be beneficial to feed particles into the froth, rather than in the pulp as is the traditional method. This could be especially beneficial with coarse particles, which may be difficult to transport into the froth from a turbulent liquid zone. Such particles may benefit by being injected directly into the froth, where they may be expected to quickly form strong links with bubbles that are of the same order of magnitude. Lambert and Jameson (2001, 2002) have described a process in which a feed containing coarse coal particles up to 2 mm in diameter is split at a size of 300–700 μm. The undersize proceeds to a flotation cell as feed in the usual way, whereas the coarse particles are mixed with clean water and are distributed over the froth. Good yields have been obtained up to at least 1.5 mm, such as with low ash. Development is continuing. In apparent contradiction to the results just described, Ross (1993); van Dyk, van Deventer, and Lorenzen (1995); van Deventer, van Dyk, and Lorenzen (2000); and van Deventer et al. (2002a, b, c) have described a process in which highly hydrophobic coarse particles appear to act as bubble breakers. If particles were introduced on top of a flotation froth, they would settle through it under gravity to be recovered in the underflow, whereas hydrophilic gangue that had been entrained in the films between bubbles would be recovered in the froth product. The explanation provided by van Deventer et al. (2002a) refers to an observation of Hemmings (1981) wherein a particle will be supported by a froth if the following relation holds: 0 < T ⁄ d p < cos θ

(EQ 37)

where T is the bubble film thickness, dp is the particle diameter, and θ is the contact angle. An equivalent diameter that included a shape factor was used to represent dp. For the same particle mass, the model predicts that particles will be floated in the froth in the order of rod, disc, sphere, flat, cone, and cube, with the rod floating best. The experimental results for model particles and for ore particles are discussed in van Deventer et al. (2002b, c). Although very promising results were obtained for coarse particles of diamonds, for example, the process overall was not very satisfactory because of losses of entrained fine particles in the entrained gangue in the froth product. Nevertheless, the process looks promising, especially where only the coarse particles are present, such as in the separation of recycled waste and in coal flotation where the fines have been separated by screens or hydrocyclones. S U M M A RY A N D C O N C L U S I O N S

Although the surface chemistry of the flotation environment obviously plays a role, the limitations on the flotation of coarse and ultrafine particles are mainly physical in nature. In the early days of flotation, many types of cells were devised. Perusal of the patent literature, or of the chapter on flotation in the classic handbook of Taggart (1945), shows an incredible range of inventiveness in the early days. As time went by, a small number of cell types began to gain favor, and by the early 1930s, agitated vessels with pressurized or self-aspirating air supply were coming to the fore. Today, mechanical cells have reached a high peak of development and provide a reliable way of floating the general run of particle sizes encountered in flotation. In the latter part of the 20th century, column cells experienced a resurgence, fueled by the pioneering work of Wheeler (1966), Boutin and Wheeler (1967), and Finch and Dobby

flotation0.book Page 365 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

365

(1990). This type of cell offers some advantages over mechanical cells, particularly in the ability to make use of the froth phase. However, the efficiency of collection is strongly influenced by the bubble size, as can be seen in Figure 4. When the bubble size increases from 200 μm to 2,000 μm, there is a twentyfold decrease in collection efficiency and, hence, in recovery. Thus, the control of bubble size is crucial in obtaining high recoveries in columns, and this measurement is rarely performed in operating concentrators. The theoretical work that has been done, and in many cases backed up by experimental laboratory work, points to the crucial influence of two factors in flotation-contacting devices: the bubble size and the intensity of energy dissipation. The bubble size obviously affects the surface area flux and, hence, the carrying capacity of the device. It also has an effect on the capture efficiency and, consequently, the kinetic rate constant, which is often dependent on the ratio of the particle diameter to the bubble diameter, and to the bubble diameter itself. The intensity of energy dissipation is a measure of the local rate of shear and also of the rotational frequency of bubbles in a suspension. Although the point does not appear to have been systematically investigated, it seems likely that in mechanical cells, the rate of capture of particles is determined in the turbulent flow in the region of the impeller. An indirect way of gauging the importance of this region of space is to study the maximum size of bubbles that is produced in a stirred vessel and to compare measured values with theoretical predictions. The results show that when the energy per unit volume, which is required in the theory, is calculated by dividing the power draw of the impeller by its swept volume, rather than the volume of aerated pulp in the cell, very good agreement is obtained. The shear flow that causes bubbles to break up is also responsible for bringing particles and bubbles into contact, and for disrupting bubble–particle aggregates, thereby setting an upper limit on the size of particles that can be captured in the cell. In columns, the rate of dissipation of mechanical energy is low and tends to be uniformly distributed, so particle capture can be regarded as occurring in a quiescent liquid. The collection efficiency of particle capture by bubbles of a given size is about one-tenth of that in a mechanical cell, so columns are typically much taller than these cells in order to compensate. To increase the rate of flotation of ultrafines, the theories point in one direction: reduce the bubble size and increase the shear rate or the rate of dissipation of mechanical energy. It has long been hypothesized that when the particle size is reduced, at some point below ~1 μm, the collection efficiency should rise, because the particles are approaching the size of large molecules, which can move by Brownian diffusion. Thus, in practical terms, a recovery-particle size curve for a mineral would show a reduction as the size was reduced from, for instance, 30 μm, but at some size in the submicron region, the recovery would start to rise again. Recently, the first experimental confirmation of this effect has been published by George, Nguyen, and Jameson (2004b, 2006). By using silica particles of sizes from 50 nm to 3 μm, the authors found a minimum in the collection efficiency at 70 nm (see Figure 8). This minimum size is somewhat smaller than had been expected from the simple model of Reay and Ratcliff (1973). However, recalling that the diffusion coefficient varies inversely as the particle size from the Stokes–Einstein equation, it is not unexpected. A 70-nm particle corresponds to a very large molecule in solution, for which the diffusion coefficient is already very low compared with those for the normal run of simple molecules or hydrated ions. With coarse particles, the limits to flotation are imposed by the mass that can be lifted by one or more bubbles into the froth, the maximum stable bubble–particle aggregate that

flotation0.book Page 366 Tuesday, January 2, 2007 7:36 PM

366

FLOTATION FUNDAMENTALS

can be achieved in a given turbulent flow field, and the disruptive energy released when the aggregate passes through the froth–liquid interface (Falutsu 1994). Thus, to increase the recovery of coarse particles, lower energy dissipation rates are needed, which cuts across the need for higher shear rates to improve the recovery of ultrafines. Because the aims are contradictory when mechanical cells are used, the best results will probably be found when the ultrafines are separated from the coarse particles and floated in different circuits. The recent discovery of the systematic formation of bubble clusters in the pulp phase in a mechanical cell (Ata and Jameson 2005) is interesting, but it has yet to be demonstrated that clusters have any real effect on the flotation process. It is evident that if clusters can be made to form, they will assist in the recovery of coarse particles by providing increased buoyancy. Cluster formation was observed to be favored by increased contact angle on the surface of the mineral. Perhaps the improved recovery of coarse coal particles with special reagents, as recently reported by Yoon, Luttrell, and Asmatulu (2002), is a consequence of cluster formation encouraged by higher contact angles. Clusters may be detrimental, however, to product grade because of the increased entrainment of gangue trapped between the bubbles in a cluster that rises preformed into the froth. The clusters already have a structure prior to entry to the froth, and it would be very interesting to know how long the bubbles in the cluster stay together before they merge with the other bubbles in the froth. In recent years, there has been increased interest in the capacities of the froth phase to capture particles, particularly in Russia (see Malinovskii et al. 1974). Because the froth has a much higher specific surface area than the pulp, it ought to provide a favorable environment for the capture of particles, especially coarse particles. Particles that are larger than the thickness of the liquid films found in the froth should be relatively easy to capture, and it may be possible to separate particles that are larger than the bubbles if they are introduced directly into the froth rather than transported there out of the liquid phase (Lambert and Jameson 2001, 2002). A significant practical difficulty is the need to distribute a feed containing coarse particles—for example, up to 2-mm in diameter—in a feed stream into the froth. Conventional water distribution systems will readily become blocked with such large particles. In the past, flotation devices have mostly been developed by practical people operating in a plant environment. In the past 20 years, several new types of flotation columns have been introduced that have found their niches, such as the Jameson Cell ( Jameson 1988), the Microcel (Yoon, Adel, and Luttrell 1991), and the pneumatic cell (Bahr, Imhof, and Ludke 1985), which are all variations on column cells; and the air-sparged hydrocyclone (Miller 1981). Mechanical cells, to a large extent, have been developed to a high degree of efficiency, through incremental improvements carried out over a long period of time. The forward approach during the next 100 years will probably be driven by theoretical advances. Key areas for research will most likely be the following: • Improved methods of making small bubbles in environments of high specific-energy dissipation (i.e., high shear rates). • Methods for bringing particles and bubbles into contact in high-shear environments at high gas fractions. Mechanical cells and columns typically operate at gas fractions in the range from 0.05 to 0.25. The practical limit for close-packed bubbles in a liquid is probably around 0.5 to 0.6. Such an environment, under the action of a high shear rate, would be highly favorable for the capture of both coarse and ultrafine particles. However, conventional theories would no longer apply, because such highly concentrated bubble suspensions would not be susceptible to turbulent or laminar

flotation0.book Page 367 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

•

•

367

flow continuum models. New methods would be required for providing the shear because impellers tend to fail at void fractions above about 0.25, due to the formation of cavities in the rear of the blades and subsequent loss of pumping action. Methods for controlling the flow of liquid and particles in the froth phase. The froth phase may prove to be a very fruitful area for research and application. Detailed knowledge of the phenomena that take place in the froth is rather imprecise, although inroads are already being made, as exemplified by the work of Neethling and Cilliers (2003). Improved design of flotation columns. Columns have a valuable, distinctive feature in that they are better fitted to control the froth phase than are mechanical cells. Columns make it easier to use wash water to reduce the concentration of entrained gangue in the flotation product, but much must be learned about the behavior of the wash water and the stability of the froth. Whenever wash water is applied at point sources in the froth, it gives rise to regions that are locally of much higher density than the froth phase, in general, so convective currents will occur that will tend to cause significant back-mixing (of high-grade material) and feed-forward (pockets of high gangue concentration). The overall consequence of internal convection in the froth will probably be deleterious to the product grade, and it will be a challenge to bring such convective cells under control.

AC K N OW L E D G M E N T S

The authors acknowledge the support of the Australian Research Council for the Centre for Multiphase Processes at the University of Newcastle under its Special Research Centres program. REFERENCES

Abrahamson, J. 1975. Collision rates of small particles in a vigorously turbulent fluid. Chem. Eng. Sci. 30:1371–1379. Ahmed, N., and G.J. Jameson. 1985. The effect of bubble size on the rate of flotation of fine particles. Int. J. Miner. Process. 14:195–215. ———. 1989. Flotation kinetics. Pages 77–99 in Frothing and Flotation. Edited by J. Laskowski. New York: Gordon and Breach. Anfruns, J.F., and J.A. Kitchener. 1977. Rate of capture of small particles in flotation. Trans. Inst. Min. Metall. 86:C9–C15. Ata, S., N. Ahmed, and G.J. Jameson. 2002. Collection of particles in the froth phase. Int. J. Miner. Process. 64:101–122. Ata, S., and G.J. Jameson. 2005. The formation of bubble clusters in flotation cells. Int. J. Miner. Process. 76:123–139. Bahr, A., R. Imhof, and H. Ludke. 1985. Applications and sizing of a new pneumatic flotation cell. In Proceedings of the XV International Mineral Processing Congress, Cannes, France. Oxford: Pergamon Press. Barbery, G., A. Bouajila, and H. Soto. 1989. Method for separation of coarse particles. U.S. Patent 4,822,493. April 18. Batchelor, G.K. 1951. Pressure fluctuations in homogeneous turbulence. Proc. Camb. Phil. Soc. 47:359–374. ———. 1972. Sedimentation in a dilute suspension of spheres. J. Fluid Mech. 52:245–268. Boutin, P., and D.A. Wheeler. 1967. Column flotation. Min. World 20(3):47–50.

flotation0.book Page 368 Tuesday, January 2, 2007 7:36 PM

368

FLOTATION FUNDAMENTALS

Brenner, H. 1961. Slow motion of a sphere through a viscous fluid towards a plane surface. Chem. Eng. Sci. 16:242–251. Christenson, H.K., and P.M. Claesson. 1988. Cavitation and the interaction between macroscopic hydrophobic surfaces. Science 239(4838):390–392. ———. 2001. Direct measurements of the force between hydrophobic surfaces in water. Adv. Colloid Interface Sci. 91(3):391–436. Collins, G.L. 1975. Dispersed air flotation of fine particles. Ph.D. thesis, University of London, Imperial College. Collins, G.L., and G.J. Jameson. 1976. Experiments on the flotation of fine particles—the influence of particle size and charge. Chem. Eng. Sci. 31:985–991. Crawford, R., and J. Ralston. 1988. The influence of particle size and contact angle in mineral flotation. Int. J. Miner. Process. 23(1–2):1–24. Cutting, G.W., S.P. Barber, and S. Newton. 1986. Effects of froth structure and mobility on performance and simulation of continuously-operated flotation cells. Int. J. Miner. Process. 16:43–61. Dai, Z., S. Dukhin, D. Fornasiero, and J. Ralston. 1998. The inertial hydrodynamic interaction of particles and rising bubbles with mobile surfaces. J. Colloid Interface Sci. 197(2):275–292. Dai, Z., D. Fornasiero, and J. Ralston. 1999. Particle–bubble attachment in mineral flotation. J. Colloid Interface Sci. 217(1):70–76. ———. 2000. Particle–bubble collision models—a review. Adv. Colloid Interface Sci. 85:231–256. Davies, J.T. 1972. Turbulence phenomena at free surfaces. AIChE J. 18(1):169–173. Diaz-Penafiel, P., and G.S. Dobby. 1994. Kinetic studies in flotation columns: Bubble size effect. Miner. Eng. 7(4):465–478. Dobby, G.S., and J.A. Finch. 1987. Particle size dependence in flotation derived from a fundamental model of the capture process. Int. J. Miner. Process. 21(3–4):241–260. Drelich, J. 1996. The significance and magnitude of the line tension in three-phase (solid-liquid-fluid) systems. Colloids and Surf. A 116:43–54. Duan, J., D. Fornasiero, and J. Ralston. 2003. Calculation of the flotation rate constant of chalcopyrite particles in an ore. Int. J. Miner. Process. 71:227–237. Dukhin, S.S. 1983. Critical value of the Stokes number and the Sutherland formula. Kolloidn. Zh. 45(2):207–218. Falutsu, M. 1994. Column flotation froth characteristics—stability of the bubble–particle system. Int. J. Miner. Process. 40:225–243. Falutsu, M., and G.S. Dobby. 1989. Direct measurement of froth drop-back and collection zone recovery in a laboratory flotation column. Miner. Eng. 2(3):377–386. Finch, J.A., and G.S. Dobby. 1990. Column Flotation. Oxford: Pergamon Press. Flint, L.R., and W.J. Howarth. 1971. The collision efficiency of small particles with air bubbles. Chem. Eng. Sci. 26:1155–1168. Fuerstenau, D.W. 1980. Fine particle flotation. Pages 669–705 in Fine Particle Processing. Edited by P. Somasundaran. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Fukui, Y., and S. Yuu. 1980. Collection of sub-micron particles in flotation. Chem. Eng. Sci. 35:1097–1105. Gaudin, A.M. 1932. Flotation. New York: McGraw-Hill. ———. 1957. Flotation. 2nd edition. New York: McGraw-Hill. George, P., A.V. Nguyen, and G.J. Jameson. 2004a. Assessment of true flotation and entrainment in the flotation of sub-micron particles by fine bubbles. Miner. Eng. 17:847–853. ———. 2004b. Modelling and validation of nanoparticle flotation. In Proceedings of the 32nd Australasian Chemical Engineering Conference Chemeca 2004, September 27–29, Sydney, Australia. Canberra: Australasian Institution of Engineers. ———. 2006. Demonstration of a minimum in the recovery of nanoparticles by flotation: Theory and experiment. Chem. Eng. Sci. 61:2494–2509. Glembotskii, V.A., V.A. Klassen, and I.N. Plaksin. 1963. Flotation. New York: Primary Sources.

flotation0.book Page 369 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

369

Goldman, A.J., R.G. Cox, and H. Brenner. 1967. Slow viscous motion of a sphere parallel to a plane wall—II Couette flow. Chem. Eng. Sci. 22(4):653–660. Goren, S.L., and M.E. O’Neill. 1971. Hydrodynamic resistance to a particle of a dilute suspension when in the neighborhood of a large obstacle. Chem. Eng. Sci. 26(3):325–338. Gu, G., K. Nandakumar, and J. Masliyah. 2003. Effects of the physical environment on induction time of air-bitumen attachment. Int. J. Miner. Process. 69(1–4):235–250. Hemmings, C.E. 1981. On the significance of flotation froth liquid lamella thickness. Trans. Inst. Min. Metall. 90:C96–C102. Hewitt, D., D. Fornasiero, and J. Ralston. 1995. Bubble–particle attachment. J. Chem. Soc. Faraday Trans. 91(13):1997–2001. Hinze, J.O. 1955. Fundamentals of the hydrodynamic mechanism of splitting in dispersion processes. AIChE J. 1:289–295. Ishida, N., T. Inoue, M. Miyahara, and K. Higashitani. 2000. Nano-bubbles on a hydrophobic surface in water observed by tapping-mode atomic force microscopy. Langmuir 16(16):6377–6380. Jameson, G.J. 1984. Physical aspects of fine particle flotation. Pages 215–232 in Principles of Mineral Flotation: The Wark Symposium. Edited by M.H. Jones and J.T. Woodcock. Melbourne: Australian Institute of Mining and Metallurgy. ———. 1988. New concept in flotation column design. Miner. Metall. Process. 5(1):44–47. Jameson, G.J., and P. Allum. 1984. A Survey of Bubble Sizes in Industrial Flotation Cells. Melbourne: AMIRA International.. Jameson, G.J., S. Nam, and M.M. Young. 1977. Physical factors affecting recovery rates in flotation. Miner. Sci. Eng. 9:103–118. Jowett, A. 1980. Formation and disruption of particle–bubble aggregates in flotation. Pages 720–754 in Fine Particles Processing. Edited by P. Somasundaran. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Klassen, V.I., and V.A. Mokrousov. 1963. An Introduction to the Theory of Flotation. London: Butterworths. Kohmuench, J.N., G.H. Luttrell, and M.J. Mankosa. 2001. Coarse particle concentration using the HydroFloat Separator. Miner. Metall. Process. 18(2):61–67. Kubota, K., and G.J. Jameson. 1993. A study of the electrophoretic mobility of a very small inert gas bubble suspended in aqueous inorganic electrolyte and cationic surfactant solutions. J. Chem. Eng. ( Japan) 26(1):7–12. Lambert, N.W.A., and G.J. Jameson. 2001. A new process for coarse coal flotation. Pages 1–10 in Proceedings of the 18th Pittsburgh Coal Conference, Newcastle, Australia, December 3–7. Pittsburgh: University of Pittsburgh. ———. 2002. Coarse coal flotation. Paper C1 in Proceedings of the 9th Australian Coal Preparation Conference, Yeppoon, Queensland, October 13–17. New South Wales, Australia: Australian Coal Preparation Society. Levich, V.G. 1962. Physicochemical Hydrodynamics. Englewood Cliffs, NJ: Prentice-Hall. Lewis, D.A., and J.F. Davidson. 1982. Bubble-splitting in shear flow. Trans. Inst. Chem. Eng. 60:283–291. Li, D., J.A. Fitzpatrick, and J.C. Slattery. 1990. Ind. Eng. Chem. Res. 29(6):955–967. Loewenberg, M., and R.H. Davis. 1994. Flotation rates of fine, spherical particles and droplets. Chem. Eng. Sci. 49(23):3923–3941. Malinovskii, V.A., N.V. Matveenko, O.M. Knaus, Y.P. Uvarov, N.N. Teterina, and N.N. Boiko. 1974. Technology of froth separation and its applications. Pages 717–726 in Proceedings of the International Minerals Processing Congress. Edited by M.J. Jones. London, England: Institute of Mining and Metallurgy. Mika, T.S., and D.W. Fuerstenau. 1968. A microscopic model of the flotation process. Page 246 in Proceedings of the 8th International Mineral Processing Congress. Volume 2. Leningrad: Mekanobr. Miller, J.D. 1981. Air-sparged hydrocyclone and method. U.S. Patent 4,279,743. Miller, J.D., Y. Hu, S. Veeramasuneni, and Y. Lu. 1999. In-situ detection of butane gas at a hydrophobic silicon surface. Colloids Surf. A 154(1–2):137–147.

flotation0.book Page 370 Tuesday, January 2, 2007 7:36 PM

370

FLOTATION FUNDAMENTALS

Mishchuk, N., J. Ralston, and D. Fornasiero. 2002. Influence of dissolved gas on van der Waals forces between bubbles and particles. J. Phys. Chem. A 106(4):689–696. Morris, T.M. 1950. Measurement of equilibrium forces between an air bubble and an attached solid in water. Min. Eng. 2:91–95. Motarjemi, M., and G.J. Jameson. 1978. Mass transfer from very small bubbles—the optimum bubble size for aeration. Chem. Eng. Sci. 33:1415. Moys, M. 1978. A study of a plug-flow model for flotation froth behavior. Int. J. Miner. Process. 5:21–28. Neethling, S.J., and J.J. Cilliers. 2003. Modelling flotation froths. Int. J. Miner. Process. 72(1–4):267–287. Nguyen, A.V. 1999. Hydrodynamics of liquid flows around air bubbles in flotation: A review. Int. J. Miner. Process. 56(1–4):165–205. ———. 2003. New methods and equations for determining attachment tenacity and particle size limit for flotation. Int. J. Miner. Process. 68:167–182. Nguyen, A.V., G.M. Evans, and H.J. Schulze. 2001. Prediction of van der Waals interaction in bubble– particle attachment in flotation. Int. J. Miner. Process. 61(3):155–169. Nguyen, A.V., S. Kmet, and H.J. Schulze. 1995. Collection events in flotation. Pages 81–85 in Proceedings of the International Congress Minerals Processing. Volume 3. Littleton, CO: SME. Nguyen, A.V., J. Nalaskowski, J.D. Miller, and H.J. Butt. 2003. Attraction between hydrophobic surfaces studied by atomic force microscopy. Inter. J. Miner. Process. 72(1–4):215–225. Nguyen, A.V., J. Ralston, and H.J. Schulze. 1998. On modelling of bubble–particle attachment probability in flotation. Int. J. Miner. Process. 53(4):225–249. Nguyen, A.V., and H.J. Schulze. 2004. Colloidal Science of Flotation. New York: Marcel Dekker. Nguyen-Van, A., and S. Kmet. 1994. Probability of collision between particles and bubbles in flotation: The theoretical inertialess model involving a swarm of bubbles in pulp phase. Int. J. Miner. Process. 40(3–4):155–169. Nguyen-Van, A.V. 1993. On the sliding time in flotation. Int. J. Miner. Process. 37(1–2):1–25. Oteyaka, B., and H. Soto. 1995. Modelling of negative bias column for coarse particles flotation. Miner. Eng. 8(1–2):91–100. Parthasarathy, R., N. Ahmed, and G.J. Jameson. 1992. Bubble breakup in stirred vessels—predicting the Sauter mean diameter. Trans. Inst. Chem. Eng. 60:295–301. Pyke, B., D. Fornasiero, and J. Ralston. 2003. Bubble–particle heterocoagulation under turbulent conditions. J. Colloid Interface Sci. 265:141–151. Ralston, J., S.S. Dukhin, and N.A. Mishchuk. 2002. Wetting film stability and flotation kinetics. Adv. Colloid Interface Sci. 95(2–3):145–236. Ramirez, J.A., R.H. Davis, and A.Z. Zinchenko. 2000. Microflotation of fine particles in the presence of a bulk-insoluble surfactant. Int. J. Multiphase Flow 26(6):891–920. Ramirez, J.A., A. Zinchenko, M. Loewenberg, and R.H. Davis. 1999. The flotation rates of fine spherical particles under Brownian and convective motion. Chem. Eng. Sci. 54(2):149–157. Reay, D., and G.A. Ratcliff. 1973. Removal of fine particles from water by dispersed air flotation: Effects of bubble size and particle size on collection efficiency. Can. J. Chem. Eng. 51:178–185. ———. 1975. Experimental testing of the hydrodynamic collision model of fine particle flotation. Can. J. Chem. Eng. 53:481–486. Ross, V. 1991a. The behavior of particles in flotation froths. Miner. Eng. 4(7–11):959–974. ———. 1991b. An investigation of sub-processes in equilibrium froths. I: The mechanisms of detachment and drainage. Int. J. Miner. Process. 31:37–50. ———. 1991c. An investigation of sub-processes in equilibrium froths. II: The effect of operating conditions. Int. J. Miner. Process. 31:51–71. ———. 1997. Particle–bubble attachment in flotation froths. Miner. Eng. 10(7):695–706. Ross, V.E. 1993. Separation method and apparatus. South African Patent 93/5,367. Scheludko, A., B.V. Toshev, and D.T. Bojadjiev. 1976. Attachment of particles to a liquid surfacecapillary theory of flotation. J. Chem. Soc. Faraday Trans. 1 72:2815. Schubert, H. 1999. On the turbulence-controlled micro-processes in flotation machines. Int. J. Miner. Process. 56:257–276.

flotation0.book Page 371 Tuesday, January 2, 2007 7:36 PM

THE FLOTATION OF FINE AND COARSE PARTICLES

371

Schulze, H.J. 1977. New theoretical and experimental investigations on stability of bubble–particle aggregates in flotation: A theory on the upper particle size of floatability. Int. J. Miner. Process. 4:241–259. ———. 1982. Dimensionless number and approximate calculation of the upper particle size of floatability in flotation machines. Int. J. Miner. Process. 9:321-328. ———. 1984. Physicochemical Elementary Processes in Flotation. Amsterdam: Elsevier. Sivarohan, R. 1990. The problem of recovering very fine particles in mineral processing-a review. Int. J. Miner. Process. 28:247-288. Somasundaran, P. 1984. Role of surface phenomena in the beneficiation of fine particles. Min. Eng. (August):1177-1186. Soto, H., and G. Barbery. 1991. Flotation of coarse particles in a counter-current column cell. Miner. Metall. Process. 8:16-21. Spielman, L.A., and J.A. Fitzpatrick. 1973. Theory of particle collection under London and gravity forces. J. Colloid Interface Sci. 43:350-373. Sutherland, K.L. 1948. Physical chemistry of flotation. XI. Kinetics of the flotation process. J. Phys. Chem. 52:394-425. Szatkowski, M., and W.L. Freyberger. 1988. The effect of bubble size distribution on selectivity of iron ore flotation. Int. J. Miner. Process. 23:213-227. Taggart, A.F. 1945. Handbook of Mineral Dressing. 4th printing of the 1927 edition. New York: John Wiley. Trahar, W.J. 1981. A rational explanation of the role of particle size in flotation. Int. J. Miner. Process. 8:289-327. Trahar, W.J., and L.J. Warren. 1976. The flotability of very fine particles-a review. Int. J. Miner. Process. 3(2):103-131. van Deventer, J.S.J. W.A. van Dyk, and L. Lorenzen. 2000. The separation of coarse particles by a moving froth bed. Pages C7-1–C7-8 in Proceedings of the 21st International Mineral Processing Congress, Rome. Italy, July 23–28. Amsterdam: Elsevier. van Deventer, J.S.J., W.A. van Dyk, L. Lorenzen, and D. Feng. 2002a. The dynamic behavior of particles in flotation froths. Part I: Model. Miner. Eng. 15:635–645. ———. 2002b. The dynamic behavior of particles in flotation froths. Part II: Density tracer tests. Miner. Eng. 15:647–657. ———. 2002c. The dynamic behavior of particles in flotation froths. Part III: Ore particles. Miner. Eng. 15:659–665. van Dyk, W.A., J.S.J. van Deventer, and L. Lorenzen. 1995. The dynamic behavior of coarse particles in flotation froths. Pages 99–103 in Proceedings of the 19th International Minerals Processing Congress. Volume 3. Littleton, CO: SME. Vinogradova, O.I. 1994. On the attachment of hydrophobic particles to a bubble on their collision. Colloids Surf. A 82(3):247–254. Wang, W., Z. Zhiang, K. Nandakumar, J. Masliyah, and Z. Zhenghe. 2004. An induction time model for the attachment of an air bubble to a hydrophobic sphere in aqueous solutions. Int. J. Miner. Process. 75(1–2):69–82. Wang, W., Z. Zhou, K. Nandakumar, Z. Xu, and J. Masliyah. 2003. Attachment of individual particles to a stationary air bubble in model systems. Int. J. Miner. Process. 68(1–4):47–69. Weber, M.E. 1981. Collision efficiencies for small particles with a spherical collector at intermediate Reynolds numbers Sep. Process. Technol. 2:29–33. Weber, M.E., and D. Paddock. 1983. Interceptional and gravitational collision efficiencies for single collectors at intermediate Reynolds numbers. J. Colloid Interface Sci. 94:328–335. Wheeler, D.A. 1966. Big flotation column mill tested. Eng. Min. J. 167(11):98–99. Yang, J., J. Duan, D. Fornasiero, and J. Ralston. 2003. Very small bubble formation at the solid–water interface. J. Phys. Chem. B 107(25):6139–6147. Ye, Y., and J.D. Miller. 1988. Bubble–particle contact time in the analysis of coal flotation. Coal Prep. 5(3–4):147–166.

flotation0.book Page 372 Tuesday, January 2, 2007 7:36 PM

372

FLOTATION FUNDAMENTALS

Yianatos, J.B., L.G. Bergh, and J. Aguilar. 2000. The effect of grinding on mill performance at Division Salvador, Codelco-Chile. Miner. Eng. 13(5):485–495. Yianatos, J.B., J.A. Finch, and A.R. Laplante. 1987. Cleaning action in column flotation froths. Trans. Inst. Min. Metall. 96:C199–C205. ———. 1988. Selectivity in column flotation froths. Int. J. Miner. Process. 23:279–292. Yoon, R.H. 1991. Hydrodynamic and surface forces in bubble–particle interactions. Aufbereit. Tech. 32(9):474–85. ———.2000. The role of hydrodynamic and surface forces in bubble-particle interaction. Int. J. Miner. Process. 58(1–4):129–143. Yoon, R.H., G. Adel, and G.H. Luttrell. 1991. Process and apparatus for separating fine particles by microbubble flotation together with a process and apparatus for generation of microbubbles. U.S. Patent 4,981,582. Yoon, R.H., and G.H. Luttrell. 1989. The effect of bubble size on fine particle flotation. Pages 101– 122 in Frothing and Flotation. Edited by J. Laskowski. New York: Gordon and Breach. Yoon, R.H., G.H. Luttrell, and R. Asmatulu. 2002. Extending the upper particle size limit for coal flotation. J. South African Inst. Min. Metall. 102(7):411–415. Yoon, R.H., and J.L. Yordan. 1991. Induction time measurements for the quartz-amine flotation system. J. Colloid Interface Sci. 141(2):374–383.

flotation0.book Page 373 Tuesday, January 2, 2007 7:36 PM

PART 3

Flotation Chemistry

373

flotation0.book Page 374 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

flotation0.book Page 375 Tuesday, January 2, 2007 7:36 PM

Flotation Reagents—A Critical Overview from an Industry Perspective D.R. Nagaraj and S.A. Ravishankar

A B S T R AC T

An overview of flotation reagents used in sulfide and nonsulfide separations is given in this chapter from an industrial perspective to highlight current applications. The historical and theoretical aspects, which are well documented in the literature, are highlighted only when necessary to support the theme of this chapter, namely, critical analysis of reagent research, design, selection, and optimization with an emphasis on relevance to plant needs and practice. The chemistry and applications of collectors, frothers, and modifiers that are currently used in the industry are briefly reviewed; pitfalls and gaps arising from somewhat arbitrary and reductionistic practice in reagent applications are identified; and the impact of environmental regulations on current and future reagent development is discussed. The importance of frothers and modifiers, which have received less overall attention than collectors, is highlighted. The critical need for a rational, holistic (total system or top-down) approach for reagent design, selection, and optimization is discussed with a view to providing remedies for pitfalls and a platform for filling gaps in reagent chemistry and applications, and to developing reagent technology that can deliver robust solutions to plant needs. Elements of such a holistic approach, specifically the definition of mineralogy and plant needs, rational reagent selection, and laboratory and plant best practices, are described. INTRODUCTION

The introduction of the third (gas) phase made flotation the most important beneficiation technique in the last 100 years. This also made possible the use of a variety of reagents, that is, collectors, modifiers, and frothers. The historical development of flotation reagents mirrors that of flotation technology in general. The chronology of development of flotation reagents of industrial relevance during the past 100 years is given in Table 1. Booth and Freyberger (1962) noted that two distinct periods (or phases) can be recognized in the historical development of flotation reagents—oil flotation (1860–1920) and chemical flotation (1921–present). A closer scrutiny would suggest three distinct periods (see Table 2), the first being still oil flotation. The second period might be characterized as discovery and expansion (1921–1950*), and the third period as rational, targeted design (1951–present). In the early days of flotation, large quantities of fatty and oily materials, up to 10%–20% of the weight of the ore, were used as a means of separating the value components from gangue minerals in ore pulps. Later, various gases replaced these large quantities of oil as a buoyant * This is an approximate date used for convenience; the shift occurred sometime in the early 1950s. 375

flotation0.book Page 376 Tuesday, January 2, 2007 7:36 PM

376

TABLE 1

FLOTATION CHEMISTRY

Chronology of reagent development*

Reagent Oils

Principle Inventors and/or Authors

Company or Institution

Oleic acid

Year 1860– 1920

Comments

1910?

CuSO4, lime, SO2, KMnO4 sulfidization

Bradford, Ralston, Owen

Soluble collectors

Perkins, Corliss, Sayre

1912– Selective flotation, activation, 1915 depression Metals Recovery Co.

1917– Short-chain amines, 1921 naphthylamine, naphthol, thiocarbanilide, and others

Chelating agents

Vivian

1921

Cyanide

Sheridan, Griswold

1922

Depression of iron sulfides

Dextrin, tannin, starch

Price

Minerals Separation Ltd.

1921

Depressants in coal flotation

Xanthate

Keller, Lewis

Minerals Separation Ltd.

1923

U.K. patent; alkali xanthate

Dithiocarbamate

Sayre

Minerals Separation Ltd.

1924

K salt for Cu sulfide and others, especially in alkaline circuit

Dyes

Berl

1924

Behavior of dye

Xanthate

Sayre

1924

Pb Zn xanthate as collector for nonacid ore pulps

Xanthate

Hallet, Ryder

1925

Determination of xanthate in solutions

Xanthate

Keller

1925 Minerals Separation North America

U.S. patent; alkali xanthate

Xanthate

Lewis

1925 Minerals Separation North America

U.S. patent; alkali xanthate

Minerals Separation Ltd.

Cresylic Whitworth, dithiophosphates (DTPs) Cyanamid

1926

Dicresyl DTPs—acid form

Xanthogen formate

Douglass

DuPont

1927

Sulfide flotation

Alkyl DTP

Christmann

Cyanamid

1928

Aliphatic DTPs—sodium salts

Mercaptobenzothiazole

Smith, Bolton

DuPont

1927– Sulfide ore 1928

Trithiocarbonate

Douglass

DuPont

1928

Cu ore (diethyl trithiocarbonate—neutral molecule)

Mercaptan

Hess

Barrett Co.

1929

Benzyl mercaptan

Dextrin, tannin, starch

Schafer

1930

Depressant in nonsulfide flotation

Mercaptan

Gaudin

1930

Amyl mercaptan for malachite, etc.

Trithiocarbonate

Gaudin

University of Utah 1930

Mercaptobenzothiazole

Gieser

1931

Sphalerite Oxidized silver ore (Table continued next page)

flotation0.book Page 377 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 1

377

Chronology of reagent development* (continued)

Reagent Trithiocarbonate

Principle Inventors and/or Authors Ferris

Dithiocarbamate

Gaudin

Amine Chelating agents

Christmann

Company or Institution

Year 1931

Comments Oxidized ore of Cu and Pb (sodium salt)

1932

Zinc ore, pH 7–10

H.Th. Bohme A.G.

1933

Quaternary amines

Cyanamid

1934

Quebracho

Orelli

Chelating agents

DeWitt, Batchelder

1939 Michigan College of Mining and Tech.

Flotation of oxides

Dextrin, tannin, starch

Booth

Cyanamid

1939

Depression of carbonaceous matter in Au and sulfide flotation

Dyes

Booth

Cyanamid

1940

Depressants in nonsulfide flotation

Hydroxamate in flotation

JPopperle

Fried Krupp Grusonwerk Akt.-Ges

1940

Sulfide and oxide ores

Cyanamid

1943

Cationic collectors for nonsulfide ores

Fatty acid polyamine condensation products

1937

Depressant

Quebracho

Lowe

Cyanamid

1943

Barite flotation

Alkyl guanidines

Jayne

Cyanamid

1944

Nonsulfide flotation

Imidazolines

Maust, Hollingsworth

Cyanamid

1944

Nonsulfide flotation

Mercaptan

Simo, O’Connor

Shell

1945

Higher alkyl, including dodecyl

Cyanamid

1945

Nonsulfide flotation

Sulfosuccinates

Gieseke

Chelating agents

Gutzeit

Petroleum sulfonates

Booth

1946 Cyanamid

1946

Nonsulfide flotation

Dextrin, quebracho

Cyanamid

1949

Depressants for iron ore in cationic silica flotation

Lignin sulfonates

Cyanamid

1949

Gangue depressant and dispersant in sulfide flotation

Ethers of polypropylene glycols

Tveter

Dow

1952

Frothers

Allyl esters of xanthate

Booth

Cyanamid

1953

MoS2, Ag, Au, and sulfidized oxides

Guar as depressant for slimes

Atwood, Borne

Duval Sulfur & Potash

1953

Potash flotation

Zinc cyanide

Booth

Cyanamid

1953

Depressant for iron and zinc sulfides for ores containing Ag and Au

Polypropylene glycol

Booth, Dobson

Cyanamid

1954

Frothers

Thionocarbamate

Harris

Dow

1954

Cu sulfide collector (Table continued next page)

flotation0.book Page 378 Tuesday, January 2, 2007 7:36 PM

378

TABLE 1

FLOTATION CHEMISTRY

Chronology of reagent development* (continued)

Reagent Carboxymethylcellulose as depressant/dispersant

Principle Inventors and/or Authors

Company or Institution Year Kalivertriebsstelle 1955

Comments Potash flotation

Synthetic hydolyzed polyacrylonitriles and polyacrylamides

Booth

Cyanamid

1955– Depressants and dispersants 1957 in sulfide flotation

Guar as depressant for Mg silicates

Drake

Sherrit Gordon Mines

1961

Talc, pyroxene, etc.

Hydroxamate in flotation

Gorlovskii

1961

Wolframite group

Hydroxamate in flotation

Gorlovskii

1961

Pyrochlore

Phosphonic acid

Wottgen, Lippman

1963

Cassiterite flotation

Dialkyl dithiophosphinates

Wystrach

Cyanamid

1964– Selective collectors for Ag, Pb, 1967 Cu, Au, polymetallic complex sulfide ores

Hydroxamate in flotation

Fuerstenau

Colorado School of Mines

1965

Chrysocolla

Sulfosuccinamates

Arbiter, Day

Cyanamid

1968

Iron and tin ores

Sulfosuccinates and sulfosuccinamates

Wang

Cyanamid

1975

Boosters for fatty acid flotation

Synthetic fuctionalized polymers

Lim

Cyanamid

1980

Depressants for nonsulfide flotation

Allyl alkyl thioncarbamates

Dauplaise

Cyanamid

1982

Selective collectors for Zn, Cu, and precious metals

Alkyl and aryl monothiophosphates

Wang, Nagaraj

Cyanamid

1982

Acid circuit sulfide collectors

Synthetic functionalized polymers

Cyanamid Rothenberg, Nagaraj, Lipp, Wang, Spitzer, Heitner

1983– Depressants for sulfide 1987 flotation

Alkoxycarbonyl thionocarbamates and thioureas

Nagaraj, Wang, Fu

Cyanamid

1985

Selective collectors for mildly alkaline circuits

Alkyl monothiophosphinates

Nagaraj, Wang

Cyanamid

1985

Sulfide and precious metal collectors for neutral circuits

Hydroxamate in flotation

Yoon, Hilderbrand

Thiele Kaolin

1986

Anatase from kaolin

Alkyl and aryl monothiophosphates

Fleming

Cyanamid

1988

Selective Au, Ag collectors in alkaline circuits

Hydroxamate in flotation

Wang, Nagaraj

Cyanamid

1989

New process for manufacture

Hydroxamate in flotation

Nagaraj

Cyanamid

1992

Precious metals flotation

Synthetic functionalized polymers

Nagaraj, Wang

Cytec

1994– Depressants for Mg silicates 1995

New sulfide collectors

Rothenberg

Cytec

2003

Cu, Cu-Au, platinum-group metals (PGMs)

New sulfide collectors

Nagaraj

Cytec

2004

Au, PGMs, Ni

Source: Nagaraj 2003. *This table captures only reagents that are, or have been, used in the industry. It is by no means exhaustive; there are a vast number of reagents that were invented or proposed in research but remained a lab curiosity and are not included.

flotation0.book Page 379 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 2

379

Three important periods in flotation reagent development

Early Days to 1920 • Oils, oleic acid, soaps, CuSO4, SO2, lime, soda ash, cresylic acids, aliphatic alcohols, aldehydes, ketones, esters, pine oils 1921–1950: Discovery and Expansion (use of chemicals developed in other industries; e.g., rubber, textile, tanning) • Collectors: Chelating agents, dithiocarbamate, dithiophosphate, fatty acids/amines, guanidines, hydroxamate, imidazolines, mercaptans, mercaptobenzothiazole, petroleum sulfonates, primary amines, sulfosuccinates, thiocarbanilide, trithiocarbonates, xanthate, xanthogen formate

• Modifiers: Dextrin, dyes, lignin, polyphosphate, sodium cyanide, sodium silicate, sodium sulfide, starch, tannin, zinc sulfate 1951–present: Rational, Targeted Design (either designed or developed for targeted use in flotation applications • Collectors: Alkoxycarbonyl thionocarbamate, alkoxycarbonyl thiourea, allyl thionocarbamates, dialkyl thionocarbamates, dithiophosphinate, hydroxamate, monothiophosphate, monothiophosphinate, phosphonic acids, sulfosuccinamates, xanthate esters

• Frothers: Polypropylene glycols and ethers, triethoxybutane • Modifiers: Synthetic functionalized polymeric modifiers, modified polysaccharides, carboxymethylcellulose, guar, polyacrylates, zinc cyanide Source: Nagaraj 2003

and separating medium, decreasing oil requirements to less than 1% of the weight of the ore. With this reduction in oil consumption, inherent differences in the frothing and collecting power of various oils were noted (Booth and Freyberger 1962). Several important modifiers such as lime, sulfur dioxide (SO2), sodium carbonate, and copper sulfate (CuSO4) were used to enhance separations. The year 1921 marked the beginning of a major revolution in flotation reagents that set the stage for the next 80 years or so. A trend was established toward reagents of definite chemical composition by the use of certain types of small organic molecules and oils containing sulfur (either naturally or by direct sulfurization). Several major patents appeared (e.g., Perkins 1921; Vivian 1921) which taught that small, water-soluble molecules could be used as collectors instead of large amounts of oil. This trend continued by the later application of N- and S-containing compounds such as naphthylamines, diamino benzene, toluidines, thiocarbanilide, and several other organic complexing (or chelating) agents for sulfide mineral flotation. Their consumption was low (0.1–0.5 lb/t). This was followed by differential flotation using sodium cyanide (NaCN) and the discovery of xanthates, dithiophosphates (DTPs), dithiocarbamates, mercaptobenzothiazole, and amines (for nonsulfides). It is important to note that the majority of chemicals used for flotation in the period from 1921 to 1950 were developed in other industries (e.g., rubber, textile, tanning). During the third period, 1951–present, the trend began for developing reagents specifically targeted for flotation application rather than merely evaluating chemicals developed for other industries (see Table 2 for examples). Modern chemical flotation is built on this foundation and it has undergone significant refinement to bring much-needed, rational, scientific rigor from developments in organic, polymer, and coordination chemistry. The ideal goal is the development of mineral-targeted, high-performance reagents that not only provide improved value recoveries and selective separation, but also provide the platform to extend flotation technique to a much greater variety of mineral and nonmineral systems and to nonflotation separations.

flotation0.book Page 380 Tuesday, January 2, 2007 7:36 PM

380

FLOTATION CHEMISTRY

A closer study of the history of flotation reagents in the past 100 years also reveals remarkable contrasts—between science and art, progress and regress, major advances in fundamental understanding and mystique, valuable knowledge base and myths, cogent concepts and misconceptions—interspersed with periods of major developments. These will be indicated in the appropriate sections in this chapter. The period between 1921 and approximately 1960* can be considered the most productive in terms of advances in the art and science of flotation reagents in particular, and flotation technology in general. This is well documented in the 50th anniversary volume of Froth Flotation (Fuerstenau 1962). Though targeted design was not prevalent in this period, many developments occurred in the chemistry of collectors, modifiers, and frothers. Excellent separation schemes were designed, and flotation was successfully applied to almost all viable ores and minerals, both sulfides and nonsulfides; relatively few additional applications have been developed after 1960. During the next two decades, the focus in the industry was on equipment, automation, and on-stream analyses. A revival of reagent development occurred in 1980, and another period of major developments ensued. Reagent design was standardized and streamlined on a rational, scientific foundation. (Nagaraj 1988, 1997) This was facilitated by advances made in surface analysis techniques that yielded a clearer understanding of reagent–mineral interactions. Synthetic polymeric reagents that could be tailor-made for a specific application or mineral were developed. The aim of this chapter is to critically analyze the development and use of flotation reagents in presenting a state-of-the-art view, to identify pitfalls and gaps in art and science, and to critically analyze research focus and approach. A conscious choice has been made to provide an overview of flotation reagents from an industry perspective rather than to repeat information that already exists in the literature with regard to historical, theoretical, and fundamental aspects. The latter will be highlighted, when necessary, to support the theme of this chapter, which is to provide a critical analysis of the art and science of flotation reagents. The emphasis will be on plant practice, which would necessarily mean a top-down or total system (holistic) view; that is, need- or mineral-centric rather than reagent-centric. T E R M I N O L O G Y A N D C L A S S I F I C AT I O N — S U L F I D E S A N D NONSULFIDES

As is typical of any technology that has evolved over a long period of time, terminology in flotation is rather loose, somewhat arbitrary, and often highlights merely one function or property, among many, of the reagent in question. There is also considerable discrepancy between industry and academe in the terminology and classification used; this is perhaps a reflection of differences in point of view between practitioners and academicians. The purpose of this section is to describe the terminology (and classification) adopted in this article and, at the same time, identify a common ground that is meaningful to both research and practice. The term flotation agent is frequently used to describe flotation reagents. The latter is more meaningful and accurate and is, therefore, preferred (see Holman 1930).

* Though arbitrary, the year 1960 was chosen for convenience and its significance; 1960 is close to the 50th anniversary of flotation and to the publication date of the 50th anniversary volume of Froth Flotation, which has been one of the most comprehensive and enduring publications.

flotation0.book Page 381 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

381

In research, collectors, for example, are variously classified into ionic (cationic and anionic) and non-ionic surfactants: sulfydryl, thio, thiol, non-thio, non-thiol, soluble, insoluble, oily, hydrolysable, ionizable, and so forth (Leja 1982; Rao 2004). In the case of sulfide minerals, these classifications are inadequate because many diverse collector families are used, collector adsorption is predominantly through surface chemical reactions (chemisorption and surface precipitation), and electrostatic effects are not very relevant. In the industry, collectors are invariably referred to and selected by their chemistry (e.g., DTPs, thionocarbamate, and the associated chain lengths of hydrocarbon substituents), which immediately suggests a link between performance and a particular combination of functional group and substituents. In other words, collectors are not selected based on the previouslymentioned academic classifications. Thus, a classification by functional group and substituents is more appropriate and scientifically meaningful in both research and practice. Similarly in nonsulfide flotation, a classification based on functional group is very useful and revealing in terms of performance characteristics. Frothers are grouped into soluble and partially (or sparingly) soluble frothers (Booth and Freyberger 1962). The soluble frothers are polyglycols and their alkyl ethers. Partially soluble or sparingly soluble frothers comprise aliphatic and aromatic alcohols, terpineols, aliphatic aldehydes, ketones, and esters. In the industry, frothers are commonly referred to and selected by their chemistry—namely, alcohols and glycols; solubility is implicit when the chemistry is specified. In view of the fact that OH is the most important functional group (other functional groups constitute a small part of the total frother usage in the industry), classification based on solubility differences is also quite acceptable, because much of the frothing properties can be readily related to solubility differences. Although molecular weight (MW) and branching play a role, they can also be linked to solubility. In nonsulfide flotation, though a separate frother is not universally used, long-chain surfactants with a variety of functional groups and short-chain alcohols and glycols are sometimes used as froth modifiers. Even in these systems, a classification by functional group is meaningful. The terminology used for modifiers is far more contentious. There is no single, generally accepted classification of modifiers. They are variously referred to as pH modifiers (or regulators), depressants, dispersants, activators, deactivators, promoters, froth modifiers, viscosity modifiers, and slime-blinding agents. Often, depressants and dispersants* are used interchangeably. The term modifier is sometimes used in a very narrow sense to represent a specific function among many; for example, a pH modifier. All such classification or categorization fails to recognize that in the complex flotation system a single modifier invariably has multiple functions, which vary in number and intensity from system to system (see section on Critical Analysis of the Status of Flotation Reagents—Modifiers later in this chapter). However, it is difficult to establish a distinct classification that comprises the entire spectrum of applications. The best recommendation is to use the term modifier in a broad sense and to recognize that it constitutes the third apex of the flotation reagent triangle (Figure 1; Nagaraj 2005), distinct from collectors and frothers. Qualitative descriptors are used routinely in research and practice to indicate reagent performance; for example, use of the terms strength and selectivity. These are relative terms * The term dispersant has different connotations in sulfide and nonsulfide systems. It is used loosely in sulfide systems where its primary function appears to be one of controlling slimes in flotation and not to provide colloidal stability to suspensions. In nonsulfide systems, its multiple functions are widely exploited.

flotation0.book Page 382 Tuesday, January 2, 2007 7:36 PM

382

FLOTATION CHEMISTRY

Collector

Flotation Reagents

Modifier

Frother

Source: Nagaraj 2005.

FIGURE 1

Flotation reagent triangle

though sometimes used as if they were absolute. They are used interchangeably to refer to strength or selectivity of adsorption (thermodynamic or kinetic) on individual minerals or of flotation recovery of a value mineral at a given dosage or under a given set of plant conditions. Thus, descriptors such as “strong or weak collector (modifier, frother)” or “selective collector (modifier, frother)” are necessarily qualitative; implicit in the description is that some reference is used. Similarly, qualitative descriptors are routinely used for frothers, for example, watery froth, dry froth (see section on Frothers and Frothing). These terms will be used as such in this chapter, for they serve a useful purpose in plant practice in the absence of quantitative descriptors. Ultimately, what matters the most in terms of plant practice is whether the selected reagents of the flotation reagent triangle are able to provide selectivity and recovery of the target value mineral species in the most economical manner. R E A G E N T S I N S U L F I D E A N D N O N S U L F I D E F L O TAT I O N — BASIC CONCEPTS

Separation schemes for the large variety of nonsulfide minerals are well recognized to be distinctly different from those for the small group of base metal sulfide minerals and precious metals (Aplan 1994; Fuerstenau 1962; Nagaraj 1988; Nagaraj, Day, and Gorken 1999). Such distinction can be readily understood by the fundamental differences that exist in physical and chemical properties between sulfide and nonsulfide minerals. These differences arise, for the most part, from differences in the chemistry between S and O. • The base metal sulfide minerals are characterized by mostly covalent or metallic bonding, low solubility in water, weakly hydrated surfaces and poor hydrogen bonding, a high degree of natural floatability, strong affinity for S-containing ligands, and pulp chemistry dominated by electrochemical reactions. • Conversely, nonsulfide minerals are generally characterized by ionic bonding, higher solubility in water, strongly hydrated surfaces and strong hydrogen bonding, strong affinity to O-containing ligands, high degree of hydrophilicity, and pulp chemistry dominated by ion exchange and electrostatic interactions. In the case of sulfide flotation, a wide variety of collectors, many of which exhibit a high degree of selectivity for a given sulfide, is available for the relatively small number of sulfide minerals and precious metals floated. In contrast, only a few collectors, many of which are rather nonselective, are available for the many nonsulfide minerals floated. Aplan and Fuerstenau (1962) noted that “Selective flotation in nonsulfide systems is often difficult in that

flotation0.book Page 383 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

383

one class of compounds alone, the carboxylic acids, may float nearly the entire array of nonsulfide minerals under the appropriate conditions. The nonsulfide mineral flotation operator is confronted with a vast array of modifying agents.” In nonsulfide flotation, there are only small differences in surface properties between a value mineral and the gangue. Indeed, modifying agents assume a greater role than collectors in nonsulfide separations, and it is common to use several modifying agents in a given separation. This obviously complicates the system considerably, puts it in the realm of art form, and makes it very challenging for a scientific study. Except for a few systems in which physical adsorption (especially electrostatic) is predominant, (e.g., adsorption of cationic reagents or sulfonates on nonsulfide minerals), the vast majority of interactions between minerals and reagents (including inorganic, small organic, and polymeric reagents) can be characterized as complexation in a broad sense of the word complex (Cotton and Wilkinson 1980). In this context, a large number of flotation reagents can be conveniently viewed as complexing agents, which can be classified based on either the donor atoms or the functional group containing all the donor atoms that participate in bonding (e.g., O– or O–O type, S– or S–N type; see Nagaraj 1988). (Chelating agents are merely a special class of complexing agents). A substantial effort has been made in the past eight decades to understand reagent– mineral interactions and, more importantly, to design or develop reagents for improving flotation separation efficiency. A brief description of the underlying concepts is given in the next few sections. Donor–Acceptor Relationships and Hard and Soft Acids and Bases

Much of the information in this section is borrowed from works by Nagaraj (1988). Although many concepts and requirements for reagent design were known and even practiced (albeit sporadically) over several decades, it was not until the late 1970s and early 1980s that a rational, scientific basis using the donor–acceptor model was fully formalized and firmly established in mineral systems. Donors, or donor atoms, or ligand atoms are those that bond directly to the metal atoms on the mineral; for example, the S atoms in xanthate and the O atoms in fatty acids. Acceptors are the metal atoms in the mineral that accept electrons from donors; for example, Cu in chalcocite and Fe in hematite. Classification of acceptors based on their affinity toward the three important donor atoms, N, O, and S, is extremely useful in reagent design and in understanding reagent–mineral interactions. Thus, elements (excluding the donor atoms and inert gases) in the periodic table fall into three groups: (1) those that show strong affinity for O (and N), (2) those that show strong affinity for S, and (3) elements in the intermediate region that show affinity for both O and S (see Figure 2). The majority of elements in the periodic table (the left and right blocks) show very strong affinity for O; a small group of elements (predominantly Cu, Au, Ag, and the platinum-group metals or PGMs) show strong affinity for S. This behavior is readily recognized in the minerals of abundance, or of economic interest, found in the earth’s crust; for example, oxides are the most important or abundant for Ti, and sulfides for Cu (this is indicated in Figure 2 under each element). A more formal and general classification of acceptors and donors is that based on Pearson’s concept of hard and soft acids and bases (HSAB), which was subsequently refined by Klopman on the basis of molecular orbitals (Nagaraj 1988). Acceptors are acids, and donors are bases; Table 3 gives examples of HSAB. The donors O and N are hard bases, and S is a soft base. The task of understanding and predicting reagent–mineral interactions has been

flotation0.book Page 384 Tuesday, January 2, 2007 7:36 PM

384

FLOTATION CHEMISTRY

H H Ox Ox Li Li Ox,Cl Ox,Cl

Be Be Ox Ox

B B Ox Ox

C C

N N

O O

F F

Na Na Ox,Cl Ox,Cl

Mg Mg Ox Ox

Al Al Ox Ox

Si Si Ox Ox

P P Ox Ox

S S

Cl Cl

K K Ox,Cl Ox,Cl

Ca Ca Ox Ox

Sc Sc Ox Ox

Ti Ti Ox Ox

V V Ox Ox

Cr Cr Ox Ox

Mn Mn Ox Ox

Fe Fe Ox Ox

Co Co Sul Sul

Ni Ni SO SO

Cu Cu Sul Sul

Zn Zn SO SO

Ga Ga Sul Sul

Ge Ge Sul Sul

As As Sul Sul

Se Se

Br Br

Rh Rh Sul Sul

Pd Pd Sul Sul

Ag Ag Sul Sul

Cd Cd Sul Sul

In In

Sn Sn Ox Ox

Sb Sb Sul Sul

Te Te

II

Ir Ir Sul Sul

Pt Pt Sul Sul

Au Au Sul Sul

Hg Hg Sul Sul

TI TI Sul Sul

Pb Pb SO SO

Bi Bi Sul Sul

Po Po

At At

Rb Rb Ox Ox

Sr Sr Ox Ox

Y Y Ox Ox

Zr Zr Ox Ox

Nb Nb Ox Ox

Mo Mo Sul Sul

Tc Tc Ox Ox

RuS RuS Sul Sul

Cs Cs Ox Ox

Ba Ba Ox Ox

La La Ox Ox

Hf Hf Ox Ox

Ta Ta Ox Ox

W W Ox Ox

Re Re Sul Sul

Os Os Sul Sul

Fr Fr

Ra Ra

Ac Ac

Sul = Ox = SO = Cl =

Donors Affinity for S. Affinity for both S and O. Both sulfide and oxide minerals common. Affinity for O, Cl, N donors. Oxide minerals most common. Sulfide Oxide, silicate, carbonate, etc. Both sulfide and nonsulfide. Chlorides and other halides.

Source: Nagaraj 2003.

FIGURE 2

TABLE 3

Periodic table of ore minerals

Some hard and soft acids and bases

Bases (nucleophiles)

Acids (electrophiles)

Hard: H2O, OH–, F– CH2CO2, PO43–, SO42– Cl–, CO32–, ClO4–, NO3– ROH, RO–, R2O NH3, RNH2, N2H4, O, N

Hard: H+, Li+, Na+, K+ Be2+, Mg2+, Ca2+ Al3+, Ga3+ Cr3+, Co3+, Fe3+ Si4+, Ti4+ Ce3+, Sn4+ SO3, CO2, NC+ HX (hydrogen bonding molecules)

Borderline: C6H5NH2, C5H5N, N3–, Br–, NO2– SO32–, N2

Borderline: Fe2+, Co2+, Ni2+, Cu2+, Zn2+, Pb2+ Sn2+, SO2, C6H5+

Soft: R2S, RSH, RS– I–, SCN, S2O32– R3P, R3As, (RO)3P CN–, RNC, CO C2H4, C6H6 S, P

Soft: Cu+, Ag+, Au+, Tl+, Hg+ Pd2+, Cd2+, Pt2+, Hg2+ Ti3+ RS+ HO+, RO+ I2, Br2, ICN, etc.

Source: Nagaraj 1988.

greatly simplified by the HSAB concepts and the knowledge of the fundamental properties of the donor atoms N, O, and S (see Table 4). Thus, metals that prefer to react with O donors also tend to preferentially adsorb O-containing reagents on their mineral surfaces. Similarly, S donors react preferentially with sulfide minerals. Donor atoms in minerals also play an important role in reagent–mineral interactions, but this is often ignored. It has been observed that metal ions interacting with soft bases become soft acids and vice versa. This is especially important for borderline cations and explains the fact that sulfidization of certain

flotation0.book Page 385 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 4

385

The properties of the major donor atoms O, N, and S

Configuration Electronegativity Valence electrons Normal valency Number of orbitals Number of bonds (valency expansion) Lone pairs pπ-pπ dπ-pπ (back-bonding) Polarizability Hydrogen bonds Bonds Steric accessibility

O

N

S

[He] 1s22s22p2 3.5 2 2 4 3 2 Strong None Nil Strong More ionic Low

[He] 1s22s22p3 3.07 5 3 4 4 1 Strong None Good Strong Less ionic Low

[Ne] 3s23p43d0 2.44 2 2 4+d 2–6 1 Poor Strong Strong Very weak Covalent High

Adapted from Nagaraj 1988.

oxide minerals increases their interaction with sulfide collectors (such as xanthate). Thus, the important contribution of donors on the mineral surfaces must be noted. Oxide minerals (which have O donors) of soft (or borderline) acid metal ions would preferentially adsorb O- and N-containing collectors (e.g., cuprite and alkyl hydroxamates or oximes), whereas their corresponding sulfides would preferentially adsorb S-containing collectors (e.g., chalcocite and DTP). The fundamental properties of the three donor atoms (see Table 4) dictate much of the relevant chemistry of flotation. A judicial combination of donors for any given acceptor, taking into account the properties of both acceptors and donors (on the mineral as well as the reagent), provides the required selectivity in flotation. In organic and polymeric reagents, the complexation (or bonding) properties of the donor atoms are readily modified by the substituents attached to them.* Substituents can directly influence the electron densities on the donor atoms and the pKa of the molecule through inductive and resonance effects (Nagaraj 1988). The electron-withdrawing substituents decrease the pKa (make the molecule a stronger acid) and, therefore, significantly alter the stability constant of the metal complex and pH range of complexation. The direction of the effect is dependent on the donor atom(s) and acceptor(s) in question. This concept has been exploited in the development of new flotation reagents; for example, sulfide collectors that function at lower pH values (Nagaraj 1997; Nagaraj, Basilio, and Yoon 1989). There are innumerable types of complexing groups, and almost all of them interact with most of the minerals; thus, the concept of absolute specificity does not exist. In fact, the differences in stability constants are only of degree rather than of kind, so that reagents are rarely specific and, often, are insufficiently selective for most practical purposes. In practice, therefore, differences in selectivity are utilized. The solubility of the mineral, in addition to that of the metal complex, has a pronounced influence on the selectivity and activity of the reagent. In the case of polymeric reagents, additional influence comes from MW, degree of substitution, presence and number of other functional groups, and molecular architecture * In inorganic compounds, the properties of the donor atoms are influenced by the counter ions (or the Lewis acids—soft, hard, or borderline); for example, the properties of Fe silicate would be quite different from those of Na silicate.

flotation0.book Page 386 Tuesday, January 2, 2007 7:36 PM

386

FLOTATION CHEMISTRY

(e.g., whether the functional group is directly on the backbone or is pendant, the backbone is rigid or flexible, how the functional groups are distributed in the polymer) and, importantly, conformation of the polymer at the solid/liquid (S/L) interface after adsorption. Much research attention has been given to chelating agents (for both collectors and modifiers) because of their ability to enhance the specificity or selectivity required in many mineral separations. Such selectivity has been demonstrated in several mineral systems (both sulfides and nonsulfides), though predictability based on extension from metal systems to mineral systems is rather tenuous and, sometimes, inaccurate because of the complexity in mineralogy and the fact that, ultimately, flotation is a (chemical-aided) physical separation of one mineral from another. Although much aura has been associated with chelating agents, it is pertinent to realize that chelating agents are merely a special case of complexing agents, in general, and that the merits of a reagent are ultimately determined by its performance in a given mineral system rather than whether it forms a single bond or a chelate with the metal atom of the mineral. Nevertheless, the majority of sulfide collectors, a few nonsulfide collectors, and several polymeric compounds have chelating-forming functional groups (Nagaraj 1988). Adsorption Processes

The driving force for, and the mechanism of, adsorption of flotation reagents on minerals comprise chemical (chemisorption, surface reaction, or complexation), electrostatic (physisorption or physical adsorption), and nonspecific forces such as van der Waals forces, hydrogen bonding, and the so-called hydrophobic force. Chemical interactions have the highest adsorption energies followed by electrostatic and nonspecific interactions. In many cases, more than one driving force is in operation. Overall adsorption energy is, therefore, a sum of all energies associated with various adsorption processes. Adsorption processes involving mainly chemical driving force provide the widest scope for investigating and modifying the chemical structure of a reagent in order to dramatically change its activity. The most important changes pertain to donor atoms of the reagent. When the adsorption process is predominantly electrostatic in nature, a change in the charge density of the mineral surface, or of the functional group, causes a noticeable change in adsorption energy or interaction energy. Pulp chemistry plays a significant role in these systems; for example, the presence or addition of inorganic ions. In general, for cationic reagents, adsorption is predominantly electrostatic. Similarly, in the case of sulfonates and sulfates, the electrostatic component is usually the predominant one (there can, however, be a chemical component also, e.g., sulfonates on iron oxides). In the case of other anionic reagents containing carboxyl or phosphonate groups, there is often a significant chemical component in the overall adsorption energy in addition to the electrostatic component. Under certain conditions, the chemical component can completely override the electrostatic component (Aplan and Fuerstenau 1962). In the case of nonspecific adsorption processes, structural aspects of the reagent molecule that can be changed include the nature and type of the hydrocarbon chain and moieties capable of hydrogen bonding. In general, such changes in the molecule cause only small changes in adsorption energy or strength of adsorption at the S/L interface. These systems are difficult to model, and structure–activity relationships tend to be empirical. Interactions between sulfide minerals and flotation reagents are dominated by chemisorption, surface chemical reaction, and surface precipitation, whereas those in nonsulfide flotation are characterized by both physical (predominant) and chemical adsorption (fatty

flotation0.book Page 387 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

387

acids, hydroxamate, and phosphonate). The majority of the sulfide collectors adsorb on sulfide minerals by forming a metal complex on the surface. Direct unequivocal evidence for surface complexes has been obtained recently from secondary ion mass spectrometry (SIMS; Nagaraj and Brinen 1995, 2001). For example, in the case of adsorption of an alcoxycarbonyl thionocarbamate on chalcopyrite, the surface species was a cuprous complex of the collector, and there was no evidence of a similar Fe complex, though both Cu and Fe atoms are in equal amounts on the surface and readily available for complexation. Table 5 gives a summary of results obtained from SIMS for the identity of adsorbed species (Nagaraj and Brinen 1997). In the case of pyrite and xanthate, there was evidence for the presence of dixanthogen (Table 5), which obviously survived the high vacuum of the instrument contrary to general belief that dixanthogen would be “pumped off ” (Kartio et al. 1992). In addition to dixanthogen, however, there was evidence for the presence of mixed ligand complexes involving Fe, xanthate, and OH; similarly for DTP. In the case of dithiocarbamate, several complexes were observed. SIMS analysis also produced clear evidence that adsorption of collectors and modifiers on natural sulfide mineral surfaces was almost always “patchy,” which is in contrast to the popular model of uniform adsorption approaching a monolayer coverage used in research. The patchy adsorption can be generalized to systems involving chemisorption and chemical reaction, and is akin to nucleation and growth of precipitates. It is quite likely that patchy adsorption occurs in systems involving physisorption, for almost all natural minerals carry impurities and imperfections, which lead to inhomogeneity in surface composition (and active sites). In systems involving chemisorption and chemical reaction, adsorption patches are also characterized by multilayers of metal–reagent complexes. The first layer in these systems is considered chemisorption (Woods 1996; Nagaraj 1988). Subsequent layers are attributed to surface reaction or surface precipitation of the reagent complex with metal ions released from the mineral itself. A substantial amount of the reagent can be precipitated in the bulk aqueous phase depending on mineral solubility and the rate of release of metal ions from the mineral (Ananthapadmanabhan and Somasundaran 1985; Nagaraj 1988). The bulk complex represents wasted reagent. Evidence was also found from SIMS analysis in some systems for mixed ligand metal complexes that were quite different from the traditionally known compositions and in which the flotation reagent was merely one of the ligands (Nagaraj and Brinen 1997, 2001), one example being that of the Fe complex with xanthate and OH. Indeed, with the preponderance of certain ligands in flotation pulps, there is no reason to expect that only the wellknown, iso–ligand complexes should form on the mineral surface. Generally, focus is only on a few such well-known complexes because either thermodynamic data are available for these or they fit the conceptual models of complexation from solutions. This could lead to erroneous conclusions and recommendations for plant practice. There is substantial evidence in nonsulfide mineral systems for reverse adsorption of the long-chain surfactants, such as fatty acids and amines (in the second layer), which makes the mineral hydrophilic and decreases flotation. In contrast, for the majority of sulfide collectors, which are typically short-chained and chemisorbing, there is no tangible evidence for reverse adsorption and reduced flotation. Sometimes flotation of a sulfide mineral as a function of concentration of collector does exhibit a maximum, but this is readily attributed to the formation and reduced buoyancy of large aggregates of the hydrophobic particles. Although chemisorption and chemical reactions characterize adsorption of reagents in sulfide flotation, physisorption is also proposed for many reagents. For example, neutral molecules, such as xanthic acid and dixanthogen, are proposed to adsorb via physisorption

flotation0.book Page 388 Tuesday, January 2, 2007 7:36 PM

388

TABLE 5

FLOTATION CHEMISTRY

Summary of SIMS analysis of adsorbed collector species on minerals Surface Species

Mineral

Collector

Diagnostic Collector Fragment

Surface Complex

Imaging

NBECTU (n-butyl ethoxycarbonyl thiourea) M = 204

Molecular ion at mass 203/205 (+ ion) M ± H

Cu–NBECTU complex at mass 267/269 (+ ion) [no complex with Fe]

Cu complex and collector molecular ion

IBECTC (isobutyl ethoxycarbonyl thionocarbamate) M = 205

Weak molecular ion signal at mass 205/206 (+ ion); mass 26 (CN) diagnostic

Weak signal at 267/269 for Cu complex? [no complex with Fe]

CN fragment

DIBDTPI (diisobutyl dithiophosphinate) M = 209

Molecular ion at mass 209 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

NBECTU M = 204

Molecular ion at mass 203/205 (+ ion) M± H

Cu–NBECTU complex at mass 267/269 (+ ion)

Cu complex and collector molecular ion

DIBDTPI M = 209

Molecular ion at mass 209 (– ion)

Weak signal at mass 481/483 for Cu–DIBDTPI 1:2 complex (– ion)

Collector molecular ion

DIBDTP (diisobutyl dithiophosphate) M = 241

Molecular ion at mass 241 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

IBX (isobutyl xanthate) M = 149

Molecular ion at mass 149 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

AX (amyl xanthate) M = 163

Molecular ion at mass 163 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

NBECTU M = 204

Molecular ion at mass 203/205 (+ ion) M ± H

1:1 Cu–NBECTU complex at mass 267/269 and 1:2 complex at mass 469/471 (+ ion)

Cu complex and collector molecular ion

AX M = 163

Molecular ion at mass 163 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

Galena

DIBDTPI M = 209

Molecular ion at mass 209 (– ion)

Pb–DTPI (dithiophosphinate) complex at mass 416, 417, 418, 419 (+ ion)

Pb complex and collector molecular ion

Pyrite

NBECTU M = 204

Very weak hydrocarbon fragments

Negligible adsorption

None

IBECTC M = 205

Very weak hydrocarbon fragments

Negligible adsorption

None

DIBDTP M = 241

Molecular ion at mass 241 (– ion)

Collector molecular ion (at 241) and fragment for Fe(DIBDTP)(OH) complex (at 314)

Collector molecular ion

IBX M = 149

Molecular ion at mass 149 (– ion)

Only the collector molecular ion (possible fragmentation of complex)

Collector molecular ion

IBX M = 172

Molecular ion at mass 149 (– ion)

Collector molecular ion at 149 (– ion)

Collector molecular ion

AX M = 163

Molecular ion at mass 163 (– ion)

All observed Collector molecular ion (at 163), fragment of dixanthogen surface (at 195), and fragment for Fe– species xanthate complex (at 275)

DBDTC (dibutyl dithiocarbamate) M = 204

Molecular ion at mass 204 (+ ion)

Fe:DBDTC 1:2 and 1:3 complexes at 464 and 668, collector dimer at 408, and collector molecular ion at 204

Chalcopyrite

Chalcocite

Quartz (Cu-activated)

Adapted from Nagaraj and Brinen 1996a, 1997, 2001.

Fe complexes and collector molecular ion

flotation0.book Page 389 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

389

Chain Length

Hydrophobicity Type of R* Groups Alkyl, Aryl... and Hetero Atoms

Branching and Mixed R* Groups

pKa

Reagent Structure

Number of Donor Atoms

Stability

Reagent Properties

Position of Donor Atoms

Type of Donor Atoms N, O, S

Physical Properties

Frothing

Selectivity Adsorption

*R = hydrocarbon substituents. Adapted from Nagaraj 1988, 2003.

FIGURE 3

Structural features of a molecule and the properties affected by changes

(Leja 1982). However, even these molecules can chemisorb or react with metal species of certain mineral by forming the metal–collector complex. This is possible because of the finite solubility, albeit small, in a wide pH range and because the electron density on the donor atoms can significantly change in the presence of a favorable acceptor such as Cu, Au, Ag, and so forth. (NOTE : Dixanthogen does not have a dissociable proton, therefore its donors are soft bases). Many of the sulfide collectors are neutral molecules and adsorb on minerals by forming a complex with the metal atoms on the surface (SIMS evidence; Nagaraj and Brinen 1996a, 1997). Structure–Activity Relationships

Structure–activity aspects become very important and offer a wide scope for reagent design and control in systems where the driving force for adsorption of flotation reagents on minerals is chemical. Because chemical interactions have the highest adsorption energies, changes in structure of the reagent molecule can potentially result in large changes in the strength of adsorption, the resultant interfacial properties, and flotation response. This has been clearly demonstrated for numerous reagent families in flotation research and practice. Structural changes fall into two categories: changes to the functional group (donor atoms or ligands) and changes to substituents attached to the functional group (see Figure 3). Of the two, the former causes far more dramatic changes than the latter. However, judicious changes in the substituents (which are easier to make than changing the functional group), especially when substituents carry additional donor atoms, are frequently adequate to obtain desired collector strength and selectivity. Substituents modify the affinity of the reagent for a given sulfide surface, the hydrophobicity conferred, kinetics of adsorption, and the pKa of the reagent molecule (when it is electronically active). Substituents can also participate in bond formation with the mineral, which may either reinforce or counter the interactions of the basic functional group with the sulfide surface (Nagaraj 1988, 1997).

flotation0.book Page 390 Tuesday, January 2, 2007 7:36 PM

390

FLOTATION CHEMISTRY

Pulp Potentials and Zeta Potentials

The behavior of sulfide minerals is analogous to that of metals (Woods 1976; Woods and Richardson 1986). Sulfide minerals and the ionic sulfide collectors readily undergo redox reactions, which can be characterized by measuring potentials directly across the mineral– solution interface. Zeta potentials are seldom used in characterizing reagent–mineral interactions in sulfide systems. Aplan and Chander (1988) noted the following: Sulfide minerals in solution are found to be charged, but understanding the data is a more difficult problem. As demonstrated by Healy and Moignard (1976), the literature gives widely differing values for the isoelectric point (IEP) of sulfides. … Fuerstenau (1980) shows, for example, that the IEP of galena changes from pH 2 in the absence of oxygen to pH 8 in its presence, whereas the IEP of pyrite changes from pH 4 in the absence of oxygen to a complicated response showing several IEP values in the presence of oxygen. … In terms of practical flotation with a strongly chemisorbed sulfhydryl collector, the exact IEP of a sulfide mineral is often academic… In sulfide systems, many of the reagent–mineral interactions occur through an electrochemical process, though equally important are the purely chemical interactions, which are often neglected in research. This has been the source of many misconceptions surrounding the role of pulp potentials in sulfide flotation (Nagaraj 2000b). The nonsulfide minerals are poor conductors; consequently, the surface potential cannot be measured directly and, instead, zeta potentials or electrophoretic mobilities are used to characterize the mineral–solution interface. Zeta potentials are very useful in systems involving electrostatic adsorption. However, their utility is limited in systems involving chemical interactions (e.g., hematite and alkyl hydroxamates, or titanium dioxide and fatty acids). Hydrophobicity

Contact angles and induction time measurements have been used for characterizing collector or modifier adsorption and hydrophobicity, for screening collectors and depressants, and for prediction of flotation. Equilibrium contact angles on well-defined, polished mineral surfaces do not provide any information about the dynamics of bubble–particle attachment, stability, and movement in the pulp and froth phases. Contact angles in single mineral systems have been correlated with mineral flotation; however, they are a poor predictor of flotation outcome in real systems with complex mineralogy. Questions such as “What contact angle (or hydrophobicity) is required for flotation in a real ore system?” will not elicit meaningful answers.* Kinetic approach is fundamental in understanding the flotation process taking place under long-term nonequilibrium conditions. For flotation, the strength of adhesion is not the important factor, but rather its ratio to strength of detachment under dynamic conditions. Induction time is a useful measure of adhesion, yet such measurements on a polished mineral surface are not very useful in ore flotation. Molecular Modeling

Molecular modeling has been attempted in systems involving chemical interactions (Pradip 1991; Pradip and Rai 2002; Natarajan and Nirdosh 2003; Leal Filho et al. 2000). This would certainly be useful in understanding such interactions in model systems; however, the

* The correct answer would be “We asked the wrong question.”

flotation0.book Page 391 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

391

complexity of mineralogy limits the utility of molecular modeling in translating sound chemistry into plant practice. It is believed, however, that molecular modeling could be a powerful screening tool in the reagent design phase. R E A G E N T S I N S U L F I D E A N D N O N S U L F I D E F L O TAT I O N — P L A N T P R AC T I C E Reagents for Base Metal Sulfides and Precious Metals

Sulfide flotation can be divided into the following four metal-value-based groups: 1. Cu, Cu-Mo, and Cu-Au ores 2. Complex, polymetallic ores such as Pb-Cu-Zn-Ag 3. Primary Au and PGM ores 4. Ni ores In groups 1 and 2, separation of sulfides from each other is necessary, and selective* reagents are invariably used. In groups 3 and 4, the separation is between value sulfides (metals) from nonsulfide gangue, and strong collectors are used to float all value sulfides and metals. In spite of this division, all four groups have much in common, which is not surprising because the important value minerals and metals in these ores are grouped together in the periodic table. Collectors. The collectors used currently, along with recently developed new collectors, are shown in Table 6. The important donor in almost all of the collectors is still the S, and the basic “active” bonding group in a majority of structures is a C=S or P=S or C–S–. The two other important donor atoms are N and O. Numerous combinations of the three donor atoms based on the donor–acceptor model have been used in developing new collector chemistries (Avotins, Wang, and Nagaraj 1994; Nagaraj 1997; Nagaraj and Avotins 1988; Nagaraj et al. 1988; Nagaraj, Basilio, and Yoon 1989), but the basic bonding group has not changed; some collectors have additional bonding groups. All of the research in the development of new collectors has confirmed these structural features. Compounds containing O and N as primary donors can be (and in many cases are) collectors for sulfides, but they are deficient in terms of selectivity and collector strength. In terms of total tonnage of flotation reagents, xanthates still dominate the industry, though the use of other collectors is steadily increasing. Much of the xanthate use is in flotation systems where selectivity between sulfide minerals is not an issue and the goal is to float most of the sulfides and precious metal values irrespective of their occurrence; for example, large quantities of xanthate are used in flotation of primary Au, PGM, and certain Ni ores. Even in these systems, additional, often more selective, collectors are invariably used along with xanthate because optimum recoveries are seldom achieved with xanthate alone. In systems where selectivity between sulfides is the primary driver—for example, in Cu, Cu-Mo and Cu-Au, and complex sulfides—other collectors such as dithiophosphinate, thionocarbamates, and DTPs are the primary collectors, and xanthate is used only in small amounts as an auxiliary collector.

* The word selective here refers to selectivity for one sulfide mineral over another. Selectivity for sulfides versus nonsulfides is generally excellent for sulfide collectors and is implicit.

flotation0.book Page 392 Tuesday, January 2, 2007 7:36 PM

392

TABLE 6

FLOTATION CHEMISTRY

Collectors used in sulfide and nonsulfide mineral flotation Sulfide Collector Families

1 Dithiocarbamate, dialkyl

Emergence in PGM ores; limited use elsewhere

S

R NH

–

C

S Na

+

R′

2 Dithiophosphate, dialkyl R

Very widely used

S

O

–

P R

S Na

+

O

3 Dithiophosphate, diaryl

Mostly for complex sulfide ores

S R

O

–

P

S Na

+

2

4 Dithiophosphinate, dialkyl

Widely used, especially for complex sulfides

S

R

–

P

S Na

+

R

5 Mercaptan, alkyl

R

SH

S

N C

6 Mercaptobenzothiazole

7 Monothiophosphate, dialkyl

R

O

–

S Na

Mostly for complex sulfides, Au ores, tarnished sulfides, and as auxiliary collectors

+

Mostly for Cu-Au and Au ores

S O–Na+

P R

Very limited use

O

8 Monothiophosphate, diaryl

Mostly for Cu-Au and Au ores

S R

O

–

P

O Na

+

2

9 Sulfide, dialkyl

R

10 Thionocarbamate, alcoxycarbonyl alkyl

S

S R

O

Widely used for Cu, Cu-Mo, Cu-Au ores

O

C

11 Thionocarbamate, allyl alkyl

Very limited use

R′

NH

C

O

R′

Widely used for Cu, Cu-Mo, Cu-Au ores

S R

O

H N

C

12 Thionocarbamate, dialkyl

CH2

CH

CH2

Widely used for Cu, Cu-Mo, Cu-Au ores

S R

O

13 Thiourea, alcoxycarbonyl alkyl

C

NH R′

S R

NH

C

14 Xanthate ester, allyl alkyl

Limited use for Cu, Cu-Mo, Cu-Au ores

O NH

C

O

R′

Widely used for Cu, Cu-Mo, Cu-Au and Au ores

S R

O

C

S

15 Xanthate, alkyl

CH2

CH

CH2

Widely used for most sulfide ores

S R

16 Xanthogen formate, dialkyl

O

C

S R

O

C

17 Diphenyl thiourea and diphenyl guanidine N

S–Na+

Mostly for Cu, Cu-Mo

O S

C

S

H

C

N

O

R′

Very limited use (complex sulfides) Very limited use (Cu-Ni matte)

H NH N H

C

N H

(Table continued next page)

flotation0.book Page 393 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 6

393

Collectors used in sulfide and nonsulfide mineral flotation (continued) Nonsulfide Collector Families

1 Fatty acids

Widely used for many nonsulfide ores

O C R

OH

2 Primary amines

Widely used for many nonsulfide ores

H R

N

+

H

H

3 Petroleum sulfonates

Widely used for many nonsulfide ores

O R

–

S O

4 Alkyl sulfates

O

Very limited use

O R

O

5 Alkyl hydroxamic acids

–

S O

O

Mostly for specialty applications

O C

OH

R

N H

6 Alkyl phosphonates

Mostly for specialty applications

O R

7 Dialkyl phosphates

–

P O

R

O

R

O

O

Very limited use

O P

8 Amphoteric collectors

–

O

Very limited use

O C

O–

CH2

R

N

C

CH3

O

9 Dialkyl sulfosuccinates

Mostly used for specialty applications

O R

C O O

R

CH2 CH SO3–

C O

10 Dialkyl sulfosuccinamates

Mostly used for specialty applications

O C – R1 O

CH2

N R2

CH C O

Adapted from Cytec Industries Inc. 2002 and Nagaraj 2003.

SO3–

flotation0.book Page 394 Tuesday, January 2, 2007 7:36 PM

394

FLOTATION CHEMISTRY

Many new collector families have been introduced in the past 25 years. These include the dialkyl dithiophosphinates, alkoxycarbonyl thionocarbamates and thioureas, dialkyl and diaryl monothiophosphates, allyl thionocarbamates, and dialkyl sulfide. The performance characteristics of all of these collectors, including the new ones, have been described in many excellent review articles and special publications (e.g., Mulukutla 1994). Frothers. Although thousands of compounds have been designed or proposed as improved frothers, in practice only the two most important families of alcohols are used, namely, the short-chain aliphatic alcohols and polyglycols (unsubstituted or alkyl monoethers) (Booth and Freyberger 1962; Klimpel and Hansen 1988). Table 7 lists the important frothers. Aldehydes, ketones, ethers, and esters are also known to be bona fide frothers, but they are usually present in relatively small amounts in industrial frothers, especially the sidestream alcohol frothers. Pine oil and cresylic acids dominated frother usage in the first 40 years or so of flotation. Their usage today is very small. Among the alcohol frothers, methyl TABLE 7

Important frothers used in the industry

Aliphatic alcohols

R

OH

C5

R

Widely used

C8

H3C CH

CH2

CH

H 3C

CH3

OH

e.g. MIBC

Poly(ethylene) or poly(propylene) glycols and their monoethers

R

O

R

O

C 3H 6

Widely used

OH

R′

R′ O

H

O

O R′

R′ = H or CH3 R = H or CH3 or higher alkyls

Alkoxy compounds

R

O

Widely used

R X

e.g., Triethoxybutane

Ketones

O R

C

Aldehydes

R′

O R

Esters

C

H

O R

C

O

Pine oil (α-terpineol)

R′

Minor amounts in aliphatic alcohol frothers Minor amounts in aliphatic alcohol frothers Minor amounts in aliphatic alcohol frothers Limited use

OH

Cresols (isomers)

CH3

OH

Source: Nagaraj 2003.

Limited use

flotation0.book Page 395 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

395

isobutyl carbinol (MIBC) is used most often, though recent concerns about its health hazards and strong odor have threatened its dominant position. It is likely that MIBC usage will gradually decline in the near future. Alkoxy-substituted paraffins, such as triethoxy butane, are also used as frothers, but to a limited extent. Hydrocarbons are often used with alcohols and cresylic acids to modify and stabilize froth structure. Frothers and the froth phase play a critical role in flotation, and a judicious choice of frother may provide a great benefit, but it is intriguing that they have not received due attention. Many attempts were made in the past to introduce well-designed specialty frothers (Klimpel and Hansen 1988; Harris and Jia 2000), but they were unsuccessful, which could perhaps be related to the commodity attitude and lower status that are given to frothers. However, the recent efforts in froth phase research in many institutions (discussed later in this chapter) may change this situation in the industry. Booth and Freyberger’s article from 1962 is still one of the best in the literature of relevance to practice; other useful articles are listed in works by Rao (2004). Modifiers. Unlike collectors, which are predominantly designed and targeted synthetic reagents, modifiers are dominated by readily available, inexpensive inorganic compounds and natural products (Nagaraj 1994; Chander 1988; Eigeles 1977). Modifiers fall into three major categories from the point of view of pure chemistry (Table 8): inorganic, small organic, and polymeric molecules. Most of the modifiers used are inorganic and were developed long ago. Small organic molecules have found limited commercial use; cost is a major factor. Until the early 1980s, the use of polymeric depressants was limited to polysaccharides and other natural products and by-products. Several synthetic, water-soluble, low-MW polymers are now in commercial use as depressants for gangue sulfides, phosphate, carbonates, and silicate minerals (Figure 4) (Nagaraj et al. 1987; Nagaraj 2000a). TABLE 8

Modifiers used in sulfide and nonsulfide mineral flotation

Inorganic Ammonium sulfide (sporadic) Ca, Mg, Al, and Fe salts Copper sulfate Hydrofluoric acid Lead nitrate Lime Nokes reagent Phosphoric acid Potassium dichromate (sporadic) Soda ash Sodium and zinc cyanide Sodium ferrocyanide (sporadic) Sodium flouride Sodium hypochlorite Sodium meta and polyphosphates Sodium metabisulfite Sodium silicate and metasilicate Sodium sulfide and hydrosulfide Sodium sulfite and sulfur dioxide Sulfuric acid Zinc sulfate

Small Organics and Oligomers Mercaptoethanol (sporadic) Organic dyes (sporadic) Polyamines (diethylenetriamine [DETA], triethylenetetramine [TETA], etc.) Sodium thioglycolate and its trithiocarbonate derivative (sporadic) Surfactants (mostly nonionic and anionic; sporadic in sulfides, more regular in nonsulfides) Tannins or quebracho (sporadic in sulfide; more regular in nonsulfide)

Adapted from Nagaraj 1994 and Cytec Industries Inc. 2002.

Organic Polymeric Carboxymethylcellulose Dextrin Guar gum, modified guars Lignin sulfonates Polyacrylates Starch Synthetic functionalized polymers

flotation0.book Page 396 Tuesday, January 2, 2007 7:36 PM

396

FLOTATION CHEMISTRY

R1

R1 (

CH2

C ) ( x C

CH2

O

C

NH2

R1

C ) ( y

CH2

O

C

NH HC

R1 CH2

C

CH2

O

Schematic Representation of Phosphate Depressant R1

C ) ( y C

CH2

O

C ) C

O

NH2

O

R1 = H or CH3 R2 = H or COOM M = Na

O

R1

C ) ( x

z

OM

R2

(

C )

z

R1 = H or CH3 M = Na

O

OM

CH2 (R

R1

R1 (

CH2

C )x( C

OH)n

CH2

O

R1

C )y( C

NH2

Schematic Representation of Silicates Depressant

CH2

O

C )z C

O

O

R1 = H or CH3 X = SH or OH M = Na

OM

CH2

R1 (

CH2

C

OH

CH2

X

Schematic Representation of Iron Sulfides Depressant

R1

C )x(

NH2

CH

O

CH2

R1

C )y(

CH2

C )z

CH2

C

NH

OM

C

R1 = H or CH3 M = Na

O

S

NH2

Schematic Representation of Sulfides Depressant (e.g., pyrite, Cu Sulfides in Cu-Mo Separation)

Source: Nagaraj 2000a.

FIGURE 4

Synthetic, water-soluble, polymeric modifiers

The use of dispersants has been sporadic in sulfide flotation. Their main usage is in addressing slime coating of sulfides and precious metals, and in dealing with high viscosity in pulps caused by certain gangue silicate minerals, such as serpentines, chlorite, and sericite. The important modifiers used in these applications are low-MW polyacrylamide-based polymers, polyacrylates, soda ash, lignin sulfonate, carboxymethylcellulose (CMC), polyphosphates, and low-MW polyethylene oxides. It is well known that dispersants can have a significant influence on the behavior and properties of the froth phase. This aspect has not received much attention in the past. Inorganic products and many of the natural products do not provide much room for structural modifications and performance improvement. Synthetic modifiers, on the other hand, are certainly attractive, because they can be tailored to specific separations (Nagaraj 2000a). Such modifiers are gaining acceptance in the industry, albeit slowly. Increasingly stringent environmental regulations may expedite their usage and may also accelerate research in this area.

flotation0.book Page 397 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

397

Reagents for Nonsulfide Minerals Collectors. Compared to sulfide flotation, the field of nonsulfide flotation is almost bewildering in terms of diversity of separations and technology employed. In stark contrast, two types of collectors, namely, fatty acids and amines, cover approximately 90% of the separations. The collectors currently used, along with the recently developed new collectors, are shown in Table 6. The important donor in almost all of the collectors is still the O, and the basic “active” bonding group in a majority of structures is a C–O, N–O, S–O, P–O, all of which behave either as hard (or borderline) bases or function via electrostatic interactions. In the commercially used collectors, N is encountered predominantly in amines, which adsorb by way of electrostatic interactions. Fatty acids dominate nonsulfide flotation. The usage of fatty acids is much higher than that of xanthates, in terms of both total tonnage and total dollar value. The usage of other collectors in nonsulfide flotation follows the order: fatty acids > amines > petroleum sulfonates > other surfactants. Fatty acids can be considered to be almost universal collectors, because they can be made to float almost all minerals. They are inherently nonselective, and the use of appropriate modifiers alleviates this problem somewhat and allows even difficult mineral separations. Fatty acids can also be used to float sulfides, but they offer no advantages over the S-containing compounds that are much more selective for sulfides than the fatty acids. Fatty acids, however, have been used for the recovery of oxide copper minerals, though this is not widespread because of the lack of selectivity and the presence of several multivalent ions in the plant water (especially alkaline-earth). Large amounts of hydrocarbon oils such as fuel oil, diesel oil, and kerosene are used in nonsulfide flotation as auxiliary collectors to enhance the flotation with fatty acids, amines, and petroleum sulfonates. Their tonnage is much more than that of even fatty acids. Primary fatty amines and ether amines constitute the bulk of cationic collector usage in the industry, essentially for silica, silicates, and potash. They are shown to be excellent collectors for sulfides, but they are nonselective and offer no major advantages over S compounds. Condensation products between polyamines and fatty acids have been used commercially in many applications (Zachwieja 1994). Sulfoxy collectors (mostly sulfonates) constitute the third most important class of collectors (after fatty acids and amines) for industrial minerals. Mixtures of sulfates and sulfonates have been used in some separations. The current usage of alkyl sulfates as collectors is limited. Other surfactants such as sulfosuccinates and sulfosuccinamates, either alone or in conjunction with fatty acids and sulfonates, are also used commercially (Cytec Industries Inc. 2002), but their tonnage is relatively low. Phosphoric acid esters are also used to a limited extent. In terms of new chemistry, as in the case of sulfide flotation, literally thousands of compounds, with various combinations of the donor atoms N and O, including numerous chelating agents, have been proposed and tested in the laboratory (mostly with pure minerals and under idealized conditions), but, except for a handful of reagents such as phosphonic acids, alkyl hydroxamates, and some amphoteric surfactants, none have been commercialized (Nagaraj, Day, and Gorken 1999). This situation will change as the demand for specialty minerals grows and the separations become more challenging. Even today, there are certain niche areas of nonsulfide flotation where the industry has already benefited from the use of alternative flotation reagents; however, effort and industry commitment are somewhat lacking.

flotation0.book Page 398 Tuesday, January 2, 2007 7:36 PM

398

FLOTATION CHEMISTRY

Research in nonsulfide flotation suffers from the same deficiencies as in sulfide flotation. As with xanthates, there is a considerable preoccupation with redundant studies on the interactions between fatty acids and selected nonsulfide minerals and between amines and quartz or silica. Frothers. In general, auxiliary frothers such as short-chain alcohols and polyglycols are not typically used in nonsulfide flotation because the majority of the collectors generate sufficient froth. However, there have been many instances where the use of an auxiliary frother resulted in significant benefits, for example, in the beneficiation of phosphate, glass sand, feldspar, and calcium carbonate. Kerosene and fuel oil are often used to effectively control the voluminous overfrothing caused on occasion by saponified fatty acids and tall oils. Modifiers. Because most of the nonsulfide collectors are rather nonselective and there is a great deal of similarity in surface properties between the various nonsulfide minerals, modifiers play a critical role. In fact, many of the mineral separations would be virtually impossible if not for the use of appropriate modifiers. The importance of the modifiers is far greater in nonsulfide flotation than it is in sulfide flotation. The modifiers currently used are given in Table 8 and Figure 4. The inorganic modifiers, once again, dominate the industrial usage. Organic modifiers are mostly natural products in the form of polysaccharides, lignin sulfonate, and tannins. Small organic molecules, such as hydroxy acids, were used briefly several decades ago, but because of their cost and the large dosages required, they are not used much today. Synthetic polymeric modifiers (Figure 4) have limited usage at present in nonsulfide flotation mainly because of the lack of concerted efforts. They are expected to gain importance in the future as the demand for, and value of, nonsulfide minerals grows. C R I T I C A L A N A LY S I S O F T H E S TAT U S O F F L O TAT I O N R E AG E N T S Collectors

It is well established that the main function of the collector is to impart sufficient hydrophobicity to the mineral surface so that the probability of bubble–mineral attachment is increased. It is also well established that a collector molecule accomplishes this function by adsorbing on the mineral surface.* Collector adsorption and the consequent development of hydrophobicity on the mineral surface does not guarantee the desired flotation outcome, for flotation is a probabilistic (or rather, pseudo-probabilistic) process.† Significant advances have been made in the past 25 years in reagent design, selection, and systematic structure–activity relationships. These are described in several notable articles and books (Aplan 1994; Aplan and Chander 1988; Herrera-Urbina 2003; Klimpel 1994; Jones and Oblatt 1984; Nagaraj 1997; Nagaraj, Basilio, and Yoon 1989; Pradip 1991; Pradip and Rai 2002; Mulukutla 1994). Some examples of structure–activity relationships that are readily apparent in the realm of sulfide collector chemistry are given in Table 9. These * A collector is not always necessary; a mineral can be floated without a collector if it is naturally hydrophobic. † Flotation is a delicate balance between buoyancy and gravitational forces; any perturbation (chemical, physical, or operational) can tip the balance. Flotation outcome cannot be predicted exactly, but probabilities of the various subprocesses can be estimated, some better than others. More accurately, flotation is pseudo-probabilistic or pseudo-deterministic. Collector, modifier, and frother work in concert.

flotation0.book Page 399 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

399

examples are only a small sample of a large body of available semiempirical structure– activity relationships that form the basis for screening of existing reagents or for designing new reagents for a targeted set of a plant’s metallurgical needs. A perusal of flotation literature on chemical aspects in the past 100 years would indicate that collectors have been accorded paramount importance and close attention at the expense of modifiers and frothers. From the perspective of the flotation plant’s overall goals of maximizing recovery, grade, and profitability, such importance to collector alone is rather distorted and reductionistic. Collectors do play a critical role in providing a part of the solution and are necessary but not sufficient. In terms of research, in sulfide flotation there is still a preoccupation with the study of interactions between xanthates (80 years after their introduction) and sulfide minerals, and elucidation of the mechanisms (more importantly, electrochemical interactions). Other collectors have received little attention, which is in stark contrast to plant practice. There is much redundancy in these studies and their findings, to the point where such research has TABLE 9

Examples of structure–activity relationships

Collector

Structure

Collector Properties

Same Substituents, Different Donors IPETC (isopropyl ethyl thionocarbamate)

S

CH3 CH

O

C

NH

CH2

CH3

Chalcocite and pyrite flotation with thionocarbamates (collector concentration: chalcocite (Cc)—1E-6M; pyrite (Py)—5E-6M)

CH3

90 S

CH3 CH

O

O

C

NH

IPECTC Cc

80

C

O

CH2

70

CH3

% Flotation

IPECTC (isopropyl ethoxycarbonyl thionocarbamate)

100

CH3

IPETC Cc

60 50 40

IPETC/Py

30 20

IPECTC/Py

10 0

2

3

4

5

6

7

8

9

10

11

12

pH

The extra C(O)O group in IPECTC increases acidity of the molecule, changes complexation on mineral surface, significantly changes collector activity. IPECTC performs better at pH DIBDTP for galena, Ag, Au, Cu, stability, selectivity, and strength of metal complexes.

P S

6 pH

S

H 3C

80

20 5

H 3C

H3C

DTP

90

FeS2 4

Diisobutyl Dithiophosphinate (DIBDTPI)

MTP

100 Au

Cu Recovery, %

Gold and Pyrite Recovery, %

O

S

H 3C

P O

S

–

CH3

H 3C

Same Donors, Different Hydrocarbons Aryl and alkyl DTPs

P

H3C

O –

S

S

H3C

P –

O

O

Aryl

S

CH3

H 3C

H 3C

DTPs

A change in hydrocarbon group (same donor atoms) caused a dramatic change in performance. Aryl DTP is excellent for PbS; alkyl DTP is poor for PbS.

H 3C

S

O

Alkyl

H3C Isobutyl (

S CH

CH2

O )

2

H3C

P S

CH3 Sec-butyl (CH3

CH2

CH

–

Collector Strength

S O )

2

P S

–

S Cresyl

Allyl and alkyl thionocarbamates

Allyl

O )

H 3C

H

S

–

Selectivity

O

Allyl double bond increases hydrophobicity and rate of flotation (Cu, activated ZnS, native Cu, PGMs, sulfidized oxide Cu), and significantly alters froth characteristics.

CH3

H2C

S H

CH3

N H3C

P

CH3

N

Alkyl

2

Steric effects (same donors, different hydrocarbons) branching and aromatic groups influence orientation/packing and hydrophobicity—therefore, selectivity and collector strength.

O CH3 S

(Table continued next page)

flotation0.book Page 401 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 9

401

Examples of structure–activity relationships (continued)

Collector

Structure

Collector Properties

Same Donors, Different Hydrocarbons Xanthate ester and xanthate

S R

O

C

S

CH2

CH

CH2

S R

S esters of other ionic collectors

O

C

S

S

S

P

P S–

R

S

S

S

C

C S

–

S

Modulating strength and selectivity in S donors; xanthate ester is very selective for Cu sulfides, MoS2, precious metals, and sulfidized Cu oxides; xanthate is nonselective.

R

S R

Modulating strength and selectivity in S donors. The esters are very selective for certain sulfides only and precious metals, do not undergo redox reactions, and are less sensitive to water chemistry changes.

R

S

H

R

Same Donors, Same Hydrocarbons, Different Geometry Oxime isomers

C9H19

C OH

Anti-isomer is a collector for Cu oxide, whereas the synisomer is not.

C9H19

C OH

N OH

Anti -isomer

N

HO

Syn -isomer

Adapted from Nagaraj 1988, 1997; Nagaraj and Avotins 1988; Nagaraj, Basilio, and Yoon 1989.

become irrelevant to the plant metallurgists and reagent developers who moved beyond xanthates many decades ago;* for example, xanthate is not the primary collector in the vast majority of copper plants that make up the bulk of sulfide flotation. Even in operations where selectivity is not critical, additional collectors are invariably used to augment performance deficiencies of xanthate. Basic chemistry would dictate that a single collector cannot be uniformly effective for the large diversity in mineralogy encountered in the plants. Plant practice supports this view. Flotation practitioners may have realized this fact of chemistry and adopted collectors according to the separation challenges faced in the plant, but in the research community, the focus is still on xanthate irrespective of the mineral separation in question. A similar vein would be the vast number of reductionistic electrochemical studies conducted over the past four decades. We cannot deny that these studies have helped in terms of gaining a better understanding of the sulfide flotation systems, but they have not had the revolutionary impact that was claimed of them time and again for plant practice. If the focus had been on the real total system, applied research would certainly have produced the necessary knowledge base to implement concepts in the plant. Indeed, the occasional limited, applied studies have already identified the pitfalls in terms of measurement, interpretation, and control of pulp potentials. The only practical method of control is using chemicals (reagents and various gases) that participate not only in electrochemical reactions but also in many * It is difficult to imagine how 80 years of research could be devoted to a single molecule. Is there so much to be learned about xanthate?

flotation0.book Page 402 Tuesday, January 2, 2007 7:36 PM

402

FLOTATION CHEMISTRY

chemical reactions. The numerous misconceptions surrounding pulp potentials have accentuated the problem (Nagaraj 2000b). The excellent fundamental research notwithstanding, very few plants have been able to exploit pulp potentials for flotation separation and optimization, given the complexity in mineralogy (composite particles, diverse mineral species, etc.), the significant routine variability in the plants, the lack of correlation between pulp potentials and mineral surface composition, interference from purely chemical reactions, and the difficulty in interpreting measured values. The landscape is either the same or worse in nonsulfide flotation where the obsessive focus is on the interactions between fatty acid or dodecylamine on selected minerals. Many of the classical problems in flotation that were recognized in the first half-century of flotation—for example, effect of water chemistry, slime coatings, recovery of coarse and fine particles, difficult mineral separations—have yet to be solved and have received relatively little attention. Even in fundamental studies related to such problems, the tendency has been to focus on xanthates and fatty acids rather than to consider alternatives that may possess certain inherently advantageous properties. Modifiers

Unlike collectors, discussion of modifiers is limited in the literature; therefore, this section gives modifiers relatively greater coverage. Selective flotation is made possible by use of a large number of modifiers. Their importance is in direct relation to complexity of mineralogy. Sutherland and Wark (1955) noted that “depressants have made possible the excellent results of selective flotation. Nevertheless, being difficult to control, they cause the metallurgist more worry than any other single class of flotation reagent.” This observation is valid even today. The chemistry of modifiers in flotation pulps is quite complex in comparison to that of collectors, and is more complex in nonsulfide systems than it is in sulfide systems. The interactions of modifiers with minerals and the mechanism by which modifiers affect mineral floatability and selectivity are summarized in Table 10. Modifiers can affect multiple factors simultaneously in flotation pulps and, therefore, have multiple functions. (Details of predominant functions of various modifiers are found in works by Eigeles [1977] and Chander [1988]). For example, addition of lime in a copper circuit can have the following effects: pH modification (and the resultant changes in interfacial chemistry, solubility of minerals, changes in water chemistry), depression of pyrite, formation of colloidal calcium sulfate precipitate and its adsorption on sulfide minerals, modification of froth characteristics and viscosity of pulp, modification of collector adsorption, precipitation of multivalent metal ions as hydroxides, removal of slime coatings from sulfides, dispersion and depression of certain gangue minerals, and reduction in grinding media wear. Thus, depending on the type, modifiers can change pH, pulp potential (Ep), composition of the aqueous phase, and surface composition of the minerals and bubbles. They can affect properties of all three interfaces. They are also known to affect dispersion and rise-velocity of bubbles. The same modifier in different circuits could produce very different flotation outcomes, or different modifiers used for pH control in a circuit can have very different results at the same pH. In addition to modifiers added deliberately, significant amounts of unavoidable ions and species released from the minerals are also present in flotation pulps, and they act as modifiers (Ananthapadmanabhan and Somasundaran 1984; Eigeles 1977), albeit inadvertently (Nagaraj and Brinen 1995, 1996b). The unavoidable ions and species formed in this process react with water, with each other, with other reagents forming complex ions and

flotation0.book Page 403 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 10

403

Interaction of modifiers with minerals and the mechanism of floatability modification

Adsorption 1

2 3 4 5

Adsorption of ions: electrostatic, specific adsorption, adsorption of potential determining ions (e.g., Mn+ on oxides and sulfides; OH– on sulfides or oxides which then affects collector adsorption; HS– on sulfides [driven by solubility product, chemical reaction]) Ion exchange and chemical reaction; Cu activation of sphalerite Entropy-driven adsorption (e.g., adsorption of polymers) Hydrogen bonding and hydrophobic bonding (e.g., adsorption of nonionic species on minerals, especially nonionic polymers) Chemisorption and chemical reaction (e.g., NaCN on sulfides)

Mechanism by which a modifier can affect mineral floatability 1 2 3 4 5 6

7

8

Selective tying up of all active sites so as to prevent collector adsorption Inadvertent activation and the consequent adsorption of collector (the active centers are not of the mineral but are of the impurity species). Concentration-dependent effect: At low concentrations, sodium sulfide can activate certain oxides and tarnished sulfides, and at high concentrations it depresses these. Adsorption of colloidal reaction products of modifiers on minerals (e.g., metal sols of sodium silicate, products of solution hydrolysis, sol formation and gel formation, or other colloidal precipitates) Prevention of heterocoagulation of slimes to minerals (e.g., with dispersants such as polyacrylate, polyphosphate) Removal of active species (sites) from the surface by complexation. This is often concentration dependent. At low concentrations, certain complexing agents can chemisorb on surfaces, but as the concentration increases, they can form soluble complexes which are then released into solution. NaCN is a good example; another example is polyphosphate. Removal of hydrophobic films from surfaces. Chemical or physical methods. For example, Na2S in Cu-Mo separation—removal of collectors from surfaces. Removal of oleate from apatite by sulfuric acid. Layers of mineral may be removed in this process. Physical means would be steaming, high-energy attrition. Removal of hydrophilic patches to increase floatability, (e.g., sodium sulfide, dispersants). For example, a thick film of limonite (approx. 300 Å) is removed from the oxidized surface of chalcopyrite by treatment with oxalic acid and subsequent washing with water. Acid treatment of many nonsulfide minerals. Also acid scrubbing of glass sands to activate iron-bearing impurities. Treatment of beryl with hydrofluoric acid leads to a solution of the silicate components such as silica and alumina and to an increase of the density of Be ions at the surface.

Adapted from Nagaraj 2003.

molecules, and also with colloidal and coarsely dispersed suspensions and precipitates. These species, in turn, adsorb on, or react with, the mineral surfaces and air bubbles and change their surface compositions. The direction and kinetics of these processes depend on the concentration of the acting components, temperature (seasonal variations), and to a large degree on hydrodynamics. The situation is complicated by large recirculating loads and recycled water in plant circuits. A major consequence of unavoidable species released from the various mineral components is inadvertent activation of certain value minerals. For example, inadvertent activation of sphalerite by Cu or Pb causes significant amounts of Zn reporting to Pb and Cu concentrates; similarly, inadvertent Cu activation of pyrite and Ca activation in nonsulfide mineral separations can occur (Eigeles 1977; Finkelstein 1997; Rao 2004). Inadvertent activation is often difficult to predict, because it is strongly influenced by mineralogy and pulp conditions that exhibit significant variability in a plant. Indeed, one of the main functions of modifiers added deliberately (e.g., lime, soda ash, sodium silicate, polyacrylates) is to control the effects of unavoidable species. Modifiers change mineral surfaces either directly by adsorbing at the S/L interface (via physical or chemical adsorption) or indirectly by changing aqueous phase composition. Modifiers affect the liquid/air interface significantly, but this aspect has been inadequately studied in research. The action of modifiers (alkalis, acids, and salts) on the dispersion of air bubbles, and the intensity of frothing, was described long ago (Gaudin 1957). The strong

flotation0.book Page 404 Tuesday, January 2, 2007 7:36 PM

404

FLOTATION CHEMISTRY

effect of pH on the stability of froth is all too well known in sulfide circuits, and the direction and magnitude of this effect is strongly influenced by gangue mineralogy and is quite variable; generalizations are difficult to make. Salts of multivalent metals, such as copper and iron, even at low concentrations, reduce the stability of the froth formed by an alcohol; the magnitude of the effect is strongly influenced by pH. Thus, copper sulfate is often used to control excessive froth in certain sulfide circuits. In a zinc circuit, after the addition of copper sulfate and xanthate, a decrease in frothing is sometimes observed, and this can be restored after the addition of lime (Eigeles 1977). Slimes and colloidal precipitates that exist in flotation pulps strongly influence the froth phase and bubble–particle interactions. Difficulties arise in the laboratory investigation of all aspects of the action of modifiers. Many of the complex effects of modifiers are evident only in real flotation pulps where the variability in mineralogy and operating conditions has a significant impact and is virtually impossible to simulate in the laboratory. Thus, effects observed in model systems used in laboratory investigation can be quite misleading. For example, single mineral studies may indicate that when lime is used as the modifier, adsorption of Ca ions (or other metal ions) on sulfide particles, such as molybdenite or pyrite, may occur, resulting in depression of these minerals (Castro and Bobadilla 1995; Raghavan and Hsu 1984). However, there is the unanswered question as to whether such adsorption occurs in the ore pulps comprising a complex aqueous phase and a large number of mineral species, especially of nonsulfide minerals. In Cu-Mo ores, the Mo content is typically on the order of 0.05%, whereas nonsulfide minerals comprise approximately 95%. It is difficult to reason that the added Ca species in the form of lime would exhibit such overwhelming specificity as to seek the very small number of molybdenite particles (and cause their depression) from an exceedingly large population of a variety of nonsulfide gangue particles on which they are known to adsorb. Of the three major reagents forming the flotation reagent triangle, modifiers have perhaps the greatest effect on rate of flotation of particles. This effect can be either positive or negative, and it can be temporary (i.e., the mineral can be refloated in another stage with relative ease, e.g., when using a nonionic polymer) or permanent (i.e., true depression; the mineral can be refloated only after drastic measures have been taken, such as a change in chemical environment; e.g., cyanide depression of pyrite or sphalerite, reversed by addition of copper sulfate). Thus, one modifier may merely decrease the flotation rate of one mineral selectively and sufficiently so that a selective flotation of another mineral is possible. In terms of new developments, advances of any commercial significance have been made in the area of development of synthetic, water-soluble polymeric depressants. The general concept is one of incorporating specific functional groups such as complexing or chelating moieties into water-soluble hydrophilic polymers (Nagaraj 1988; Nagaraj et al. 1987). These depressants offer many advantages over the traditionally used depressants, such as better performance at a lower dosage and lower treatment cost (exploiting the efficiency of a large molecule), ease of handling, lower toxicity, lower transportation and storage hazards, ease of structural modification to suit different applications and ore variability, and consistency from batch to batch. The chemistry and applications of the new, synthetic polymeric modifiers were discussed by Nagaraj (2000a). Certain synthetic water-soluble polymers (e.g., polyacrylates) are used in the minerals industry as dispersants or selective flocculants for clays and slimes (Lin and Burdick 1988). The use of polysaccharides is reviewed in many articles (Mackenzie 1986; Pugh 1989). In polymeric modifiers, structure–activity relationships take on added dimensions in terms of complexity and possibilities. In addition to changes in functional group, factors

flotation0.book Page 405 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

405

Froth Phase Composition

Solids

Sulfides Carbonates

Liquid

Silicates Oxides

Recycled H2O

Fresh Water

Air

Silica

Collector

Frother

Modifiers

C1 C2 C3

F1 F2 F3 F4

M1 M2 M3

R1 R2 R3 R4 Source: Nagaraj 2003.

FIGURE 5 The complexity of the three-phase froths (R1, R2, etc., represent major species in recycled water; C1, C2, and C3 represent three different collector types; similarly for frothers, F, and modifiers, M)

such as the MW, degree of substitution, location, and distribution of functional groups become critical. The first successful synthetic polymeric depressant appears to be the phosphate depressant introduced in 1982 (Cytec 2002; Nagaraj et al. 1987). Low-MW ( other isomers; aliphatic alcohols > aromatic alcohols; saturated > unsaturated (opposite in terpene-based)

Polyglycols and Glycol Ethers Strongest frothers; maintain frothing power without stage addition

Bubble size

Larger bubble size, less compact structure

Smaller bubble size, compact structure

Kinetics

Faster kinetics

Slower kinetics

Sensitivity to other conditions

More sensitive to pH; tolerant to slimes

Less sensitive to slimes

Selectivity

Lower water retention (greater froth drainage) resulting in greater selectivity; less downstream effects because of selectivity and low persistence

More water retention, lower selectivity, greater downstream effect because of persistence

Stability

Brittle froth; less tenacity, less persistence; requires staged addition; may cause effervescence at high dosages

Persistent at low dosages; more persistent; do not require staged addition; Break down readily in launders; do not cause effervescence at high dosages

Coarse particle recovery and load capacity

Less

More

Adapted from Booth and Freyberger 1962, Klimpel and Hansen 1988, Cytec Industries Inc. 2002. NOTE: These are merely crude generalizations; exceptions can always be found. *TEB produces froths resembling those of pine oil (small-bubble, closely knit texture, lower drainage, lower grades, higher recoveries; breaks down readily in launders); however, like polyglycols, TEB does not flatten froth and effervesce at high dosages. †The froths resulting from fatty acids, sulfonates, and amine collectors in nonsulfide flotation are closely textured, stablebubble aggregates that frequently are difficult to disintegrate even by the application of water jets and sprays. Flotation operators seek to avoid voluminous overfrothing and permanent, lathery froth structures because these are not conducive to the production of high-grade concentrate and are difficult to handle. The addition of specific frothers along with the nonsulfide collectors will at times improve these froth conditions, particularly if large quantities of slimes are present in the ore pulp.

flotation0.book Page 410 Tuesday, January 2, 2007 7:36 PM

410

FLOTATION CHEMISTRY

poorer selectivity.” Statements and generalizations, such as these and many others that are prevalent in the industry, are certainly useful though unsubstantiated by fundamental link and should be viewed with caution because they may provide good information for specific systems and sets of conditions only. Klimpel and Hansen (1988) further noted, It is sometimes assumed (especially in laboratory studies) that the choice of frother type and dosage is not as crucial as some of the other factors such as collector type and dosage. Plant practice does not support this assumption. Significant improvements can be achieved by managing froth phase, i.e., frother type and dosage. The scientific fundamentals involved in froth development and stability in the presence of particles and turbulence are extremely complex and not yet fully developed; consequently, a priori predictions of performance cannot be made for a given frother. Sometimes, a higher frother dosage, rather than higher collector dosage, is used to improve value recovery because frothers generally have lower unit price, in spite of the fact that this invariably produces poor selectivity. Rao (2004) observed the following: “It is obvious that despite a considerable amount of attention being paid in the last 100 years to the characteristics of bubbles, of thin layers separating these bubbles in foams and froths, and of the lifetime of froths, there is no unanimity regarding parameters that govern the stability and the collapse of froths.” Therefore, research is critically needed in the three-phase froths with direct relevance to plant systems. T E C H N O L O G Y T R A N S F E R — L A B O R AT O R Y T O P L A N T

Literally thousands of new flotation reagents—collectors, modifiers, and frothers—have been proposed for sulfide and nonsulfide flotation over the past eight decades since the beginning of the use of small organic molecules and soluble surfactants in flotation. This is obviously driven by needs arising from the extreme complexity of mineral systems and, perhaps, by a realization that reagents are the easiest to change and control in flotation to achieve desired goals. Thus, there has been an insatiable urge from the beginning to develop reagents that are extremely selective, provide better recoveries, and are inexpensive, though an attempt to maximize all three qualities is unrealistic and is perhaps a reflection of the naiveté associated with the understanding of flotation process and mineralogy. In stark contrast, the number of flotation reagents used in practice is a tiny fraction of the total number proposed or developed in the laboratory. Acceptance in the industry for novel reagents is still rather slow. Many in the reagent development business also recorded the slow pace of acceptance of new technology (Cappuccitti 1994; Nagaraj 1994; Malhotra 1994; Klimpel 1994) and identified many factors responsible for the lack of successful translation of laboratory findings into commercial application. These factors are summarized in Table 12. One of the more important factors is the somewhat arbitrary, reductionistic, and bottom-up approach taken in reagent selection and optimization, despite the understanding that mineral systems are complex, and that flotation outcome is determined by complex interactions between physical and chemical factors. A holistic (or top-down) approach is necessary in order to develop robust solutions to plant needs.

flotation0.book Page 411 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

TABLE 12

411

Critical factors in reagent development and plant use

Factor Cost Choice of reagent

Application knowledge

Research

Issue

Solution

Focus on pricetag. Perceived high cost of the proposed alternative reagent. Not based on realistic needs, goals, commercial and technical viability, but based on personal experience, gut feeling, anecdotes. Inadequacy. Available mostly for a handful of “old” reagents (e.g., xanthate, fatty acid, NaCN, SO2, lime, etc.). Inadequate for the majority of the remaining traditional and new reagents. Insufficient effort to develop fundamental applications knowledge for the traditional and new chemistries. Highly reductionistic approach under idealized conditions. Narrow and obsessive focus on xanthates, fatty acids, and a few wellknown modifiers. Redundant, reductionistic, and irrelevant research. Many of the long-recognized problems in flotation (such as the effect of water chemistry, slimes coatings, fine and coarse particle flotation, to name a few) remain unsolved.

Focus on delivered value, not the unit cost of the reagent. Focus on “total system”; holistic selection criteria.

Need a concerted effort to develop applications knowledge on natural ores under realistic conditions and, certainly, followed by larger-scale testing.

Research should be driven by the needs in the industry. Should be solution based.

Objectives

Testing objectives that don’t meet or are not consistent with plant needs. Not all plant problems are solvable by chemical means.

Plant needs and objectives must be clearly understood, defined, and agreed upon before a single test is run or a solution recommended.

Environmental

Increasingly stringent environmental regulations. Long and expensive procedure and protocols to qualify products for commercial use.

Return on investment

Lead time for commercialization is very long, and return on investment in R&D is slow.

Reagent R&D and application must be driven by industry needs and commitment; otherwise, such R&D will be a wasted effort. Reagent R&D and application must be driven by industry needs and commitment; otherwise, such R&D will be a wasted effort.

Lab and plant testing

Inadequate or improper testing that yields misleading information regarding the efficacy of a flotation reagent, and expensive failures in plant. Lack of standardization of laboratory flotation testing best practices and statistical tools.

The trend toward the implementation of quality management principles and best practices such as Six Sigma is firmly in place in most industries today, but not all are embracing these concepts to the same degree.

Industry commitment and commoditization of technology

Lack of plant’s commitment to the testing of new flotation reagents. The industry is often seen as resistant to change chemistry and resistant to consider alternative conditions to accommodate the new chemistry. A strong tendency to commoditize new technology.

If the needs are clearly defined and the benefits clearly demonstrated, then plant’s commitment would be facilitated. Economic calculations for new technology must be holistic, not just based on pricetag.

Nonchemistry factors

Chemistry is not everything. Many other factors determine reagent development (e.g., manufacturing, environmental, handling, hazards, downstream effects). Shortage of qualified metallurgists; declining education in mineral processing. Pressure on plants to find ways to further cut costs to be competitive in the new global economy.

Need to consider all technical and commercial factors and constraints from the beginning and throughout reagent development process—holistic approach. Industry and academe should work closely to promote mineral industry education. Must demonstrate improved value and efficiency.

Education and expertise

Economic pressures

Adapted from Nagaraj 1994; Cappuccitti 1994; Malhotra 1994; Klimpel 1994.

flotation0.book Page 412 Tuesday, January 2, 2007 7:36 PM

412

FLOTATION CHEMISTRY

MARKETING ASPECTS AND TRENDS*

From a marketing point of view, flotation reagents can be segmented into commodities (e.g., xanthate, fatty acid) and specialty chemicals. Specialty chemicals can be further broken down into mature and novel specialties. In sulfide flotation, products such as DTPs, thionocarbamates, xanthogen formates, mercaptobenzothiazole, and mercaptan can be classified as mature specialties, depending on logistical and geographical factors combined with service components. Products such as dithiophosphinates, monothiophosphates, functionalized thionocarbamates, and thioureas developed more recently are considered novel specialties. Among modifiers, the recently developed, functionalized, synthetic polymers are novel specialty reagents; most other modifiers are commodities. Among frothers, certain targeted frother compositions can be considered a mature specialty. The majority of collectors and modifiers used for nonsulfide flotation have been commodities for many decades. However, a number of collectors that have been known for several decades do enjoy specialty status today by virtue of their use in niche applications and the ability to demonstrate value (e.g., hydroxamates, morpholines, phosphate esters, amphoteric collectors, and phosphonic acids). High-performance, specialty products have been successful in replacing the more traditional reagents, including xanthates, in a cost-effective manner in many large plants. However, this has been possible only with significant advances in applications technology that the manufacturer has developed in-house, and has come after a large investment in resources and time, and only with a willing plant management that values technology and is committed to quality and best practices. The growth of the specialty reagent market overall has been slow because of a pricetag mind set and the lack of adequate commitment from the industry. China is now the largest producer of xanthate and IPETC (isopropyl ethyl thionocarbamate), supplied globally through distributors. As a result, there has been significant price erosion of other reagents. Several plants also obtain liquid xanthate that is produced in close proximity. Similarly, amines and fatty acids are supplied from plants close to the mines. Despite significant price erosion and commoditization of many of the flotation reagents, the market for specialty reagents and the associated application technology is still healthy owing to the following factors. Plants are coming under intense pressure from global economic and environmental factors to maximize productivity and minimize effluents that have the potential to be hazardous. This will, then, force the plants to look for advanced technology to go beyond what traditional products can offer. In many mines, the highgrade, easy-to-treat ores have been mostly exhausted. The new deposits are significantly more challenging to treat and require the use of alternative chemistries. Also, flotation plants that have used the same products over a long period of time, irrespective of significant changes in ore types (mineralogy) and plant conditions, would not be optimized and would have large swings in plant performance. The survival of such plants will be predicated upon how readily (and quickly) they can implement new technology and use robust performance products that are designed to handle changes in mineralogy and routine plant conditions. The need for ever-increasing productivity improvements and the eventual emergence of the less-developed countries into the global market economy will require the higher performance benefits produced by specialty reagents. Also, with the large increase in raw material costs in recent years, the profit margins have diminished dramatically for xanthates, and are * Information on marketing aspects of flotation reagents presented in this section is largely borrowed from work by Cappuccitti (1994) but has been updated and expanded.

flotation0.book Page 413 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

413

now at a point where the selling prices are barely above the level of the raw material costs; consequently, xanthate producers will no longer be able to defend the price against the benefits of the specialties. Because the industry will not be able to reduce costs on the basis of price of reagents, they will have to look at other means. A holistic, solution-based approach is needed to take the performance to the next level. In many cases, this may require plants to adopt conditions and reagents different from those used traditionally. Developing new technology is long and expensive. Many new chemical species that were developed out of the R&D program during the past two decades still required 5–10 years from concept to commercialization and more than $5–$10 million in investment to develop. They are now being accepted more broadly by the industry and are beginning to replace and augment the more traditional reagents as a result of sustained technical and commercial effort that required the allocation of many technical field resources to establish the efficacy of these new products in plant operations. Many new products have been developed as a result of continued R&D programs, but they have not been fully evaluated in plants. To date, the return on these investments can be described as marginal, at best, for the suppliers of novel products who have ongoing R&D programs. It is important to recognize that advances made in flotation reagents, and in physicomechanical and operational factors, can result in large improvements in performance only when there is concerted effort to bring these factors together. Although there is increased recognition that synergistic partnerships between reagent developers and mining companies are necessary to improve profitability and technology transfer in a global economy, realization of such partnerships will occur only with a total system or holistic approach and sharing of best practices. E N V I R O N M E N TA L , T O X I C I T Y, A N D R E G U L AT O R Y I S S U E S *

There are many regulations that govern the introduction of new chemical substances into commerce. Several countries have implemented chemical control law legislation for new substances and have established national inventory lists of chemical entities that have been approved for manufacture and/or import: • Australia—Australian Inventory of Chemical Substances (AICS) • Canada—Domestic Substances List (DSL) • China—Inventory of Existing Chemical Substances (IECSC) • The European Union (EU)—European Inventory of Existing Chemical Substances/ European LIst of Notified Chemical Substances (EINECS/ELINCS) • Japan—Existing and New Chemicals and Substances (ENCS), also known as the Ministry of Economy, Trade and Industry (METI) • Korea—Existing Chemicals List (ECL) • the Philippines—Philippines Inventory of Chemicals and Chemical Substances (PICCS) • the United States—Toxic Substances Control Act (TSCA) The legislation is different from country to country but outlines the tests (chemical/ physical properties, mammalian toxicity, aquatic toxicity, mutagenicity, and environmental fate) that are required in order to add a chemical substance to a country’s inventory. A product that contains a component not on the inventory cannot be manufactured or sold in * Information in this section was produced by the Toxicology Department of Cytec Industries, Inc.

flotation0.book Page 414 Tuesday, January 2, 2007 7:36 PM

414

FLOTATION CHEMISTRY

that country without some type of registration or exemption. For the most part, regulations are for distinct chemical species and the product as sold. Therefore, if a product contains several new chemicals, each one would require regulatory approval, and, in some cases, the complex mixture, including impurities, would require regulatory approval. For new chemical substances in the United States, submission to the U.S. Environmental Protection Agency (EPA) of a Premanufacture Notification (PMN) is the sole requirement for inclusion in the TSCA Inventory. Currently, there are no mandatory toxicology testing requirements set forth by the EPA, but the EPA can require data after their initial review process. As part of the Product Stewardship initiatives, and the internal standard operating procedure for new product introduction, the reagent manufacturer must conduct a preliminary set of toxicity tests to assess the hazards of products as part of the TSCA registration process. The manufacturer must also routinely perform a mutagenicity test and must test for acute mammalian toxicity, skin and eye irritation, aquatic toxicity, and perform preliminary environmental fate testing for new materials. The EU, however, requires a Premarketing Notification (PMKN) with extensive testing based on production volumes. Australia, Canada, China, Japan, and Korea also require extensive testing as part of their registration processes. The higher the production/import volume or potential exposure, the greater is the potential testing requirement. Typical testing costs range from $60,000 to $65,000 in the United States to $175,000 to $300,000 in the EU for each listed chemical or complex mixture. Canadian approval ranges from $50,000 to $300,000. A typical Japanese METI registration can range from $25,000 to $400,000. Meeting the testing requirements for Europe will satisfy the requirements for most countries because of the mutual acceptance of studies conducted by Organisation for Economic Co-operation and Development (OECD) test guidelines. An additional test or two would be required for approval in Canada and Australia. Requirements for registration in Japan are more unique, can become quite costly, and do not easily equate to those of other countries. Water solubility, bacterial biodegradation (mandatory), aquatic toxicity, and potential for bioaccumulation (mandatory) are the main determinates for completing a registration package for Japan. China’s testing requirements are evolving but currently appear to be a combination of requirements for EU and Japan with some in-country testing requirements for aquatic toxicology studies. Efforts to harmonize registration requirements are continuing, but this effort will take many years to accomplish. Most of the traditional collectors that have been used in the mining industry for decades have been grandfathered onto the individual national inventories. But with new chemicals, the large investment in R&D requires that there be a significant global market (at a minimum of 100 metric tons per chemical per year) to justify the development, which could require as much as $2 million per chemical to be approved for commerce in the eight countries/regions where a national inventory currently exists. This will not obviously apply to areas of the world that do not have chemical control regulations that govern toxic or potentially toxic substances, which may result in these areas of the world gaining potential competitive advantage over developed countries. R E A G E N T S E L E C T I O N A N D O P T I M I Z AT I O N

In the 1920s and 1930s, the number of commercially used flotation reagents was relatively small, and reagent selection for a given application was relatively straightforward. As the number and diversity of commercially available flotation reagents steadily increased to meet

flotation0.book Page 415 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

415

the ever-increasing technical and economic challenges, reagent selection became gradually more complicated. Four other factors have made the situation more acute: 1. The obsessive research focus on “old” reagents (e.g., xanthate, fatty acid, cyanide) meant that the chemistry and applications knowledge for all other reagents remained woefully inadequate in the published literature. 2. Science did not keep pace with technology; the increasing emphasis on reductionistic research over the decades meant a widening gap between research and practice. 3. Application knowledge developed by in-house research in the chemical companies remained largely inaccessible in the public domain. 4. There has been a steady decline of expertise in flotation reagents in the industry over the decades. Consequently, at present, the process of reagent selection and optimization can be characterized as rather informal, reductionistic, and frequently based on extension of personal experience from one mineral system or plant to another, gut feeling, anecdotes, and myths (Nagaraj and Bruey 2003; Nagaraj 2005). There is no recognized standard practice, and the informal process is fraught with pitfalls.* For a given separation, at first glance the pool of reagent classes—and homologues within each class—available for selection appears to be large and bewildering. This is partly the result of the evolutionary nature of reagent development over the decades. But more importantly, given the complexity of the flotation system, a particular reagent class or a specific homologue just happens to produce the optimum results for the separation and ore type in question. General guidelines are available for reagent selection in the form of an accumulated knowledge base in reagent developers’ handbooks and product literature (e.g., the Mining Chemicals Handbook [Cytec 2002]). This serves to narrow the selection for a given application; however, the metallurgist must still use additional filters to refine the selection to a manageable level in order to move to the laboratory testing phase. Historically, this stage of reagent selection has tended to be reductionistic and informal. Problems in complex systems invariably have multiple solutions, each with its degree of desirability. Even an arbitrary process of reagent selection will provide a solution to a plant’s problem. However, such a solution may not be sufficiently robust, that is, the selected reagent will have a narrow, unoptimized window of performance—and may only be a temporary fix—and its nonrobustness may not even be discovered for a long time, given the significant routine variability in plant performance.† This approach is very costly in the long run and adds little to the knowledge base. If the objective is to develop a robust solution to a problem in the plant, and to increase value or profitability in the operation, a holistic or total system approach is necessary. * For example, an immediate temptation to fix a recovery problem in a plant might be to begin a laboratory testing program to seek an alternative collector, keeping all other variables constant. The presumption here is that the recovery problem is somehow linked directly (and only) to a deficiency in the collector, and that the performance of this collector can somehow be treated independently of other reagents and factors. Similarly, a grade problem may be linked solely to a modifier, and a frothing problem to a frother. In all these cases it is quite possible that fortuitously an alternative reagent may provide an apparent or temporary fix to the plant problem, but such a solution may not be robust and sustainable. † A rigorous, statistical analysis of plant performance is needed in order to discern significant changes in plant performance outside of random or auto-correlated plant noise.

flotation0.book Page 416 Tuesday, January 2, 2007 7:36 PM

416

FLOTATION CHEMISTRY

Minerals (Ore Types) Chemical Factors

Chemical Factors Interfacial Chemistry

Flotation System Operational Factors

Physical– Mechanical Factors

Aqueous Phase Reagents

Air

Best Practices

Ore Types

Speed

Collector

Modifier

Plant Needs

Quality

Testing Strategy

Cost

Recovery Plant Performance

Reagents

Grade

Knowledge Base

Cost People

Implementation

Frother

Process

Tools

Source: Nagaraj 2005.

FIGURE 7 Expanded holistic view of the flotation system in reagent selection and optimization

Although the schematic representation (shown in Figure 6) of the flotation system is holistic, it is an inadequate representation. An expanded holistic view of the flotation system, with particular importance to reagent selection and optimization and depicting additional triangles and trade-offs, is shown in Figure 7. The expanded view of chemical factors reveals the flotation reagent triangle interacting with mineralogy (ore types).* The resulting tetrahedron forms the basis for reagent selection and evaluation, in combination with triangles depicted for best practices, plant needs, and knowledge base—all emphasizing trade-offs. “Reagent optimization” has many connotations. To the mill operators, for example, it may mean working with the reagent currently in use to find conditions (e.g., dosages, addition points) that provide optimum performance with respect to certain targeted metallurgical goals (e.g., improved recovery/grade, or cost-effectiveness). Or it may mean searching the market for an alternative reagent that provides performance better than the one currently in use. To the chemical supplier, it may mean modification of existing, or designing new, structures to provide reagents that perform better (for a given application, e.g., copper flotation) than currently marketed reagents, not just in any particular mill but across all mills in the industry. The ultimate goal is a robust solution to the needs in the dynamic plant system with all its variability. Although laboratory testing is an integral part of the process of optimization, its goal is merely to identify potential solutions, problems, and benefits.

* Note that only the chemical factors are expanded. The rationale is that much of the reagent selection process occurs in the laboratory stage where a plant’s physical and operational factors are difficult to simulate. These factors are necessarily included when the reagent, identified in the laboratory, is taken to the plant.

flotation0.book Page 417 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

417

Discovery/ Definition

Plant Needs/Objectives Mineralogy/Ore Types Plant Operation and Constraints

Reagent Selection

Performance Knowledge Base (Reagent Developer) Fundamental Structure–Activity Relationships Technical Factors and Constraints

Lab Testing—Diagnostic Tests, Screening Phase Wide Scope of Chemical and Operational Factors, and Ore Types Establish Suitability of Ore Samples and “Controls” Laboratory Testing Lab Testing—Main Laboratory Test Program, Quantitative Phase Important Variables and Ore Types

Plant Testing

Plant Trial and Optimization

Successful Implementation

Build Knowledge Base

Source: Nagaraj 2005.

FIGURE 8 Rational process for reagent selection and optimization (arrows indicate that the overall process is highly iterative)

A rational, holistic process is necessary in order to implement the total system approach described in Figure 7. Such a process comprising four critical phases is schematically represented in Figure 8. Although these four phases proceed in a logical sequence, they are indeed highly iterative. The discovery/definition phase provides all the necessary information for reagent selection and sets the objectives, goals, and success criteria. The reagent selection phase, then, begins with preliminary screening of available reagents using knowledge base and expertise of the reagent developer to arrive at a subset of reagents that meet the requirements established in the discovery/definition phase. The selected reagents can then be screened in the laboratory by an iterative process. Historically, the preliminary screening phase has been rather simplistic, as it involved merely selecting representative candidates from several different families, and essentially ignoring other important reagents in the circuit and operational variables. Thus, if the objective is to screen collectors for a given separation, the candidate products might be a set of representative collectors from several families, keeping all other reagents and conditions constant. The assumption in this one-reagent-ata-time approach is that reagents perform independently of one another and independent of the operating conditions. A holistic approach for reagent selection is based on considering more than just the chemistry of one reagent in isolation or more than just one type of reagent for a given mineralogy. In fact, all the chemical and operational factors are considered simultaneously. A summary of these factors is given in Table 13. They are conveniently grouped as technical factors and constraints, and each group serves as a filter to narrow down the candidates into a manageable subset.

flotation0.book Page 418 Tuesday, January 2, 2007 7:36 PM

418

TABLE 13

FLOTATION CHEMISTRY

Important factors that dictate reagent selection and trade-off

Technical Factors Reagent chemistry Ore types, mineralogy, mineral chemistry Frothing characteristics Water chemistry and gangue minerals Stability (both short term and long term) Particle size distribution Downstream effects Co-value minerals recovery Plant circuit, process conditions, plant constraints Compatibility with other reagents

Constraints Reagent regulatory approvals Handling Logistics Toxicity and environmental issues Odor Manufacture Cost Strategic fit

Source: Nagaraj 2005.

Factors grouped under constraints in Table 13 are relatively straightforward to deal with in the reagent selection process, but they can often have an overriding impact on reagent selection. For example, a reagent selected on the basis of its superior metallurgical performance can still be rejected because its odor is objectionable, or it was not registered for use in the country where the plant is located, or it might be perceived to present environmental problems, or its unit price is perceived to be unacceptable though the overall treatment cost would have been significantly lower. Technical factors (see Table 13) present the greatest challenge in reagent selection, because many of these are difficult to assess fully in the preliminary reagent screening phase prior to laboratory evaluation. Consequently, a combination of available knowledge base (experience) and empirical diagnostic laboratory testing is used in an iterative process to assess the impact of the technical factors in reagent selection. The objective is to reduce the number of potential reagents and conditions that need to be evaluated in the laboratory. A holistic approach using highly efficient experimental designs and rigorous data analysis will greatly facilitate the reagent selection process and the chances of rapidly arriving at a robust solution. Recently, a rule-based expert system reagent selector was developed by Cytec Industries (Franzidis 2005). This is the first of its kind and is part of a comprehensive package, called Flotation Matrix 100, that is designed to incorporate all the elements of the holistic approach, namely, definition of mineralogy and plant needs, reagent selection, and laboratory and plant best practices. The available performance knowledge base and published literature follows the order: collectors >> modifiers > frothers. Thus, collector selection is easier than that of modifiers and frothers. General guidelines are, however, available for all three classes of reagents, and these form the basis of narrowing the selection to a given application (Cytec 2002). The available pool of reagents is large for collectors and modifiers. In contrast, the choice of a frother is limited to three types of compounds in terms of fundamental chemistry. Frequently, formulations are used to fine-tune a frother for a given specific application. Consequently, frother selection is invariably based on empirical testing, most often in the plant. Laboratory evaluation of frothers is generally (though somewhat inaccurately) considered to be inadequate; this is not surprising given that the froth phase in a plant operation is very dynamic, and its properties are variable and strongly influenced by mineralogy and physical, operational, and chemical factors.* Nevertheless, laboratory testing, when conducted properly, can still be useful for an empirical ranking of frothers and for assessing their potential benefits and problems in a plant circuit. Historically, fundamental research on characterization of frothers and froth phase has been of little value in frother selection or optimization.

flotation0.book Page 419 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

419

Initial selection of a frother for any particular flotation operation will be contingent upon how closely a compound fulfills the basic requirements in terms of several “appearance attributes” such as froth volume, texture, persistence, gangue and water carryover, mineral loading, and flow. Numerous qualitative descriptors (listed previously) are used in the plants. In this initial selection stage, the same technical factors and constraints that were shown in Table 13 are used. Selection of modifiers is far more challenging than that of collectors and frothers. General guidelines are documented in chemical handbooks (Cytec 2002). Unfortunately, information regarding performance of modifiers in the published literature is applicable when dealing with the traditional collectors such as xanthates, fatty acids, and amines. When an alternative collector is used, new information on selecting and optimizing appropriate modifiers has to be developed using empirical testing on ores. However, the task becomes less daunting if a holistic approach is used in which all the flotation reagents are considered simultaneously and all the technical factors and constraints are included in the screening phase. The ultimate choice of any flotation reagent is invariably based on empirical testing in the laboratory followed by rigorous testing in the plant,* and to a large extent, on the overall or global performance of the reagent in the plant (i.e., not just in one small part of the whole plant circuit). CONCLUDING REMARKS

The development of flotation reagents in the last 100 years mirrors that of flotation technology in general; this is necessarily so because reagents played a critical role in making flotation the most widely used physical separation method. Three distinct periods of reagent development can be observed: the early days of flotation (prior to 1920); the discovery and expansion period (1921–1950), characterized by application of chemicals developed in other industries, namely, rubber, textile, tanning, and so forth; and the rational, targeted design period (1951–present) in which the trend began for developing reagents specifically targeted for flotation application. A closer study of the history of flotation reagents also reveals significant contrasts. Remarkable progress has been made in the understanding of the science of flotation. Excellent separation schemes have been designed using a wide variety of reagents, and flotation has been applied to almost all viable ores and minerals. Yet, flotation art has always preceded flotation science. The first 50 years appears to have established an excellent knowledge base, but its use in the next 50 years appears to be rather arbitrary, sporadic, and reductionistic. This could indicate a widening gap between science and technology. Although great effort and resources have gone into the study and application of new sulfide reagents, progress has been slow in the practical application of new technology in actual mill practice. There has been a more concerted effort to develop reagents based on sound scientific concepts and targeted design, but the industry appears to be less receptive to such new technology. Major advances have been made in the understanding of flotation * Based on an extensive survey of plant usage of frothers, Booth and Freyberger noted in 1962 that “the same frother may be used in different plants in a given geographical area, but treating quite dissimilar ore types; conversely, different frothers may be used by plants treating almost the same ore types and similar plant conditions.” This finding is valid even today, and many examples can be found for frother selection that are not based on purely metallurgical considerations. * Details of the laboratory testing and plant testing phases in the process of reagent optimization are discussed elsewhere (Nagaraj and Bruey 2003; Nagaraj 2005).

flotation0.book Page 420 Tuesday, January 2, 2007 7:36 PM

420

FLOTATION CHEMISTRY

chemistry using state-of-the-art analytical techniques, especially surface analysis, yet there are still many unknowns and unresolved problems, and there appears to be a certain degree of mystique about reagent chemistry in the industry, indicating a problem in technology transfer and research focus. There are many misconceptions about reagents and their functions. There is a tendency to apply knowledge of one reagent (such as xanthate) to all other reagents. There is a large disparity between research and practice. For example, 90% of the research focus is on xanthate, but in practice, xanthate accounts for roughly 55% of the total collector usage; in Cu flotation, it accounts for only about 16% of the total collector usage. Despite a general recognition that flotation is a complex system and the outcome is determined by complex interactions between physical, chemical, and operational factors, the approach in research and practice appears to be mostly reductionistic (one factor at a time and assumption of linearity in effects of variables). Although there is general recognition that all three types of flotation reagents—that is, collectors, frothers and modifiers—are critical to the success of flotation in a plant and have equal importance, there is a disproportionate amount of attention given to collectors. The industry need is to improve the efficiency of plant operation with a special emphasis on value recovery in the most profitable manner. Solutions to this overarching need lie in the total system or holistic approach, first for the whole operation comprising geology, mine, comminution, concentrating units, and downstream operations. The focus next would be within an individual unit operation, without losing sight of interlinks between unit operations, and incorporating the physical, chemical, and operating factors simultaneously rather than optimizing one factor at a time. There is also a need to recognize that reagent chemistry is one of approximately 20 factors that dictate reagent selection and optimization. Thus, a holistic approach is necessary to incorporate all the factors and to view reagents in terms of a flotation reagent triangle. In flotation unit operation, the choice of a reagent is critical, but the current default mode is selection of traditional workhorses (e.g., xanthate in sulfide flotation, fatty acid or amine in nonsulfide flotation) without a total system approach and irrespective of the need and mineralogy. This is essentially a status quo and fails to exploit the advantages and benefits offered by other chemistries. However, in order to promote other chemistries, a deeper understanding from both fundamental and application research is necessary. Unfortunately, such understanding is inadequate, again related to a narrow focus on a few reagents. It is clear that the approach, thus far, has been bottom-up or reagent-centric; for example, in sulfide flotation the question appears to be, “Given xanthate, how do we make a separation and how does it perform for all the diversity of mineralogy?”—rather than a question from a top-down approach: “Given a separation, what are the most appropriate reagents and conditions?” In this context, the industry must clearly communicate the need, not merely state the problem. When the need is clearly defined, research focus efforts will be targeted toward the need rather than reconfirming the stated problem. Therefore, the focus should really be mineralogy-driven in order to truly meet the industry needs; the appropriate reagents and conditions would then follow. This is the top-down approach that is needed for both research and practice. The industry will be better served by taking this “big picture view” and by making a commitment to support and direct relevant research. As a result of the high cost of developing new reagents, which most likely will apply only to specialized niche market segments of low volume potential, the prohibitive registration costs associated with the new environmental regulations, the increasing trend toward commoditization, and future R&D will necessarily be highly focused and driven by industry

flotation0.book Page 421 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

421

needs. Significant effort will be made to develop holistic, sophisticated, and rigorous testing protocols and scale-up methodologies to ensure successful commercialization. In order to develop sustainable, robust solution and profitable implementation in the plant, a holistic approach is necessary; it is also the most cost-effective. Future R&D will also be focused more on the application of the many new and exciting technologies that have already been partially developed but not completely evaluated. And finally, strategic partnerships between reagent developers, equipment developers, the research organizations, and plants are critical in advancing the art and science of flotation, sharing best practices and resources, and effective technology transfer. REFERENCES

Ananthapadmanabhan, K.P., and P. Somasundaran. 1984. The role of dissolved mineral species in calcite-apatite flotation. Miner. Metall. Process. 1:36. ———. 1985. Surface precipitation of inorganics and surfactants and its role in adsorption and flotation. Colloids Surf. 13:151. Aplan, F.F. 1994. Reagents for the flotation of non-sulfide ores. Pages 103–117 in Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Aplan, F.F., and S. Chander. 1988. Collectors for sulfide mineral flotation. In Reagents in Mineral Technology. Surfactant Science Series 27. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Aplan, F.F., and D.W. Fuerstenau. 1962. Principles of nonmetallic mineral flotation. Pages 170–215 in Froth Flotation, 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Avotins, P.V., S.S. Wang, and D.R. Nagaraj. 1994. Recent advances in sulfide collector development. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Booth, R.B., and W.L. Freyberger. 1962. Froths and frothing agents. In Froth Flotation, 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Cappuccitti, F.R. 1994. Current trends in the marketing of sulfide mineral collectors. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Castro, S.H., and C. Bobadilla. 1995. The depressant effect of some inorganic ions on the flotation of molybdenite. Pages 95–103 in Proceedings of the UBC-McGill Bi-Annual International Symposium on Fundamentals of Mineral Processing, 1st, Processing of Hydrophobic Minerals and Fine Coal. Edited by J.S. Laskowski and G.W. Poling. Quebec: Canadian Institute of Mining, Metallurgy and Petroleum. Chander, S. 1988. Inorganic depressants for sulfide minerals. Chapter 14 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Cotton, F.A., and G. Wilkinson. 1980. Advanced Inorganic Chemistry. New York: Wiley Interscience. Crozier, R.D. 1992. Flotation, Theory, Reagents and Ore Testing. Oxford: Pergamon Press. Cytec Industries Inc. 2002. Mining Chemicals Handbook. West Paterson, NJ: Cytec Industries. Dippenaar, A., P.J. Harris, and M.J. Nicol. 1978. The Effect of Particles on the Stability of Flotation Froths. Report. South Africa: National Institute for Metallurgy. Eigeles, M.A. 1977. Modifiers in the Flotation Process. Moscow: Nedra. Finkelstein, N.P. 1997. The activation of sulfide minerals for flotation: A Review. Int. J. Miner. Process. 52:81–120. Franzidis, J.P. 2005. Integration of science and practice in mineral flotation. Min. Mag. (May): 12–17. Fuerstenau, D.W., editor. 1962. Froth Flotation, 50th Anniversary Volume. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Fuerstenau, M.C. 1980. Page 7 in The Physical Chemistry of Mineral–Reagent Interactions in Sulfide Flotation. IC 8818. Edited by P.E. Richardson, G.R. Hyde, and M.S. Ojalvo. Washington, DC: U.S. Bureau of Mines. Gaudin, A.M. 1957. Flotation. 2nd edition. New York: McGraw-Hill.

flotation0.book Page 422 Tuesday, January 2, 2007 7:36 PM

422

FLOTATION CHEMISTRY

Harris, G.H. and R. Jia. 2000. An improved class of flotation frothers. Int. J. Miner. Process. 58:35–43. Healy, T.W., and M.S. Moignard. 1976. Page 275 in Flotation—A.M. Gaudin Memorial Volume. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Herrera-Urbina, R. 2003. Recent developments and advances in formulations and applications of chemical reagents used in froth flotation. Miner. Process. Extr. Metall. Rev. 24:139–182. Holman, B.W. 1930. Flotation reagents. Inst. Min. Metall. Bull. No. 314. Jones, M.J., and R. Oblatt, editors. 1984. Reagents in the Mineral Industry. London: The Institution of Mining and Metallurgy. Kartio, I., K. Laajalehto, E. Suoninen, S. Karthe, and R. Szargan. 1992. Technique for XPS measurements of volatile adsorbed layers: Application to studies of sulfide flotation. Surf. Interface Anal. 18:807. Klimpel, R.R. 1994. A discussion of traditional and new reagent chemistries for the flotation of sulfide minerals. Pages 59–66 in Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Klimpel, R.R., and R.D. Hansen. 1988. Frothers. In Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Leal Filho, L.S., P.R. Seidl, J.C.G. Correreia, and L.C.K. Cerqueira. 2000. Molecular modeling of reagents for flotation processes. Miner. Eng. 13(14–15):1495–1503. Leja, J. 1982. Surface Chemistry of Froth Flotation. New York: Plenum Press. Lin, K.F., and C.L. Burdick. 1988. Polymeric depressants. Chapter 15 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Mackenzie, M. 1986. Organic polymers as depressants. In Chemical Reagents in the Mineral Processing Industry. Edited by D. Malhotra and W.F. Riggs. Littleton, CO: SME. Malhotra, D. 1994. Reagents in the mining industry: Commodities or specialty chemicals? In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Mulukutla, P.S, editor. 1994. Reagents for Better Metallurgy. Littleton, CO: SME. Nagaraj, D.R. 1988. The chemistry and applications of chelating or complexing agents in mineral separations. Chapter 9 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. ———. 1994. A critical assessment of flotation agents. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. ———. 1997. Development of new flotation chemicals. Trans. Indian Inst. Met. 50(5):355–363. ———. 2000a. New synthetic polymeric depressants for sulfide and non-sulfide minerals. In Proceedings of the XXI International Mineral Processing Congress, Rome, Italy, July 23–27. Developments in Mineral Processing Series 13. Edited by P. Massacci. Amsterdam: Elsevier. ———. 2000b. Pulp redox potentials: Myths, misconceptions and practical aspects. SME Annual Meeting. Littleton, CO: SME. ———. 2003. Internal training materials. Cytec Industries. ———. 2005. Reagent selection and optimization: The case for a holistic approach. Paper presented at the Reagents ’04 Symposium, Falmouth, U.K. Miner. Eng. 18(2):151–158. Nagaraj, D.R., and P.V. Avotins. 1988. Development of new sulfide and precious metals collectors. Page 399 in Proceedings of the 2nd International Minerals Processing Symposium. Edited by Y. Aytekin, U. Ipekoglu, and M. Akdag. Izmir, Turkey: Eylul University. Nagaraj, D.R., C. Basilio, and R.H. Yoon. 1989. The chemistry and structure-activity relationships for new sulfide collectors. Pages 157–166 in Processing of Complex Ores. Edited by G.S. Dobby and S.R. Rao. Toronto: Pergamon. Nagaraj, D.R., and J.S. Brinen. 1995. SIMS study of metal ion activation in gangue flotation. Pages 253–257 in Proceedings of the XIX International Mineral Processing Congress. Littleton, CO: SME. ———. 1996a. SIMS study of adsorbed collector species on mineral surfaces: Surface metal complexes. Preprint 96-181. SME Annual Meeting, Phoenix. Littleton, CO: SME.

flotation0.book Page 423 Tuesday, January 2, 2007 7:36 PM

FLOTATION REAGENTS

423

———. 1996b. SIMS and XPS study of the adsorption of sulfide collectors on pyroxene. Colloid Surf. 116:241–249.

———. 1997. SIMS studies of mineral surface analysis: Recent studies. Trans. Indian Inst. Met. 50(5):365–376.

———. 2001. SIMS study of adsorption of collectors on pyrite. SME Annual Meeting, Denver, CO. Preprint 97-171. Littleton, CO: SME. Also published in Int. J. Miner. Process. ( June). Nagaraj, D.R., and F.S. Bruey. 2003. Reagent optimization in base metal sulfide flotation—pitfalls of standard practice. Flotation and flocculation—from fundamental to applications. Page 257 in Proceedings of the Strategic Conference and Workshop, Kona, Hawaii, August 2002. Edited by J. Ralston, J.D. Miller, and J. Rubio. Adelaide, Australia: University of South Australia, Ian Wark Research Institute. Nagaraj, D.R., A. Day, and A. Gorken. 1999. Non-sulfide minerals flotation: An overview. In Proceedings of the Symposium Honoring M.C. Fuerstenau. Littleton, CO: SME. Nagaraj, D.R., M.E. Lewellyn, S.S. Wang, P.A. Mingione, and M.J. Scanlon. 1988. New sulfide and precious metals collectors: For acid, neutral and mildly alkaline circuits. Pages 1221–1232 in Developments in Minerals Processing. Volume 10B. Amsterdam: Elsevier. Nagaraj, D.R., A.S. Rothenberg, D.W. Lipp, and H.P. Panzer. 1987. Low molecular weight polyacrylamide based polymers as modifiers in phosphate beneficiation. Int. J. Miner. Process. 20:291–308. Natarajan, R., and I. Nirdosh. 2003. Application of topochemical, topostructural, physicochemical and geometrical parameters to model the flotation efficiencies of N-arylhydroxamic acids. Int. J. Miner. Process. 71:113–129. Perkins, C.L. 1921. Amino compounds to promote ore flotation separation. U.S. Patent 1,394,639. Pradip. 1991. On the design of selective reagents for mineral processing applications. Met. Mater. Process. 3(1):15–36. Pradip, and B. Rai. 2002. Design of tailor-made surfactant for industrial application using a molecular modeling approach. Colloids Surf. A. Physicochem. Eng. Asp. 205:139–148; and 2003, Int. J. Miner. Process. 72:95–110. Pugh, R.J. 1989. Macromolecular organic depressants in sulfide flotation—A review, 1. Principles, types and applications. Int. J. Miner. Process. 25(1–2):101–130. Raghavan, S., and L.L. Hsu. 1984. Factors affecting the flotation recovery of molybdenite from porphyry copper ores. Int. J. Miner. Process. 12(1–3):145–162. Rao, S.R., editor. 2004. Surface Chemistry of Froth Flotation. New York: Kluwer Academic Press/Plenum. Richardson, P.F. 1986. Dispersants in mineral processing applications. Modifying regents. Chapter 3 in Chemical Reagents in the Mineral Processing Industry. Edited by D. Malhotra and W.F. Riggs. Littleton, CO: SME. Sheridan, M.S., D.R. Nagaraj, D. Fornasiero, and J. Ralston. 2002. The use of a factorial experimental design to study collector properties of N-allyl-O-alkyl thionocarbamate collector in the flotation of a copper ore. Miner. Eng. 15:333–340. Sutherland, K.L., and I.W. Wark. 1955. Classification of flotation reagents. Chapters 1, 2, 6, 7, 9, 14, and 18 in Principles of Flotation. Melbourne: Australasian Institute of Mining and Metallurgy. Taggart, A.F. 1945. Flotation. In Handbook of Mineral Dressing. New York: John Wiley. Vivian, A.C. 1921. Concentrating ores. GB Patent GB 186,760. Woods R. 1976. Page 298 in Flotation—A.M. Gaudin Memorial Volume. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. ———. 1996. Chemisorption of thiols on metals and metal sulfides. Mod. Aspects Electrochem. 29:401–453. Woods, R., and P.E. Richardson. 1986. The flotation of sulfide minerals—electrochemical aspects. Chapter 9 in Advances in Mineral Processing. Edited by P. Somasundaran. Littleton, CO: SME. Zachwieja, J.B. 1994. An overview of cationic reagents in mineral processing. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME.

flotation0.book Page 424 Tuesday, January 2, 2007 7:36 PM

424

FLOTATION CHEMISTRY

BIBLIOGRAPHY

Chander, S. 2003. A brief review of pulp potentials in sulfide flotation. Int. J. Miner. Process. 72:141–150. Chen, H.T., S.A. Ravishankar, and R.S. Farinato. 2003. Rational polymer design for solid–liquid separations in mineral processing applications. Int. J. Miner. Process. 72:75–86. Fuerstenau, M.C., editor. 1976. Flotation—A.M. Gaudin Memorial Volume. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Fuerstenau, M.C., and K.N. Han. 1988. Activators. Chapter 13 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Glembotski, V.A., V.I. Klassen, and I.N. Plaksin. 1963. Flotation. Translated by R.E. Hammond. New York: Primary Sources. Hackley, V.A., P. Somasundaran, and J.A. Lewis, editors. 2002. Pages 135–156 in Polymers in Particulate Systems Properties and Applications. New York: Marcel Dekker. Klimpel, R.R. 1988. The industrial practice of sulfide mineral collectors. Chapter 21 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Klimpel, R.R., R.D. Hansen, G. Garcia-Huidobro, J. Broitman. 1988. New Frother and Collector Chemistry for Sulfide Mineral Flotation. Pages 105–19 in Proceedings of Copper 87. Volume 2. Edited by A. Mular, G. Gonzalez, and C. Barahona. Santiago: University of Chile. Laskowski, J.S. 1988. Dispersing agents in mineral processing. Pages 1–16 in Developments in Mineral Processing. Volume 9, Froth Flotation. Edited by S. Castro and J. Alvarez. Amsterdam: Elsevier. Lee, J.S., D.R. Nagaraj, and J.E. Coe. 1998. Practical aspects of oxide copper recovery with alkyl hydroxamates. Miner. Eng. 11(10):929–939. Lovell, V.M. 1982. Pages 73–90 in Industrial Flotation Reagents, Principles of Flotation. Edited by R.P. King. Johannesburg: South African Institute of Mining and Metallurgy. Malhotra, D., and W.F. Riggs. 1986. Chemical reagents in the mineral processing industry. In Proceedings of the Symposium on Chemical Reagents in the Mineral Processing Industry. Littleton, CO: SME. Marabini, A.M., M. Barbaro, and V. Alesse. 1991. New reagents in sulfide mineral flotation. Int. J. Miner. Process. 33(1–4):291–306. Nagaraj, D.R., J.S. Brinen, R.S. Farinato, and J.S. Lee. 1992. A study of the interaction of decresyl monothiophosphate with noble metals using electrochemical, wetting and spectroscopic methods. Langmuir 8(8):1943–1949. Nagaraj, D.R., S.S. Wang, P.V. Avotins, and E. Dowling. 1986. Structure-activity relationships for copper depressants. Trans. Inst. Min. Metall. 95:C17–C26. Parekh, B.K., and J.D. Miller, editors. 1999. Advances in Flotation Technology. Littleton, CO: SME. Riggs, W.F. 1994. Reagent interactions. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME. Rogers, J. 1962. Page 139 in Froth Flotation, 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Somasundaran, P., and B.M. Moudgil, editors. 1988. Reagents in Mineral Technology. Surfactant Science Series 27. New York: Marcel Dekker. Taggart, A.F., T.C. Taylor, and C.R. Ince. 1929. Experiments with Flotation Reagents. Technical Publication No. 204. New York: American Institute of Mining and Metallurgical Engineers. Wottgen, E., H. Baldauf, and A. Rosenbaum. 1978. Developmental trends in the field of flotation reagents. Freiberg. Forschungsh A593:49–73. Yang, D.C. 1988. Reagents in iron ore processing. In Reagents in Mineral Technology. Surfactant Science Series 27. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Yordan, J.L., R.H. Yoon, and T. Hilderbrand. 1994. Hydroxamate vs. fatty acid flotation for the beneficiation of Georgia Kaolin. In Reagents for Better Metallurgy. Edited by P.S. Mulukutla. Littleton, CO: SME.

flotation0.book Page 425 Tuesday, January 2, 2007 7:36 PM

Sulfide Mineral Flotation Maurice C. Fuerstenau, Subhash Chander, and Ronald Woods

INTRODUCTION

Froth flotation was developed at the Broken Hill mine, Australia, a century ago with the flotation of the common sulfide mineral, sphalerite. With this development, billions of tons of worthless rock containing a variety of valuable metals became ore. Anchoring this revolutionary development was the later introduction of xanthate as a collector for sulfide minerals. Probably more than any other aspect, xanthate entrenched froth flotation’s role in the utilization of the world’s natural mineral resources. Sulfide minerals are the largest of the groups of minerals floated. Today, more than a billion tons of sulfide ores are concentrated annually throughout the world with this technology. As a group, these minerals possess a number of unusual properties that are utilized in their flotation concentration. They are conductors of electrons, they develop a potential when placed into a solution, their surfaces are readily oxidized by dissolved oxygen, and their contained metals form insoluble collector compounds with short-chained collectors. These properties enable categorization of sulfide minerals. Chander (1985) has proposed categorizing them into two classes: reversible and irreversible (passivated) sulfides. In general, the properties of reversible sulfide systems can be predicted from thermodynamic considerations, and their potential response in aqueous solution can often be predicted by the Nernst equation. Minerals that fall into this category are galena, chalcocite, and sphalerite. In contrast, irreversible sulfides are often covered with products of oxidation–reduction reaction, and the properties of these systems require close scrutiny of time effects and the history of the mineral surface. Pyrite, chalcopyrite, and arsenopyrite are examples of minerals in this system. Some sulfide minerals (e.g., molybdenite) can be floated without collector addition in the presence of air, while others can be floated in the absence of a collector under conditions in which mild oxidation of the contained sulfide to elemental sulfur or polysulfide occurs. Each of these categories is discussed separately in this chapter. C O L L E C T O R L E S S F L O TAT I O N

Several investigations have demonstrated that many sulfide minerals exhibit native floatability and can be floated without a collector. The reasons for collectorless flotation vary with mineral type. Some minerals exhibit native floatability because the surface is created by rupture of weak bonds. Such surfaces do not have high affinity for polar water molecules and attach readily to air bubbles. Other surfaces undergo chemical reactions producing elemental sulfur or a metal-deficient layer that also exhibits native floatability.

425

flotation0.book Page 426 Tuesday, January 2, 2007 7:36 PM

426

FLOTATION CHEMISTRY

Native Floatability in the Absence of Oxygen

A few sulfides are naturally floatable without a collector, whereas others are considered to have intrinsic hydrophobic character in the absence of oxidation (Ravitz 1940; Fuerstenau and Sabacky 1981). The native floatability is based on the assumption that sulfide lattice ions are expected to be weakly hydrated and do not interact strongly with water molecules. The critics of this hypothesis argue that sulfide minerals are thermodynamically unstable and sufficient oxygen remains in the system to cause oxidation (Miller 1988), presumably leading to the formation of elemental sulfur (a hypothesis originally proposed by Wark [1938] and later suggested by several others). Buckley, Hamilton, and Woods (1985) have proposed that a metal-deficient layer is formed under mildly oxidizing conditions; however, the minimum quantity of the hydrophobic entity (elemental sulfur or metal-deficient layer) needed for complete flotation has yet to be established. Other reasons for native floatability of sulfides have been postulated in the past. Gaudin (1957) considered that surfaces formed by rupture of van der Waals bonds are naturally hydrophobic. Chander and Fuerstenau (1972) postulated that molybdenite retains its hydrophobic character because the product of oxidation of the lattice metal ion is a soluble anion, which does not have the same hydration characteristics as other metal cations. In other words, many sulfides may acquire surface hydrophilic character through dissolution and readsorption of hydroxylated cations. These arguments suggest that the reasons for the floatability of sulfide minerals observed in the absence of collectors may vary from mineral to mineral. In most cases, the flotation behavior depends on the nature of the surface, and this might be readily altered by chemical/electrochemical reactions. Collectorless Flotation Under Oxidizing Conditions

Many investigators consider that sulfide minerals can be floated under mild to modest oxidizing conditions, but the reasons given for such a response vary considerably. As early as 1949, Plaksin (1949) proposed that adsorbed oxygen decreases surface hydration, thereby making a mineral hydrophobic. In contrast, Heyes and Trahar (1977) observed that modest oxidation is required for collectorless flotation of chalcopyrite. Although several investigators have suggested that elemental sulfur is responsible for flotation, quantitative correlations have been difficult to establish. This aspect has been discussed in detail by Heyes and Trahar (1977). One reason for the lack of correlation between experimental results and theoretical calculations is that unit activity of elemental sulfur is assumed. In a study of oxidation of sulfide ions at a gold electrode, Chander and Briceno (1989) distinguished three forms in which sulfur could be present in its zero oxidation state. The three forms of sulfur are atomic sulfur, S0, sulfur in polysulfides (which in fact are a mixture of S in two oxidation states), and elemental sulfur in S8 form. It was observed that both polysulfides and elemental sulfur make the surface hydrophobic. In another X-ray photoelectron spectroscopy (XPS) study, a metal-deficient sulfide layer was proposed by Buckley, Hamilton, and Woods (1985). In addition, electrochemical studies show that the iron oxides which form when chalcopyrite oxidizes under certain conditions do not adhere to the surface (Pang and Chander 1992). This observation agrees with previous findings that a metal-deficient layer, particularly an iron-deficient layer for chalcopyrite, is responsible for flotation without a collector. These observations were further confirmed by Chander and Khan (2000) who measured contact angles of chalcopyrite as a

flotation0.book Page 427 Tuesday, January 2, 2007 7:36 PM

SULFIDE MINERAL FLOTATION

427

function of potential for several pH values. The contact angle was nonzero under mildly oxidizing conditions associated with formation of an iron-deficient layer at the surface. Collectorless Flotation in the Presence of Air

Sulfur-bearing minerals that exhibit native floatability in the presence of air are sulfur, molybdenite, and stibnite. Chander, Wie, and Fuerstenau (1975) suggested that native hydrophobicity of a solid depends on the crystal structure, the nature of charged hydrophilic sites present on the mineral surface, and the anisotropic behavior of the mineral surface. Sulfur occurs in several forms, the usual form consisting of molecules in the form of puckered rings of eight sulfur atoms. In the crystal, the sulfur rings are held together by van der Waals bonds that readily cleave to produce the surface of sulfur particles. Similarly, in the case of molybdenite (MoS2), the crystal structure is such that a layer of molybdenum atoms is sandwiched between layers of sulfur atoms on each side. The sandwich layers are held together by van der Waals bonds that give rise to a well-defined cleavage plane for molybdenite. The surface of molybdenite created by the cleavage plane is highly hydrophobic, and the plane perpendicular to it is only weakly hydrophobic (Chander and Fuerstenau 1972). The atomic structure of stibnite (Sb2S3) consists of atoms of Sb bonded with atoms of S to form an unusual chain structure where Sb and S atoms are bound in infinite hexagonal sheets. The hexagonal sheets are held together by van der Waals bonds. The breakage of these bonds produces a cleavage plane that gives rise to the hydrophobic properties of this mineral. Although these materials are naturally floatable, oftentimes hydrocarbons or oily collectors are added to increase their rate of flotation. Collectorless Flotation in the Presence of Sulfide Ion

Chalcopyrite and sphalerite can be floated in the presence of sodium sulfide in the absence of collectors (Yoon 1981). Yoon considered that oxidizing conditions are not necessary, and the role of sodium sulfide (Na2S) is to clean the surface of chalcopyrite (i.e., free it of oxidation products). Because oxidation of sulfide ion is catalyzed by sulfide minerals (Mitrofanov, Kurochkina, and Sokolava 1954), and no effort was made in Yoon’s experiments to exclude oxygen, it is likely that oxidation products, possibly polysulfides, might have played a role. Indeed, recovery in the collectorless flotation of pyrite in the presence of sulfide ion was shown both by Walker, Walters, and Richardson (1984) and by Heyes and Trahar (1984) to correlate with oxidation of hydrosulfide ion to elemental sulfur on the mineral surface. Figure 1 shows the relationship between flotation recovery of chalcopyrite and potential (relative to the standard hydrogen electrode, or SHE) reported by Heyes and Trahar (1977), Gardner and Woods (1979), and Luttrell and Yoon (1984). It can be seen that flotation only occurs in the higher potential region and that there is good agreement with respect to the potential of the onset of floatability. In the first and last of these works, potential was controlled by the addition of redox reagents. Gardner and Woods (1979), on the other hand, controlled potential with a potentiostat; a large particle size (295 μm) was used in the investigations of these authors to maintain contact with the feeder electrode, and this accounts for the lower recoveries obtained. These authors showed that the onset of flotation correlated with the commencement of oxidation of the chalcopyrite surface. Figure 2 shows the results reported by Walker, Walters, and Richardson (1984) in which voltammetric investigations of sulfide ion oxidation on pyrite are compared with

flotation0.book Page 428 Tuesday, January 2, 2007 7:36 PM

428

FLOTATION CHEMISTRY

100

Flotation Recovery, %

80

60

40

20

0 –0.6

–0.4

–0.2

0.0

0.2

0.4

0.6

0.8

Potential, V (vs. SHE) Data from Heyes and Trahar 1977 Data from Gardner and Woods 1979 Data from Luttrell and Yoon 1984

FIGURE 1 Potential dependence of flotation recovery of chalcopyrite in the absence of collectors

Voltammogram

Contact Angle

Flotation

–1.2

–1.0

–0.8

–0.6

–0.4

–0.2

0.0

0.2

0.4

0.6

Potential, V (vs. SHE)

Source: Walker, Walters, and Richardson 1984.

FIGURE 2 Correlations of potentials for sulfur deposition, initiation of hydrophobic contact angle, and flotation of pyrite in 10–2 M HS– at pH 9.2

flotation0.book Page 429 Tuesday, January 2, 2007 7:36 PM

SULFIDE MINERAL FLOTATION

429

measurements of contact angle and flotation recovery. It can be seen that the onset of hydrophobicity correlates with the beginning of an anodic current, which is due to oxidation of HS– to S0. E L E C T R O C H E M I S T R Y I N S U L F I D E M I N E R A L F L O TAT I O N

It was not until midway through the 20th century that the electrochemical nature of the interaction of flotation collectors with sulfide mineral surfaces was recognized. Nixon (1957) was the first to propose that the collector being adsorbed as an uncharged species was the result of an anodic process, and that the electron transferred to the mineral in this process was returned to the solution phase through the simultaneous cathodic reduction of oxygen. Several mechanisms for collector–mineral interaction had been previously put forward and strongly promoted by their advocates, but each contained difficulties. Nixon saw that the electrochemical approach explained the observed phenomena and that it also solved the problems inherent in the previous theories and allowed them to be reconciled. An important consequence of the electrochemical mechanism is that flotation recovery is a function of the electrochemical potential across the mineral–solution interface. As will be discussed later, measurements of potential, often referred to as Eh, are now regularly used in the development of flotation strategies and in appraising plant performance. Reactions that Impart Hydrophobicity to Sulfide Mineral Surfaces

It is the anodic oxidation reaction involving the collector that imparts hydrophobicity to the sulfide mineral surface. The cathodic reaction simply allows that process to proceed. Three types of hydrophobic species have been identified for different minerals and thiol collectors (Woods 1984). These are (1) a chemisorbed thiol in which the anodic reaction for a thiol ion (X–) can be represented by: X – → X ads + e –

(EQ 1)

2X – → X 2 + 2e –

(EQ 2)

(2) a dithiolate:

and (3) a metal thiol compound in which the metal atoms are derived from the mineral, for example: X – + MS → MX + S + e –

(EQ 3)

Reaction 3 can also occur by prior oxidation of the mineral followed by ion exchange to form the thiol compound, for example: MS + H 2 O → MO + S + 2H + + 2e – ; MO + 2X – → 2MX –

(EQ 4)

With uncharged thiols, such as the copper sulfide mineral collector, O-isopropyl-N-ethylthionocarbamate, chemisorption and metal compound formation both involve the release of a hydrogen atom from the collector molecule to form a hydrogen ion in solution (Woods and Hope 1999). In this way, a charge transfer process can occur with both the collector and the resulting surface species being neutral.

flotation0.book Page 430 Tuesday, January 2, 2007 7:36 PM

430

FLOTATION CHEMISTRY

Electrochemical studies of the chemisorption of thiols on sulfide minerals have been augmented by infrared and electron spectroscopic investigations in order to provide information on the chemical composition of the adsorbate and its bonding to the surface (Buckley, Hope, and Woods 2003). These techniques have distinguished among the initial chemisorbate, the metal compound, and the dithiolate. Ultraviolet-visible (UV-Vis) spectroscopy has also been applied to determine the extraction of a collector from the solution when chemisorption takes place, and surface-enhanced Raman scattering (SERS) spectroscopy of the adsorption of collectors on the coinage metals has shown that the integrity of the collector molecule is retained in the chemisorbed molecule (Woods, Hope, and Watling 2000). All of these spectroscopic observations support a model of chemisorption in which the thiol is bonded through a sulfur atom to a metal atom in the mineral surface retained in the mineral lattice. This contrasts with compound formation in which the metal atom moves out of the mineral lattice and forms a separate metal collector phase. It may be expected that chemisorption would not occur on a preoxidized surface but rather that the metal collector compound would be formed by Reaction 4. XPS and timeof-flight secondary ion mass spectrometry (TOF-SIMS) investigations of the galena/butylxanthate and the chalcocite-diethyl dithiophosphate systems have shown (Buckley et al. 2003), however, that the products of interaction of the preoxidized mineral with the collector are the same as those observed with freshly exposed surfaces. It was concluded that the preoxidized mineral surface reacts with the collector by ion exchange and that this is followed by reorganization of the metal thiolate in the surface layer to form the thermodynamically more favorable species, the chemisorbed thiol. Voltammetric studies on some systems do not show a prewave characteristic of chemisorption. In particular, the interaction of xanthates with pyrite only gives rise to currents in the potential region expected for the formation of dixanthogen (Richardson and Walker 1985). XPS studies (Szargan, Karthe, and Suoninen 1992; Leppinen et al. 1995) have confirmed the formation of dixanthogen on pyrite, and this finding is in agreement with infrared studies (Fuerstenau, Kuhn, and Elgillani 1968; Leppinen et al. 1995). Other Fourier transform infrared (FTIR) spectroscopy investigations (Fornasiero and Ralston 1992; Wang 1995), however, have indicated that an iron xanthate species can also be produced in this system. Potential Dependence of Hydrophobicity

Numerous studies have been conducted to determine how each of the species identified using electrochemical procedures and supporting spectroscopic techniques influences the wettability of the mineral surface. This has involved determining wettability as a function of the potential across the mineral–solution interface. Wettability has been determined by measuring contact angles and flotation recovery as a function of potential. In flotation, nitrogen is used as the carrier gas, and the potential is controlled either with a potentiostat or by the addition of redox reagents to the flotation pulp whereby the potential is measured with a platinum electrode inserted into the pulp. The relationship between flotation recovery and potential for chalcocite in the presence of ethyl xanthate is shown in Figure 3. In the case of the data reported by Richardson and Walker (1985), potential of a particulate mineral bed electrode was controlled with a potentiostat. Heyes and Trahar (1979) used a modified Denver flotation cell and controlled the potential by the addition of the redox reagents dithionite and hypochlorite. Basilio, Pritzker, and Yoon (1985) used a Hallimond tube cell and controlled the potential in the

flotation0.book Page 431 Tuesday, January 2, 2007 7:36 PM

SULFIDE MINERAL FLOTATION

431

100

450

150 60

0

–150

40

Current, μA

Flotation Recovery, %

300 80

–300 20 –450

0

–600 –0.4

–0.2

0.0

0.2

Potential, V (vs. SHE) Data from Richardson and Walker 1985 Data from Heyes and Trahar 1979 Data from Basilio, Pritzker, and Yoon 1985 Voltammogram

FIGURE 3 Relatiohship between flotation recovery and potential for chalcocite in the presence of ethyl xanthate

same manner as Heyes and Trahar. It can be seen that the potential dependence of flotation recovery found by the three groups is in general agreement. A voltammogram recorded from the chalcocite bed (Richardson and Walker 1985) is also presented in Figure 3. The voltammogram shows the prewave region between –0.3 and 0 V, which corresponds to chemisorption. The anodic current above 0 V is due to copper xanthate formation. It can be seen that the onset of flotation occurs in the prewave potential region. Thus, chemisorbed xanthate initiates flotation. Flotation continues at an efficient rate when copper xanthate is also formed, but the presence of the bulk compound is not necessary for flotation. Flotation recovery and surface coverage as a function of potential for three mineral/ collector systems are presented in Figure 4. The flotation recoveries for galena/ethyl xanthate in Figure 4a is from Heyes and Trahar (1979), and the surface coverage is from Buckley and Woods (1991). For chalcocite/ethyl xanthate in Figure 4b, recovery is from Leppinen, Basilio, and Yoon (1989), and coverage is from Woods, Young, and Yoon (1990). The flotation recovery for the chalcocite/diethyldithiophosphate system in Figure 4c is from Woods, Kim, and Yoon (1993), and the surface coverage is from Buckley and Woods (1993). It can be seen that for each system, the initiation of flotation corresponds to the chemisorption region. Significant flotation recovery is observed at low thiol coverage. About 50% recovery occurs at a fractional coverage of tertiary > secondary > primary, corresponding with the order of head-group radius from quaternary to primary amine (Nishimura et al. 2000). Only in the last decade, with the development of the AFM soft-contact imaging technique, has direct observation of “wormlike” micelle structures been made at atomically smooth mica surfaces. Before that, most reported research discussed the cationic ammonium surfactant adsorption at mica surfaces as being in a monolayer or multilayer conformation (Rutland, Waltermo, and Claesson 1992). Since 1995, a number of AFM soft-contact images in the literature have provided direct evidence that meandering micelles are formed at mica surfaces at concentrations below the solution CMC values of cationic amine molecules. Figure 24 shows the micellar structure for CTAB at mica surfaces. For CTAB at 0.1-mM concentration, the meandering linear structures are believed to be cylindrical surface micelles, whereas for DTAB, the CMC value is about 15 mM; at a concentration of 20 mM, a micellar structure similar to that shown in Figure 25 was observed by Ducker and Wanless (1996). The meandering micellar structures of CTAB and DTAB at mica surfaces are different from the micellar structures of CTAB and DTAB at the atomically smooth graphite surface, where linear and parallel hemicylindrical micellar structures were observed. This difference indicates that the topographic structure of the substrate template has a significant influence on the formation of surface micelles. Also, these AFM image analyses are supported by measurement of the diameter of these micelle structures, about 5 nm, which corresponds to the total length of two surfactant molecules. SDS bears a negatively-charged head group. SDS is not expected to adsorb at the negatively-charged surface of mica, and no surfactant surface structures were observed in the case of SDS. It is thus concluded that electrostatic interaction plays an important role for the adsorption of cationic surfactants at muscovite mica surfaces. In contrast, at a smooth SiO2 surface, spherical micelle structures are observed for CTAB below the solution CMC (0.1 mM) as shown in Figure 25.

flotation0.book Page 484 Tuesday, January 2, 2007 7:36 PM

484

FLOTATION CHEMISTRY

300

300

200

200

100

100

0

0 0

100

200

300

FIGURE 24 Micellar structures of CTAB at mica surfaces (measured in nanometers)

0

100

200

300

FIGURE 25 Soft-contact AFM images (measured in nanometers) for a SiO2 surface in 1.0-mM CTAB solutions. Spherical surface micelles are observed at the silica surface with a diameter of approximately 6.2 nm.

New Flotation Chemistry. In addition to studies of adsorption and molecular conformations of cationic amines at mica surfaces, the separation of mica from galena using dextrin as a depressant was investigated by Rath and Subramanian (1998). From pH 6.4 to 11.3, the adsorption density of dextrin for galena was observed to be quite high compared to that for mica. Both adsorption isotherms exhibited Langmuir behavior. Flotation results showed that dextrin is a depressant for galena but not for mica. In alkaline solution at about pH 11.0, maximum adsorption of dextrin at the galena surfaces occurs. Separation of mica from galena is achieved above pH 8.5. Because the adsorption of amine by quartz is a well-established physical adsorption process, the zeta potential of quartz particles is known to play a critical role in adsorption (Gaudin and Fuerstenau 1955b). Figure 26 shows the correlation of flotation recovery of quartz with DAA as collector with the corresponding zeta potentials at pH 5.0 and 9.8 (Takeda and Usui 1987). Complete recovery of quartz particles corresponds well with the PZR values at two different pH values. Takeda and Usui found that at pH 5.0, complete recovery of quartz is achieved at an adsorption density of DAA of only a few percentages of monolayer, whereas at pH 9.8, complete flotation requires a surface coverage of DAA of more than a complete monolayer. Quartz cannot be easily recovered using anionic collector without activation in neutral and basic solution because quartz particles are negatively charged. Multivalent metal cations have been found to serve as activators in the flotation of quartz using anionic surfactants such as carboxylates, sulfates, and sulfonates. M.C. Fuerstenau and Palmer (1976) have systematically studied the adsorption of cations, such as Fe3+, Al3+, Pb2+, Mg2+, Mn2+, and Ca2+, at quartz surfaces and their influence on quartz flotation response with anionic surfactants such as sulfonate and lauric acid as collector. As shown in Figure 27, activation of quartz occurs in the pH range in which metal ions hydrolyze to hydroxy complexes. The pH at which the first hydroxy complex becomes significant in concentration is shown, together with the edges outlining minimum values in pH for flotation for each of the cations. Flotation occurs, then, in a similar fashion to that with which insoluble oxides and silicates are floated by surface reaction with anionic collectors (Figures 8 and 9). The complete range of pH in which flotation is obtained is shown in Figure 28. The absence of flotation above the

flotation0.book Page 485 Tuesday, January 2, 2007 7:36 PM

485

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

100

Recovery, %

80 60 40 pH 9.8 pH 5.0

20 0 10–5

10–4

10–3

–80 pH 9.8 pH 5.0

Zeta Potential, mV

–60 –40 –20 0 20 40 10–5

10–4

10–3

Equilibrium Concentration of DAA, M

FIGURE 26 Flotation recovery and zeta potential of quartz at pH 5.0 and 9.8 as a function of the equilibrium concentration of DAA in the presence of 1 × 10–3 M KCl pH of Hydroxy Complex Formation FeOH2+

AlOH2+

PbOH+

MnOH+

MgOH+

CaOH+

100

Flotation Recovery, %

80

60

Fe3+

Al3+

Pb2+

4

6

Mn2+

Mg2+

Ca2+

40

20

0 0

2

8

10

12

14

pH Source: M.C. Fuerstenau and Palmer 1976.

FIGURE 27 Minimum flotation edges of quartz as a function of pH (Conditions: 1 × 10–4 M sulfonate, 1 × 10–4 M metal ion)

flotation0.book Page 486 Tuesday, January 2, 2007 7:36 PM

486

FLOTATION CHEMISTRY

100

Flotation Recovery, %

80 Pb2+

Mn2+

60

Fe3+ 40

20

0 0

2

4

6

8

10

12

14

pH

Source: M.C. Fuerstenau and Palmer 1976.

FIGURE 28 Flotation recovery of quartz as a function of pH (Conditions: 1 × 10–4 mol/L sulfonate, 1 × 10–4 mol/L metal ion) 100 CaCl2 Addition, mol/L 5 × 10–4 3 × 10–4 1 × 10–4

Flotation Recovery, %

80

60

40

20

0 10–5

Ca5

Ca3

Ca1

10–4

10–3

Lauric Acid Addition, mol/L

Source: M.C. Fuerstenau and Cummins 1967.

FIGURE 29 Flotation recovery of quartz as a function of lauric acid and calcium chloride (CaCl2) additions at pH 11.5

maximum flotation edge is due to the stability of the metal hydroxide relative to the metal collector salt and below the minimum flotation edge to an insufficient amount of hydroxyl complex required for flotation. Precipitation of the metal collector has been involved in some of these systems. As shown in Figure 29, flotation of quartz occurred only after precipitation of calcium laurate occurred in solution. Arrows indicate the activity of laurate at which calcium laurate precipitated in each system. The critical effect of collector addition relative to that of activator has been shown by Gaudin and Rizo-Patron (1942) and Cooke and Digre (1949). In systems where

flotation0.book Page 487 Tuesday, January 2, 2007 7:36 PM

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

487

precipitation of the metal collector has occurred, at constant collector addition, increasing the metal ion addition by an order of magnitude reduced the minimum pH at which flotation occurred by one unit. James and Healy (1972) provided an excellent analysis of these systems in terms of competing energy changes as an ion approaches an interface. The attractive energy is the electrostatic free energy, possibly supplemented by short-range forces. The opposing energy involves the secondary solvation energy changes as parts of the solvation sheath are rearranged or replaced. Their analysis shows that solvating energy change is much more favorable for a hydroxy complex than for a hydrated divalent ion with a solid of low dielectric, such as quartz. Hence, the overall free energy of adsorption will be much more favorable. Flotation separation of Na-feldspar from K-feldspar has been studied by Demir, Abramov, and Celik (2001). Because of the similarities in the physical and physicochemical properties of Na- and K-feldspars, their flotation separation was considered very challenging. However, it was found that with addition of monovalent salt, such as NaCl or KCl, and bivalent salts, such as CaCl2 or BaCl2, satisfactory separation using froth flotation technique can be achieved (Demir, Abramov, and Celik 2001; Demir et al. 2003). The fundamental nature of this separation was considered to be the result of ion exchange between the added cations and those in the crystal lattice. Talc and Pyrophyllite Surface Structure and Properties. Both talc and pyrophyllite have three-layer sheet structures. In the case of talc [Mg3(Si4O10)(OH)2], no replacement of silicon by aluminum occurs. The brucite layer has the positive charges to neutralize the two hexagonal networks of silica tetrahedra to give the crystal the sandwich construction shown in Figure 30. There is no net charge on the three-layer sheets. The three-layer sheets are held together only by van der Waals forces so that the layers are capable of slipping easily over one another. Therefore, talc is a very soft and smooth material. As shown in Figure 31, pyrophyllite has a threelayer sheet structure similar to talc, in which the magnesium atoms are replaced by aluminum atoms to neutralize the sheet by forming a gibbsite layer sandwiched between two tetrahedral sheets. In the case of pyrophyllite [Al2(Si4O10)(OH)2], each three-layer sheet is also

FIGURE 30 Talc layer structure showing silica tetrahedra and magnesia octahedra

FIGURE 31 Pyrophyllite layer structure showing silica tetrahedra and alumina octahedra

flotation0.book Page 488 Tuesday, January 2, 2007 7:36 PM

488

FLOTATION CHEMISTRY

electrically neutral, and the bases of these sheets are held together only by weak van der Waals forces. The lack of surface polarization provides talc and pyrophyllite with natural floatability. Surfactant Adsorption. The faces of talc and pyrophyllite are naturally hydrophobic, as expected, based on their crystalline structure. However, the edges of the talc and pyrophyllite particles are created by the breakage of the Si–O or Al–O bonds, and, consequently, the edges are hydrophilic. The ratio of hydrophilic/hydrophobic surface sites is expected to affect the electrokinetic properties and the flotation behavior of these minerals. The PZC for pyrophyllite is reported as pH 2.4 (Hu, Liu, and Xu 2003), whereas for talc the PZC is reported to be pH 3.0 (M.C. Fuerstenau, Lopez-Valdivieso, and Fuerstenau 1988). Because talc has natural floatability, collectorless flotation of talc is expected to be achieved. Yehia and Al-Wakeel (1999) achieved about 60% recovery of talc using only frother in laboratory flotation experiments at neutral pH. By column flotation using methyl isobutyl carbinol (MIBC), an 83% recovery of talc was reported by Kho and Sohn (1989). Erdemoglu and Sarikaya (2002) reported that 60% to 70% of pyrophyllite can be recovered using diethylhexanol as frother. 3+ 3+ 3+ New Flotation Chemistry. The effect of the hydrolyzed Al , Cr , and Fe on hydrophobic talc particles was examined by Fuerstenau and colleagues with electrokinetic and microflotation studies (M.C. Fuerstenau, Lopez-Valdivieso, and Fuerstenau 1988). At a concentration of 1 × 10–4 M, all the cations were found to reverse the zeta potential of talc from negative to positive. Because talc is comprised of polar and nonpolar surfaces, the change in zeta potential is believed to be due to the adsorption of metal hydroxy complex species at the polar edges. Microflotation results showed that talc flotation is depressed in a specific pH range, where the polyvalent cations precipitate as metal hydroxides. From the analysis of species distribution diagrams, it was concluded that uptake of metal hydroxide precipitate occurs on both the polar and nonpolar surfaces at pH values below the PZC of the hydroxides, resulting in the loss of natural hydrophobicity. Depression with Organic Colloids. Talc depression is important because in many cases talc is considered as a gangue mineral in the flotation separation. This is particularly important in the case of sulfide flotation. For example, talc is a common component of the gangue minerals associated with the platinum-group-metal sulfide ores in South Africa. Because of its natural hydrophobicity, significant amounts of talc minerals are entrained in the froth phase, resulting in a decrease in the concentrate grade. Parolis, Groenmeyer, and Harris (2004) investigated the effect of metal cations on the adsorption of carboxymethylcellulose. They found that the presence of Ca2+ or Mg2+ reduces the electrostatic repulsion between the carboxymethylcellulose and the talc edge, resulting in a higher adsorption density of carboxymethylcellulose at talc surfaces. The high ionic strength also has the effect of allowing the polymer to form in a more tightly coiled conformation so that a higher adsorption density is achieved. These researchers also found that an increase in the polymer molecular weight did not significantly affect adsorption at the talc surfaces. Poly(oxyethylene)alkyl ethers (C12(EO)n, n = 8, 5, 3) and polysaccharides such as guar gum and carboxymethylcellulose have been studied as depressants for talc by other research groups (Pugh and Tjus 1990; Pugh 1991; Shortridge et al. 2000). Three poly(oxyethylene)dodecyl ether derivatives with a definite number of oxyethylene units (n = 8, 5, or 3) were found to have an increasing depression effect on talc in the order of C12(EO)8 > C12(EO)5 > C12(EO)3, which can be related to size of the ethoxy unit. The flotation recovery minimum was reached at a critical concentration for all three derivatives. By careful control of the depressant concentration, a

flotation0.book Page 489 Tuesday, January 2, 2007 7:36 PM

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

489

Nonpolar Region

Polar Region

FIGURE 32 Geometry of dextrin molecule using MP3 simulation showing polar and nonpolar molecular regions

satisfactory separation of galena from talc could be achieved, as revealed from the flotation experiments (Pugh and Tjus 1990). The predominant interaction forces between poly(oxyethylene)alkyl ether and talc were proposed as hydrophobic interactions and/or hydrogen bonding occurring at faces and edges. Recent research has shown the significance of hydrophobic bonding for dextrin adsorption at the nonpolar talc surface. Figure 32 shows the dextrin molecular structure optimized using PM3. The OH group and oxygen atoms are intended to align at one side of the dextrin molecule. This side of the molecule is expected to have strong polar properties though the other side of the molecule reveals a nonpolar region. The adsorption of dextrin molecules at talc surfaces may be explained as a result of hydrophobic attraction between dextrin molecules and talc surfaces. Shortridge et al. (2000) reported about the study of polysaccharide depressants on the flotation of talc. They found that guar was a much more effective depressant than carboxymethylcellulose. Also, as the molecular weight of the guars increased, their depressing ability on talc increased accordingly. SUMMARY AND CONCLUSIONS

Some substantial research progress has been made during the last two decades. First, reverse flotation of aluminosilicate minerals from bauxite ores has been used to remove aluminosilicates such as kaolinite, pyrophyllite, and illite. The surface characteristics and flotation behavior of kaolinite, pyrophyllite, and illite have been studied and analyzed by Hu’s group. New collectors and depressants were developed with success. A satisfactory product can be obtained for the further Bayer process for recovery of diasporic bauxite resources. Second, the surface micelle structure at the surface of mica has been extensively studied using the AFM soft-imaging technique and interaction force measurements. These studies provide straightforward information about how the surfactant molecules adsorb and pack at the mineral surfaces and what the driving forces are for adsorption. Visualization of micellar structure at atomically smooth mica surfaces and at the surface of silicon wafers verified the long-held assumption that surfactant molecules are intended to form a micelle structure at the mineral surfaces and a good flotation response could be achieved at the concentration of surface CMCs, which are usually smaller than the solution CMCs. Third, progress in the depression of naturally hydrophobic talc has been made. Many types of long-chain organic molecules or polymers and the influence of some cations on the adsorption of these large molecules at talc surfaces have been investigated. However, because of the complexity of the depression process, new discoveries and techniques are required to efficiently depress talc particles. This review shows the relationship between surface properties of aluminosilicates and their flotation behavior. The hydrophobic or hydrophilic nature

flotation0.book Page 490 Tuesday, January 2, 2007 7:36 PM

490

FLOTATION CHEMISTRY

of an aluminosilicate is determined by its intrinsic crystal chemistry properties of the cleavage surface in water. Because a wide range of crystal structures are present in aluminosilicate minerals, studies of their surface chemistry and crystal structure provide the basis for understanding their flotation chemistry. Traditionally, aluminosilicates were considered as the flotation gangue materials. With the developments and requirements of economy and technology, several aluminosilicate minerals have been recovered worldwide during the last two decades. The further study of surface chemistry and new flotation technology will be expected and required for better utilization of aluminosilicate resources.

Kaolin Flotation Practice C. Basilio and S. Mathur

BAC K G RO U N D

Kaolin is a naturally occurring, relatively white clay material composed predominantly of kaolinite, a hydrous aluminum silicate mineral. Kaolinite can be formed via residual weathering, hydrothermal alteration, or sedimentation. Kaolins are generally classified as primary or secondary deposits (Prasad, Reid, and Murray 1991). Primary kaolin is formed in situ, usually by alteration of crystalline rocks like granite or gneiss. The alteration can be by surface weathering or action of hydrothermal fluids. A good example of primary kaolin is the deposit in Cornwall, England. On the other hand, secondary kaolin is a sedimentary mineral that has been eroded, transported, and deposited as beds or lenses associated with other sedimentary rocks. Most of the secondary deposits, such as those found in Georgia (United States) and Brazil, were formed by the deposition of kaolinite that had been formed elsewhere. The whiteness, brightness, and platy shape of kaolin makes it useful as coatings and fillers for paper. Kaolin gives opacity, gloss, and good printability to coated paper and board. Other properties of kaolin, such as chemical inertness, ultrafineness, platy shape, and so forth, make it an excellent filler, carrier, opacifier, or diluent in a variety of industrial products such as paints, fiberglass, plastics, cement, rubber, fertilizers, textiles, pharmaceuticals, cosmetics, and detergents. In 2003, about 25 Mt of kaolin were produced in the world with the United States being the largest producer. Production in the United States was about 7.9 Mt and valued at $960 million (Virta 2004). The average value of U.S. kaolin produced was estimated to be $212 per ton. About 50% of the world’s kaolin production was for the paper industry. In the United States, 55% of total domestic sales for kaolin was for paper applications, 16% for refractory products, 5% for fiberglass, 4% for catalyst, 3% for rubber, and 17% for other applications. Generally, kaolin clay, as mined, contains a variety of mineral impurities such as iron oxides, titaniferous minerals, silica, feldspar, mica, sulfides, and sometimes organic matter such as lignite, graphite, and so forth. Crude clays require purification from associated impurities in order to make them acceptable for the uses mentioned previously. The colored impurities, primarily the titaniferous minerals (iron-anatase), must be reduced to acceptable levels to produce high-brightness clay products. Several process applications are currently used on an industrial scale. The first stage of processing includes washing and degritting to remove the coarse impurities, such as silica and mica. Next, the discoloring impurities are

flotation0.book Page 491 Tuesday, January 2, 2007 7:36 PM

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

491

removed by several beneficiation processes, such as froth flotation, high-gradient magnetic separation, selective flocculation, and leaching. High-gradient magnetic separation (HGMS) has been widely used in the kaolin industry to remove colored impurities, mainly titaniferous minerals (iron-anatase). HGMS involves the use of a magnetic field to remove the impurities with magnetic susceptibility, such as anatase, rutile, hematite, mica, and pyrite. However, this method is not very effective for submicron particles and is limited in its capability to produce high-brightness kaolin products. A major breakthrough in HGMS technology is the development of the superconducting HGMS (Clark 1985). Using superconducting technology, extremely high magnetic fields (up to 50 kilogauss) can be achieved with very minimal power consumption. Selective flocculation is an effective process for recovering fine to ultrafine kaolin minerals that respond poorly to conventional beneficiation processes, such as flotation and magnetic separation. This process has been successful in beneficiating fine-grained kaolins and involves activating impurities with polyvalent cations (Maynard, Skipper, and Millman 1968; Mercade 1972, 1975; Sheridan 1974), conditioning with ammonium salt (Shi 1986), or fatty acid and polyvalent cations (Behl, Willis, and Young 1996), and then selectively flocculating the impurities with anionic polymers. The drawback is the relatively low recoveries obtained in this process. The leaching of kaolin clay involves the use of iron-reducing reagents, such as zinc or sodium hydrosulfite. This leaching method is limited to removing iron contaminants only. Typically, this process is carried out at very acidic conditions. Other known leaching reagents and/or processes are not economical for removing titanium impurities. Froth flotation is regarded as one of the most effective methods in the kaolin industry for removing titaniferous discoloring impurities from kaolin clays. Typically, kaolin clay to be beneficiated by froth flotation is first blunged in the presence of a dispersant and pH modifier to prepare a stable slurry. Next, the slurry is conditioned with a collector to render the impurity minerals hydrophobic. The conditioned impurities are then removed by flotation via the attachment of hydrophobic impurities to air bubbles. There are a number of kaolin flotation technologies using a variety of process strategies to make the impurities hydrophobic. The discoloring impurities found in kaolin clays are mainly titaniferous minerals, specifically iron-anatase and iron oxides. Although commercial anatase is pure white with brightness close to 100, the iron-anatase naturally occurring in kaolin is beige to dark reddish brown in color. This is due to the substitution of iron within the lattice of the anatase at levels typically less than 5%. Therefore, removal of the iron-anatase from kaolin results in an increase in the brightness of kaolin clays. The iron oxides are typically removed by acid leaching in reducing environments. Brightness is a measure of the blue reflectance of pigments. The brightness scale is calibrated with respect to pure magnesium carbonate (MgCO3), which has been arbitrarily assigned a value of 100. Brightness is measured at an effective wavelength of 457 nm and is distributed throughout the spectral range of 400 to 500 nm. Kaolin clays, unlike most minerals beneficiated by flotation, are very fine with particle sizes of about 50% to 90% passing 2 microns. It is known from the literature that flotation of most minerals is optimum in the particle size range of 100 to 10 microns (Wills 1992). Below 10 microns, flotation is considered very difficult, although there has been no evidence of a critical size below which particles will not respond to flotation. In the case of kaolin flotation, the gangue minerals (i.e., iron-anatase) are actually the ones floated, so the

flotation0.book Page 492 Tuesday, January 2, 2007 7:36 PM

492

FLOTATION CHEMISTRY

majority of the fine particles (i.e., kaolinite) are hydrophilic and remain in suspension. The reverse flotation of kaolin may partly help explain the success found in the flotation of such fine particles. Flotation Strategies

All kaolin deposits do not respond to flotation in the same way. The presence of various trace impurities, particle size distribution of kaolin and iron-anatase, and the extent of liberation of the coloring impurities are key factors that govern the selectivity and yield in the flotation of kaolin clays. Coarse soft clays found in the Middle Georgia area of the United States respond better to flotation than the East Georgia kaolin deposits. The East Georgia kaolin deposits have a finer particle size, higher surface area, and higher TiO2 content than those of the Middle Georgia kaolin clays. Flotation of kaolin clays from other major secondary deposits, such as Brazil, is currently not practiced. The various surface chemical interactions involved between collector molecules and the minerals—specifically, iron-anatase in providing selectivity in the flotation processes—are provided in the following sections. Carrier Flotation. Carrier flotation, also known as ultraflotation or piggyback flotation, was the first successful commercial flotation process in the kaolin industry (Greene, Duke, and Hunter 1961). In this process, the kaolin clay is slurried using sodium silicate as a dispersant and degritted using 325-mesh screens. The slurry is then conditioned with tall oil in the presence of calcium carbonate. The dissolved calcium ions from the calcium carbonate act as an activator for the iron-anatase surface to absorb the tall oil. The fine coloring impurity (i.e., iron-anatase) “piggybacks” to the calcium carbonate, which acts as the carrier mineral (see Figure 33). Because of the relatively coarser particle size of the carrier mineral, flotation is enabled at reasonable rates. The carrier used is calcite (typically –325 mesh), which is also conditioned with tall oil to make its surface hydrophobic. Calcite is a satisfactory carrier mineral because of its low cost, excellent flotation response, availability, and ease of removal from the froth. Frother is not needed because the tall oil provides sufficient froth. Wang and Somasundaran (1980) studied the mechanism of carrier flotation by measuring the zeta potential of anatase, kaolinite, and calcite as a function of sodium oleate concentration. Their results showed both anatase and calcite to be negatively charged in the Carrier Mineral

Air Bubble

Source: D.W. Fuerstenau 1980.

FIGURE 33

Schematic of carrier flotation

Slime Mineral

flotation0.book Page 493 Tuesday, January 2, 2007 7:36 PM

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

493

presence of oleate collectors. This suggests that the “piggyback phenomenon” of attachment of anatase onto calcite is due to hydrophobic interaction between the oleate layers adsorbed on the mineral surfaces, not electrostatic attraction. In addition, Wang and Somasundaran (1980) provided an explanation for the beneficial effect of high-speed agitation during the conditioning process in ultraflotation. Dissolved calcium ions adsorb on both kaolinite and anatase surfaces, facilitating the adsorption of tall oil. However, their data showed that the zeta potential of kaolinite decreases with increasing high-shear agitation but that of anatase does not change. The authors concluded that collector adsorption by kaolinite is relatively weak as compared to its adsorption by calcite and anatase. Therefore, upon high-shear agitation, the collector coating on kaolinite is readily removed though it remains on iron-anatase, allowing for selectivity to be achieved in the flotation process. Carrier flotation has been practiced successfully to overcome the fine particle size problems associated with the coloring impurity and to be able to produce high-brightness kaolin clays. However, this process has certain inherent disadvantages: (1) high reagent consumption is required because of the need to condition the carrier mineral; (2) low pulp density is required for greater efficiency; and (3) carrier mineral in the kaolin clay product needs to be removed completely and needs to be recovered in the froth phase for recirculation. The recovery by carrier flotation may be improved by centrifuging the froth to separate the coarse calcium carbonate–titania aggregates from the entrained kaolin (Gantt et al. 1994). A mechanical separation process, such as a centrifuge or a hydrocyclone, will not accomplish the separation if all of the particles are dispersed. The presence of a flotation collector to agglomerate the impurities is essential for mechanical separation and to enhance the overall recovery of kaolin. Fatty Acid Flotation. Cundy (1969) developed the first carrierless flotation process for the removal of titaniferous impurities from kaolin. The two essential features of this process are high-energy scrubbing of the kaolin slurry at 40% to 60% solids and the presence of activator ions, such as calcium. According to Cundy, the scrubbing action resulting from the high-shear agitation at high-solids content cleans the mineral surfaces from contamination, resulting in an improved difference between the different mineral particles. It is, however, more likely that this process facilitates liberation of the colored impurities from kaolin and thus prepares the slurry for conditioning with the oleic acid in presence of calcium ions. As discussed for the carrier flotation process, the high-speed agitation probably also leads to the collector coating the iron-anatase particles only, which results in flotation selectivity. In addition, the collector-coated iron-anatase particles may be selectively coagulated under high-speed agitation, which effectively increases the particle size and allows flotation. In the Cundy process, the high-energy scrubbing is conducted at relatively high solids content in two stages. Initially, the kaolin crude is scrubbed in the presence of a dispersant (e.g., sodium silicate) and a pH modifier (e.g., ammonium hydroxide). The second stage of scrubbing involves conditioning with calcium salts and fatty-acid-type collector (e.g., oleic acid). After conditioning, the pulp density is reduced to about 15% to 20%, and conventional mechanical flotation cells are then used to remove the colored impurities. Again, similar to carrier flotation, the fatty acid collector provides the frothing action required for flotation. The fatty acid flotation by Cundy was modified by Young, Morris, and Brooks (1985) to what is called the titanium removal and extraction process (TREP). This carrierless flotation uses oleic acid as a conditioning reagent in the presence of a calcium activator but under

flotation0.book Page 494 Tuesday, January 2, 2007 7:36 PM

494

FLOTATION CHEMISTRY

acidic conditions. A major difference between the TREP and the previous kaolin flotation processes is that in the TREP, flotation is accomplished at the same high solids as during conditioning, that is, in excess of 25%, the benefit of which is higher throughput and lower dewatering costs. Other unique features of TREP compared to previous flotation methods are the novel high-intensity conditioner (Bacon and Brooks 1984) and column-like flotation cell (Bacon 1984). In the TREP conditioner, the temperature of the conditioned slurry reaches in excess of 93°C (200°F). The conditioned product is then treated with a dispersant (i.e., sodium polyacrylate) prior to being floated in the column flotation cell described in the patent by Bacon (Bacon 1984). The high conditioning temperature facilitates higher solubility of oleic acid and also decreases the pKa of the oleic acid. As a consequence, even though the pH of the conditioner feed is in the range of 6.1 to 6.3, interaction between the oleic acid and the ironanatase particles occurs through a bridging mechanism with calcium ions. The conditioning under acidic pH distinguishes the TREP from the Cundy process. A typical result of TREP flotation is given in Table 3. The separation efficiency is sensitive to the dispersant added prior to conditioning and flotation. Young, Morris, and Brooks (1985) showed that organic dispersants, such as polyacrylates or inorganic dispersants such as sodium polyphosphates, added prior to conditioning inhibit flotation (see Table 3). The results show dramatically that the addition of a polyacrylate dispersant prior to conditioning is very detrimental to TiO2 removal. The polyacrylate salt added before conditioning provides only one-fourth of the TiO2 removal obtained otherwise. However, the conditioned product could be best dispersed for flotation with polyacrylate rather than an inorganic dispersant, as illustrated in Table 4. These results show that the polyacrylate dispersant added to the conditioned product results in an almost twofold TiO2 removal during flotation compared to no dispersant. Further, the inorganic dispersants offer only a marginal improvement as compared to no dispersant addition prior to flotation. The effects of dispersant added after conditioning are shown in Table 5. A major drawback to TREP is that the high-intensity conditioning is time intensive, which significantly increases the processing cost. In addition, equipment wear is more pronounced, resulting in higher maintenance costs. Efforts to reduce the conditioning time have resulted in inadequate flotation separation and/or low kaolin recovery. TABLE 3

Effect of TREP on beneficiation of kaolin

Sample Feed TREP product

% TiO2 1.76 0.53

GE Brightness 84.7 89.1

Source: Young, Morris, and Brooks 1985.

TABLE 4

Effects of dispersant addition prior to conditioning

Sample Feed TREP product TREP product

Dispersant Added Before Conditioning — Sodium silicate Sodium silicate Sodium polyacrylate

Source: Young, Morris, and Brooks 1985.

Dosage, kg/ton — 1.15 1.15 0.85

Product, % TiO2 1.83 0.53 1.50 —

GE Brightness 85.6 90.9 87.0 —

% TiO2 Removal — 71.0 18.0 —

flotation0.book Page 495 Tuesday, January 2, 2007 7:36 PM

495

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

TABLE 5

Effects of dispersant addition after conditioning

Sample Feed TREP product TREP product

TREP product TREP product

Dispersant Added After Conditioning — None Trisodium polyphosphate/ sodium carbonate (Na2CO3) Sodium silicate Sodium polyacrylate

Dosage, kg/ton — — 1.0

Product, % TiO2 1.76 1.12 1.06

GE Brightness 84.7 86.0 86.8

% TiO2 Removal — 36.4 39.8

1.0 1.0

0.95 0.53

87.3 89.1

46.0 69.9

Source: Young, Morris, and Brooks 1985.

Carrier flotation and the fatty acid flotation technologies are based on the use of fatty acid– or tall oil–type collectors and require the use of divalent or trivalent cations for activation. The presence of activator ions makes the process sometimes difficult to control because of the necessity to maintain a proper balance between the amounts of collector and activator added. For instance, an excessive use of activator can induce coagulation of the clay particles and makes the separation difficult. Further, activators may also cause the flotation of the clay particles themselves rather than the colored impurities, resulting in poor selectivity and a decrease in clay recovery. It is therefore desirable to have a collector for colored impurities that does not require activators. The use of alkyl, aryl, or alkylaryl hydroxamates in the flotation of minerals that chelate with hydroxamate is also known in the industry (Nagaraj 1988). Hydroxamates are powerful collectors in flotation due to their ability to selectively chelate at the surfaces of minerals that contain titanium, yttrium, lanthanum, cerium, niobium, tantalum, tin, iron, manganese, and copper. Mixtures of minerals containing copper and iron have been successfully beneficiated by flotation using hydroxamates as the collector (Peterson et al. 1965; M.C. Fuerstenau, Miller, and Gutierrez 1967; M.C. Fuerstenau, Harper, and Miller 1970). Yoon and Hilderbrand (1986) first patented a successful kaolin flotation process based on hydroxamate collectors. The hydroxamate collectors can be used effectively at pH values above 6, under which conditions clay dispersion is readily achieved. The amounts of these reagents required for flotation are considerably less than those typically used in the conventional tall oil flotation process. Also, the hydroxamate collectors possess frothing properties so that frother addition is not required for flotation. An improved manufacturing process for hydroxamates was patented by Wang and Nagaraj (1989). This process is used for the commercial production of the alkyl hydroxamate (Aero 6493 Promoter) currently used in the kaolin industry. Flotation with hydroxamate collectors consists of similar basic steps as those described in the other processes, such as dispersing the clay slurry and conditioning with the collector. Conditioning solids can be as high as 70%, and the flotation solids can be between 15% and 45% with the use of hydroxamate collector. Activators such as calcium ions are not required for hydroxamate flotation. The distinguishing features with the hydroxamate collectors are relative insensitivity to the dispersant type and high solids during conditioning and flotation (Yoon et al. 1992). The conditioning pH is generally maintained between pH 8 and 10, because the process is not as efficient at lower pH values and more alkaline pH results in excessive frothing that inhibits effective separation. In many flotation systems that use hydroxamate as collectors, the optimum pH has been between pH 9 and 9.5, which is similar Hydroxamate Flotation.

flotation0.book Page 496 Tuesday, January 2, 2007 7:36 PM

496

TABLE 6

FLOTATION CHEMISTRY

Tall oil flotation vs. hydroxamate flotation

Collector Feed Tall oil Octyl hydroxamate

Dosage, kg/ton — 1.5 0.75

Product, % TiO2 1.42 0.48 0.16

GE Brightness — 88.7 92.4

% TiO2 Removal — 66.2 88.7

Source: Yoon et al. 1992.

to the pKa of hydroxamic acid (M.C. Fuerstenau, Miller, and Gutierrez 1967; M.C. Fuerstenau, Harper, and Miller 1970; D.W. Fuerstenau 1980). Table 6 shows the results of two laboratory flotation tests comparing tall oil flotation to potassium octyl hydroxamate flotation (Yoon et al. 1992). The tall oil flotation used 1.28 kg/t of calcium acetate as an activator while the hydroxamate flotation did not use any activator. As shown, the use of hydroxamate gave significantly better TiO2 removal than the tall oil. The hydroxamate kaolin clay product also has much higher brightness than the product obtained from tall oil flotation. The hydroxamate collector renders the iron-anatase more hydrophobic, resulting in a faster flotation rate, better selectivity, and improved recovery. The chelation of the hydroxamate with either the titanium or iron on the surface of ironanatase is responsible for the strong collecting action of this reagent. New, modified hydroxamate collectors were developed (Basilio et al. 2000) to overcome some of the shortcomings of the commercially available alkyl hydroxamate (Cytec Aero 6493 promoter). The alkyl hydroxamate activities of the modified reagents are higher than that of Aero 6493 Promoter (i.e., >30% active). These reagents also have different carrier solvents that allow the alkyl hydroxamate to remain soluble at ambient temperatures. In addition, some of the reagents use a carrier solvent that has some frothing properties; thus reducing the frother requirement for flotation. These modified hydroxamates can be tailored via the use of different carriers and methods of synthesizing the reagent. Flotation tests conducted on a kaolin crude sample from Georgia in the United States show that reagent S-8706 promoters give better flotation performance than Aero 6493 promoter and tall oil at a lower collector dosage (see Table 7). Another recent development in hydroxamate flotation is the use of hydroxamate for TREP flotation, described previously. Mathur, Brooks, and Finch (2002) used hydroxamate instead of tall oil/calcium chemistry in the TREP conditioner, which resulted in a decrease in the conditioning time by at least 50% and a lower temperature. Because of these benefits, lower processing and maintenance costs were obtained. Table 8 provides the flotation test results comparing the use of oleic acid and hydroxamate collector in the TREP flotation system. As shown, about 75% reduction in conditioning time is obtained without significant change in flotation performance. Co-collector Flotation. The very high selectivity of hydroxamate flotation without the need of an activator has made this technology a good alternative for kaolin flotation. The main disadvantage of hydroxamate as a collector is its relatively poor frothability (as compared to fatty acids), which makes its application in column cell flotation difficult because of the deep stable froth required. Consequently, the use of a frother is required, which results in a more complicated reagent-addition system. Excessive foaming of the flotation product has detrimental downstream processing effects and must be avoided. In addition, the reagent cost of hydroxamate is higher than that of tall oil.

flotation0.book Page 497 Tuesday, January 2, 2007 7:36 PM

FLOTATION CHEMISTRY AND TECHNOLOGY OF NONSULFIDE MINERALS

TABLE 7

497

Modified hydroxamate vs. Aero 6493 promoter and tall oil

Collector Dosage, kg/ton Tall oil* 1.0 1.0 Aero 6493 promoter† Reagent S-8706 promoter 0.625

Product, % TiO2 0.79 0.27 0.33

% Clay Recovery 67 80 96

% TiO2 Removal 43.6 80.7 76.4

Source: Basilio et al. 2000. *0.25 kg/ton CaCl2 is added as an activator. †0.2 kg/ton of propylene glycol–based frother is added.

TABLE 8

Hydroxamate vs. oleic acid in TREP flotation Conditioning Time, min 80 20

Collector Oleic acid Aero 6493 promoter

Conditioning Temperature, ºC 93 54

Product, % TiO2 0.51 0.47

% Clay Recovery 86 84

Source: Mathur, Brooks, and Finch 2002.

TABLE 9

Co-collector flotation vs. tall oil and hydroxamate-only flotation

Collector Tall oil* Alkyl hydroxamate† Tall oil/alkyl hydroxamate

Dosage, kg/ton 1.5 1.0 0.5/0.25

Product, % TiO2 0.40 0.41 0.27

% Clay Recovery 81.6 96.4 84.8

% TiO2 Removal 74.2 73.5 82.6

Source: Shi and Yordan 1996. *0.25 kg/ton CaCl2 is added as an activator. †0.2 kg/ton of propylene glycol–based frother is added.

The use of a combination of fatty acid–type collector (i.e., tall oil) and hydroxamate was more effective in removing titaniferous impurities from kaolin clays than using either collector alone (Shi and Yordan 1996). Lower amounts of collectors are needed to obtain improved or equivalent flotation performance than when either reagent is used alone. In addition, a more stable froth is obtained with this “co-collector flotation chemistry” than that with hydroxamate alone. The process is essentially similar to that used for hydroxamate flotation. The only difference is the addition of the two collectors, hydroxamate and tall oil, during conditioning. This has resulted in the effective application of hydroxamate flotation to column cells. Table 9 shows the improved flotation performance obtained with the use of the co-collector system compared to either the hydroxamate- or tall-oil-only flotation chemistries. In addition, the hydroxamate dosage has been significantly reduced for the co-collector chemistry. Interestingly, the co-collector chemistry circumvents the need for an activator that is typically required for tall oil flotation. SUMMARY

For more than 40 years, flotation has been used in the kaolin industry to improve the brightness and color of kaolin clays for pigment applications. Kaolin flotation, unlike most flotation systems, involves minerals with very fine particle sizes ranging from about 50% to 90% passing 2 microns. Flotation of such very fine particles has been difficult with most minerals, but with the adoption of various technologies, successful separation has been obtained

flotation0.book Page 498 Tuesday, January 2, 2007 7:36 PM

498

FLOTATION CHEMISTRY

in the kaolin industry. Contributing to this success is the fact that kaolin flotation utilizes reverse flotation which only requires the flotation of a very small fraction of the feed material. Currently, flotation is the most efficient method for removal of discoloring impurities from kaolin and production of high-brightness kaolin clay pigments. A review of the current flotation technologies used in the kaolin industry reveals a variety of innovative processes, from carrier flotation to the use of hydroxamate collectors. Interestingly, carrier flotation, which was developed more than 40 years ago, is still being used in the kaolin industry. In terms of flotation machines, the kaolin industry has adapted well with the use of both conventional mechanical flotation cells and column cells. Despite the kaolin industry’s ability to adapt the various flotation technologies to fine particle processing, many technological challenges need to be addressed, including • Variable flotation response of kaolin clays from different sources, • High conditioning-energy requirements, • Low flotation pulp densities that entail a large amount of water removal downstream, and • Removal of the iron oxide impurities. A fundamental understanding of the industrial practice is certainly needed to refine the existing processes and to develop new flotation technologies. The move to specific collector chemistry such as hydroxamates has shown a lot of promise in designing robust flotation processes. The hydroxamate collector gives more flexibility, because it can be used for different kaolin crude sources and for more varied process conditions. However, the current downside is the higher cost associated with the use of this reagent.

Chemistry of Iron Oxide Flotation K.H. Rao and K.S.E. Forssberg

INTRODUCTION

World iron ore production totaled 1,045 Mt in 2001 and is forecast to reach 1,119 Mt by 2007 based on current mine capacities, planned expansions, and the development of new iron ore mines. The largest production increases will come from traditional sources in Australia and Brazil, and from the growth regions of India and South Africa. The bulk use of iron ore is for the production of pig iron to make steel and other alloys. Demand for iron ore, therefore, depends on the necessity for a vital construction material—steel and its alloys. Steel is still the core of the building industry regarding transportation by rail, ship, and road vehicles (including roads and bridges). The importance of iron and steel, and therefore the need to process iron ore, continues to be a major factor for the medium-grade iron ores as well as for improving high-grade iron ores in the overall economic development of all nations. Direct reduction steelmaking uses 4 with sodium dodecylsulfonate • pH < 4 with dodecylamine Screening Phosphate Depressants for Floating Coarse Silica from Phosphate

Five potential phosphate flotation depressants were subjected to evaluation during cationic flotation of coarse silica using the quaternary/oil collector “Q” process. IMC/Agrico’s Kingsford plant spiral feed was used for all tests. The potential depressants included sodium tripolyphosphate (STPP), fluosilicic acid (FSA), DPA, starch, and sodium silicate. An abbreviated dry-screen analysis of the spiral feed sample showed the following particle size distribution: Tyler Mesh +14 14/20 20/35 –35 Total

% Wt 6.5 24.9 56.1 12.5 100.0

Cumulative % Wt 6.5 31.4 87.5 100.0

For a phosphate depressant to be considered effective, its use in flotation should result in an increase in phosphate (P2O5) recovery without a significant parallel increase in the percentage of Insol in the concentrate. Also, the performance of a “true” depressant is often dependent on the type and level of collector used. Referring to Figures 1–3, only STPP, DPA, and, to a lesser degree, FSA appeared to exhibit useful depressant activity when used with the quaternary/oil collector combination. The results are presented in Figures 1–3 in accordance with the following legend: Figure 1 2 3

Phosphate Depressant STPP FSA DPA

Conditioning pH 6.7–6.9+ 3.0–4.0+ 4.1–5.7

Flotation pH 7.0–7.2 5.3–6.9 5.8–6.9

flotation0.book Page 557 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

B. 0.25 kg Cationic Collector/Ton Feed

100

100

80

80 % P2O5 % Insol % P2O5 Recovery

60

Weight, %

Weight, %

A. 0.2 kg Cationic Collector/Ton Feed

40

20

0 0.0

557

% P2O5 % Insol % P2O5 Recovery

60

40

20

0.5

1.0

1.5

0 0.0

0.5

Sodium Tripolyphosphate, kg/t feed

FIGURE 1

1.0

1.5

Sodium Tripolyphosphate, kg/t feed

Flotation concentrate grade/recovery using various STPP levels 100

Weight, %

80

% P2O5 % Insol % P2O5 Recovery

60

40

20

0 0.0

0.25

0.5

0.75

Fluosilicic Acid, kg/t feed

FIGURE 2

Flotation concentrate grade/recovery using various FSA levels

Figures 1a and 1b illustrate that the use of 0.125 kg STPP per ton of feed with the lower collector level produced a 30.6% P2O5/10.9% Insol concentrate at 96.5% P2O5 recovery compared to a 31.1% P2O5/10.0% Insol concentrate at 88.8% P2O5 recovery when no depressant was used. Using the higher collector level, a better-grade concentrate analyzing 31.6% P2O5/7.4% Insol was produced at 94.5% P2O5 recovery. However, the STPP required to maintain high P2O5 recovery increased to about 0.5 kg/t of feed. Using only 0.125–0.25 kg STPP per ton of feed yielded concentrates analyzing 31.7%–32.2% P2O5/ 6.1%–6.6% Insol at 84.9%–85.3% P2O5 recovery. When no depressant was used, the concentrate analyzed 31.0% P2O5/9.5% Insol at only 77.0% P2O5 recovery. Figure 2 shows that FSA in the range of 0.125–0.5 kg/t of feed was effective in depressing phosphate and resulted in P2O5 recoveries exceeding 92% for all tests performed. However, no concentrates analyzing 30% P2O5 or higher were produced. The use of more than 0.25 kg of collector is indicated as necessary to produce a 30+% P2O5 concentrate. Using the 0.25-kg collector level with 0.125 kg of FSA per ton of feed produced a phosphate concentrate reported to analyze 29.1% P2O5/9.3% Insol at 92.4% P2O5 recovery. When no

flotation0.book Page 558 Tuesday, January 2, 2007 7:36 PM

558

FLOTATION CHEMISTRY

100

Weight, %

80

% P2O5 % Insol % P2O5 Recovery

60

40

20

0 0.0

0.25

0.5

0.75

Diophosphonic Acid, kg/t feed

FIGURE 3

Flotation concentrate grade/recovery using various DPA levels

depressant was used, the concentrate analyzed 31.0% P2O5/9.5% Insol at 77.0% P2O5 recovery, indicating a possible analytical error. Figure 3 indicates that DPA was an effective phosphate depressant when used at a level of 0.25 kg/t of feed with this high collector dosage. A phosphate concentrate was produced analyzing 31.15% P2O5/6.1% Insol at 91.8% P2O5 recovery. Using less depressant resulted in a significant decrease in P2O5 recovery, and using more depressant produced a lowergrade concentrate. DPA appears to be very sensitive to the level used and is considered to be inferior to STPP as a selective phosphate depressant when used with the quaternary/oil collector combination used to float coarse silica. In contrast, neither Westvaco starch nor N-Brand sodium silicate performed as selective phosphate depressants with this flotation process. At all usage levels tested, lower-grade phosphate concentrates were produced compared to the standard no-depressant tests. Sodium silicate was expected to be detrimental to coarse silica flotation because of possible chemical reaction with the quaternary collector. The use of a higher quantity of starch plus quaternary/oil collector levels could possibly yield a 30+% P2O5 concentrate at 90+% P2O5 recovery. Screening Phosphate Depressants for Floating Fine Silica from Phosphate

The preceding section summarized laboratory cationic flotation test results obtained using a quaternary/oil collector, with and without potential phosphate depressants, to float “coarse” quartz from IMC/Agrico’s Kingsford plant spiral feed. This section presents the results obtained using an amine condensate/diesel fuel collector, with and without potential phosphate depressants, to float “fine” quartz from IMC/Agrico’s Kingsford plant amine feed. Gelatinized starch, STPP, DPA (Dequest 2010), FSA, and OPA were evaluated as potential phosphate depressants during laboratory amine flotation of fine quartz from phosphate using a commercial amine condensate/diesel fuel collector combination to process the previously cited plant amine feed. From an overall cost and performance standpoint, starch was considered to be the best phosphate depressant. STPP was considered to be almost equally as effective but more costly in comparison to starch as a selective phosphate depressant. Using starch or STPP, concentrates were produced analyzing 32%–33+% P2O5/3%–4% Insol at 95%–96% P2O5 recovery using as little as 0.125 kg of depressant per ton of feed. Similar results were obtained using DPA at a level of 0.125–0.25 kg/t of feed provided that

flotation0.book Page 559 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

559

the amine collector level was more carefully controlled. Using starch or STPP at a higher level (0.25–0.5 kg/t of feed) had almost no detrimental effect on flotation concentrate grade provided that an adequate quantity of amine collector was used. When no depressant was used, about 12% P2O5 recovery loss occurred when the amine collector level was increased from 0.2 to 0.25 kg/t of feed. The proper use of starch, STPP, or DPA prevented this loss. FSA and OPA failed to effectively control P2O5 recovery when the amine level was increased from 0.2 to 0.25 kg/t of feed. These two acids were considered to be ineffective as phosphate depressants within the range of test conditions employed. Use of Depressants in Carbonate/Apatite Systems

Numerous phosphate mineral depressants have been reported in the literature for use during anionic flotation of dolomite from carbonate fluorapatite in slightly acidic circuits (see Table 1). For example, citric acid functions as a dolomite depressant by the binding of its carboxylate groups with calcium and magnesium and the formation of insoluble citrate compounds on the dolomite surface. Another example is CMC, which adsorbs on dolomite and forms a barrier for collector adsorption. Its depressing action is attributed to the presence of multifunctional carboxylate groups in addition to its large molecular volume. The laboratory tests described in this section are intended to compare the effectiveness of several of the reported depressants to process rod-milled, deslimed, high-MgO (magnesium oxide) Florida phosphate pebble from the southern mining area’s Four Corners reserves, currently mined and processed by IMC/Agrico. In order to minimize the detrimental effects of hard water on flotation efficiency using fatty acids or their soaps, sulfonated oleic acid soap (plus oil) was used as the dolomite collector for all tests. The use of this collector is described in U.S. patents assigned to International Minerals & Chemical Corp (Snow 1982; Lawver, McClintock, and Snow 1983). Ten potential phosphate depressants were evaluated during laboratory anionic flotation of dolomite from phosphate in a slightly acid circuit using a sulfonated oleic acid soap plus oil as the dolomite collector. The flotation feed used for this study was prepared by wet rod milling and desliming Florida pebble (25.8% P2O5, 2.6% MgO) to yield a –48 +325 mesh product (26.5% P2O5, 1.6% MgO) for testing. The –325 mesh grinding slimes (27.0% wt) contained 23.6% of the total P2O5 and 54.5% of the total MgO present in the original pebble sample: Product 48/325 feed –325 slime

Wt % 73.0 27.0

% P2O5 26.52 22.17

% Insol 11.80 5.42

% MgO 1.60 5.20

Dolomite Flotation at Different pH and Collector Levels

In other tests, dolomite was floated at different pH and collector levels, and the results are presented in Figures 4a–4c. The data in Figure 4a shows that about 1.125–1.250 kg of sulfonate OA-5R collector per ton of feed were required to produce a phosphate concentrate containing less than 0.9% MgO. Another test produced a phosphate concentrate analyzing 26.3% P2O5, 0.78% MgO, and MgO/P2O5 ratio = 0.030 at 65.9% P2O5 recovery when 1.25 kg of collector per ton of feed was used. Figure 4b shows that using the same collector level at the lower flotation pH range, yielded a phosphate concentrate analyzing 27.3% P2O5, 1.00% MgO, and MgO/P2O5 ratio = 0.037 at 82.9% P2O5 recovery. Using the lower pH conditions and the same collector level, P2O5 recovery was substantially improved at the expense of a higher concentrate percentage of MgO. Essentially, the same concentrate grade

flotation0.book Page 560 Tuesday, January 2, 2007 7:36 PM

560

FLOTATION CHEMISTRY

B

A 40

7 pH 6 pH 5

6

pH 6 pH 5

MgO in Tails, %

Floated, %

30

20

5

4

10 3

0 0.0

0.25

0.5

0.75

1.0

1.25

1.5

0 0.0

0.2

Sulfonate Dosage, kg/t feed

0.4

0.6

0.8

1.0

1.2

1.4

Sulfonate Dosage, kg/t feed

C 100 pH 6 pH 5 P2O5 Recovery, %

90

80

70

60 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Sulfonate Dosage, kg/t feed

FIGURE 4 Effect of collector dosage on (a) the amount floated, (b) MgO in tails, and (c) flotation recovery

and P2O5 recovery were obtained with only 0.75 kg of collector using the higher pH range during conditioning/flotation. Consequently, the initial flotation tests comparing potential phosphate depressants were performed using 1.25 kg of sulfonate OA-5R and a flotation pH = 5.5–6.0 in order to have the best chance of producing phosphate concentrates with MgO/P2O5 ratios of 0.030 or less. In the series of laboratory flotation tests that were performed comparing the effectiveness of 10 potential phosphate depressants, depressant addition levels ranged from 0.5 to 1.5 kg/t of feed. The depressants evaluated were as follows: STPP, DPA, sodium hexametaphosphate (SHMP), tetrasodium pyrophosphate (TSPP), OPA, aluminum tartrate complex, disodium hydrogen phosphate (DSHP), starch (CCD-2112), FSA (2%), and Alizarin Red S. The tartrate complex consisted of two parts by weight of aluminum sulfate plus one part by weight of sodium potassium tartrate. Sulfuric acid for pH regulation and the selected depressant were added first to the flotation cell containing feed slurry (~25% solids) and conditioned for 1 minute. Sulfonate OA-5R solution (5%) and Philflo oil were added, and conditioning continued for 1 minute before introducing air to start flotation. A 3-minute flotation time with small sulfuric acid additions for pH control was used again for all flotation tests. Most of the flotation results are plotted as concentrate % MgO, % P2O5 recovery, and % MgO recovery versus depressant level used.

flotation0.book Page 561 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

100

100 80

% P2O5 % Insol % P2O5 Recovery

60

Weight, %

Weight, %

80

40 20 0 0.0

561

% P2O5 Recovery % MgO Recovery % MgO × 10 in Concentrate

60 40 20

0.5

1.0

1.5

2.0

2.5

3.0

0 0.0

3.5

0.5

Sodium Tripolyphosphate, kg/t feed

FIGURE 5 Effect of STPP on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition

1.5

2.0

FIGURE 6 Effect of DPA on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition

100

100

80

80

% P2O5 Recovery % MgO Recovery % MgO × 10 in Concentrate

60

Weight, %

Weight, %

1.0

Diphosphonic Acid, kg/t feed

40

% P2O5 Recovery % MgO Recovery % MgO × 10 in Concentrate

60 40 20

20 0 0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Sodium Hexametaphosphate, kg/t feed

Tetrasodium Pyrophosphate, kg/t feed

FIGURE 7 Effect of SHMP on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition

FIGURE 8 Effect of TSPP on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition

1.6

Performance curves for the most effective phosphate depressants (STPP, DPA, SHMP, and TSPP) are presented in Figures 5–8, respectively. In all of these tests, P2O5 recoveries were considerably superior to those where no depressant was used (concentrate = 0.78% MgO, MgO/P2O5 = 0.030%, and 65.9% recovery P2O5). DPA was considered to be the least powerful and most expensive of the four effective depressants for the test conditions employed. There appeared to be very little difference between STPP, SHMP, and TSPP with respect to their performance as good phosphate depressants. Results of testing OPA, and aluminum tartrate indicate that these reagents are practically noneffective phosphate depressants (Figures 9 and 10). In addition to the previously tested reagents, numerous polymers may be used as depressants and froth modifiers. Table 2 summarizes some typical applications of polymers as depressants. Structures for guar and CMC are shown in Figures 11 and 12. Figure 13 (Lin and Burdick 1988) indicates that a small amount of CMC as phosphate depressant increased recovery by about 10 percentage points in floating silica from phosphate.

flotation0.book Page 562 Tuesday, January 2, 2007 7:36 PM

562

FLOTATION CHEMISTRY

100

Weight, %

60

% P2O5 Recovery % MgO Recovery % MgO × 10 in Concentrate

40

20

0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Orthophosphoric Acid, kg/t feed

FIGURE 9 Effect of OPA on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition 100

Weight, %

60

% P2O5 Recovery % MgO Recovery % MgO × 10 in Concentrate

40

20

0 0.0

0.5

1.0

1.5

2.0

Aluminum Tartrate, kg/t feed

FIGURE 10 Effect of aluminum tartrate on depressed phosphate concentrate at 1.25 kg/t of sulfonate collector addition

TABLE 2

Examples of polymers used as depressants

Polymer Type Starch Cationically modified polysaccharides Cationically modified guar Carboxymethyl cellulose Polyphosphates

Minerals Depressed Phosphate Silica Silica Calcite, dolomite, apatite, and talc Siliceous gangue

A depressant must have functional groups that exhibit a preference attraction to the gangue minerals and a strong hydrophilicity by virtue of either the same or other functional groups in its molecular structure. Simultaneously, the depressant molecules should not have functional groups that compete effectively with the collector for the surface of the minerals to be floated.

flotation0.book Page 563 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

CH2OH HO

CH2OH O

H OH

Galactose Units

H

H

OH

HO O

CH2OH

OH

OH

H

H

H

H

H

O

H

HO O

CH2OH

CH2

H

H

H H

H

O

HO

H

O

563

O H

O

CH2 O

H

H

O

O

H

H

OH

OH

OH

OH

OH

OH

H

H

H

H

H

H

H

O

H

Mannose Chain

FIGURE 11

Guar structure CH2COONa O

Sodium Carboxymethyl Cellulose

CH2 H

H

OH

H

O

H

H

FIGURE 12

OH

CMC structure

Apatite Recovery, %

100

90

80

70 0.0

0.5

1.0

1.5

2.0

Polymer Dosage, kg/t feed

FIGURE 13 Reverse flotation of silica from apatite with CMC as phosphate depressant at 0.11 kg/t of amine

The following major factors should be considered in selecting a depressant: charge density, molecular weight, and dosage. High-charge density polymers are not desirable depressants because most collectors are either positively or negatively charged. Hydroxyl-bearing water-soluble polymers, such as starch and guar, have been used as depressants. The hydroxyl groups on these polymer molecules impart strong hydrophilicity as well as a fairly good

flotation0.book Page 564 Tuesday, January 2, 2007 7:36 PM

564

FLOTATION CHEMISTRY

affinity to gangue minerals. However, they are not nearly as surface active as most charged collector functionalities and, therefore, compete less effectively for the active sites on the surface of the minerals to be floated. A basic rule for depressant selection is to consider which one would be a good collector for the particular mineral that needs to be depressed, and then to select a depressant that includes the same functionality as the collector but in which the hydrophobe is replaced with a hydrophile. For example, fatty acids, which contain carboxylate functionality, are effective collectors for calcite, dolomite, apatite, and hematite. Therefore, a carboxylated depressant such as CMC would be an effective depressant for these minerals. Siliceous minerals, such as silica and kaolinite clay, are floated by cationic amine collectors. Therefore, depressants based on cationically modified polysaccharides are effective depressants for siliceous minerals. Nevertheless, because of their relatively high cost, cationic polymeric depressants are not popular on an industrial scale. Instead, underivatized polysaccharides, such as starch and guar gum, are commonly used to depress silica or “blind” slime. Polyglycols of high molecular weight are effective in depressing siliceous materials but could cause excessive foaming. Polyacrylamides and dextrines are also effective but costly. Selection of molecular weight also depends on the mechanism of depression. If depression is through flocculation, higher molecular weight is desirable, whereas low-molecularweight polymers work better when dispersion is desired. The optimal dosage varies with the type of ore, the particle size and size distribution, the type of depressant, the molecular weight of the depressant, and the influence of other species, particularly those in the water. As a general rule, higher dosages are needed for lower-molecular-weight polymers. The effectiveness of a depressant is highly dependent on the mineral system. For example, polyphosphates are effective in depressing siliceous gangue in base metal flotation but are not ideal depressants when floating phosphate from silica. Polymeric depressants may increase the recovery of coarse particles. This is particularly true for a slimy feed. Without the proper use of a polymeric depressant, the high surface area of the gangue can prevent the coarse particles from adsorbing an adequate share of the collector to gain floatability. If more collector is added in an attempt to float these coarse particles, an intolerable amount of gangue will also be floated, thus rendering a poor-grade concentrate. A recent study by Miller (1998) showed the significant benefit of adding nonionic polymer to fatty acid. Nonionic polyethylene oxides (PEOs) having a molecular weight between 1,000 and 8,000 are particularly effective. Outside this molecular-weight range, the effect of PEO is insignificant. For example, in the case of a coarse feed (16 × 35 mesh), in order to achieve 85% recovery, 1,200 g/t of the fatty acid/fuel oil blend is required, but only 500 g/t is required when PEO is used. At about the same collector addition, phosphate recovery can be improved by more than 10% with PEO addition. In order to achieve one-step anionic flotation of phosphate, Nagaraj (1985) developed siliceous depressants of copolymers or terpolymers derived from 85%–95% acrylamide and about 10% N-acrylamidoglycolic acid. Polyacrylamide (I), containing both OH and COOH functional groups obtained by reaction with aldehydes and ketones, and several related (I)-based copolymers proved to be selective phosphate depressants in amine flotation of silica (Nagaraj 1987). The addition of this modifier at even very low dosages (5–20 g/t) resulted in a large increase in phosphate recovery (by 10% at 20 g/t polymer).

flotation0.book Page 565 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

565

100

Apatite Recovery, %

95

90

85

80

75

70 0.0

0.05

0.10

0.15

Depressant Dosage, kg/t feed

FIGURE 14 fatty acid

Flotation of apatite from silica with cationic guar as silica depressant at 1.1 lb/t of

Flotation of phosphate was significantly improved by adding polymeric surfactants such as partial polyacrylic acid ester or polyphosphates (Smith 1982). Figure 14 (Lin and Burdick 1988) shows an example of polymer being used as silica depressant in floating phosphate. Recovery was improved by 10 percentage points by adding 2%. Phosphate ore flotation froths are not affected by oil contents as high as 4%. The reason for this difference in behavior may be because the frothing agent in phosphate flotation is the ionized fatty acids soap species. Leja (1982) stated, “A significant feature of all frothers used in flotation systems is their nonionic polar group.” Excessive frothing is produced by ionic species because of the electrostatic repulsion that prevents the thinning of the liquid film. Two species may act as natural froth modulation agents (antifoamers): (1) collector oil molecules that were not converted to fatty acid or not saponified in the preparation stage; and (2) oil molecules present in the starch depressant (Peres and Guimarães 1999). Unmodified cornstarches are not soluble in cold water. The solubilization procedure is known as “gelatinization.” After having been mixed with water, caustic soda (NaOH) at a weight ratio of 3:1 (starch/NaOH) is added to the starch. The final concentration of the cornstarch solution is 3% w/w, and the total gelatinization time is about 30 minutes. Cornstarches may be chemically modified for different purposes. Dextrines are modified cornstarches that are soluble in cold water. The molecular weight of dextrines is, by far, lower than that of unmodified starches. Dextrines keep the hydrophilization ability of largemolecular-weight starches but act as dispersants rather than flocculants. Potato starches are also used extensively for industrial purposes in Europe, but there are no records of its use in the mining industry. In addition, potato degrades much faster than corn. In Brazil and other tropical areas, cassava (also known as manioc or yuca) is a common vegetable that is rich in starch with high amylopectin plus amylose and low oil content. The low oil content prevents the risk of froth suppression, and the cost of production is lower compared with corn. The viscosity of the gums produced by gelatinizing cassava is higher than that of gelatinized cornstarch. The viscosity of the gum correlates with the molecular weight of the starch in solution and is a good indicator of the depressant ability of the reagent. A less-pure product, “cassava scrap,” is achieved by grinding the root with its inner skin. The depressant action of this inexpensive product is still acceptable. Cassava has attracted the attention of plant operators for many years, but commercial problems have

flotation0.book Page 568 Tuesday, January 2, 2007 7:36 PM

568

FLOTATION CHEMISTRY

prevented its wide utilization. Though the Brazilian production of cassava is fairly extensive and no special requirements are necessary for growing the vegetable, there are no large groups growing cassava in Brazil and co-operative efforts to supply the amounts necessary to meet the demand have not been successful. This lack of major producers prevents its extensive use in the mining industry. It has been employed intermittently in iron ore flotation, but there are no records of its use in phosphate ore flotation. Additives used in phosphate flotation are synthetic organic species. They are produced by the ethoxylation of fatty alcohols. The alcohols are obtained from vegetable oils or animal fats. Ethylene oxide comes from the petroleum industry. These reagents may present variable molecular composition and number of carbon atoms in the hydrocarbon chain, as well as the presence of double bonds, different stereochemistry (cis-trans isomerism), and also several levels of ethoxylation. Additives employed in phosphate ore flotation present different lengths of carbon chain, with a predominance of 18 carbon atoms. The average number of ethylene oxide groups in the molecule represents the ethoxylation level. Best results have been achieved with three or four groups. The dosage of additives is 5% with respect to the collector dosage, reaching 10% under special conditions. The additive is dosed either in a collector-conditioning tank or in the flotation cell in the case of mechanical machines. The additive dosage depends on the ore mineralogy. They enhance the flotation selectivity in the presence of silicate minerals and also act as froth modulators. The use of additives is a tool for the control of the surface tension of the flotation system, significantly facilitating froth transfer through pumping. Barite preflotation is performed in two Brazilian concentrators: Bunge–Araxá and Ultrafértil–Catalão. Microflotation studies by Albuquerque (1995) indicated that pure cornstarch depresses fine barite (–43 μm) more effectively than coarse barite (+43 μm). Barite preflotation of natural fines (fines produced after rod milling) was eliminated at Bunge’s Araxá concentrator when a column-only circuit replaced the mechanical cells circuit. Barite is depressed by cornstarch, together with other gangue minerals, with apatite being floated in the presence of rice bran oil soap. The pH required for effective fine barite depression reaches 12. At Ultrafértil’s concentrator, barite is prefloated in all circuits. D E P R E S S A N T S I N I R O N O R E F L O TAT I O N

Starches are the universal depressants of iron oxides in the cationic reverse flotation of iron ores, ether amines being employed as quartz collectors. Starch may be extracted from several vegetable species, such as corn, cassava, potato, wheat, rice, arrowroot, etc. In the mineral industry, cornstarches are by far the most widely used species. In Brazil, cornstarch has been used in iron ore flotation since 1978. The trade name of the reagent was Collamil, consisting of a very fine and very pure product. The amylose plus amylopectin content is 98% to 99% (dry basis), the balance being represented by minor contents of fibers, mineral matter, oil, and proteins. This starch was used at Samarco and also at phosphate concentration plants. There were no technical reasons to search for alternatives to Collamil. On the other hand, one company held the monopoly of its supply. Serious commercial problems arose from this monopoly, and Samarco performed laboratory-scale investigations on the performance Collamil alternatives (Viana and Souza 1988). One product that has been utilized in beer making was available at attractive commercial conditions, the

flotation0.book Page 569 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

569

grits starch. The terms conventional starch and nonconventional starch will be used for Collamil and grits, respectively, in this chapter (refer to Table 3). Results from plant practice showed that the use of nonconventional starch did not impair the metallurgical performance in terms of iron recovery and contaminant content of the concentrate. The price of the alternative depressant was approximately one-half the price of conventional starch. There was a strong competition among eight suppliers, a scenario by far more comfortable than the previously mentioned monopoly. Despite the practical industrial evidence that both types of starch yielded similar performances, the suppliers of conventional starch claimed that the protein content might be harmful to the flotation performance. Figure 16 shows the floatability of hematite as a function of the ether amine concentration for zein and other depressants. Therefore, the adequate industrial performance of unconventional starch was not accidental. Pinto, Dearaujo, and Peres (1992) observed from microflotation experiments presented in Figure 17 that amylopectin is the starch component 100 Starch Amylose Amylopectin

Floatability, %

80

60

40

20

0 0

20

40

60

Depressant Concentration, mg/L

FIGURE 16

Floatability of hematite as a function of depressant concentration 20

Floated, %

16

Starch Gluten Amylose Amylopectin Zein

12

8

4

0 0.0

0.05

0.10 Amine, mg/L

FIGURE 17

Depression action of various depressants on quartz

0.15

flotation0.book Page 570 Tuesday, January 2, 2007 7:36 PM

570

FLOTATION CHEMISTRY

that depresses more effectively the mineral hematite. One corn product supplier developed a genetically modified corn species, “waxy corn,” presenting an amylopectin content of 96%, higher than the 75%/25% amylopectin/amylose ratio in the regular yellow corn. The benefit of using the waxy cornstarch was not observed at industrial scale, and the product was also fairly expensive. The demand for corn grits by the snack food industry, at a much higher price than the mineral industry could afford, pushed the corn industry to offer another food segment product, locally known as fubá. Fubá is finer than grits and presents a larger oil content. The corn grains are initially degermed, because the germ containing basically proteins and oil is a valuable product for the food industry. The degermed grains are then abraded for removal of the pericarp or hull and dry-ground in hammer mills, producing different size fractions. Because the germ and the portion of the endosperm near the germ are softer than the rest of the grain, the finer fractions are richer in oil. The fact that oils are froth and foam inhibitors is well known, according to theories on the elasticity of films surrounding gas bubbles, yet some plant operators have gained such knowledge the hard way. Occasionally, some small suppliers do not find a market for the germ fraction and therefore decide to grind the whole corn kernel. The result is a starch with extremely high oil content that may surpass 3%. The consequence on the flotation machine operation is complete froth suppression, representing many hours of production interruption. Oil contents higher than 1.8% in starches are considered to represent a risk concerning froth stability. The risk increases when the loss-onignition ore content is high. Concerning the solubilization of cornstarch, there are two possibilities: heating the suspension of starch in water at 56°C or adding NaOH. Because of the hazards of employing hot water in a concentrator, as used in the first operation, all companies now use caustic soda. Because of the high cost of NaOH and frequent price fluctuations, the thermal route deserves attention and may become an attractive alternative again. Among depressants from other sources, CMC presents a strong potential. Technically, this reagent was approved as an alternative to starch. Several laboratory test programs, with different iron ores from the Iron Ore Quadrangle, have already been performed with commercial-grade CMCs of varying degrees of substitution and diverse molecular weights. In general, all CMCs tested gave concentrate grades of lower silica than starch, but iron grades of the tailings are slightly higher for CMCs tested so far (Viana and Araujo 2003). Flotation test results with three types of CMC are presented in Figure 18. To be competitive in terms of operational cost, the CMC dosage must be at least 5 times smaller than that of starches. Tested dosages were in the range from 1/10 to 1/5 starch. Some CMC dosages presented fairly good results even when used at 1/10 of the starch dosage. Another option under investigation is the use of synthetic polymers, employed as flocculants, as partial replacement for starch (Turrer 2004). Anionic, cationic, and nonionic polyacrylamides are being tested at laboratory scale. The much smaller addition level may counteract the much higher price of these reagents. D E P R E S S A N T S F O R F L O TAT I O N O F O T H E R N O N S U L F I D E ORES

Typical reagents types and dosages utilized in potash flotation are as follows: • Collector: hydrogenated alkyl amine acetate, 10–180 g/t;

flotation0.book Page 571 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

571

12

100

Iron Recovery, %

CMC1 CMC2 CMC3

8

60 6

Silica Content, %

10 80

40 4

20 0

50

100

150

200

2 250

Concentrate, g/tonne feed

FIGURE 18 Iron recovery and silica content in the concentrate as a function of dosage for three types of CMC (CMC1 is slightly cationic; CMC2 and CMC3 are a mixture of anionic CMCs with varying degrees of substitution) 30

100

80

Grade, %

60

40

Recovery, %

20

10 20 Grade Zn Recovery Zn 0

0 Dextrine

FIGURE 19

Sodium Silicate

Cornstarch

CMC

Potassium Dichromate

Effect of depressants on calamine flotation

Frother: methylisobutyl carbinol (MIBC), 15–20 g/t; Gangue depressant/dispersant: dextrine, 80–100 g/t, in the case of gangue consisting predominantly of clay minerals In the flotation of graphite ores, sodium silicate is frequently used as a gangue depressant/ dispersing agent. The effect of several gangue depressants (cornstarch, dextrine, CMC, sodium silicate, and potassium dichromate) was tested in bench-scale flotation of a calamine zinc ore (Pereira and Peres 2005). CMC and sodium silicate were slightly more effective, as illustrated in Figure 19. In general, the presence of depressants did not significantly improve the flotation selectivity. During two decades, direct anionic flotation of magnesite ore was performed in a Brazilian concentrator. More than 90% by weight of the feed was floated with fatty acids as collector. Following a laboratory bench-scale investigation (Santana and Peres 2001), the reverse flotation of gangue, consisting predominantly of quartz and silicates, was implemented in July 2004, where cornstarch was employed as the magnesite depressant. • •

flotation0.book Page 572 Tuesday, January 2, 2007 7:36 PM

572

FLOTATION CHEMISTRY

BIBLIOGRAPHY

Abramov, A.A., Al.Al. Abramov, G. Onal, S. Atak, and M.S. Celik. 1993. Mechanism of reverse flotation calcareous phosphate ores. Pages 281–288 in Beneficiation of Phosphate: Theory and Practice. Edited by H. El-Shall, B.M. Moudgil, and R. Wiegel. Littleton, CO: SME. Albuquerque, R.O. 1995. Studies on barite floatability. M.Sc. thesis, Curso de Pós-Graduação em Engenharia Metalúrgica/Universidade Federal de Minas Gerais, Brazil (in Portuguese). Baumann, A.N., and R.E. Snow. 1980. Processing techniques for separating MgO impurities from phosphate products. Pages 269–280 in Proceedings of the 2nd International Congress on Phosphorus Compounds, Boston. April. Morocco: Institut Mondialdu Phosphate. Borisov, V.M. 1956. Conditions for the beneficiation of the difficult-to-enrich phosphate ores of the Kara-Tau deposits. Khim. Prom. 13–19. Clerici, C. 1984. Flotation of a phosphate rock with carbonate-quartz gangue. Pages 221–225 in Reagents in the Mineral Industry. Edited by M.J. Jones and R. Oblatt. London: Institute of Mining and Metallurgy. Davis, B.E., T.O. Liewellyn, and C.W. Smith. 1984. Pages 10–13 in Continuous Beneficiation of Dolomitic Phosphate Rocks. Report of Investigations 8903. Washington, DC: U.S. Bureau of Mines. El-Shall, H., P. Zhang, N.A. Khalek, and S. El-Mofty. 2004. Beneficiation technology of phosphates: Challenges and solutions. Miner. Metall. Process. 21(1):17–26. Fu, E., and P. Somasundaran. 1986. Alizarin Red S as a flotation modifying agent in calcite-apatite systems. Int. J. Miner. Process. 18:287–296. Fuerstenau, M.C., and D.A. Rice. 1968. The influence of sodium silicate in non-metallic flotation systems. Trans AIME 241:319–323. Gieseke, E.W., editor. 1985. Florida phosphate rock. Chapter 21, pages 2–5 in SME Mineral Processing Handbook. New York: SME-AIME. Good, P.C. 1976. Pages 1–17 in Beneficiation of Unweathered Indian Calcareous Phosphate Rock by Calcination and Hydration. 1976-603-755/129. Washington, DC: U.S. Government Printing Office. Gruber, G., and P. Somasundaran. 1996. Understanding the Basics of Anionic Conditioning in Phosphate Flotation. FIPR Publication 02-090-121. Bartow, FL: Florida Institute of Phosphate Research. Gruber, G.A. 1987. Adapting technology to beneficiate a low-grade phosphorite ore. Miner. Metall. Process. (February): 14–18. Guimarães, R.C., and A.E.C. Peres. 1998. Relevant aspects of barite separation of a phosphate ore via flotation. Pages 285–289 in Innovations in Mineral and Coal Processing. Edited by S. Atak, G. Önal, and M.S. Çelik. Rotterdam: A.A. Balkema. Hollingsworth, C.A. 1961. Beneficiation of phosphate rock. U.S. Patent 3,013,664. Houot, R. 1982. Beneficiation of phosphatic ores through flotation: Review of industrial applications and potential developments. Int. J. Miner. Process. 9:353–384. Houot, R., and J.L. Polgaire. 1980. Inverse flotation beneficiation of phosphate ores. Pages 231–246 in Proceedings of the 2nd International Congress on Phosphorus Compounds, Boston. April. Morocco: Institut Mondialdu Phosphate. Hsieh, S.S., and J.R. Lehr. 1985. Beneficiation of dolomitic Idaho phosphate rock by the TVA diphosphonic depressant process. Miner. Metall. Process. (February): 10–13. Jones, D.A., and W.A. Jordan. 1975. Flotation beneficiation of phosphate ores. U.S. Patent 3,862,028. Lawendy, T.A.B., and G.H. McClellan. 1993. Flotation of dolomitic and calcareous phosphate ores. Pages 231–243 in Beneficiation of Phosphates: Theory and Practice. Littleton, CO: SME. Lawver, J.E. 1982. Beneficiation of dolomitic Florida phosphate reserves. Paper presented at XIV International Mineral Processing Congress, Toronto, October. Lawver, J.E., W.O. McClintock, and R.E. Snow. 1978. Beneficiation of phosphate rock: A state of the art review. Miner. Sci. Eng. 10(4):278–294. ———. 1983. Method of beneficiating phosphate ores containing dolomite. U.S. Patent 4,372,843. February 8.

flotation0.book Page 573 Tuesday, January 2, 2007 7:36 PM

DEPRESSANTS IN NONSULFIDE MINERAL FLOTATION

573

Lawver, J.E., and R.E. Snow. 1980. Method of beneficiating phosphate ores. U.S. Patent 4,189,103. February 19. Lawver, J.E., R. Snow, and R. Wiegel.1984. New techniques in beneficiation of the Florida phosphate of the future. Miner. Metall. Process. 8:89–105. Leal Filho, L.S. 1988. Page 165 in Contribution to the study of depressants for the anionic direct flotation of Jacupiranga’s phosphate. M.Sc. thesis, Curso de Pós-Graduação em Engenharia Metalúrgica/Universidade Federal de Minas Gerais, Brazil (in Portuguese). Leal Filho, L.S., and A.P. Chaves. 1993. The influence of corn starch on the separation of apatite from gangue minerals via froth flotation. Pages 147–155 in Beneficiation of Phosphates: Theory and Practice. Edited by H. El-Shall, B.M. Moudgil, and R. Wiegel. Littleton, CO: SME. Leal Filho, L.S., P.R. Seidl, J.C.G. Correia, and L.C.K. Cerqueira. 2000. Molecular modelling of reagents for flotation processes. Mineral. Eng. 13(14–15):1495–1503. Lehr, J.R., and S.S. Hsieh. 1981. Beneficiation of high carbonate phosphate ores. U.S. Patent 4,287,053. Leja, J. 1982. Surface Chemistry of Froth Flotation. New York: Plenum Press. Lin, K.F., and C.L. Burdick. 1988. Polymeric depressants. Pages 471–483 in Reagents in Mineral Technology. Edited by P. Somasundaran and B. Moudgil. New York: Marcel Dekker. Llewellyn, T.D. 1982. Beneficiation of High-Magnesium Phosphate from Southern Florida. Report of Investigations 8609. Washington, DC: U.S. Bureau of Mines. Miller, J.D. 1998. Improved Phosphate Flotation with Nonionic Polymers. Final Report, FIPR Publication 96-02-113R. Bartow, FL: Florida Institute of Phosphate Research. Moudgil, B.M. 1986. Separation of Dolomite from the South Florida Phosphate Rock. FIPR Publication 02-023-051. Bartow, FL: Florida Institute of Phosphate Research. ———. 1992. Flotation of Florida Phosphate Rocks Using Anionic Collectors. Progress Report, FIPR Publication 91-02-087. Bartow, FL: Florida Institute of Phosphate Research. Moudgil, B.M., and P. Somasundaran. 1986. Advances in phosphate beneficiation. Pages 426–441 in Advances in Mineral Processing, Arbiter Symposium. Littleton, CO: SME. Nagaraj, D.R. 1985. Flotational dressing of nonsulfide minerals. U.S. Patent 4,720,339. ———. 1987. Low molecular weight polyacrylamide-based polymers as modifiers in phosphate beneficiation. Int. J. Miner. Process. 20:291–308. Pereira, C.A., and A.E.C. Peres. 2005. Reagents in calamine zinc ores flotation. Miner. Eng. 18(2):275–277. Peres, A.E.C., and M.I. Correa. 1996. Depression of iron oxides with corn starches. Miner. Eng. 9(12):1227–1234. Peres, A.E.C., and R.C. Guimarães. 1999. The use of polymers in the Brazilian mineral industry. Pages 19–29 in Polymers in Mineral Processing. Edited by J.S. Laskowski. Quebec: Metallurgical Society of the Canadian Institute of Mining, Metallurgy and Petroleum. Pinto, C.L.L., A.C. Dearaujo, and A.E.C. Peres. 1992. The effect of starch, amylase and amylopectin on the depression of oxi-minerals. Miner. Eng. 5(3–5):469–478. Qi, G.W., C. Klauber, and L.J. Warren. 1993. Mechanism of action of sodium-silicate in the flotation of apatite from hematite. Int. J. Miner. Process. 39(3–4):251–273. Rao, D.V. 1985. Flotation of calcareous mussorie phosphate ore. Int. J. Miner. Process. 14:57–66. Rao, K.H., B.M. Antti, and E. Forssberg. 1989. Flotation of phosphatic material containing carbonatic gangue using sodium oleate as collector and sodium silicate as modifier. Int. J. Miner. Process. 26(1–2):123–140. Reis, R.L.R., A.E.C. Peres, and A.C. Araujo. 1988. Corn grits: A new depressant agent for the flotation of iron ores and phosphate rocks. Pages 389–397 in Proceedings of the II International Mineral Processing Symposium. Izmir, Turkey: Dokuz Eylül University. Rule, A. 1970. Removal of Magnesium Impurities from Phosphate Rock Concentrates. Report of Investigations 7362. Washington, DC: U.S. Bureau of Mines. ———. Flotation of Carbonate Minerals from Unaltered Phosphate Ores of Phosphoria Formation. Report of Investigations 7864. Washington, DC: U.S. Bureau of Mines.

flotation0.book Page 574 Tuesday, January 2, 2007 7:36 PM

574

FLOTATION CHEMISTRY

Rule, A., D. Kirby, and D. Dahlin. 1978. Recent advances in beneficiation of western phosphates. Min. Eng. 1:37–40. Rule, A.R. 1982. Application of carbonate-silica flotation techniques to western phosphate materials. Report of Investigations 8728. Washington, DC: U.S. Bureau of Mines. ———. 1985. Beneficiation of complex phosphate ores containing carbonate and silica gangue. Proc., XVth Int. Miner. Process. Cong. 3:380–389. Santana, A., and A. Peres. 2001. Reverse magnesite flotation. Miner. Eng. 14(1):107–111. Santhana, V., and Y. Attia. 1988. Polymeric flocculants. Pages 485–518 in Reagents in Mineral Technology. Edited by P. Somasundaran and B. Moudgil. New York: Marcel Dekker. Shaw, D.R. 1987. Selective flocculation process for the recovery of phosphate. U.S. Patent 4,690,752. Sis, H., and S. Chander. 2003. Reagents used in the flotation of phosphate ores: A critical review. Miner. Eng. 16:577–585. Smani, S., J.M. Cases, and P. Blazy. 1975a. Beneficiation of sedimentary Moroccan phosphate ore. Part 3. Selective flotation and recovery. Trans. SME-AIME 258:176–180. ———. 1975b. Beneficiation of sedimentary Moroccan phosphate ore. Part 4. Depression of phosphate oolites and calcite flotation. Trans. SME-AIME 258:181–182. Smith, E.L. 1982. Process of phosphate ore beneficiation in the presence of residual organic polymeric flocculants. U.S. Patent 4,309,282. Smith, R. 1988. Cationic and amphoteric collectors. Pages 219–256 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Snow, R.E. 1979. Beneficiation of phosphate ore. U.S. Patent 4,144,969. March 20. ———. 1982. Flotation of phosphate ores containing dolomite. U.S. Patent 4,364,824. December 21. ———. 1990. Sodium silicate as a phosphate flotation modifier. U.S. Patent 4,904,375. Tanaka, Y., N. Katayama, and S. Arai. 1988. Reagents in phosphate flotation. Pages 645–661 in Reagents in Mineral Technology. Edited by P. Somasundaran and B.M. Moudgil. New York: Marcel Dekker. Turrer, H.D.G. 2004. Page 79 in Utilisation of high molecular weight polyacrylamides in the reverse cationic flotation of iron ore. M.Sc. thesis, Curso de Pós-Graduação em Engenharia Metalúrgica/ Universidade Federal de Minas Gerais, Brazil (in Portuguese). Viana, P.R.M., and A.C. Araujo. 2003. Confidential report. Viana, P.R.M., and H.S. Souza. 1988. The use of corn grits as a depressant for the flotation of quartz in hematite ore. Pages 233–244 in Proceedings of the 2nd Latin-American Congress on Froth Flotation, Developments in Mineral Processing, 9. Edited by S.H.F. Castro and J.M. Alvarez. Amsterdam: Elsevier. Xiao, L., and P. Somasundaran. 1989. Interactions between oleate collector and alizarin modifier in dolomite/francolite flotation system. Miner. Metall. Process. 5:100–103. Zellars-Williams Co. 1989. Anionic Flotation of Florida Phosphate. FIPR Publication 02-063-071. Bartow, FL: Florida Institute of Phosphate Research. Zhang, P., Y. Yu, and M. Bogan. 1997. Challenging the Crago double float process II. Amine-fatty acid flotation of siliceous phosphates. Miner. Eng. 10(9):983–994.

flotation0.book Page 575 Tuesday, January 2, 2007 7:36 PM

Flotation of Precious Metals and Their Minerals J.T. Woodcock, G.J. Sparrow, and W.J. Bruckard

INTRODUCTION

This chapter deals with the concentration by flotation of the precious metals (gold, platinum, and silver), their naturally occurring alloys, which vary greatly in composition, and their “true” minerals, which also differ greatly in composition. In addition, because the precious metals often occur in submicroscopic form in common sulfide minerals such as pyrite, pyrrhotite, arsenopyrite, pentlandite, galena, and stibnite, the flotation of those minerals is also taken into account where appropriate. Precious metal minerals are dense, with specific gravities ranging from 10 up to 20 or more, and are often very hard (e.g., some platinum group elements [PGE] minerals) or very soft and ductile (e.g., gold and silver) so that they are not readily ground fine enough to float. Hence, gravity concentration sections are often included in ore treatment grinding circuits to recover early in the circuit those coarse precious metal minerals that are not readily ground fine enough to float. Furthermore, hydrometallurgical methods, such as cyanidation or leaching with some other lixiviant, are often included in a circuit to recover fine particles of gold and silver (e.g., 11 with lime), calcium ions, cyanide (CN– ions or cuprocyanides), sodium sulfite or SO2, sodium sulfide, sodium silicate, tannin and related compounds derived from mine timber or vegetation, starch, and heavy metal ions. These can possibly act by competitive adsorption with collector ions or in various other ways. However, there is not good agreement between the various authors on exactly how the depressants function.

flotation0.book Page 580 Tuesday, January 2, 2007 7:36 PM

580

FLOTATION CHEMISTRY

Activators are rarely needed because clean native gold does not need activation to be floated with the most commonly used collectors, such as xanthates and dithiophosphates. However, an activator could be useful if dirty or coated gold is present. Sulfur dioxide has been used to activate cyanide-depressed native gold and pyrite. Copper sulfate can perform the same function and might also activate gold-iron composites formed during grinding, but the copper then consumes cyanide in any subsequent cyanidation stage; for example, on a roasted flotation concentrate. Teague, van Deventer, and Swaminathan (2000) suggest that a higher recovery of gold can be obtained by adding copper sulfate after the addition of the collector, which is contrary to the usual procedure. Oxygen is needed for the xanthate-gold reaction to occur, but as noted earlier, plant measurements by Woodcock and Jones (1970) indicate that gold flotation pulps are usually almost fully saturated with oxygen. However, with an ore very high in reactive pyrrhotite, oxygen levels could be low. General Flotation Conditions. General flotation conditions such as pulp density, pulp temperature, and flotation cell type do not seem to be particularly important in native gold flotation, and no examples where these parameters were controlled in operating plants were found. However, flotation pulps tend to contain 25%–45% solids in rougher circuits and 10%–20% solids in cleaner circuits as discussed by Broekman, Carter, and Dunne (1987) and Lins and Adamian (1993a). O’Connor, Dunne, and Botelho de Sousa (1984) obtained much better grade-recovery curves for pyrite at 52°C than at 3°C, and similar effects may also apply to native gold and other gold-bearing minerals as well, but this does not seem to have been reported as having been tried in a plant. Many flotation cell types from different manufacturers have been used to float native gold as well as pyrite from a wide variety of ores, and it seems that all these machines can be operated to achieve acceptable results on a specific ore. It has been suggested, however, that shallow cells may be preferable for floating coarse gold. It has also been suggested that mechanically agitated cells may be preferable to air-agitated column cells. Chong et al. (1999) found that replacing flotation columns with mechanically agitated cells for gold scavenging duty in a Canadian plant improved gold recovery by 5%. Dorr and Bosqui (1950) reported that Denver cells were used for flash flotation of ball mill discharge, but other types of cells can also be used. Flotation cell impeller speed, air intake, air dispersion, and bubble size may be important, and Deglon, Egya-Mensah, and Franzidis (2000) have considered these factors in relation to platinum flotation plants in South Africa. Teague, van Deventer, and Swaminathan (1999) have discussed some related factors for gold flotation. Flotation of Gold Tellurides

Gold tellurides and other mineral tellurides have a high degree of natural floatability and are readily floated with frother alone. For example, at the Emperor gold mine in Fiji, where a gold ore containing tellurides and auriferous pyrite is treated (Colbert 1980a), the pH of the ground ore is adjusted with lime to give a value of 9 to depress pyrite, and the tellurides are then floated for about 2 minutes with 40 g/t of the frother Teric 407 in a 5-cell bank of Fagergren cells. Rougher concentrate is cleaned in three stages in Agitair cells to give a final telluride concentrate assaying 2,500 g/t Au. A small amount of sodium silicate is added to the last two cleaning stages to act as a slime gangue depressant and froth modifier. Telluride concentrate is treated in batches for gold and tellurium recovery. Sodium carbonate (0.61 kg/t) is

flotation0.book Page 581 Tuesday, January 2, 2007 7:36 PM

FLOTATION OF PRECIOUS METALS AND THEIR MINERALS

581

added to the telluride rougher tailing to reactivate any pyrite depressed by the lime previously added. The tailing is then floated in a rougher-scavenger-cleaner circuit to give a cleaner pyrite concentrate assaying 85 g/t Au. This concentrate is roasted and cyanided. In addition, valuable information on the flotation and treatment of gold tellurides is given by Field (1963), Kleeman (1963), A.M. Smith (1963), and L.I. Smith (1963). Flotation of Minor Gold Minerals

A few relatively rare gold minerals are known to exist, as listed in Table 1. These include native amalgam, auricupride, aurostibite, maldonite, and gold-carbon associations. Little is known about the flotation properties of these minerals, but it is thought that most of them do float readily with xanthate and dithiophosphate collectors, and some may have natural floatability. Mercury droplets, if small enough, are known to float readily, and flotation is sometimes suggested for cleaning mercury-contaminated soils; therefore, it is assumed that gold amalgam would also float readily. Gold-carbon associations float readily and, when present with gold-free carbonaceous material, cause problems in cyanidation circuits because the carbon tends to collect dissolved gold ions. However, if all the carbonaceous material is floated first as a waste product, this can lead to a loss of gold in the carbonaceous concentrate. Maldonite and/or auricupride occurred in the copper-gold ore at New Occidental Gold Mines N.L., Cobar, New South Wales, Australia, and reported in the copper-gold bulk flotation concentrate. Although the extra gold was welcome, if the bismuth level in the concentrate was above 0.2%, the custom smelter to which it was sent imposed a severe penalty. Aurostibite occurs with stibnite in antimony-gold ores and is floated with the stibnite in flotation circuits. Free native gold and submicroscopic gold in the stibnite are also commonly present in such ores. Flotation of Gold-bearing Sulfide Minerals

Gold in gold-bearing sulfide minerals ranges in size from submicroscopic particles, possibly occurring as individual interstitial atoms in the mineral lattice, up to particles several hundreds microns in size. These minerals, when unoxidized, mostly float readily with xanthates and/or dithiophosphates and a synthetic frother such as MIBC. Thus, the iron sulfides (pyrite, pyrrhotite, and arsenopyrite) rarely cause any problems. They can be depressed by cyanide in cyanidation circuits, but can be reactivated by SO2 conditioning or by the addition of copper sulfate. Iron sulfides are depressed at high pH levels (>10), especially in lime circuits, or by low pH levels, but this can be rectified by a small addition of sodium carbonate to maintain the pH at about 8. This is the most favored pH. Gold Flotation Practice and Flowsheets

Flotation has played an important role in the recovery of gold from many, but not all, types of gold-bearing ores. Much information is available on the practice of gold ore treatment, and only a small fraction can be noted here. Information on Australian practice is given by Blaskett and Woodcock (1953), Elvey and Woodcock (1965), Woodcock (1965), Woodcock (1980), and Woodcock and Hamilton (1993). Details of Canadian practice are given by Carter (1957, pp. 91–162) and Pickett (1978, pp. 45–79). Details of South African practice are given by King (1949), Adamson (1972), and Stanley (1987). Practice in other countries is discussed in Taggart (1945), Dorr and Bosqui (1950), Michell (1950), and Marsden and House (1992).

flotation0.book Page 582 Tuesday, January 2, 2007 7:36 PM

582

FLOTATION CHEMISTRY

It is important to recognize that besides flotation, many other techniques are necessary to recover gold from a particular ore; these include gravity concentration, cyanidation, and smelting. Many crushing and grinding circuits are used, but these are not discussed here. Many types and sizes of flotation machines are used in gold ore processing but are not discussed in detail here as all of them can be operated to give an acceptable result. Attention is directed to the function of flotation in the overall flowsheet for treating: (1) low-sulfide (5% S) refractory gold ores (i.e., those containing gold tellurides and/or submicroscopic gold in pyrite and/or arsenopyrite); and (3) gold-containing base metal ores such as copper-gold ores and stibnite-gold ores. Free-Milling Gold Ores. Free-milling gold ores can be either oxide (or oxidized) ores or sulfide-containing ores. A general flowsheet for this type of ore is shown in Figure 1. Flotation is rarely applied to oxidized or oxide ores because they commonly contain a relatively high proportion of very fine gold, well below the range of application of flotation. Such ores, which may occur at the earth’s surface, are soft and easily mined; some ores containing as little as 1 g/t Au can be profitably mined and treated by cyanidation. Gravity concentration of such ores is commonly used in the grinding circuit to ensure that any relatively coarse gold present is recovered as a separate product, because such gold may not have time to dissolve during the retention time of a typical cyanidation circuit. Sulfide-containing, free-milling ores, however, are commonly treated either by direct cyanidation or by flotation of the fine free gold and auriferous sulfides (mainly pyrite and arsenopyrite) with cyanidation of the flotation concentrate as shown in Figure 1. Gravity concentration in the grinding circuit is usually necessary. The gravity equipment used depends on the size of the gold and includes jigs, Knelson concentrators, Knudsen bowls, and strakes. On many Central Victorian ores in Australia, where walnut-size gold nuggets are common, jigs may be preferred because they can recover the very coarse gold as a bed concentrate and finer gold as a hutch concentrate (Clarke and Thompson 1967). The sulfides and fine free gold in such ores float freely with a xanthate or dithiophosphate collector and MIBC frother, and high recoveries in the concentrate are obtained. At the Wattle Gully mine (Clarke and Thompson 1967), for example, a gold ore containing about 2% sulfides and assaying 10.5 g/t Au was treated by fine grinding with recovery of coarse gold by jigging and amalgamation. The sulfides (pyrite and arsenopyrite) were then floated with xanthate and a frother. The flotation tailings assayed as little as 0.2 g/t Au. The concentrate was cyanided, and the cyanidation tailing was stockpiled for possible future recovery of the gold remaining in the sulfides. Overall gold recovery was 95.3%. Refractory and Mixed Oxide-Sulfide Ores. Refractory ores (i.e., those containing submicroscopic gold in pyrite and arsenopyrite and/or gold tellurides) and mill feed containing mixed oxide and sulfide ores require more complex circuits for satisfactory gold recovery, as indicated in Figure 2. The types of ore involved are (1) mixed oxide-sulfide ores with refractory sulfides, (2) medium- or high-sulfide ores with refractory sulfides and low tellurides, and (3) medium- or high-refractory sulfides with high tellurides. Figure 2 shows three methods used for oxidation of auriferous sulfides (especially pyrite, pyrrhotite, and arsenopyrite) to release the submicroscopic gold present for dissolution by cyanide solution; namely, roasting, pressure oxidation, and bacterial oxidation. Chemical oxidation, such as the use of calcium hypochlorite or potassium permanganate, is not used.

flotation0.book Page 583 Tuesday, January 2, 2007 7:36 PM

FLOTATION OF PRECIOUS METALS AND THEIR MINERALS

583

Low-Sulfide Free-Milling Gold Ore ( highvolatile B bituminous (hvBb) > sub-bituminous A—MIBC quickly loses its effectiveness, first to 2-ethyl hexanol, then to texanol and glycol frothers (e.g., DF-1012). On the hvBb coal, 2-EH (ethyl hexanol) floated 15% more coal than MIBC. Wheeler’s results confirm that though short-chain aliphatic alcohols possess only frothing properties, other frothers

flotation0.book Page 624 Tuesday, January 2, 2007 7:36 PM

624

FLOTATION CHEMISTRY

Ambient Air

Vortex Recirculation Path Disperser Hood

Disperser

Rotor

Draft Tube

Courtesy of Dorr-Oliver Eimco. ©GL&V 2006. All rights reserved.

FIGURE 10

Wemco flotation cell

also exhibit collecting properties and the properties of oil emulsifiers. Therefore, the first conclusion is that in the flotation of lower rank and oxidized coals, a single reagent is not sufficient (see Figure 5). In most cases, a combination of a properly selected frother and an oil, as well as good emulsification, leads to satisfactory flotation. Also, the use of specifically selected promoters may be helpful. Flotation Technology Conventional Machines. Most of the industrial flotation machines used in the coal industry are mechanical (or conventional) cells. These machines consist of a series of agitated tanks (usually 4–8 cells) through which fine coal slurry is passed. The agitators are used to ensure that larger particles are kept in suspension and to disperse air that enters down through the rotating shaft assembly (Figure 10). Air is either injected into the cell using a blower or drawn into the cell by the negative pressure created by the rotating impeller. For coal flotation, trough designs that permit open flow between cells along the bank are more common than cell-to-cell designs that are separated by individual weirs. Some of the major manufacturers of flotation equipment include Wemco, Metso, Svedala, and Outokumpu. The commercial units are very similar in basic design and function, although some slight variations exist in terms of cell geometry and impeller configuration. Machines with large specific surface areas are generally preferred for coal flotation because of the fast flotation kinetics of coal and the large froth solids loadings. Flotation machines with individual cell volumes of up to 28 m3 are commonly used because of advantages in terms of capital, and operating and maintenance costs. Some manufacturers also offer “tank” machines, which consist of cylindrical tanks equipped with conventional impellers. The simplified structural design, which allows these machines to be much larger, can offer significant savings in terms of capital and power costs for some installations. Tank cells with volumes as large as 100 m3 are already in operation at coal plants in Australia. Circuit Variations. Coal flotation is typically performed in a single bank of cells arranged in series with no attempt to reprocess the reject or concentrate streams. As a result, the nonselective recovery of ultrafine clay by hydraulic entrainment can greatly diminish the quality of the froth product. In the minerals industry, the entrainment problem has traditionally been handled using multiple stages of cleaner flotation (Figure 11a). Although this approach has been successfully used for coal flotation, it is generally not attractive in practice because of the low unit value of coal. A more popular circuit (Figure 11b) uses smalldiameter (150-mm) cyclones ahead of flotation to remove a substantial amount of the

flotation0.book Page 625 Tuesday, January 2, 2007 7:36 PM

COAL FLOTATION

625

Reagents

Flotation Bank Deslime Cyclones

Classifying Cyclones

Classifying Cyclones

Flotation Bank

Flotation Bank

Reagents

Spirals

Spirals Fine Sieves

Fine Sieves Coal Rock

Coal Rock B

A

FIGURE 11 Circuits used to minimize the nonselective entrainment of ultrafine (2.5 can cause disruption of the froth for most minerals. The bubble diameter is affected by frother and rotor speed. The measurement is difficult, but new procedures are being developed to achieve a relative size comparison. If too much air is added to the machine, pulp density decreases and coarse particle settling can occur.

flotation0.book Page 679 Tuesday, January 2, 2007 7:36 PM

MECHANICAL FROTH FLOTATION CELLS

679

Area (A)

Surface (S)

Airflow (Q)

FIGURE 4

Bubble-surface-area flux

Tank Design Criteria

The evolution of flotation machines has gravitated from rectangular and U-shaped tanks to cylindrical vessels. Benefits derived from cylindrical tanks include reduced particle sanding due to tank corner elimination, fabrication cost reductions, and decreased power consumption per unit volume. Although the tank shape has changed, the cell’s aspect ratio, height of the pulp to tank diameter (H/D), has remained relatively constant. For the Wemco product line, H/D ranges from 0.68 to 0.75. In rougher applications, H/D ranges from 0.7 to 0.9 for the Dorr-Oliver machines, and for cleaner applications, the H/D ranges from 0.9 to 1.2. Both the Dorr-Oliver and Wemco cylindrical vessels incorporate beveled bottoms in their tank designs. This feature has been found to greatly reduce oversize particle settling and increase mixing within the cells. Another important design consideration is the tank-totank open connection. Changes in pulp velocity through this opening have an effect on pressure or head drop. This effect is expected to vary by the square of the velocity. Therefore, accurate flow rates through the flotation circuits are required to properly design this connection. As cell volumes increase, the designer must be aware of the cubic relationship in volume compared to the square relationship of the froth surface area. The result is that retention time increases more rapidly than froth surface area. In some applications, such as cleaner circuits or coal flotation, the limiting factor on recovery may not be residence time but froth removal rate. External peripheral launders can be used in place of the standard internal peripheral launder to increase froth surface area. In addition, most cylindrical cells now incorporate radial launders to increase weir length and reduce the froth travel distance. REFERENCES

Arbiter, N., C.C. Harris, and R.F. Yap. 1969. Hydrodynamics of flotation cells. Trans. SME-AIME 244:134–148. Bull, W.R., and D.J. Spottiswood. 1974. A study of mixing patterns in a bank of flotation cells. Q. Colo. School of Mines, 69:1–26. Degner, V.R. 1987. 3000 Cubic foot flotation machine development. Internal document of DorrOliver Eimco. Degner, V.R., and H.B. Treweek. 1976. Large flotation cell design and development. Pages 816–837 in Flotation. Volume 2. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Dinsdale, J.D., and Y. Berube. 1972. A characteristic of hydrodynamics in a 700-cu.ft. Maxwell flotation cell. Can. Metall. Q. 11:507–513

flotation0.book Page 680 Tuesday, January 2, 2007 7:36 PM

680

FLOTATION CELLS, MODELING, AND SIMULATION

Gardner, R.P., H.M. Lee, and B. Yu. 1980. Development of radioactive tracer methods for applying the mechanistic approach to continuous multiphase particle flotation processes. Pages 922–943 in Fine Particle Processes, Proceedings of the International Symposium. Volume 1. Edited by P. Somasundaran. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Gorain, B.K., J.P. Franzidis, and E.V. Manlapig. 1997. Studies on impeller type, impeller speed and airflow rate in an industrial scale flotation cell. Part 4: Effect of bubble surface area flux on flotation performance. Miner. Eng. 10(4):367–379. Harris, C.C. 1976. Flotation machines. Pages 753–815 in Flotation. Volume II. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Harris, C.C., and A. Cuadros-Paz. 1978. Species interaction in flotation: A laboratory scale semibatch study. Int. J. Miner. Process. 5:267–283. Kallionen, J. 1999. Advances in application driven design of flotation cells. Pages 29–39 in Proceedings of the Copper 99/Cobre 99 International Environmental Conference. Volume II: Mineral Processing/Environment, Health and Safety. Edited by B.A. Hancock and M.R.L. Pon. Warrendale, PA: Minerals, Metals, and Materials Society. Klimpel, R.R. 1980. Selection of chemical reagents for flotation. Pages 907–934 in Mineral Processing Plant Design. Edited by A.L. Mular and R.B. Bhappu. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Lawrence, G.A., and J.C. McHardy. 1985. Performance characteristics of an improved flotation mechanism. Presented at the 17th Canadian Mineral Processing Conference, Ottawa, Canada. Levenspiel, O. 1962. Mixed models to represent flow of fluids through vessels. Can. J. Chem. Eng. 40:135–162. ———. 1989. The Chemical Reactor OmniBook. Corvallis, OR: Oregon State University Book Stores. Nelson, M.G., and D. Lelinski. 2000. Hydrodynamic design of self-aerating flotation machines. Miner. Eng. 13(10–11):991–998. Weber, A., C. Walker, L. Redden, D. Lelinski, and S. Ware. 1999. Scale-up and design of large-scale flotation equipment. Pages 353–369 in Advances in Flotation Technology. Edited by B.K. Parekh and J.D. Miller. Littleton, CO: SME. Wen, C.Y., and L.T. Fan. 1975. Models for Flow Systems and Chemical Reactors. New York: Marcel Dekker.

flotation0.book Page 681 Tuesday, January 2, 2007 7:36 PM

Column Flotation J.A. Finch, J. Cilliers, and J. Yianatos

A B S T R AC T

The flotation column is one of a class of pneumatic cells whose history is covered elsewhere in this book. Column flotation appears to be the term introduced in the early 1960s following the patent of Boutin and Tremblay, which launched the Column Flotation Company of Canada. The extensive trials at various locations led to commercial applications, initially at Les Mines Gaspé, then worldwide. The success of columns helped stimulate the new generation of flotation machines, from the Jameson to the various tank cell designs. The objective of this chapter is not to review design or operating practices, which are adequately dealt with elsewhere but rather to provide some fundamental background material on which the interested student can build. After a brief introduction to the “conventional” flotation column, the chapter is organized into three divisions: Section I, the Collection Zone (by J.A. Finch); Section II, the Froth Zone (by J. Cilliers); and Section III, Column Control (by J. Yianatos). T H E C O N V E N T I O N A L F L O TAT I O N C O L U M N

This review builds on three extensive previous ones (Finch and Dobby 1990; Rubenstein 1995; and Finch, Uribe-Salas, and Xu 1995) and focuses on the conventional flotation column as patented by Boutin and Tremblay (1963), which is now used worldwide. Figures* 1a and 1b show a photograph and a schematic of the conventional column, respectively. Feed enters at one-third of the distance from the top and descends against a rising swarm of bubbles generated at a variety of bubble generators or spargers. The collected (hydrophobic) particles form bubble–particle aggregates that rise to establish a froth. This defines the two major zones (both in terms of hydrodynamic characteristics and metallurgical function): the collection (slurry or pulp) zone and the froth zone. A key feature is the addition of wash water into the froth, which promotes rejection (cleaning) of entrained particles from the float product. This action provides the alternative name for the froth, the cleaning zone. The target is to maintain a net downward flow of wash water, referred to as a “positive bias,” to maximize the cleaning action. Figure 1 includes some typical dimensions and flow rates. The latter are expressed as a superficial velocity, Ji = Q i/A, where Q i is the volumetric flow rate of phase i (i.e., air or gas, g; slurry, sl; water, w; etc.), and A is the cross-sectional area of the column. Typical units of Ji are centimeters per second (i.e., Qi is in cubic centimeters per second and A in square centimeters). Superficial rates are employed because, to a large extent, values are independent of machine size (i.e., A), thus, even laboratory- and full-scale machines can be compared on the basis of superficial velocities. * Figures, tables, and equations are numbered separately in each section. References are listed at the end of each section, respectively. 681

flotation0.book Page 682 Tuesday, January 2, 2007 7:36 PM

682

FLOTATION CELLS, MODELING, AND SIMULATION

A

B Wash Water (0.05–0.3 cm/sec) Froth Zone, Hf Interface

Concentrate

Feed (1 cm/sec)

Collection Zone, Hc Bubbles (Diameter 0.5–3 mm)

Dc Tails (1 cm/sec)

FIGURE 1 (a) McGill gas dispersion team about to probe a column at Agnico-Eagle’s LaRonde concentrator, Quebec, Canada; and (b) a general schematic of a conventional flotation column

Section I: The Collection Zone J.A. Finch

INTRODUCTION

The collection (pulp or slurry) zone is one of interacting bubble and particle dispersions. There is considerable literature on these dispersions separately but relatively little on their interaction under the unique conditions of flotation, namely, the presence of reagents that stabilize the bubbles against coalescence and the presence of particles ranging in size (often down to 181, corresponding to Db > 1.15 mm. Thus, it appears that bubbles of flotation size carry a volume of water greater than the volume of the bubble. Pursuing the connection suggests that the wake is the source of water entering the froth. Smith and Warren (1989) have raised reasonable objections to the wake being the source, proposing instead a “bubble swarm” effect where water is pushed into the froth by advancing layers of bubbles. This source is also challenged in the “Water-Carrying Rate” subsection. Predicting Sb–Jg and Sb–εεg. The same cases given in Table 1 can be readily extended to predict Sb –Jg and Sb –εg relationships (Figures 16 and 17, respectively). Figure 16 (Sb –Jg ) includes two solutions for case 4, Db = 2.5 mm (A) and Db = 3.5 mm (B) as, unlike gas holdup, Sb is dependent on bubble size for Db > 2 mm. Because Db is fixed in cases 1 to 4, the slope is constant (reciprocal of Db). Case 5, where bubble size increases with increasing gas rate, gives the more common trend (Figure 13). This can be modeled by combining the definition of Sb with the general form of Equation 5a, namely 6J g S b = -----------------------D 0 + CJ g n

(EQ 8)

140 1 2 3 4A 4B 5

Surface Area Flux, Sb, sec –1

120 100 80 60 40 20 0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Gas Rate, Jg, cm/sec

FIGURE 16

Bubble surface area flux vs. gas rate (up to Jgmax) 160 1 2 3 4A 4B 5

Surface Area Flux, Sb, sec –1

140 120 100

Sb

=

ε 5.5

g

80 60 40 20 0 0

5

10

15

20

25

30

Gas Holdup, εg (%)

FIGURE 17

Bubble surface area flux vs. gas holdup (including correlation Sb = 5.5 εg)

flotation0.book Page 697 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

697

Figure 17 shows the Sb –εg relationship. Apart from case 4 (i.e., the absence of frother), the trend approximates Sb (sec–1) ≈ 5.5 εg (%) (Finch et al. 2000). With knowledge of gas holdup and bubble size, the relationship with interfacial area (= 6 εg/Db) could be explored as well. This opens a mass transfer approach to modeling flotation, which is briefly considered in the “Flotation as a Rate Process” subsection. Estimating Bubble Size from εg–Jg Data. To illustrate the estimation of bubble size from εg–Jg data, two examples are taken from Banisi and Finch (1993) that include the widest range in bubble size. The estimated bubble size (Table 2) corresponds well to the measured Sauter mean of the distribution. The use of drift flux analysis assumes a unique relationship among gas holdup, bubble size, and gas rate, one that is independent of frother type, for instance. This has been challenged by Zhou, Egiebor, and Plitt (1993). There are some issues that need addressing, for example, the Db –UbT relationship. There is a complication in measuring terminal velocity in surfactant (frother) solutions; the velocity shows a profile from the point of bubble generation, initially rapidly increasing, reaching a maximum, then decreasing to reach a steadystate value (Sam, Gomez, and Finch 1996; Krzan and Malysa 2002). The steady-state value is the measure of terminal velocity. The time to reach the steady-state value increases as concentration decreases, and it is possible that some measurements have not allowed sufficient bubble rise time. Sam, Gomez, and Finch (1996), using a 400-cm column, concluded that there was no concentration effect, suggesting that given enough rising distance, the bubble will eventually reach sufficient surface coverage to attain the same terminal velocity. Other researchers, using shorter columns, have reported a dependence of UbT on concentration (Krzan and Malysa 2002; Zhou, Egiebor, and Plitt 1992; Fuerstenau and Wayman 1958). Regardless, for conditions relevant to flotation (frother concentrations greater than at least 1–2 ppm), the evidence shows that terminal velocity is independent of frother concentration. An effect of frother type on velocity may also be related to the profile; measuring velocity over a given distance for frothers having marked differences in profile will produce an apparent “frother effect.” The question as to whether frother type influences the terminal velocity (i.e., final, steady-state velocity in the profile) is an intriguing one. The data of Krzan and Malysa (2002) for a series of n-alcohols suggest no differences, but Sam, Gomez, TABLE 2

Estimation of bubble size

Jg, cm/sec

Jl, cm/sec

ε g, %

UbT, mm, Equation 6

Db, mm, Equation 4a

Db, mm, measured

0.91 1.00

9.50 12.3

14.08 6.77

1.06 0.58

1.20 0.62

1.0 0.5 Source: Banisi and Finch 1993

TABLE 3

Effect of frother type on terminal velocity Terminal Velocity, cm/sec, at Bubble Diameter, mm

Frother Type MIBC Dowfroth 250 Pine oil

0.9 12.0 ± 0.3 11.5 ± 0.4 11.0 ± 0.3

Adapted from Sam, Gomez, and Finch 1996.

1.5 16.5 ± 0.3 15.6 ± 0.5 14.8 ± 0.2

2.2 19.3 ± 0.3 18.2 ± 0.4 17.0 ± 0.3

2.7 21.7 ± 0.3 20.8 ± 0.5 18.0 ± 0.3

flotation0.book Page 698 Tuesday, January 2, 2007 7:36 PM

698

FLOTATION CELLS, MODELING, AND SIMULATION

and Finch (1996) concluded there was a difference in the case of three commercial frothers (Table 3). A broader range of frothers needs to be examined. Although the emphasis in this subsection was to show the use of drift flux to predict bubble size, the calculated size has potential use on its own. It could be referred to as the “drift flux equivalent” bubble size. Swarm Velocity

The average bubble rise velocity in a swarm or swarm velocity is given by Jg/εg (Zhou, Egiebor, and Plitt 1993). Generally, Jg/εg decreases with increasing Jg , as predicted by the drift flux model (e.g., consider Equation 6b reflecting the growing hindering effect because of increasing gas holdup), or is roughly constant, following the common observation that the εg–Jg relationship is linear. There are a few independent measures of the swarm velocity. One method is to relate the average velocity with the buoyancy velocity (Ub0), which is the velocity of the bubble front created when the gas is shut off (as introduced in Figure 11), where Jg ---- = U b0 + J g + J l εg

(EQ 9)

Zhou, Egiebor, and Plitt (1993) found that Equation 9 was obeyed for water only but not in the presence of frother. Shen and Finch (1996) used a different approach. By adapting the technique introduced to measure Ub0, step changes in gas rate were introduced, and the velocity of the front created was determined. By interpolating the velocity to zero step change, a measure of swarm velocity was made. To distinguish from Jg/εg , this was called the “hindered” velocity, UbH. In a study on batch water, they found that UbH < Jg/εg. The argument advanced was that Jg/εg may be the velocity associated with perfect bubbly flow (unisized bubbles) and that UbH is the practical measure for the real situation, imperfect bubble flow. A third method is given by Yianatos et al. (1994). They measured the mean residence time of gas (τg ) using a radioactive tracer technique; thus, the average bubble velocity is given by Hc/τg. In an operating column (i.e., frother and solids are present), Hc/τg is roughly constant (~5.9–6.3 cm/sec) as a function of Jg (1.3–2.1cm/sec), and, like Shen and Finch (1996), found this measure to be less than Jg/εg (9.5–11.3 cm/sec). Resolving how to estimate swarm velocity is worth pursuing because understanding the impact of factors as diverse as slurry rheology, process chemistry, and changes in cell geometry (e.g., due to insertion of launders, froth crowders) ultimately returns to the effect on swarm velocity. F L O TAT I O N A S A R AT E P R O C E S S

There are two approaches to flotation as a rate process: analogy with chemical kinetics and with mass transfer (Ityokumbul 1994). The mass transfer approach requires knowledge of the bubble loading (fraction of bubble surface covered with particles) and bubble interfacial area (= 6 εg/Db), which has probably deterred using this route. The kinetic model remains central to flotation simulation (Harris et al. 2002) and flotation column scale-up (Dobby 2002).

flotation0.book Page 699 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

699

Particle Collection

A common starting point in developing a kinetic model of flotation is to calculate a rate of collection by considering the number of particles in the path of a rising single bubble, applying a collection probability (or efficiency) factor (fraction of particles in the path that are collected), and multiplying by the number of bubbles. The various formulations ( Jameson, Nam, and Moo-Young 1977; Finch and Dobby 1990; King 2002) can be transformed to 3 Jg rate of collection = ⎛ -- ------P k⎞ C ⎝ 2 Db ⎠

(EQ 10a)

1 rate of collection = ⎛ --S b P k⎞ C ⎝4 ⎠

(EQ 10b)

or

where Pk is the particle collection efficiency and C is the particle concentration (particles per unit volume of cell). The result shows the process to be first order with respect to C with a rate constant, k, given by 1 k = --P k S b 4

(EQ 11)

The physics of the process thus analyzed (rising bubbles encountering particles) appears to apply more to columns than to mechanical machines but remains an inevitable simplification. One assumption is that the availability of bubble surface is not rate limiting; alreadycollected particles do not interfere with subsequent collection events (i.e., the bubble remains lightly loaded). This is difficult to establish for any given system. If particles are collected on the upstream side of a bubble and are swept to the rear and accumulate, then perhaps surface for collection remains available even up to 50% loading (see Figure 8 in Section II). A recent study suggests loadings well below this amount (~20%), even when the carrying capacity is reached (Li, Del Villar, and Gomez 2004). Whether the kinetic model is applicable must be evaluated in each case. Figure 18 shows a test of Equation 10 in de-inking recycled paper using laboratory- and pilot-scale columns (Hernandez, Gomez, and Finch 2003). The conditions used to estimate k are appropriate to first-order kinetics, namely, low particle concentration (~1% w/w) to restrict bubble loading. The linear k–Sb relationship supports the observation of Gorain et al. (1998), expressed by k = PS b

(EQ 12)

where P is particle “floatability” (and is equal to Pk/4), which—similar to collection efficiency, Pk—will depend on variables such as particle size and hydrophobicity. The formulation has the attraction of isolating machine-related factors (which control Sb) from ore/ chemistry factors (which control P). The linearity expressed in Equation 12 implies that P is not a function of bubble size, which is debatable. Analysis of the collection subprocesses—namely, collision, attachment, and detachment—suggests P α 1/Dbn where n may range up to 2 (Ahmed and Jameson 1989). This is supported by single bubble–particle-encounter experiments (Anfruns and

flotation0.book Page 700 Tuesday, January 2, 2007 7:36 PM

700

FLOTATION CELLS, MODELING, AND SIMULATION

Flotation Rate Constant, k, min –1

1.4 4-in. Laboratory 20-in. Pilot

1.2 1.0 0.8 0.6 0.4 0.2 0.0 0

20

40

60

80

100

120

140

160

Bubble Surface Area Flux, Sb, sec –1

Adapted from Hernandez, Gomez, and Finch 2003.

FIGURE 18

Test of the k–Sb model for the collection zone in two columns

Kitchener 1976; Yoon and Luttrell 1989). It may be that systems comprising bubble and particle swarms (i.e., the flotation condition) obscure the dependence. Diaz-Penafiel and Dobby (1994) advanced a similar argument. At present, the simple linear k–Sb model, although vigorously challenged (Heiskenan 2000), seems worth retaining until evidence forces a change. This is the view taken, for instance, in the development of JKSimFloat software (Harris et al. 2002). Transport

To apply the kinetic model, particle transport (or mixing) in the cell is necessary, described by a residence time distribution (RTD). The common assumption is that particle RTD approximates to liquid RTD, which can be more readily obtained (e.g., injecting salt tracer and monitoring conductivity at the discharge). There are two extremes: plug flow (also applicable to batch conditions) and fully (or perfectly) mixed flow. The former is distinguished by all particles having the same residence time “t” and the latter by particles exhibiting a residence time decaying exponentially with a mean residence time, “τ.” The extremes provide for simple solutions to model recovery, R. For plug flow, the recovery equation is R = 1 – exp ( – kt )

(EQ 12a)

kτ R = ------------------( 1 + kτ )

(EQ 12b)

and for fully mixed flow:

The models assume that maximum recovery (Rmax) is 100% and that k is a single value. Models allowing for Rmax < 100% and a distribution in k can be substituted, perhaps the most successful being that by Klimpel (1980). In columns, one key element to the transport behavior is the height-to-diameter ratio, Hc/Dc. The plug flow approximation holds for many laboratory columns where Hc/Dc > 20:1, and the fully mixed flow approximation is appropriate for industrial columns for which Hc/Dc < 5:1. Mixing characteristics of columns (and other flotation machines) have recently been reviewed by Yianatos et al. (2005). (See also Section III.)

flotation0.book Page 701 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

701

Noting the advantage of plug flow—namely, that for a given residence time, the recovery and selectivity will be higher—flotation columns were often baffled to create sections with higher Hc/Dc. A series of diagnostic tests indicated that significant differences in gas holdup can easily occur between the sections because it is almost impossible to arrange an even distribution of air and feed to each (Tavera et al. 1997). The corresponding difference in bulk density then drives circulation of slurry between the sections, degrading performance. (In the case study in question, the baffles were subsequently removed.) Combining Equations 11 and 12a yields the first element in building the JKSimFloat flotation simulator (Harris et al. 2002), the recovery of true floated particles in the collection zone: PS b τ R = -------------------1 + PS b τ

(EQ 13)

To develop the model further requires inclusion of phenomena such as bubble loading (carrying capacity), entrainment, and interaction with the froth zone. Two Applications of Equation 13 Effect of Changing Sparger Type. Columns often operate at quite low stage recoveries (see Section III), which operations may want to increase. One option is to consider alternative sparger types. There is potential to approximately double the Sb in some cases. The impact on recovery can be predicted to indicate whether the change is worthwhile. The trade-off between increasing Sb and adding more cells can then be evaluated. Optimum Sb. Optimum Sb addresses the Sb to maximize some aspect of performance, such as selectivity defined by the difference in recovery for two minerals (A and B). The recovery difference is given by

PA Sb τ PB Sb τ R A – E B = ----------------------- – ----------------------1 + PA Sb τ 1 + PB Sb τ

(EQ 14)

The maximum difference in recovery then occurs (differentiating and setting to zero) at 1 S b = -------------------τ PA PB

(EQ 15)

Thus, for given residence time and mineral floatabilities, there is a value of Sb that maximizes selectivity. This optimum is “local,” and it has proved difficult to identify conditions that optimize selectivity for circuits ( Jowett and Sutherland 1985). Water-Carrying Rate

In addition to particles, bubbles transport water which controls the nonselective entrainment of particles and often limits the grade of the float product. Section II addresses transport through the froth; the concern here is the water carried into the froth. This is called the water-carrying rate and is quoted on a per-unit area of cell basis, Jwf. Two methods are considered to estimate Jwf : one is based on the volume of wake per bubble (Volw), and the second is based on an equivalent water layer thickness per bubble.

flotation0.book Page 702 Tuesday, January 2, 2007 7:36 PM

702

FLOTATION CELLS, MODELING, AND SIMULATION

Adapting the definitions in Figure 2, the first approach gives ( nVol w ) J wf = ------------------A

(EQ 16a)

Substituting for n and Volw using Equation 7 arrives at Qg J wf = ⎛ ---------------⎞ Vol b ( 0.045Re b – 0.314 ) ⎝ AVol b⎠

(EQ 16b)

J wf = J g ( 0.045Re b – 0.314 )

(EQ 16c)

which simplifies to

Over the applicable range in Reb of Equation 7 (approximately 0.5- to 1.5-mm bubbles), the ratio Jwf/Jg increases from ~0.08 to 1.6, which is opposite to that which is usually observed, namely, that the volume of water entering the froth increases as bubble size decreases. The wake does not seem to be the source of water carried into the froth. The second approach assumes that water is carried as a layer on the bubble surface (Xu, Uribe-Salas, and Finch 1991; Bascur and Herbst 1982), which gives, assuming uniform thickness δ nSδ J wf = --------- = S b δ A

(EQ 17)

where n is the number of bubbles per unit time, S is the bubble surface area, and A is the column cross-sectional area (see Figure 2). Xu, Uribe-Salas, and Finch (1991), who introduced Sb for the purpose of quantifying water transport, showed that the water-carrying rate did vary with Sb. Bascur and Herbst (1982) tried to predict δ. By measuring Jwf and Sb, δ could be estimated. A first attempt was compromised because at the frother dosages used (>20 ppm), gas holdup increased, but bubble size did not materially decrease (the identical situation discussed with regard to Figure 9), and therefore Sb was constant. Substituting εg for Sb gave the result shown in Figure 19. For the family of n-alcohols, Jwf is clearly dependent on εg—and, significantly, frother type—increasing as chain length expands. From Equation 17, this implies that film thickness δ increases with chain length. Direct measurement of the thickness of water film on bubbles blown with these same alcohols supports this conclusion (Gélinas, Finch, and Gouet-Kaplan 2005). The findings indicate a necessity to include chemical as well as physical properties to model how water is carried into the froth. This would modify the “bubble swarm” effect model of Smith and Warren (1989) mentioned previously. B U B B L E G E N E R AT I O N Current Situation

The search for a reliable bubble generation system in pneumatic cells continues. A reasonable expectation is to disperse air into bubbles of a size comparable to mechanical cells (approximately 0.5–2 mm) at similar gas rates (0.5 < Jg < 2 cm/sec) with minimum maintenance. The evolution is described generally in the following paragraphs (Finch 1995).

flotation0.book Page 703 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

703

Water Carrying Rate, Jwf, cm/sec

0.5 Pentanol Hexanol Heptanol Octanol

0.4

0.3

0.2

0.1

0.0 10

15

20

25

30

35

Gas Holdup, εg, %

Data courtesy of P. Moyo, current research, McGill University, Montreal, Canada.

FIGURE 19 Water-carrying rate vs. gas holdup for four n-alcohols (Jwf is measured at a froth depth of 7 cm)

Initially, spargers were made using various flexible porous media (rubber, cloth, etc). These were often high-maintenance items due to fouling by particles and precipitates. Attempts were made to jet a mix of air and water through orifices along a tube inserted in the column. The addition of water enhanced fine bubble production, but maintenance again proved limiting. At many sites, the water was switched off (and gas rate increased to compensate for the larger bubbles produced to try to preserve Sb). Currently, the two principal sparger systems are based on jetting and slurry–air contacting external to the column. Jet-type spargers rely on forcing air at high pressure (30–100 psig) through annular or circular orifices, typically ~1 mm diameter (Dobby 2002). Shear along the jet surface generates the bubble dispersion. The second method relies on pumping slurry, drawn from the column, through a contactor where it mixes and shears air into a bubble dispersion that is injected into the column. A commercial system is the Microcel where the contactor is an in-line mixer (Brake et al. 1996). A variation using a Venturi contactor has been used in laboratory columns (Xu and Stratton-Crawley 1996; Hardie, Gomez, and Finch 1999). Both the jet and external contactor types are designed for online replacement and maintenance (i.e., there is no need to shut down the column). Some Characteristics of the Sparging Systems

Ideally, at this point, a comparison of gas dispersion properties of the two commercial systems would be introduced. To do so, however, requires testing under the same chemical and rheological conditions. Presently, there are few such data. Figure 20 summarizes the Sb –Jg trends measured at two concentrators: plant A where a jet sparger and a Microcel were compared in the same column, and plant B where a jet sparger is compared with a mechanical cell. (NOTE : It is important to vary gas rate over a wide range to be confident in the trends in order to compare systems.) Plant B process water has a high salt content that probably provides most of the anticoalescence function, so although the two machine types are in different parts of the circuit, the comparison is considered valid. The trends obtained from drift flux analysis (cases 4A and 5) are included for reference. The Microcel appears to yield a higher slope (i.e., smaller D32) than the jet

flotation0.book Page 704 Tuesday, January 2, 2007 7:36 PM

704

FLOTATION CELLS, MODELING, AND SIMULATION

Surface Area Flux, Sb, sec –1

120 100

Plant A In-Line Mixer Jet Type

Plant B Mechanical Cell Jet Type

80 Case 5 60 40 Case 4A

20 0 0.0

0.5

1.0

1.5

2.0

Gas Rate, Jg, cm/sec

FIGURE 20 Bubble surface area flux vs. gas rate: Comparison of sparger types and mechanical cell (cases 4A and 5 from Table 1 are included for reference)

spargers,* but neither approaches the performance of the mechanical cell. A more definitive comparison of these spargers has yet to be accomplished. Another comparison of the spargers involves a series of works on flotation de-inking of recycled paper. On the basis of ink recovery versus Sb, Hardie, Gomez, and Finch (2004) observed that the in-line mixer gave equivalent performance to a porous (stainless-steel) sparger, although Finch et al. (1999) found that a jet sparger could not match the performance of a porous sparger. Together, these observations again hint that the in-line mixer produced finer bubbles (equivalent to a porous media) than the jet system. Extensive testing of jet-type spargers suggests that there are two bubble-production mechanisms (Bailey, Gomez, and Finch 2005), namely, shear along the jet surface that is responsive to frother (i.e., coalescence is controlling) and turbulent breakup at the end of the jet that produces large bubbles (approximately 5–10 mm) independent of frother (i.e., coalescence is not a factor). The large bubbles generated from the second source limit the Sb that is attainable. A similar detailed evaluation of the in-line mixer system is needed. Given the long-standing concern over the bubble-generating system in pneumatic cells, a concerted characterization effort is overdue. CONCLUDING REMARKS

The properties of the collection zone of a flotation column have been reviewed. There is limited understanding of the interaction of the bubble and particle swarms. Use has been made of the gas holdup–gas rate relationship to qualitatively describe the impact of common variables. Techniques ranging from drift flux analysis to swarm velocity measurements were applied to try to interpret the behavior. Lack of measurement tools that restrict the analysis have been noted and the new gas dispersion sensors, which may provide some insights, have been described. It has been observed that the unique conditions of flotation, primarily the use of frothers to help produce small bubbles, means that mineral processing engineers are largely “on their own” when it comes to relevant studies. Some thoughts on the action of frothers have been offered. * At a comparable gas rate of ~0.6 cm/sec, the in-line mixer exhibits about twice the Sb of the jet type; this observation was used in the “Effect of Changing Sparger Type” subsection.

flotation0.book Page 705 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

705

Models (drift flux analysis and particle collection kinetics) are used to provide some practical inputs to the process. The problem of reliable bubble generation remains. The limited data on the two common commercial systems suggests that bubble size still does not match that achieved in mechanical cells. AC K N OW L E D G M E N T S

The bulk of the work on which this review is based was conducted under a succession of grants from the Natural Sciences and Engineering Research Council of Canada (NSERC), sponsored by, at various times, Inco, Falconbridge, Noranda, Teck Cominco, COREM, SGS Lakefield Research, and Amira International under the P9 program. REFERENCES

Ahmed, N., and G.J. Jameson. 1989. Flotation kinetics. Pages 77–100 in Frothing in Flotation, A Volume in Honor of J. Leja. Edited by J. Laskowski. New York: Gordon and Breach Science Publishers. Anfruns, J.P., and J.A. Kitchener. 1976. The absolute rate of capture of single particles by single bubbles. Pages 625–637 in Flotation, A.M. Gaudin Memorial Volume. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Bailey, M., C.O. Gomez, and J.A. Finch. 2005. Development of an image analysis method for wide bubble size distributions. Miner. Eng. 18:1214–1221. Banisi, S., and J.A. Finch. 1993. Reconciliation of bubble size estimation methods using drift flux analysis. Miner. Eng. 7:1555–1559. Banisi, S., A.R. Laplante, J.A. Finch, and M.E. Weber. 1995. Effect of solid particles on gas holdup in flotation columns, part II. Investigation of mechanisms of gas holdup reduction in presence of solids. Chem. Eng. Sci. 50:2335–2342. Bascur, O.A., and J.A. Herbst. 1982. Dynamic modelling of a flotation cell with a view toward automatic control. 14th International Mineral Processing Congress, Toronto, Canada. CIM 111:11.1–11.22. Bazin, C., and C. B-Chapleau. 2005. The difficulty associated with measuring slurry rheological properties and linking them to grinding mill performance. Miner. Eng. 79:93–99. Bouaifi, M., G. Hebrard, D. Bastouf, and M. Roustan. 2001. A comparative study of gas holdup, bubble size, interfacial area and mass transfer coefficients in stirred gas-liquid reactors and bubble columns. Chem. Eng. Process. 40:97–111. Boutin, P., and R.J. Tremblay. 1963. Method and apparatus for the froth flotation of ores. G.B. Patent GB970,841. Brake, I., G. Eldridge, G. Luttrell, and R.H. Yoon. 1996. The design of industrial MicrocelTM sparging systems. Pages 13–23 in Column ’96: Proceedings of an International Conference on Column Flotation. Edited by C.O. Gomez and J.A. Finch. Montreal: Canadian Institute of Mining, Metallurgy and Petroleum. Cho, Y.S., and J.S. Laskowski. 2002. Effect of frothers on bubble size and foam stability. Int. J. Miner. Process. 64:69–80. Clift, R., J.R. Grace, and M.E. Weber. 1978. Pages 321–347 in Bubbles, Drops and Particles. New York: Academic Press. Dahlke, R., C.O. Gomez, and J.A. Finch. 2005. Operating range of a flotation cell from gas holdup vs. gas rate. Miner. Eng. 18:977–980. Diaz-Penafiel, P., and G.S. Dobby. 1994. Kinetic studies in flotation columns: Bubble size effect. Miner. Eng. 7:465–478. Dobby, G., J. Yianatos, and J.A. Finch. 1988. Estimation of bubble diameter in flotation columns from drift flux analysis. Can. Metall. Q. 27:85–90.

flotation0.book Page 706 Tuesday, January 2, 2007 7:36 PM

706

FLOTATION CELLS, MODELING, AND SIMULATION

Dobby, G.S. 2002. Column flotation. Pages 1239–1252 in Proceedings, Mineral Processing Plant Design, Practice and Control. Volume 1.. Edited by A.L. Mular, D.N. Halbe, and D.J. Barratt. Littleton, CO: SME. Fan, L.S., and K. Tsuchiya. 1990. Pages 143–178 in Bubble Wake Dynamics in Liquids and LiquidSolid Suspensions. Boston: Butterworth-Heinemann. Finch, J.A. 1995. Column flotation: A selected review. Part IV—novel flotation devices. Miner. Eng. 8:587–602. Finch, J.A., and G.S. Dobby. 1990. Column Flotation. Oxford: Pergamon Press. Finch, J.A., C.O. Gomez, C.A. Hardie, and G. Leichtle. 1999. Bubble surface area flux: A parameter to characterize flotation cells. Pages 200–209 in Proceedings of the 31st Annual Meeting of the Canadian Mineral Processors, Ottawa. Montreal: Canadian Mineral Processors, Division of CIM. Finch, J.A., A. Uribe-Salas, and M. Xu. 1995. Column flotation. Pages 291–330 in Flotation Science and Engineering. Edited by K.A. Matis. New York: Marcel Dekker. Finch, J.A., J. Xiao, C. Hardie, and C.O. Gomez. 2000. Gas dispersion properties: Bubble surface area flux and gas holdup. Miner. Eng. 13:365–372. Fuerstenau, D.W., and C.H. Wayman. 1958. Effect of chemical reagents on the motion of single air bubbles in water. Min. Eng. ( June): 694–699. Gandhi, B., A. Prakash, and M.A. Bergounou. 1999. Hydrodynamic behavior of slurry bubble columns at high solids concentrations. Powder Technol. 103:80–94. Garibay, R.P., P.M. Gallegos, A. Uribe-Salas, and F. Nava. 2002. Effect of collection zone height and operating variables on recovery of overload flotation columns. Miner. Eng. 15:325–331. Gélinas, S., J.A. Finch, and F. Cappuccitti. 2005. Frother analysis: Procedure and plant experience. Pages 569–576 in Proceedings of the 37th Annual Meeting of the Canadian Mineral Processors of CIM. January 18–20. Montreal: Canadian Mineral Processors, Division of CIM. Gélinas, S., J.A. Finch, and M. Gouet-Kaplan. 2005. Comparative real-time characterization of frother bubble thin films. J. Colloid Interface Sci. 291:187–191. George, P., A.V. Nguyen, and G.J. Jameson. 2004. Assessment of true flotation and entrainment in the flotation of submicron particles by fine bubbles. Miner. Eng. 17:847–853. Gomez, C.O., F. Cortés-López, and J.A. Finch. 2003. Industrial testing of a gas holdup sensor for flotation systems. Miner. Eng. 16(6):493–501. Gomez, C.O., and J.A. Finch. 2002. Gas dispersion measurements in flotation machines. CIM Bull. 95(1066):73–78. Gorain, B.K., T.J. Napier-Munn, J.P. Franzidis, and E.V. Manlapig. 1998. Studies on the impeller type, impeller speed and air flow rate in an industrial scale flotation cell, Part 5: Validation of the k–Sb relationship and effect of froth depth. Miner. Eng. 11:615–626. Hardie, C.A., C.O. Gomez, and J.A. Finch. 1999. A venturi aerated column cell for de-inking: Effect of design and operating parameters. Prog. Pap. Recycl. 8:33–41. ———. 2004. Comparison of internal and external mixer spargers in a laboratory de-inking flotation column. Can. J. Chem. Eng. 82:504–509. Harris, M.C., K.C. Runge, W.J. Whiten, and R.D. Morrison. 2002. Pages 461–478 in Proceedings, Mineral Processing Plant Design, Practice and Control. Volume 1. Edited by A.L. Mular, D.N. Halbe, and D.J. Barratt. Littleton, CO: SME. Heiskanen, K. 2000. On the relationship between flotation rate and bubble surface area flux. Miner. Eng. 13:141–149. Hernandez, H., C.O. Gomez, and J.A. Finch. 2003. Gas dispersion and de-inking in a flotation column. Miner. Eng. 16:739–744. Hernandez-Aguilar, J., S.R. Rao, and J.A. Finch. 2005. Testing the k–Sb relationship at the microscale. Miner. Eng. 18(6):591–598. Hernandez-Aguilar, J.R., R.G. Coleman, C.O. Gomez, and J.A. Finch. 2004. A comparison between capillary and imaging techniques for sizing bubbles in flotation systems. Miner. Eng. 17(1):53–61. Ityokumbul, M.K. 1994. Design and scale-up of column flotation. Pages 187–199 in Proceedings of the Innovations in Mineral Processing Conference, June 6–8. Edited by T. Yalcin. Sudbury, Canada: Laurentian University.

flotation0.book Page 707 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

707

Jameson, G.J., S. Nam, and M. Moo-Young. 1977. Physical factors affecting recovery rates in flotation. Miner. Sci. Eng. 9:103–118. Johnson, S.B., G.V. Franks, P.J. Scales, D.V. Boger, and T.W. Healy. 2000. Surface chemistry-rheology relationships in concentrated mineral suspensions. Int. J. Miner. Process. 58:267–304. Jowett, A., and D.N. Sutherland. 1985. Some theoretical aspects of optimizing complex mineral separation systems. Int. J. Miner. Process. 14:85–109. King, R.P. 2002. Pages 193–194 in Modeling and Simulation of Mineral Processing Systems. Oxford: Butterworth-Heinemann. Klimpel, R.R. 1980. Selection of chemical reagents for flotation. Pages 907–934 in Mineral Processing Plant Design. 2nd edition. Edited by A.L. Mular and R. Bhappu. Littleton, CO: SME. Krzan, M., and K. Malysa. 2002. Profiles of local velocities of bubbles in n-butanol, n-hexanol and n-nonanol solutions. Colloids Surf. A 207:279–291. Langberg, D.A., and G.J. Jameson. 1992. The co-existence of the froth and liquid phases in a flotation column. Can. J. Chem. Eng. 81:63–69. Laskowski, J.S., Y.S. Cho, and K. Ding. 2003. Effect of frothers on bubble size and foam stability in potash flotation systems. Can. J. Chem. Eng. 81:63–69. Lessard, R.R., and S.A. Zieminski. 1971. Bubble calescence of gas transfer in aqueous electrolyte solutions. Ind. Eng. Chem. Fundam. 10(2):260–269. Li, H., R. Del Villar, and C.O. Gomez. 2004. Reviewing the experimental procedure to determine the carrying capacity in flotation columns. Can. Metall. Q. 43:513–520. Machon, V., A.W. Pacek, and A.W. Nienow. 1997. Bubble sizes in electrolyte and alcohol solutions in a turbulent stirred vessel. Trans. IChemE 75:339–348. Mena, P.C., M.C. Ruzicka, J.A. Teixeira, and J. Drahos. 2005. Effect of solids on homogeneousheterogeneous flow regime in bubble columns. Chem. Eng. Sci. 60:6013–6026. Molerus, O. 1993. Principles of Flow in Disperse Systems. New York: Chapman and Hall. Nesset, J.E., J.R. Hernandez-Aguilar, C. Acuna, C.O. Gomez, and J.A. Finch. 2005. Some gas dispersion characteristics of mechanical flotation machines. Pages 243–249 in Proceedings of the Centenary of Flotation Symposium, June 6–9, Brisbane, Australia. Edited by G.J. Jameson. Melbourne: Australasian Institute of Mining and Metallurgy. Powell, A., J.P. Franzidis, and E.V. Manlapig. 2000. The characterization of hydrodynamics conditions in industrial flotation cells. Pages 243–255 in Proceedings of the 7th Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Rodrigue, D. 2001. A general correlation for the rise velocity of single gas bubbles. AIChE J. 47(1):39–44. Rowe, P.N. 1987. A convenient empirical equation for estimation of the Richardson–Zaki exponent. Chem. Eng. Sci. 47:2795–2796. Rubenstein, J.B. 1995. Column Flotation; Processes, Designs and Practices. Volume 2. New York: Gordon and Breach Science Publishers. Sam, A. C.O. Gomez, and J.A. Finch. 1996. Axial velocity profiles of single bubbles in water-frother solutions. Int. J. Miner. Process. 47:177–196. Shen, G., and J.A. Finch. 1996. Bubble swarm velocity in a column. Chem. Eng. Sci. 51:3665–3674. Shen, G., H. Nawfal, J. Watson, J.A. Finch, and S. Banisi. 1995. Measurement of bubble swarm velocity in a slurry. Pages 781–785 in Proceedings of the CAMI ’95: 3rd Canadian Conference on Applications of Computers in the Mineral Industry. Edited by H.S. Mitri. Montreal: McGill University, Ecole Polytechnique, CIM. Shi, F.N., and T.J. Napier-Munn. 1996. A model of slurry rheology. Int. J. Miner. Process. 47:103–123. Smith, P.G. and L.J. Warren. 1989. Entrainment of particles into flotation froths. Pages 123–146 in Frothing in Flotation. Edited by J. Laskowski. New York: Gordon and Breach Science Publishers. Tang, C., and T.J. Heindel. 2004. Time-dependent gas holdup variation in an air-water bubble column. Chem. Eng. Sci. 59:623–632. Tavera, F.J., R. Escudero, C.O. Gomez, and J.A. Finch. 1997. Gas holdup and slurry conductivity as process diagnostics in column flotation. Pages 3–20 in Processing of Complex Ores, Proceedings of the International Symposium. Edited by J.A. Finch, S.R. Rao, and I. Holubec. Montreal: MetSoc CIM.

flotation0.book Page 708 Tuesday, January 2, 2007 7:36 PM

708

FLOTATION CELLS, MODELING, AND SIMULATION

Voigt, P. February 2003. Generation and measurement of bimodal bubble size distributions. Internal report. Montreal: McGill University. Wallis, G.B. 1969. One-Dimensional Two-Phase Flow. New York: McGraw-Hill. Warsito, M. Ohkawa, N. Kawata, and S. Uchida. 1999. Cross-section distributions of gas and solid hold-ups in slurry bubble column investigated by ultrasonic computed tomography. Chem. Eng. Sci. 54:4711–4728. Weber, T., C.O. Gomez, and J.A. Finch. 2003. A frother concentration meter. Pages 639–652 in Proceedings of the 35th Annual Meeting of the Canadian Mineral Processors division of CIM, January 21–23. Montreal: Canadian Institute of Mining, Metallurgy and Petroleum. Xu, M., and J.A. Finch. 1989. Effect of sparger type and surface area on bubble size in a flotation column. Can. Metall. Q. 28:1–6. Xu, M., and R. Stratton-Crawley. 1996. Rate constant and selectivity measurement in a flotation column. Pages 81–94 in Column ’96: Proceedings of an International Conference on Column Flotation. Edited by C.O. Gomez and J.A. Finch. Montreal: Canadian Institute of Mining, Metallurgy and Petroleum. Xu, M., A. Uribe-Salas, and J.A. Finch. 1991. Maximum gas and bubble surface rates in column flotation. Int. J. Miner. Process. 32:233–250. Yianatos, J.B., L.G. Bergh, F. Diaz, and J. Rodriguez. 2005. Mixing characteristics of industrial flotation equipment. Chem. Eng. Sci. 60:2273–2282. Yianatos, J.B., L.G. Bergh, O.U. Duran, F.J. Diaz, and N.M. Heresi. 1994. Measurement of residence time distribution of the gas phase in flotation columns. Miner. Eng. 7:333–344. Yoon, R.H., and G.H. Luttrell. 1989. The effect of bubble size on fine particle flotation. Pages 101– 122 in Frothing in Flotation, A Volume in Honor of J. Leja. Edited by J. Laskowski. New York: Gordon and Breach Science Publishers. Zhou, Z.A., N.O. Egiebor, and L.R. Plitt. 1992. Frother effects on single bubble motion in a water column. Can. Metall. Q. 31:11–16. ———. 1993. Frother effects on bubble motion in a swarm. Can. Metall. Q. 32:89–96.

Section II: The Froth in Column Flotation J. Cilliers

INTRODUCTION

The froth phase in column flotation plays a very important role in determining the mineral recovery and the grade of concentrate produced. The role of the froth has only recently been recognized as important, and as being more than a simple mass transfer mechanism for carrying particles attached to bubbles from the pulp–froth interface to the concentrate launder. As the size of flotation cells has continued to increase, the importance of the distance that bubbles have to travel to the launder becomes evident, and more complex launder and crowder designs have been implemented. Additionally, the use of wash water in column flotation is common, and both internal and froth surface washing is used, without either being regarded as superior. All of these operating and design variables have been changed largely by empirical observation and trial and error. In this section, the role of froth in column flotation will be detailed in an attempt to explain the reasons for many of the empirical observations made, and to give engineers and operators a fundamental basis for their actions. The section will first introduce the physics of flotation froths. It is important to understand that the froth structure involves the components of the froth which carry liquid, and that particles attached to bubbles which move freely are affected differently by changes in

flotation0.book Page 709 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

709

design and operation. Next is a description of the motion of the bubbles, the liquid, and the solids. Finally, the effect of various changes in the structure—most importantly, the bubble size and froth stability—will be detailed. Equations have been developed and described to show how various factors interact to determine the overall froth behavior. These will be expanded by considering a number of simulations that consider the effect of launder design and wash-water distribution. This section concludes with an industrial wash-water case study that illustrates the importance of understanding how the froth structure affects the flotation performance. The intent of this section is to encourage engineers and operators to consider the froth as a complex system with properties that can be manipulated to produce the separation required. F ROT H S T RU C T U R E Structural Components—Lamellae, Plateau Borders, and Vertices

Foams and froths have very well-defined internal structures and very specific properties that determine these structures. Consider a vertical cross section through a typical two-phase foam, as shown in Figure 1. The foam is formed by bubbles rising freely through the liquid until they meet the foam–liquid interface. In the lowest part of the foam, the bubbles are round and essentially a collection of close-packed spheres. The liquid content in the lower portion of the foam will, therefore, correspond closely to that of a close-packed system, and, for mono-disperse bubbles, will be on the order of 30%. For mixed bubble sizes, the number may be somewhat lower, but this does not significantly affect the froth behavior. Furthermore, the liquid content very rapidly decreases and, as can be observed in the image, within a few bubble diameters, the foam is significantly drier.

FIGURE 1

Vertical cross section through a typical flowing foam

flotation0.book Page 710 Tuesday, January 2, 2007 7:36 PM

710

A

FIGURE 2

FLOTATION CELLS, MODELING, AND SIMULATION

B

Plateau borders: (a) simulation, and (b) micrograph of a solidified liquid foam

Further away from the liquid–foam interface, the bubbles take on an angular shape, with apparently flat lamellae separating them. These polyhedral bubbles initially are of the same volume as those entering the foam, but rapidly start coalescing to form larger bubbles. This occurs when the lamella separating two bubbles fails. Plateau borders are channels formed where three lamellae meet. Because of surface tension, the angle between three lamellae forming a Plateau border is always 120°, and only three lamellae form a stable Plateau border. Plateau borders have a well-defined shape, with a curvature determined by the liquid content and bubble size. Figure 2 shows the shape of a Plateau border. In the picture of the foam (Figure 1), the Plateau borders can be seen as a network of lines, which become visibly thinner higher up in the foam, indicating that the foam is drier. Four Plateau borders meet in a vertex at the tetrahedral angle. Only four Plateau borders can form a stable vertex. Figure 2a shows the simulated shape of a vertex, based on minimum surface area calculations, and Figure 2b is a micrograph of a solidified liquid foam. The close association is striking. The Plateau borders and vertices form an interconnected network of channels through the foam. The curvature of the gas–liquid interface of Plateau borders and vertices exerts a negative pressure on the lamellae and draws liquid out of the lamellae. Therefore, the network of Plateau borders and vertices contains virtually all the liquid in a foam or froth and, for most calculations, it can be assumed that none of the liquid is in the lamellae. Because the link between gangue recovery and water recovery has been known for many years, the importance of the Plateau border and vertex network on flotation is evident. To fully appreciate the role of the froth in flotation, the froth should, therefore, be regarded as a network of liquid channels, the dimensions of which vary with liquid content (Figure 3). The dimensions of Plateau borders in a flotation froth are on the order of millimeters, and can readily be observed on the surface of the froth, between the bubbles. The lamellae in a froth have a thickness similar to the particle diameter. In a three-phase flotation froth, the lamellae separating the bubbles contain the selectively attached hydrophobic particles. The Plateau borders and vertices contain a slurry of

flotation0.book Page 711 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

FIGURE 3

711

Simulated foam structure (1% liquid content)

liquid and freely moving particles. Failure of a lamella produces bubble coalescence within the froth or bursting on the surface, which releases previously attached, hydrophobic particles into the Plateau borders and vertices. Though all gangue particles enter the Plateau border network by entrainment from the pulp, hydrophobic particles are both entrained from the pulp and enter the Plateau borders by lamellae failure. When within the Plateau borders, the motion of both the hydrophobic and hydrophilic particles is determined solely by their physical properties rather than their surface properties. The transfer of particles from being attached to unattached is a key process in the froth, determined by the change in bubble size from the pulp–froth interface to the froth overflow, as well as the fraction of air entering the froth that bursts on the surface. Bubble Size and Shape

The bubble size and shape in the froth determines the Plateau border length associated with each bubble. This is required for estimating the local liquid fraction. The bubble size changes through the froth because of coalescence. In general, bubble size change by diffusion of air from smaller to larger bubbles is insignificant in flotation froths as the rate is slow, considering the residence time, and only bubble growth by coalescence will be considered. The bubbles in a froth are volume filling and will minimize the total surface area to minimize the surface energy. The distribution of bubble sizes and shapes in a real foam is complex, even for bubbles of the same volume. For modeling purposes, an ideal bubble shape must be defined. In 2-D, the honeycomb structure produces the shortest line length. There is no equivalent proof in 3-D. To date, no single size and shape bubble structure has been found that has a more efficient packing than the Kelvin shape, or tetrakaidecahedron (Figure 4). The tetrakaidecahedron has fourteen faces: six four-sided and eight six-sided. The Weaire–Phelan structure comprises two different-shaped bubbles of the same size and is 0.3% more efficient than the Kelvin cell (Weaire and Hutzler 1999). In general, the number of faces in a foam bubble increases as the bubble size increases, either by diffusion or coalescence.

flotation0.book Page 712 Tuesday, January 2, 2007 7:36 PM

712

FLOTATION CELLS, MODELING, AND SIMULATION

FIGURE 4

Kelvin bubbles, or tetrakaidecahedra

FIGURE 5 volume

Simulated vertices in a foam at three liquid contents: 0.1%, 1%, and 10% liquid by

Relationship Between Bubble Size and Plateau Border Length

The geometry of the Kelvin and Weaire–Phelan cells allows the edge length per unit volume to be estimated. These shapes are, however, geometrically complex. A simple volume-filling packing structure, in which all the bubble faces are the same shape, is the dodecahedron with 12 pentagonal faces. The dodecahedron will be used for geometric convenience in the following paragraphs. The edge length per volume, λ, for a dodecahedron is found from geometry as 1.7 λ ≅ ------R2

(EQ 1)

The edge length per unit volume, λ, is inversely proportional to the square of the bubble radius, R (Neethling, Cilliers, and Woodburn 2000). In a coalescing froth, therefore, a doubling in bubble size will reduce the Plateau border length by a factor of four. This has a significant impact on the liquid and gangue recovery to the concentrate, as will be shown later in this section. Relationship Between Liquid Content and Curvature

For a foam of constant bubble size, the Plateau border and vertex radii of curvature, and, hence, their surface area, change as the liquid content changes. This can be visualized when considering a single vertex in a unit volume. The visualizations in Figure 5 show that, even for relatively dry foams (1% liquid), the dimensions of the Plateau borders are substantial.

flotation0.book Page 713 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

713

There are two independent variables: the bubble size and radius of curvature of the Plateau borders, which can be changed independently of each other to produce a foam with a given liquid content. This, in part, explains the difficulty that has been found in estimating, for example, concentrate flow rate from bubble size, because information about an additional variable is required. The bursting rate can be used to estimate Plateau border radius of curvature at the froth surface, which will be shown later in this section. M OT I O N O F T H E P H A S E S Motion of the Bulk Foam

The time-average motion of the bubbles in a flotation froth approximates flow of a continuous phase through a relatively simple geometry. It is possible, therefore, to describe the bubble motion using a continuum-flow model. Several researchers (Moys 1984; Murphy, Zimmerman, and Woodburn 1996; Neethling and Cilliers 1999a) have reported that the motion of the bubbles within a flotation froth can be closely approximated using Laplace’s equation: Δ2Ψ = 0

(EQ 2)

where Ψ is a stream function. This implies that, for the flow rates and geometries found in flotation froths, the bubbles are assumed to be incompressible and irrotational. The gas in a flotation froth can quite reasonably be assumed to be incompressible because most flotation froths have relatively low liquid fractions, and so the change in the internal pressure of the bubbles is quite small as they rise. The irrotational assumption can be made for two reasons. First, flotation froths experience slip at the column cell wall, and little shear stress is experienced. Second, the flow geometry is relatively simple, and the bubbles do not experience sharp directional changes that cause swirling motion. If a flotation froth has a complex geometry or experiences high flow rates, then the effect of the apparent froth viscosity (as opposed to that of the liquid within the froth) must be included, and the assumption of irrotational flow is no longer valid. If the flow rates are quite high, or the froth is very dry, the amount of shear exerted by the wall can also become appreciable. In general, the additional complexity introduced by the apparent froth viscosity is not required, and Laplace’s equation is generally adequate. Water and the Drainage Equation

The motion of liquid and the local liquid content of a flowing flotation froth is one of the most important aspects that determines the overall separation performance. Fortunately, it is also one of the most readily calculated, as the physical relationships and equations have been studied for many years. Liquid draining through foams and froths shows flow behavior that is markedly different from that of liquid draining through, for example, a packed bed of solid particles. This is because the Plateau borders in the foam, through which the liquid drains, change dimension when the liquid velocity changes. In packed beds, the channel diameters are fixed. Therefore, a brief introduction to drainage theory and the drainage equations is necessary. The foam drainage equation is obtained from a force balance on the liquid in the Plateau

flotation0.book Page 714 Tuesday, January 2, 2007 7:36 PM

714

FLOTATION CELLS, MODELING, AND SIMULATION

FIGURE 6 Foam drainage performed with a dye solution, showing that the drainage front is horizontal

borders. The three forces in balance are the force of gravity acting downward, viscous drag as the liquid moves relative to the gas bubbles, and capillary suction. The capillary suction is brought about by changes in the curvature of the Plateau border interfaces as the Plateau border area, A, changes. This is the key difference between foam drainage and drainage through a packed bed. Drainage of liquid through a foam can be observed experimentally by adding a dye solution to a stationary foam column. Figure 6 demonstrates that the liquid dye interface moving downward is flat horizontally, and shows the balance between capillarity and gravity. The one-dimensional drainage equation (Equation 3) that describes the force balance was originally discovered by Leonard and Lemlich (1965), but went mainly unnoticed until independent rediscovery by Verbist, Weaire, and Kraynik (1996). It has been extended for 2-D flowing foams: k 2 ∂A - ------- + u y v y = – k 1 A – ------A ∂y

(EQ 3)

where vy is the upward velocity of the liquid (or slurry in the case of flotation froths), A is the cross-sectional area of the Plateau border, and uy is the upward velocity of the gas. In essence, in Equations 3 and 4, k1 is a result of the balance between gravity and viscosity, whereas k2 is a result of the balance between capillary suction and viscosity. ρg k 1 = ---------------3C PB μ ⎛ ⎞ ⎜ 3–π --- ⎟ γ 2⎠ ⎝ k 2 = ----------------------------6C PB μ

(EQ 4)

(EQ 5)

flotation0.book Page 715 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

715

Courtesy of G. Morizot, Bureau de Recherches Géologiques et Minières, France.

FIGURE 7

Photograph of two particle-coated bubbles being forced together

In the drainage equations, ρ, μ, and γ are the density, viscosity, and surface tension of the slurry, respectively, in the Plateau borders. CPB is the viscous drag coefficient in the Plateau borders and varies with the mobility of the interface. For immobile interfaces, CPB ≈ 49 and is, therefore, appropriate for flotation froths where solids coat the interfaces. The local Plateau border area is the result, which allows the local liquid content to be calculated. The derivation of the drainage equation in the 2-D form given in Equation 3, with the gas velocity term included, can be found in the paper by Neethling, Cilliers, and Woodburn (2000). Motion of the Solids

All hydrophobic or hydrophilic particles in the froth must be considered as one of two types: those particles attached to the bubble lamellae or those particles that are unattached and free to move through the Plateau borders. The attached particles will generally only be hydrophobic, whereas the unattached particles may be hydrophobic or hydrophilic. The Attached Particles. The attached particles enter the froth attached to the bubble and in the lamellae; they are generally hydrophobic. Importantly, the bubbles are significantly larger than the particles in most cases, often as much as two orders of magnitude. Figure 7 illustrates the relative scale of the bubbles and particles in a typical flotation system. The attached particles will move on the same trajectory as the bubbles, as described by Laplace’s equation. It is, therefore, relatively simple to estimate the contribution of the lamellae to the collection of hydrophobic material in the concentrate by approximating the surface area overflowing the weir and measuring the mass loading per unit area. Attached hydrophobic particles detach from bubble lamellae and become unattached because of bubbles bursting at the top surface or through coalescence in the froth. When bubbles coalesce or rupture, some of the bubble surface area is lost, and the attached material on that lamella transfers to the Plateau borders. Because there is a tremendous number of coalescence events occurring in the froth, and also a significant fraction of the air bursting on the surface, the role of the unattached particles is often much more significant than that of the attached.

flotation0.book Page 716 Tuesday, January 2, 2007 7:36 PM

716

FLOTATION CELLS, MODELING, AND SIMULATION

The Unattached Particles. The unattached particles move through the Plateau border network and will tend to follow the liquid. However, these particles also move relative to the liquid by means of one of three mechanisms: geometric dispersion, Plateau border dispersion, and hindered settling under gravity. Geometric dispersion occurs because of the geometry of the Plateau border network through which the particles move and causes particles to spread out perpendicular to the direction of net liquid motion. In the Plateau network, four Plateau borders meet at a vertex, and a particle entering a vertex from one Plateau border can exit from three other Plateau borders. The probability of a particle exiting a vertex through a particular Plateau border is proportional to the fraction of the liquid exiting the vertex into that Plateau border. Plateau border dispersion is caused by the liquid velocity profile in the Plateau borders. A liquid velocity profile exists because, as in pipe flow, the liquid in the center of a Plateau border moves faster than the liquid at the walls. The particles move with the liquid, so particles in the Plateau border spread out parallel to the flow direction. The particles also experience hindered settling under gravity, affected by the local solids concentration, particle size, and density. When dispersion and settling effects are combined with the effect of liquid flow and the release of attached particles due to coalescence, the concentration of unattached solids (Csi) at each position in the froth is obtained by solving the following equation for each different type of solid within the system:

Addition Through Coalescence

–

+

∂(uxSB) ∂x

+

∂(vyCsiAλ)

Liquid Motion

∂y

∂(uySB)

Sconc,i = –

∂y +

Geometric Dispersion

∂x

∂y

∂y

– DG

∂x

∂Csi ∂y

∂y ∂ dpAλ|vx – ux|

– Dp

∂Csi

∂y ∂ dpAλ|vy – uy|

Plateau Border Dispersion

Particle Settling

∂Csi

∂ dbAλ|vy – vSettling, i – uy|

– Dp

∂y

∂(vxCsiAλ)

∂ dbAλ|vx – ux| – DG

∂(vSettling, iCsiAλ)

∂Csi ∂x

∂y

(EQ 6)

flotation0.book Page 717 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

717

where vSettling, i = hindered settling velocity of particle type i SB = specific surface area of the bubbles Sconc,i = mass of attached particles per surface area of bubbles dp = effective Plateau border diameter db = bubble diameter DG, DP = dimensionless dispersion coefficients The result of a simulation using this equation, in combination with the equations describing the other phases, is the concentration of each type of solids at each point within the system, as well as the flux of solids at each point. A more complete description of the model has been published elsewhere (Neethling and Cilliers 2002b). Other Issues. Many unanswered questions remain regarding the behavior of particles in the froth. The first of these is whether there is reattachment of hydrophobic particles from the Plateau borders onto the lamellae, once they have become unattached, possibly through coalescence or surface bursting. There is evidence that attachment of particles onto the lamellae of a particle-barren froth is possible, and has been used as a method for upgrading coarse particles (Ata, Ahmed, and Jameson 2002). On the other hand, the surface area of the Plateau borders is relatively small, and because of the low pressure caused by the curvature, attached particles are likely to move from the lamellae onto the Plateau border surfaces, making further reattachment unlikely. The second question is whether there is upgrading of the lamellae grade because of coalescence in the froth, as it can be postulated that the particles that detach are the least hydrophobic (Gourram-Badri, Conil, and Morizot 1997). This has been investigated by two-bubble coalescence experiments but remains an open question. It can be reasonably postulated, however, that neither of these mechanisms constitute a major effect in the role of the froth. B U B B L E S I Z E T H R O U G H T H E F R O T H A N D F R O T H S TA B I L I T Y

The lower boundary of the froth is the upper surface of the pulp. Bubbles entering the froth have a hydrophobic particle load associated with them. The particle load that a bubble can carry is determined by its size, and the size, density, and coverage of the particles. The bubble–particle conglomerate must still be buoyant (i.e., able to rise up through the pulp). The particles are free to move on the surface of the bubble, and are found on the lower hemisphere of the bubble rising through the pulp. Images of particle-laden bubbles have been taken using high-speed photography; an example is shown in Figure 8. The bubble size entering the froth is a key variable and affects the flotation behavior significantly, not only through pulp particle–bubble interaction kinetics but also because of the effect on liquid drainage in the froth. Theoretical Stability of Solid-Containing Lamellae

The liquid layers between bubbles can be stabilized by a coating of fine solid particles. The stability of foams containing particles depends very strongly on the contact angle, and particles having a contact angle of approximately 100° increase the foam half-life significantly, whereas larger contact angles destabilize foams (Aveyard et al. 1994; Johansson and Pugh 1992).

flotation0.book Page 718 Tuesday, January 2, 2007 7:36 PM

718

FLOTATION CELLS, MODELING, AND SIMULATION

Courtesy of Z. Xu.

FIGURE 8

Photograph of rising –38 μm coal-particle-coated bubbles

This has recently been extended to the stabilization of emulsions, where the particles adsorb at the oil–water interface and form a coating around the droplet, protecting it from coalescence. As for foams, the particle contact angle plays a critical role, and for an oil–water contact angle of approximately 90°, the energy required to remove the particle can be several times kT, and the adsorption can be regarded as irreversible. It has, however, been pointed out that “theory developed to date leaves unexplained much of the corpus of detailed experimental findings” (Aveyard, Binks, and Clint 2003). Ali, Rossen, and Gauglitz (2000) have analyzed theoretically the stability of a 2-D film containing circular particles. Surprisingly, they suggest that the 3-D liquid film between bubbles containing one row of solid particles will be unstable unless all the solids touch. This can be observed on the surface of flotation cells, where expanding bubbles burst as soon as a clear “window” forms on the bubble surface. A further study by Kam and Rossen (1999) using the same theoretical system indicated that particle-coated surfaces can withstand significant pressures, which explains the high stability. It must be emphasized that the Rossen theory is essentially 2-D and does not realistically describe 3-D spherical particle systems. A recent emulsion-particle study by Kruglyakov, Nushtayeva, and Vilkova (2004) has extended the theory to expanding lamellae. Coalescence Between Bubbles

In spite of the theoretical developments on the stability of solid-containing lamellae, prediction of coalescence between bubbles in the froth and the rate of bursting on the surface of the froth remains highly complex, with only limited success in elucidating relationships between, for example, the loading of particles on the lamellae and the air recovery. The importance of understanding coalescence and what affects the rate of coalescence cannot be overstated. Flotation froths are apparently quite unstable, as is usually evident from their instantaneous collapse when overflowing into the concentrate launder. Consider a typical froth, 100-mm thick, being formed continuously from a superficial gas velocity of 20 mm/sec. A bubble, therefore, has an average lifetime in the froth of 5 seconds. Then assume a pulp bubble size of 0.5 mm and a surface bubble size of 16 mm. The number of

flotation0.book Page 719 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

719

coalescence events required for the bubble diameter to grow from the pulp size (dbubble, in) to the surface size (dbubble, out) is ⎛ bubble,out ⎞ 3 ln ⎜d --------------------- ⎟ ⎝ d bubble,in ⎠ Coalescences = -----------------------------------ln ( 2 )

(EQ 7)

Each bubble, therefore, has to coalesce approximately 15 times while passing through the froth—three coalescence events every second. Every bubble has approximately 14 lamellae, but each lamella is shared between two bubbles. Therefore, each lamella has approximately less than a 40% chance of failing while passing through the froth. Individual lamellae are, thus, quite stable and give credence to the theoretical predictions that particle-stabilized lamellae are very stable. A probabilistic approach to describing dynamic foam stability has been successfully demonstrated for unstable two-phase foams (Cilliers et al. 2002), which predicted the experimental foam height from microscale two-bubble coalescence data. Theoretical equations have also been developed that link the probability of lamella failure to the predominant disturbances and modes of failure in the froth (Neethling and Cilliers 2003). These have not been rigorously tested but have established a mathematical framework for further experimental investigations. Nonetheless, film failure and bubble coalescence remain as some of the most challenging aspects of understanding the froth behavior in detail. Bubble Size on the Surface—Measurement and Modeling

The bubble size of the overflowing froth clearly affects the flotation process significantly, both by the length and volume of Plateau borders containing the unattached particles and by the lamella surface area containing the attached particles. Measuring the overflowing bubble size has received increasing attention during recent years. Several automated image analysis systems have been developed that are able to estimate online the size of the lamellae that make up the surface pattern. These methods, in general, use some form of watershed algorithm to find the edges of the bubbles. Though some technical issues remain, in general, image analysis of the froth can result in a reasonable estimate of the size distribution of the exposed lamellae on the froth surface. The lamellae size distribution is not the same as the bubble size distribution, and will depend not only on the bubble size, but also the shape. In addition, the very top surface of the froth may not be representative of that which overflows the weir. These issues are being investigated currently. An interesting approach is that of S.J. Neethling (personal communication), who is studying 3-D computer-generated foam structures. He defines very specific bubble size distributions inside the froth and then predicts the associated surface lamella size distribution. This potentially will allow the image analysis results to be interpreted more completely and be of greater utility. An example of such a simulation is shown in Figure 9. Air Recovery—Importance, Measurement, and Interpretation

“Air recovery” is defined as the fraction of air entering the pulp that exits the cell by overflowing the weir (i.e., as unburst bubbles). The air recovery is, in general, surprisingly low, and values greater than 50% are rarely observed, mostly in highly stable cleaner and recleaner columns. In rougher and scavenger cells, values below 20% and as low as 5% are commonly observed.

flotation0.book Page 720 Tuesday, January 2, 2007 7:36 PM

720

FLOTATION CELLS, MODELING, AND SIMULATION

FIGURE 9 Simulation of a foam consisting of a single bubble size and a free surface (all the bubbles have the same volume, but the surface lamellae have many sizes)

The significance of the air recovery in the modeling of flotation froths, and as a measurement in industrial operation for estimating performance, has only recently become evident. Air recovery has been recognized empirically as determining the balance between the valuable particle recovered by attachment to the lamellae and those recovered from the Plateau borders (e.g., Woodburn, Austin, and Stockton 1994; Ventura-Medina and Cilliers 2002). The air recovery is the key variable that determines the upper boundary condition when solving the Laplace equation for froth motion. In addition, the air recovery determines, to a large extent, the water recovery to concentrate and, because water recovery and gangue recovery are related, the concentrate grade, which will be discussed in subsequent paragraphs. In this subsection, methods for measuring the air recovery are detailed, and the current understanding of the relationship between the air recovery and operating variables are discussed. The importance of the air recovery is a relatively new finding and many of the results shown are based on relatively sparse data. The air recovery, α, is the fraction of air overflowing the weir that entered the cell: α = Q air, concentrate ⁄ Q air, pulp

(EQ 8)

It is assumed that the air flow rate into the cell is known, either by direct measurement or from gas holdup techniques. The volumetric flow rate of air overflowing the weir can be estimated directly by measuring and multiplying the froth velocity, vfroth, the froth depth overflowing the weir, hfroth, and the total lip length, L: Q air, concentrate = v froth ⋅ h froth ⋅ L

(EQ 9)

The velocity on the surface of the froth overflowing the weir can readily be estimated using image analysis, or by simple techniques such as dropping pieces of paper onto the froth and timing their progress toward the lip. Care must be taken to ensure that the froth velocity measured on the overflowing surface is representative of the average overflowing velocity by observing the froth from various viewpoints. The depth of the froth overflowing the weir can be measured directly. A second method for inferring air recovery has recently been developed, based on the height that a froth will rise in a nonoverflowing “froth stability column” (Barbian, VenturaMedina, and Cilliers 2003). This is based on the Bikerman dynamic foam stability (1973), which is the ratio of the maximum froth volume and the air rate. In general, froths show the following behavior: At low air rates, the equilibrium froth height increases with air rate, but at higher air rates, the froth becomes increasingly unstable and the equilibrium height either

flotation0.book Page 721 Tuesday, January 2, 2007 7:36 PM

721

8

80

7

70

Measured Air Recovery, %

Estimated Air Recovery, %

COLUMN FLOTATION

6 5 4 3 2 1 0

60 50 40 30 20 10 0

0

2

4

6

8

10

12

14

Air Flowrate Q, m 3/min

FIGURE 10 Air recovery predicted using the froth stability column (note the peak in stability at 8 m3/min)

0

2

4

6

8

10

Estimated Air Recovery, %

FIGURE 11 Correlation between predicted and measured air recovery using the froth stability column

becomes constant or can even decrease. It should be clear that at an operating froth depth equal to the maximum froth depth, no froth will overflow, and the air recovery will be zero. For froth depths that are a progressively smaller fraction of the maximum, a greater fraction of air will be recovered. Hence, the air recovery at any froth depth can be predicted from the froth stability at that air rate alone. Barbian, Ventura-Medina, and Cilliers (2003) showed that, at laboratory scale, the prediction of air recovery very closely matched the air recovery obtained at various froth depths. Further studies at industrial scale showed that the same behavior is observed: for a given froth depth at low air rates, the air recovery increases with air rate, but at higher air rates, the air recovery decreases significantly (Figure 10). Although the fraction of air recovered could not be predicted directly, the trends in air recovery as a function of air rate measured using the froth stability column were exactly the same as measured by image analysis, as shown in Figure 11. The distinct maximum in air recovery, as a function of air rate, can be interpreted qualitatively as follows. At very low air rates, the bubbles in the froth are very thoroughly coated with particles and resist bursting and coalescence. However, because the froth velocity is also low, froth residence time is high and the bubbles burst before overflowing. At very high air rates, the bubbles are not armored with particles and therefore coalesce and burst readily. In spite of the low residence time, the high bursting rate leads to low air recoveries. There is an intermediate air rate at which the bursting rate is reduced by particle loading, though the residence time is also low enough for a substantial proportion of the bubbles to overflow unburst. This maximum air recovery has important implications for control using air rate, as it affects the separation behavior differently at air rates below and above the maximum. In conclusion, it is clear that manipulating the air recovery, techniques to measure air recovery, and theory to interpret the importance of air recovery will play an increasingly important role in understanding the behavior of the froth, how it affects the separation, and how it can be optimized. E X A M P L E S A N D R E L AT I O N S H I P S

The physics of flotation froths and the fundamental equations describing the motion of liquid and solids are of great practical and industrial value. In the first instance, a complete description of all the phases and their motion through the froth allows design and operational modifications to be made to improve the separation performance. Second, the equations can

flotation0.book Page 722 Tuesday, January 2, 2007 7:36 PM

722

FLOTATION CELLS, MODELING, AND SIMULATION

be simplified with suitable assumptions and boundary conditions to give a clear indication of which variables are key and must be manipulated to move the system in the desired direction. Both these cases will be illustrated in the following subsections. Interplay Between Air Recovery, Bubble Size, and Water Recovery

The equations that describe the Plateau border dimensions can be solved explicitly with suitable boundary conditions for the surface of the froth, which provides a method for estimating the liquid flow rate to the concentrate. This is a very important variable, as it has been shown both experimentally (Engelbrecht and Woodburn 1975) and theoretically (Neethling and Cilliers 2002a) that the gangue rate to the concentrate closely follows the water rate. A direct estimate of the water rate is, therefore, an immediate estimate of the concentrate grade. The liquid drainage equations are combined with relationships between the bubble size overflowing the weir (dbubble) and Plateau border dimensions, and solved using suitable boundary conditions. The solution predicts the water rate to the concentrate and has two solutions, depending on the value of the air recovery (α): For α < 0.5 (unstable froth, more typical of roughers and scavengers) v g2 water rate to concentrate ∝ A column --------------- ( 1 – α ) α 2 d bubble

(EQ 10)

For α ≥ 0.5 (stable froth, more typical of cleaners and recleaners) v g2 water rate to concentrate ∝ A column --------------2 d bubble

(EQ 11)

These equations indicate many important proportionalities. The following paragraphs consider each in turn, assuming all other variables remain the same. The first important variable, conspicuous by its absence, is the froth depth. The Plateau border area in the froth is relatively constant between the froth surface and a short distance from the pulp–froth interface. Hence, changes in performance attributed to an increase in froth depth are generally due to secondary effects as a result of an increase in residence time, increased coalescence, larger overflowing bubbles, and lower water recovery; and associated changes in bubble loading, froth stability, and air recovery. When the air recovery is low, as is typical of roughers and scavengers where values less than 20% are typical, the water rate to concentrate depends on the air recovery, α. Although this is nonlinear, it is monotonically increasing, and an increase in air recovery (i.e. froth stability) will increase the water rate. The influence of air recovery decreases as it approaches 0.5, and when the froth becomes stable enough that more than 50% of the air is recovered to concentrate, it has no further effect. More significant are the effects of air rate and the overflowing bubble size, both of which have a squared dependence. As the air rate is increased, the water rate increases significantly, and the grade is expected to decrease. This is in accordance with general industrial experience. Also, a decrease in the overflowing bubble size will increase the water rate significantly, as is also commonly found.

flotation0.book Page 723 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

723

The interactions between the variables make the overall separation behavior more difficult to predict uniquely. For example, an increase in air rate increases the pulp bubble size entering the froth. The increased air rate and increased pulp bubble size will most likely increase the recovery rate of valuable minerals, because of an increase in pulp bubble surface area. On the other hand, the bubbles will have a lower loading of particles, which makes them more susceptible to coalescence. The air recovery is a strong function of air rate, often showing a maximum (as shown in Figure 10) and, hence, the increase in air rate can either increase or decrease the air recovery. Additionally, an increase in air recovery increases the froth residence time and will increase the overflowing bubble size because of increased coalescence, enhanced by the lower loading. Should the air recovery increase as air rate increases, the grade of the concentrate is likely to increase because of increased valuable mineral recovery due to the increased air recovery and decreased water and gangue recovery attributable to the larger bubble size. If air recovery decreases as the air rate increases, the grade will decrease, given that the valuable recovery is decreased because of decreased air recovery, and the water recovery increased as there is less coalescence. Figure 12 shows the effect of a change in air rate on the air recovery down a bank of rougher cells, and also on the grade-recovery curves. In this case, the air recovery decreases down the bank, and also shows clearly that an increase in air rate decreases the air recovery, and leads to a decrease in separation performance (Figure 13). The differences in loading on the bubbles on the surface of the froth can clearly be observed in the images taken at the three air rates (Figure 14). This emphasizes the importance not only of measuring the air

10 D

Air Recovery, %

8

C 6

B A

4 A

2 C 1,070

1,170

40 35 30 25 20 15 0

B 0

Cumulative Grade, % Cu

45

Cell

D 1,270

3

Air, m /hr

FIGURE 12 Air recovery measured down a rougher bank and at different air rates

20

40

60

80

100

Cumulative Recovery, % Cu 1,070 m3/hr

1,170 m3/hr 3

1,270 m /hr

FIGURE 13 Grade-recovery curves showing improved performance at lower air rates

FIGURE 14 The froth surface as the air rate increases, showing the reduction in particle loading (left to right): 1,070 m3/hr; 1,170 m3/hr; and 1,270 m3/hr

flotation0.book Page 724 Tuesday, January 2, 2007 7:36 PM

724

FLOTATION CELLS, MODELING, AND SIMULATION

recovery and overflowing bubble size, but also whether a change in the operating variable will increase or decrease the air recovery and the overflowing bubble size. A P P L I C AT I O N C A S E S T U D I E S

The mathematical description of all the phases in a flotation froth, using equations that describe accurately the froth structure and physics, provides two important advantages over empirical equations. First, the role of each variable is explicit, and its relative importance is immediately clear from the equation form. Interactions between variables are clear, and the confounding of variables commonly found when experimenting is eliminated. This was previously illustrated when the equations were solved to predict explicitly the water recovery from an overflowing froth. Second, only a mathematical description of this type is satisfactory when design modifications are required. The boundary conditions represent real dimensions and values, and can be manipulated. This is illustrated in two case studies: the effect of changes in launder layout and of internal wash-water distribution. These case studies are chosen because they illustrate that the results could not realistically have been obtained in a cost-effective manner by any alternative means. Design of Weir and Crowder Layouts

Two different launder and crowder layouts are commonly encountered in large, circular column cells: the so-called donut launder, where the froth is crowded both inward and outward and collected in a single internal launder; and the double launder, in which the froth is crowded only outward by a central crowder and an outwardly sloped internal launder. By only changing the physical boundary conditions of the simulations, these two layouts can be compared under identical operating conditions. These results have been previously reported in greater detail (Neethling and Cilliers 2003), and only the key findings are shown here. Figure 15 shows the two launder designs in cross-section: as a vertical section through the froth, and from the pulp froth interface to the bursting surface. The center of the cell is on the left in each of the graphs. Each graph also shows the predicted relative value of the grade at all points in the froth.

A

B

FIGURE 15 Donut launder layout and relative grade (a) through the froth, in comparison with (b) internal and external launders with crowding

flotation0.book Page 725 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

725

The most significant result is immediately evident when comparing the mineral grade of the froth at the surface for the systems in which the froth is crowded both inward and outward (donut launder) and where it is crowded only outward (double launder). The mineral grade of the froth at the surface, and, hence, also the concentrate grade for the donut launder is significantly different in the inner and outer parts of the froth. However, both overflows are collected in the same launder and cannot be subsequently treated separately. In contrast, the double-launder design has very similar surface grades, and the concentrates can be combined. The reasons for the grade differences in the donut system are clear when considering the physics of the bulk flow and the liquid drainage within the froth. For the froth that is crowded outward, the volume expands, whereas the inward crowding forces the froth into a progressively smaller volume. By reducing the flow area when crowding the froth inward, the froth velocity is increased and the liquid drainage is reduced, producing a wetter concentrate with a lower grade. This is avoided when crowding both sections outward, reducing the froth velocity, increasing drainage and the grade. The differences in grade in the froth for the inward- and outward-flowing froths in a donut launder cell design have been confirmed experimentally (Neethling and Cilliers 2003). Wash-Water Distribution

The use of wash water in flotation columns is widespread. Wash water is added to the froth to create a downward liquid flow through the froth. If the wash-water rate is greater than the concentrate water rate, this is referred to as “positive bias”, whereas a “negative bias” occurs when the wash-water rate is less than the concentrate water rate. The wash-water rate is one of the key variables in washed flotation columns although usually an approximately neutral bias is maintained (i.e., adding wash water at the concentrate liquid flow rate). In terms of physical design, there are two options for how to add the wash water: either on top of the froth or internally. There are no clear guidelines for which addition method is most suitable under which conditions. Additionally, a variable that has been not been studied extensively is the effect of the distribution of the wash water across the froth area, that is, whether it is uniformly distributed or added at specific points. These issues can be resolved using a complete mathematical description of the froth, as the wash-water rate and addition point is a boundary condition of the simulation. A conclusive answer as to whether internal or surface froth washing should be used will not be attempted here. Instead, the motion of the wash water for the two cases will be investigated, and it will be shown that the two methods require different distribution patterns. An industrial case study will illustrate the potential benefit. Surface Froth Washing. Wash water can be added uniformly across the froth surface or restricted to part of the froth. For example, it can be added preferentially toward the center of the cell or at the overflow weir. In addition, the water addition rate can be varied, but this will not be considered in this chapter. As an illustration, two cases of water addition at a single point on the froth surface are shown in Figure 16: wash water addition at the weir and addition toward the center of the cell. Figure 16 shows the motion of the liquid from the point of addition through the froth. For the wash-water addition at the weir, it is clear that a significant proportion of the water added simply flows over the weir and does not wash the gangue from the froth. A small fraction does flow downward along the front wall of the cell and causes a recirculation of liquid carrying gangue from the center of the cell back into the pulp at the front wall.

flotation0.book Page 726 Tuesday, January 2, 2007 7:36 PM

726

A

FLOTATION CELLS, MODELING, AND SIMULATION

B

FIGURE 16 Trajectories of wash water through the froth when added on the surface: (a) wash water added toward the lip of the cell, and (b) wash water added approximately one-third from the center (or back wall) of the cell

A

B

FIGURE 17 Liquid flow velocity vectors for internal froth washing: (a) water distribution only in the center of the cell, and (b) an even distribution across the cell diameter

In contrast, when the wash water is added at a single point, approximately one-third of the way from the center of the cell (or rear cell wall), the wash-water action is significantly more effective. Very little of the water short-circuits to the weir, and a clear horizontal liquid motion is established in the lower portion of the froth that allows the gangue to drain back into the pulp. Internal Froth Washing. Internal froth washing has gained favor because it does not produce a wetter concentrate or reduce the recovery of valuable minerals to the same extent as surface washing. However, internal washing has severe practical difficulties, most significantly that blockage of wash-water sprays cannot be directly observed. The wash-water mechanism must therefore be removed regularly from the cell for maintenance. The distribution of wash water inside the froth is as important as that for surface washing. However, the way in which the water moves through the froth is significantly different. Two cases with equal wash-water rates are considered here: uniform water distribution across the froth and water distribution predominantly toward the center of the cell. Figure 17 shows a cross section through the froth and the simulated results for the pulp motion. Obviously, for a perfectly uniform distribution of internal wash water, the pulp

flotation0.book Page 727 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

727

motion is even and the gangue is removed equally at all points. In contrast, for a wash-water distribution toward the center of the cell, there is a significant upward motion of pulp toward the outer edge of the cell and over the weir into the concentrate. In this region, the wash water plays no role, and the gangue is able to sidestep the downward flow. Though not shown here, a distribution of wash water predominantly toward the outer cell wall produces very similar performance results regarding uneven distributions toward the center. Clearly, in contrast to surface wash-water addition, an even internal water distribution is required for optimal results. Maldistribution of wash water has a greatly diminished effect if the water bias is upward. This is, of course, seldom the case in washed column cells. This is difficult industrially, as the wash-water distributors often block preferentially at the ends of pipes (i.e., at the cell wall), and also there is significant space allowed between the wash-water distributors and the cell wall to allow removal without damage. This often results in an uneven distribution toward the center of the cell. The following case study shows the effect of improving the internal wash-water distribution on a zinc flotation column performance. The Red Dog Problem. A detailed simulation of a single column cell at the Red Dog zinc mine in Alaska was performed to explain the reason for excess gangue recovery. This column utilizes internal froth washing. Detailed data from the column and operating conditions were used to set up the simulations. It was initially discovered that the grade, recovery, and concentrate rate predicted by the equations described the operating data very poorly. Subsequently, further simulations were performed using maldistributed wash water, as it was postulated that an uneven distribution may lead to gangue bypassing the washed zones and flowing into the concentrate. When distributing the wash water into the central portion of the column only, the simulated results matched (within approximately 20% for all mineral flow rates) the industrial data very closely. Figure 18 from the simulations shows how the wash water wets the froth vertically but does not disperse outward toward the cell walls. The relatively small bubbles in the froth and the high liquid addition rate mean that the horizontal capillary suction is substantially slower than the gravity-driven vertical drainage. This allows gangue to bypass the downward moving water near the outer cell walls and move upward to overflow the lip (Figure 19). Simulations indicated that a considerable decrease in gangue recovery could be achieved by modifying the wash-water distributors to completely cover the full cell cross section. Following modification and testing, a 36% reduction in gangue recovery was achieved, without a significant decrease in valuable minerals recovery. The increase in the grade achieved resulted in an annual economic benefit of approximately $1.5 million. CONCLUSIONS

In this section, the importance of the froth phase on flotation performance has been discussed. Moreover, it has been shown that the behavior of the froth and its components can be described mathematically from a fundamental understanding of physics. These equations are complex but tractable. Simplified solutions give a clear indication of which variables are most important and how they should be manipulated to make process changes. The complete set of equations can also be solved simultaneously to give a complete description of the motion and concentration of all the components of the froth at all positions:

flotation0.book Page 728 Tuesday, January 2, 2007 7:36 PM

728

FLOTATION CELLS, MODELING, AND SIMULATION

FIGURE 18 Plateau border area (m2) as a function of position within the froth. This is proportional to the liquid content in the froth. (Each division on the x- and y-axis represents 4 mm.)

FIGURE 19 Concentration of gangue in the Plateau borders (kg/m3)

the bubbles, the liquid, and all particles. This has tremendous utility in cell design and operating conditions. This was illustrated with a number of case studies, including industrial improvement. Clearly, this approach has value and a concerted effort must be made to try and understand the remaining subprocess, the coalescence of particle-coated bubbles, the reattachment of particles to bubbles in the froth, and the factors that affect the air recovery. BIBLIOGRAPHY

Ali, S.A., W.R. Rossen, and P.A. Gauglitz. 2000. Stability of solids-coated liquid layers between bubbles. Ind. Eng. Chem. Res. 39:2742–2745. Arditty, S., C.P. Whitby, B.P. Binks, V. Schmitt, F. Leal-Calderon. 2003. Some general features of limited coalescence in solid-stabilized emulsions. Eur. Phys. J. E 12:355–355. Ata, S., N. Ahmed, and G.J. Jameson. 2002. Collection of hydrophobic particles in the froth phase. Int. J. Miner. Process. 64:2–3, 101–122. Aveyard, R., B.P. Binks, and J.H. Clint. 2003. Emulsions stabilised solely by colloidal particles. Adv. Colloid Interface Sci. 100–102:503–546. Aveyard, R., B.P. Binks, P.D.I. Fletcher, T.G. Peck, and C.E. Rutherford. 1994. Aspects of aqueous foam stability in the presence of hydrocarbon oils and solid particles. Adv. Colloid Interface Sci. 48:93–120. Barbian, N., E. Ventura-Medina, and J.J. Cilliers. 2003. Dynamic froth stability in froth flotation. Miner. Eng. 16(11):1111–1116. Bikerman, J.J. 1973. Foams. New York: Springer-Verlag. Binks, B.P., and C.P. Whitby. 2004. Silica particle-stabilized emulsions of silicone oil and water: Aspects of emulsification. Langmuir 20:1130–1137. Cilliers, J.J., S.J. Neethling, M.T. Spyridopoulos, and S.J.R. Simons. 2002. The failure of thin films between bubbles—from micromanipulation to foam stability. Pages 169–178 in Proceedings of Flotation and Flocculation: From Fundamentals to Applications, Strategic Conference and Workshop, Hawaii, July 28–August 2. Edited by J. Ralston, J. Miller, and J. Rubio. Medindie, South Australia: Snap Printing. Engelbrecht, J.A., and E.T. Woodburn. 1975. The effect of froth height, aeration rate and gas precipitation on flotation. J. South African Inst. Min. Metall. 10:125–132.

flotation0.book Page 729 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

729

Gourram-Badri, F., P. Conil, and G. Morizot. 1997. Measurements of selectivity due to coalescence between two mineralized bubbles and characterization of MIBC action on froth flotation. Int. J. Miner. Process. 51(1–4):197–208. Johansson, G., and R.J. Pugh. 1992. The influence of particle size and hydrophobicity on the stability of mineralized froths. Int. J. Miner. Process. 34(1–2):1–21. Kam, S.I., and W.R. Rossen. 1999. Anomalous capillary pressure, stress and stability of solids-coated bubbles. J. Colloid Interface Sci. 213:329–339. Kruglyakov, P.M., A.V. Nushtayeva, and N.G. Vilkova. 2004. Experimental investigation of capillary pressure influence on breaking of emulsions stabilized by solid particles. J. Colloid Interface Sci. 276:465–474. Leonard, R., and R. Lemlich. 1965. Laminar longitudinal flow between close-packed cylinders. J. Chem. Eng. Sci. 20:790–791. Moys, M.H. 1984. Residence time distributions and mass transport in the froth phase of the flotation process. Int. J. Miner. Process. 13:117–142. Murphy, D.G., W. Zimmerman, and E.T. Woodburn. 1996. Kinematic model of bubble motion in a flowing froth. Powder Technol. 87:3–12. Neethling, S.J., and J.J. Cilliers. 1999a. A visual kinematic model of flowing foams incorporating coalescence. Powder Technol. 101:249–256. ———. 1999b. Visualisation and drainage of coalescing, flowing foams. Pages 117–124 in Foams and Films. Edited by D. Weaire and J. Banhart. Bonn, Germany: MIT-Verlag. ———. 2002a. Entrainment of solids into flotation froths. Int. J. Miner. Process. 64:123–134. ———. 2002b. Solid motion in foams. Chem. Eng. Sci. 57:607–615. ———. 2003. Modelling flotation froths. Int. J. Miner. Process. 72(1–4):267–287. Neethling, S.J., J.J. Cilliers, and E.T. Woodburn. 2000. Prediction of the water distribution in a flowing foam. Chem. Eng. Sci. 55:4021–4028. Neethling, S.J., H.S. Lee, and J.J. Cilliers. 2002. A foam drainage equation generalised for all liquid contents. J. Phys. Condens. Matter 14:331–342. Ventura-Medina, E., and J.J. Cilliers. 2002. A model to describe flotation performance based on physics of foams and froth image analysis. Int. J. Miner. Process. 67:79–99. Verbist, G., D. Weaire, and A.M. Kraynik. 1996. The foam drainage equation. J. Phys. Condens. Matter 8:3715–3731. Weaire, D., and S. Hutzler. 1999. Physics of Foam. Oxford: Clarendon Press. Woodburn, E.T., L.G. Austin, and J.B. Stockton. 1994. A froth based flotation kinetic model. Chem. Eng. Res. Des. 72(A2):211–226.

Section III: Flotation Column Control J. Yianatos

INTRODUCTION

Flotation columns typically outperform conventional mechanical cells in cleaning operations because of their particular design with a single and deep froth, which is also provided on top with wash water. Figure 1 shows the comparison of mechanical and column flotation results for Red Dog lead rougher concentrate. The most significant advantage for columns is the concentrate upgrade to moderate recoveries (60%–70%), as shown in Figure 1. When columns are pushed for high recoveries, the main advantage of having higher upgrades than mechanical cells becomes less significant. The column feed characteristics such as flow rate, pH, values grade, solids content, mineralogy, particle size distribution, liberation, and reagent concentration (collectors, frothers,

flotation0.book Page 730 Tuesday, January 2, 2007 7:36 PM

730

FLOTATION CELLS, MODELING, AND SIMULATION

70

Lead Grade In Concentrate, %

Cell Column

60

50

0 30

60

90

Lead Recovery, %

Source: Yianatos and Murdock 1991.

FIGURE 1

Column vs. mechanical cells

Residence Time Distribution, E(t)

0.12 Data LSTS* Model 0.09

0.06

0.03

0.00 0

10

20

30

40

50

Time, min *LSTS = large-small tanks-in-series.

FIGURE 2

Residence time distribution in a rectangular column 2 × 6 × 13 m

etc.) are usually determined from previous grinding operations, flotation stages (rougher and scavenger), and conditioning tanks. The main disadvantage of flotation columns with respect to conventional mechanical cells is the large spread of results, mainly in terms of recovery, which is usually compensated by a high circulating load and large volumes (overdesign). The variability in performance of single large-size columns can be partially attributed to the mixing conditions. Despite the presence of baffles, the mixing conditions in large industrial columns are closer to a single perfect mixer. For example, Figure 2 shows the residence time distribution of flotation columns at the El Salvador division, Codelco Chile (Corporación Nacional del Cobre de Chile). The circuit consists of two rectangular columns, 2 × 6 × 13 m, operating in parallel, with a common scavenger circuit of mechanical cells (Yianatos et al. 2005). Thus, commonly, the column operates in a recovery condition (around 60%–70%) similar to that of the first mechanical cell in a bank. Consequently, columns must to be arranged in circuits in order to reach a target recovery for the entire cleaning operation.

flotation0.book Page 731 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

731

0

Feed Rougher

Cleaner

Final Concentrate Regrind

Surge Thickener

Scavenger

Final Tail

FIGURE 3

Mo/Cu flotation circuit at Andina (Chile)

COLUMN CIRCUITS

Some column-in-series circuits, such as cleaner–scavenger arrangements, have been reported in copper and complex ore concentrators (Dobby, Amelunxen, and Finch 1985; EspinosaGomez et al. 1989). More recent developments in sulfide copper concentrators have shown that a better compromise is the use of column cleaner–mechanical cell scavenger (CS) circuits, where the overall performance is achieved using a single stage of columns to produce the final concentrate grade in a closed loop with a scavenger of mechanical cells to achieve the overall cleaning recovery (Figure 3). The typical column flotation performance yields a copper column recovery of 60%– 70%, whereas the scavenger recovery is about 92%–95%, which gives an overall cleaning recovery of 95%–98%. The use of columns in series in a cleaner–cleaner (CC) arrangement was reported originally in Canada at Mines Gaspé (Coffin and Miszczak 1982). In the last decade, some new projects have attempted to build Mo/Cu separation circuits considering four or five stages of flotation columns only. In all cases, the plant experiences have shown that the first rougher and cleaner stages do not perform well, and large circulating loads are generated that are difficult to manage, despite the use of intermediate thickeners. Thus, modifications have been introduced to improve metallurgical performance. Basically, the rougher, first cleaner, and sometimes the second cleaner stages have been replaced by mechanical cells, whereas the third and fourth cleaning stages are typically developed in columns. Thus, simple and more stable circuits with lower circulating loads have been developed as shown in Figure 4. CONTROL OBJECTIVES

The primary control objectives are the column concentrate grade and recovery, which represent the indices of product quality and process productivity. Often, the cleaner final concentrate grade is the main target, and a secondary objective is the cleaner–scavenger tailings grade, which has a minor impact on the overall recovery compared to changes in the rougher tailings. The online estimation of these indices usually requires a significant amount of work in maintenance and calibration of onstream analyzers in order to maintain good accuracy and high availability (Bergh, Yianatos, and Cartes 1996). Therefore, a common industrial practice

flotation0.book Page 732 Tuesday, January 2, 2007 7:36 PM

732

FLOTATION CELLS, MODELING, AND SIMULATION

Feed

Rougher

Final Tail

Cleaner 2

Cleaner 3

Mo/Cu Bulk Concentrate

Cleaner 1

FIGURE 4

Mo/Cu flotation circuit at Andina

is to control secondary objectives, such as pH, froth depth, air flow rate and wash-water flow rate. These are usually implemented as local controllers or under a distributed control system, or DCS. If the secondary objectives are under control and the primary objectives are not measured, cascade control of gas holdup, εg , or bubble surface area flux, Sb (using gas flow-rate control), and bias (using wash-water flow-rate control) became intermediate objectives. Froth characteristics, such as color, form, speed, and size, which can also be considered as intermediate objectives, depend on the regulation of the secondary objectives and the feed characteristics. In both cases, the problem is how to relate these intermediate objectives with concentrate grade and process recovery. C O N T R O L S T R AT E G I E S

Original control strategies considered the control of indirect variables such as bias. The “bias” has been defined as the “net downward flow of water through the froth” or “net difference of water flow between the tailings and feed” (Finch and Dobby 1990). Another way to quantify the bias is by the ratio between the wash-water addition and the water recovered into the concentrate, called “displacement washes” (Redfearn and Egan 1989). For a proper interpretation of the bias results, the operation should be under steady state and, ideally, without froth mixing. In plant practice, entrainment into the concentrate can be observed working even under a positive bias condition, because of the froth mixing. Bias has been estimated at laboratory column scale using neural network modeling techniques (i.e., Del Villar, Gregoire, and Pomerleau 1999) where calibration was derived from mass balance assuming steady state. The actual trend is to measure the target variables directly. If onstream analysis is available, then grade can be controlled directly. For example, online measurement of the concentrate grade, together with automatic manipulation of the air flow rate, wash-water flow rate, froth depth, and reagents dosage, provides space for process optimization. On the other hand, as column circuits grow in size, it is common to find four, six, or eight large-size columns in parallel. Then it is unlikely that each unit will have onstream analysis, which makes other methods based on bias or gas holdup worth evaluating. However, alternative approaches such as the cascade control of air holdup and air flow rate, and the cascade control of bias and wash-water flow rate, are still in development, and they are not widely used in plant practice.

flotation0.book Page 733 Tuesday, January 2, 2007 7:36 PM

COLUMN FLOTATION

733

P L A N T O P E R AT I O N Pulp Conditioning

Density of pulp feed or solid weight percentage is regulated by tank conditioners or thickeners in more complex column flotation circuits, such as selective copper-molybdenum separation, where several column flotation stages are required. Online measurements of pulp density can be obtained from nuclear density devices; however, it is rarely included in the whole control strategy because of limitations in calibration and maintenance. Particle size distribution or some other physical property of the pulp entering the column is not usually available, and they are considered disturbances. Chemical reagents such as collectors, depressants, and frothers are usually added before the cleaning stage. Measurement of pH is important, and its regulation is usually made independent of the flotation column operation in the previous stages of the process. A minor variation in pH does not significantly affect the column performance. Feed grade can strongly affect the column concentrate grade, so that feed grade variations due to changes in the ore are important disturbances that can be only partially compensated with a proper management of the control tools. For example, changes in sulfide copper ore mineralogy can be observed through the Cu/Fe ratio. Thus, continuous variations of the Cu/Fe ratio from 1 to 3 (i.e., because of the change in the proportion of chalcopyrite, chalcocite, and pyrite) can dramatically change the final copper concentrate grade between 30%–50% for the same copper feed grade. This poses a great challenge in terms of control and a good opportunity for using more efficient (i.e., supervisory, optimizing) control strategies (Bergh and Yianatos 2003). Column Variables Interaction

Significant interaction between manipulated variables has been observed in plant practice. For instance, the effect of adding more wash water generally increases the concentrate grade, though recovery decreases; however, below a certain critical froth depth, the trends can reverse, as shown in Figure 5. For froth depths >50 cm, the concentrate grade increases while increasing the wash water, as expected. However, for froth depths chalcopyrite > pyrrhotite. By maintaining an atmosphere of nitrogen above the pulp, better recovery of nickel and rejection of pyrrhotite can be achieved (Kelebek 1993). In the INCO matte separation process, the bulk nickel-copper concentrates are flashsmelted to a Bessemer matte, and by slow cooling of the matte, copper sulfide crystallizes as chalcocite, nickel sulfide crystallizes as heazlewoodite (Ni3S2), and a Cu-Ni-Fe alloy is formed. After grinding, the alloy is removed and chalcocite is separated from the heazlewoodite by flotation at pH 12 with addition of lime using diphenyl guanidine as both collector and frother (Boldt 1967; Xu, Wells, and Wong 2003). Russian Practice

In Russia, the copper-nickel sulfide deposits in the Noril’sk-Talnakh Area (Taimyr peninsula in northwestern Siberia) are processed. The major minerals are chalcopyrite, pentlandite, cubanite, and pyrrhotite, which occur in massive or disseminated ores. The massive ore containing 70% by volume sulfides is crushed, and selective flotation yields a copper concentrate, a nickel concentrate, and a pyrrhotite concentrate (all containing some PGEs). In copper flotation, sodium butyl dithiophosphate is used as a collector and pine oil as a frother, whereas in nickel flotation, potassium butyl xanthate is the collector, T-80 is the frother, sodium dimethyl dithiocarbamate is a depressant, sodium bisulfite (NaHSO3) is a pH modifier, and CaO is used to control solution alkalinity. With the disseminated copper-nickel sulfide ores, gravity concentration precedes flotation. The gravity concentrate is ground, and a bulk float is made using potassium butyl xanthate and sodium butyl dithiophosphate as collectors, T-80 as frother, and CaO as a depressant (Kozyrev et al. 2002). Better selectivity and metal recovery for flotation under nitrogen has been demonstrated with the Noril’sk disseminated ore (Volkov et al. 1991). In the plant, rougher flotation

flotation0.book Page 809 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

809

under nitrogen with butyl xanthate (70 g/t) as collector and a frother (12 g/t) gave increases in nickel, copper, and noble metals recovery (7.2%, 5.5%, and 16%, respectively), although there was only a small increase in grade (0.5% for nickel). F L O TAT I O N O F C O P P E R - M O LY B D E N U M O R E S

Copper porphyry ores often contain small amounts of molybdenum as well as copper. At levels as low as 0.01% Mo, it may be economic to recover the molybdenum in a separate concentrate as a by-product with the copper. The major molybdenum mineral of economic importance is molybdenite (MoS2). Its natural floatability is utilized in commercial practice as it is first floated with the copper minerals, and then a selective float is used to separate the molybdenite from the copper minerals. Two general methods are used to achieve the separation. The molybdenite may be depressed and the copper minerals floated or, more commonly, the copper and iron sulfides are depressed and the molybdenite is floated. Depending on the copper mineralogy, several reagent regimes have been used to achieve this latter separation. The copper–molybdenum separation is usually followed by several, or often many, stages of cleaning of the molybdenite concentrate to increase the molybdenum concentrate grade. Several typical operations are described in the following subsections. Kennecott Magna Operations, Utah

The copper-molybdenum separation is achieved by depressing the molybdenite and floating the copper minerals at the Kennecott Magna operations in Utah, United States (Tveter and McQuiston 1962). The ore contains chalcopyrite with minor amounts of chalcocite and bornite. Bulk rougher Cu-MoS2 flotation is achieved in two stages with addition of lime for pH control, cyanide, sodium dicresyl dithiophosphate as collector, MIBC and cresylic acid as frothers in the first stage, and fuel oil and cresylic acid as frothers in the second stage. To the first-stage concentrate (which contains the bulk of the copper), dextrin, xanthate, and frother are added to depress the molybdenite and float a final copper concentrate. The tailing is combined with the second-stage concentrate and roasted to destroy residual reagents on the minerals. Talcose gangue minerals are then floated by addition of a frother with lime for control of pH. Molybdenite is recovered from the tailing by flotation with fuel oil, sodium silicate, and a frother, with lime for pH control. Chuquicamata Deposit, Chile

With the chalcocite- and covellite-rich deposits in Chile (e.g., the Chuquicamata deposit), the copper rougher concentrates that contain the molybdenite are obtained with isopropyl xanthate (22.5 g/t) and Aerofloat 238 (2.5 g/t) as collectors, Dowfroth 250 and pine oil (2.5 g/t) as frothers, and CaO (1,900 g/t) for pH control. The rougher concentrate is thickened and reground prior to two stages of cleaning. A scavenger concentrate from the cleaner tailing is returned for regrinding. The cleaner copper concentrates are thickened and washed to remove excess copper flotation reagents and then conditioned at 60 wt % solids with 4,500–6,000 g/t Nokes reagent (phosphorus pentasulfide, P2S5; and sodium hydroxide, NaOH) to depress the copper and iron sulfides, and the molybdenite is then floated. The rougher molybdenite concentrate is cleaned up to seven times (including a regrind) and leached with NaCN (750–1,100 g/t) to remove residual copper to yield a concentrate containing 90%–93% MoS2 (Tveter and McQuiston 1962).

flotation0.book Page 810 Tuesday, January 2, 2007 7:36 PM

810

TABLE 15

FLOTATION PLANT PRACTICE

Metallurgical results for San Manuel copper-molybdenum operation (United States) Assay, %

Product Flotation feed Bulk Cu-MoS2 flotation concentrate Bulk flotation tailing Feed to Cu-MoS2 separation Molybdenite flotation concentrate Copper flotation concentrate

Cu 0.79 27.8 0.14 27.8 0.91 28.0

Distribution, % MoS2 0.018 0.59 0.004 0.59 95.4 0.16

Cu 100.0 83.3 16.7 100.0 NA NA

MoS2 100.0 77.6 22.4 100.0 72.5 27.5

NA = Not available.

San Manuel Operation, Arizona

The San Manuel operation in Arizona, United States (Tveter and McQuiston 1962), is a porphyry deposit with the molybdenite associated with chalcopyrite and nonsulfide copper minerals (mainly chrysocolla). The copper flotation flowsheet included rougher flotation; regrinding of the rougher concentrates with two stages of cleaning using Minerec A (9.0 g/t), Aero 404 (7.5 g/t), and isopropyl xanthate (5.5 g/t) as collectors; MIBC (26.5 g/t) and stove oil (10.0 g/t) as frothers; and lime (1,400 g/t) for pH control. The copper concentrate was thickened and treated with sodium hypochlorite (6,900 g/t NaOH and 6,200 g/t Cl2 [chlorine]) to partially oxidize the residual flotation reagents. In the subsequent selective float, sodium ferrocyanide (675 g/t) was added to depress the copper minerals, and stove oil (600 g/t) was added to promote molybdenite flotation. The rougher concentrate was given eight stages of cleaning at pH 8.0 and a low pulp density (5 wt % solids) to yield a product containing 95.4% MoS2. Metallurgical data for the operation are given in Table 15. Highland Valley Copper Operations, British Columbia

In porphyry copper deposits in British Columbia, Canada, molybdenite is associated with chalcopyrite at the Brenda Mines operations (Bradburn 1978) and with chalcopyrite and bornite in the Highland Valley Copper operations at Logan Lake ( Johnston, Simkus, and Caines 2000). Sodium hydrogen sulfide is used in the copper-molybdenum separation to depress the copper and iron minerals in both operations. In the Highland Valley Copper operations ( Johnston, Simkus, and Caines 2000), fuel oil (0.1 g/t), potassium amyl xanthate (2.8 g/t), sodium hydrogen sulfide (0.2 g/t), Dowfroth 250 (7.2 g/t), and pine oil (5.5 g/t) are added during grinding and various stages of flotation to produce a bulk copper-molybdenum concentrate (Figure 9). Lime is added to the grinding circuit to maintain the pH at 9.2 and to the cleaning circuit to obtain a pH of 10.5. The bulk flotation concentrates are thickened to 60 wt % solids and conditioned with sodium hydrogen sulfide (45 g/t) to depress the copper minerals in molybdenite flotation. Shirley (1981) has suggested conditioning for 24 hours when chalcopyrite is the major copper mineral. Fuel oil (0.2 g/t) is added as a collector in the rougher and scavenger flotation for molybdenite. Carbon dioxide is introduced in the first flotation cell to maintain a pH of 9.0. The remaining flotation cells use nitrogen/air as the flotation carrier gas. The molybdenum rougher concentrate is reground and refloated in column cells to give a final molybdenum flotation concentrate, which is sent to a leaching plant. The molybdenum scavenger tailing is the final copper concentrate. The molybdenum scavenger concentrate is recycled to the molybdenum conditioner.

flotation0.book Page 811 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

811

Crushed Ore Water SAG Mills

Molybdenum Conditioner

Screen

Oversize

Undersize

Molybdenum RougherScavenger Flotation

Molybdenum Molybdenum Rougher Scavenger Conc. Tail

Cyclone and Regrind Mill

Cyclone Overflow

Regrind Mill

Bulk Cu-Mo RougherScavenger Flotation

Rougher Conc.

Scavenger Scavenger Conc. Tail

Scavenger Conc.

Final Copper Conc.

Molybdenum Column Flotation

Tail

Conc.

Final Tail Cyclone and Regrind Mill

Molybdenum Conc. to Leach Plant

Cyclone Overflow

Bulk Cleaner and Recleaner Flotation

Bulk Cleaner Tail

Bulk Recleaner Conc.

Thickener

Overflow

Underflow

Final Tail

FIGURE 9 (Canada)

Flowsheet of copper-molybdenum ore treatment at the Highland Valley Copper mill

The molybdenum concentrate contains 2.5%–4% Cu and is selectively leached using ferric chloride, with chlorine added to reoxidize ferrous ions, to remove the copper to 10–11 by the addition of lime, the judicious addition of sodium cyanide, and the addition of an appropriate collector. If the arsenopyrite contains gold, it may be possible to float the arsenopyrite separately and send it to a gold recovery plant. However, in some deposits, such as those at Lepanto Consolidated Mining Co., at Lepanto in the Philippines (Tveter and McQuiston 1962), arsenic occurs in the copper minerals themselves, and a different approach is necessary. At Lepanto, the main economic minerals are enargite and luzonite, both of which have the formula 3Cu2S·As2S3 (see Table 1), and they account for 90% of the copper content. Chalcopyrite is also present as well as tetrahedrite and tennantite. Gold is present mainly as gold tellurides, but native gold is also present. Gangue minerals are mainly quartz, chalcedony, and sericite. Very fine grinding was necessary to liberate the copper and gold minerals. The treatment flowsheet used at Lepanto (Figure 10) is divided into two main sections. In the first section, the ore is ground and then floated to produce a bulk concentrate while producing a low-grade scavenger tailing for discard. In the second section, the bulk concentrate is upgraded by further grinding and reflotation. Reagent usage is shown in Table 17; metallurgical results are summarized in Table 18. Bulk flotation on primary classifier overflow is conducted in acid circuit at pH 5.6 and 42% solids. Bulk rougher concentrate is cleaned and thickened to 70%–80% solids, and then given a 30-minute aeration-attritioning-conditioning treatment with lime to scour the mineral surfaces and enable subsequent pyrite depression. Attritioner discharge is floated at pH 10.5 in a three-stage upgrading circuit to produce a final concentrate. Only pyrite depressants are used, without collector or frother, in this stage (Table 17). Tailings from the upgrading section were classified, reground, and floated in another three-stage upgrading circuit to produce more final concentrate and final tailing. A description of the rebuilding of the operation after its destruction during World War II is given by Murray and Bein (1951). No information was found on the treatment of the copper-arsenic concentrate to recover copper and/or arsenic. F L O TAT I O N O F C O P P E R - U R A N I U M O R E S

Although copper-uranium ores are not common, the Olympic Dam deposit of the Olympic Dam Joint Venture at Roxby Downs in South Australia is a copper-uranium ore with reserves of more than 500 Mt. The plant was commissioned in 1988 as a joint venture between Western Mining Corporation Ltd. and BP Minerals. Subsequently, it was operated by WMC Resources Ltd. and from 2005 by BHP Billiton Ltd. In 1998, 1.8 Mt of ore was treated at grades of 3.3% Cu and 1,000 ppm U3O8 (Vonk 1993), whereas in 2004, 9 Mt of ore was treated at grades of 1.1% Cu and 400 ppm U3O8. There are two forms of copper mineralization: chalcocite-bornite and chalcopyritebornite mineralization. The major uranium mineral is uraninite (UO2) with smaller

flotation0.book Page 813 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

813

Crushed Ore Water Grinding and Classification

Bulk Flotation

Bulk Rougher Bulk Rougher Conc. Tail

Cleaner Flotation

Bulk Cleaner Conc.

Bulk Cleaner Tail

To Circuit

Scavenger Conc.

Scavenger Tail

Discard

Regrind

Thickener

Overflow

Scavenger Flotation

Underflow (70%–80% solids)

Aeration Attritioning

Rougher-CleanerRecleaner Flotation

Recleaner Conc.

Rougher Tail

Regrind and Refloat In Rougher-CleanerRecleaner Circuit

Refloat Recleaner Conc.

Final Conc.

Refloat Rougher Tail

Discard

FIGURE 10 Flowsheet of copper-arsenic ore treatment at Lepanto Consolidated Mining Co. (Philippines)

TABLE 17 Reagent consumptions at the copper-arsenic flotation plant of Lepanto Consolidated Mining Co. (Philippines)

Reagent Minerec 27 Aerofloat 25 Pine oil Reagent 404 Zinc sulfate Sodium cyanide Lime Butyl xanthate

Bulk Flotation, g/t of ore 40 35 85 75 — — — —

Upgrading Section g/t of ore — — — — 40 50 330 4

g/t of bulk concentrate — — — — 260 190 2,000 27

flotation0.book Page 814 Tuesday, January 2, 2007 7:36 PM

814

FLOTATION PLANT PRACTICE

TABLE 18 Metallurgical results at the copper arsenic flotation plant of Lepanto Consolidated Mining Co. (Philippines) Assay Product Feed Bulk concentrate Bulk tailing Upgraded concentrate Upgraded section tailing Final mill tailing Copper concentrate

Cu, % 3.29 19.07 0.15 30.5 0.55 0.175 NA

Recovery Au, g/t 4.45 24.4 10.7 32.5 6.2 1.3 NA

Cu, % 100.0 NA NA NA NA NA 95.1

Au, % 100.0 NA NA NA NA NA 76.5

NA = Not available.

amounts of coffinite (U[SiO4,(OH)4]) and brannerite [(U,Ca,Y,Ce)(Ti,Fe)2O6]. The major gangue minerals are hematite, sericite, potassium feldspar, and quartz. There is no pyrite in the ore (Vonk 1993). A simplified flowsheet of the Olympic Dam Joint Venture flotation circuit is shown in Figure 11, but this does not show the main uranium leaching circuit. Typical metallurgical data are given in Table 19. The ore is ground to a P80 of 75 μm and fed to a flotation circuit that consists of a bulk sulfide float and three stages of cleaning. Sodium ethyl xanthate (40 g/t) and Interfroth 75 (40 mL/min) are added in flotation. The first cleaner concentrate is reground to a P80 of 25 μm for two additional cleaning stages. A column cell is used in the final cleaning stage. Amdel in-stream analysis probes are used to monitor the % Cu, % Fe, and solids pulp density in the flotation circuit (Vonk 1993). The thickened concentrate is leached with sulfuric acid (150 kg/t) in an inert atmosphere at 65°–75°C for 12 hours to remove uranium from the concentrate (Ragozzini and Sparrow 1987). The leached concentrate is washed with dilute sulfuric acid and water, repulped, neutralized with caustic soda, and refloated using sodium ethyl xanthate and Interfroth 75. The leaching liberates a significant amount of gangue from the copper concentrate, and the refloat increases the copper concentrate grade (Vonk 1993). The leached copper concentrate is then fed to the smelter. F L O TAT I O N O F C O P P E R - T I N O R E S

Copper-tin ores are not common, but the treatment of such an ore at Cleveland Tin Ltd., Luina, Tasmania, Australia, was described by Robinson (1980). The ore contained 0.5% Sn, mostly as cassiterite (SnO2), but some stannite (Cu2S·FeS·SnS2) was also present. Some 10%–20% sulfides such as pyrite and pyrrhotite were present with about 0.5% Cu present as chalcopyrite. This ore was treated at the rate of 450,000 t/a using the flowsheet shown in Figure 12. Most of the cassiterite was recovered by gravity methods, but fine cassiterite (minus 106 μm) was recovered by flotation. Sulfides were removed by flotation, and a copper sulfide concentrate assaying about 20%–25% Cu and 2%–3% Sn was produced. Pyrrhotite that built up in the magnetic ferro-silicon medium used in the heavy-medium separation section of the plant was removed in periodic campaigns by flotation. Thus, the circuit shown in Figure 12 contained three separate flotation sections: a pyrrhotite flotation circuit to remove sulfides from the ferro-silicon heavy medium, a bulk sulfide flotation circuit plus a chalcopyrite recovery circuit, and a cassiterite flotation circuit on fine sulfide-free tailings.

flotation0.book Page 815 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

815

Run-of-mine Cu-U Ore Water Autogenous Mill

Trommel

Oversize

Undersize

Crusher

2 Ball Mills

Second and Third Cleaner Flotation

Second Cleaner Tail

Cyclones

Third Cleaner Conc.

Overflow Underflow

Thickener

Underflow Overflow

Rougher-Scavenger Flotation To Circuit Rougher Conc.

Scavenger Scavenger Tail Conc.

First Cleaner Float

Thickener Residue Overflow Underflow

First Cleaner Tail

First Cleaner Conc.

Concentrate Leach

To Circuit

Regrind

Flotation Tailing to Main Cu–U Leach

Solution

Rougher-Scavenger Flotation

Scavenger Tail

Rougher Conc. To Smelt

To Cu-U Recovery

Cyclone

Overflow Underflow

FIGURE 11 (Australia)

Flowsheet of copper-uranium ore treatment at the Olympic Dam Joint Venture

Crushed ore (minus 45 mm) was screened at 15 mm and the plus-15-mm fraction was sent to heavy-medium separation. Coarse washed “float” was discarded. Coarse washed “sink” was combined with the minus-15-mm ore and ground to 80% minus 250 μm using a rod mill in closed circuit with DSM (Dutch States Mines) screen. Screen undersize was thickened and sent to a conditioner where sodium ethyl xanthate and potassium amyl xanthate (110 g/t in total) plus 40 g/t of copper sulfate were added, and the pH was adjusted to 5.5 with sulfuric acid (1,770 g/t). An unspecified frother was added

flotation0.book Page 816 Tuesday, January 2, 2007 7:36 PM

816

FLOTATION PLANT PRACTICE

TABLE 19

Typical metallurgical data for flotation of Olympic Dam (Australia) copper-uranium ore Assay

Product Flotation feed Flotation concentrate Flotation tailing Concentrate after leach and reflotation

Weight, % 100 6.5

Cu, % 3.32 46.6

S, % 1.6 19.5

93.5 NA

0.23 56.8

0.4 22.5

Distribution U3O8, ppm 1,000 1,110

Cu, % 100.0 96.5

S, % 100.0 77.2

U3O8, % 100.0 7.9

6.5 NA

22.8 NA

92.1 NA

902 213

NA = Not available.

Crushed Ore (–45 mm)

Crush and Wet Grind and Screen

Screen

Ferrosilicon Makeup +15 mm

–15 mm

Screen Undersize (–0.5 mm)

Heavy-Medium Separation

Float

Washing

Washed Float

Washings

Thickener

Sink

Underflow

Washing

Washings

Overflow

Conditioner

Washed Sink

To Circuit

Bulk Sulfide RougherScavenger Flotation

Discard Rougher Conc. Ferrosilicon Medium Recovery

Pyrrhotite Flotation

Tail

Scavenger Tail To Recovery of Cassiterite by Gravity and Flotation Methods

Regrind

Conc.

Discard Copper Rougher Flotation

Rougher Tail (Pyrite)

Rougher Conc. (Chalcopyrite)

Discard

3-Stage Cleaning

Third Stage First Stage Conc. Tailing Final Copper Conc.

FIGURE 12

Flowsheet of copper-tin ore treatment at Cleveland Tin Ltd. (Australia)

flotation0.book Page 817 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

817

and a bulk sulfide concentrate produced in a rougher-scavenger flotation circuit. Scavenger tailing was sent to the cassiterite circuit where cassiterite concentrates were produced by gravity separation of the coarser material and flotation of the finer material. Bulk sulfide rougher concentrate was reground in a ball mill/cyclone circuit to 80% minus 75 μm and pumped to the copper flotation section where a copper rougher tailing and a copper rougher concentrate were produced. The copper rougher tailing, which was mainly pyrite, was discarded. The copper rougher concentrate was treated with lime (500 g/t) to raise the pH to 11 and then cleaned in three stages. The first-stage tailing was recycled to the copper rougher float, and the third-stage concentrate, assaying 20%–25% Cu and 2%– 3% Sn, was the final copper concentrate. The tin in this product was mainly present as stannite, with minor amounts of cassiterite in composites.

Section II: Flotation of Zinc-Lead-Silver Ores N.W. Johnson

INTRODUCTION

This section focuses on plant designs for the Mount Isa and George Fisher deposits, and for the Red Dog, Century, and McArthur River deposits, which are relatively new, major producers and for which there has been much process development. These deposits are classified as stratiform deposits (Waltho, Alnutt, and Radojkovic 1993). Some relevant background for complex sulfide ores was published recently by Johnson and Munro (2002). GEORGE FISHER ORE PROCESSING

Much development of the lead-zinc concentrator of Mount Isa Mines Ltd. (now Xstrata) occurred between 1982 and 2000. The reason for this is because the zinc recoveries to the zinc concentrate had traditionally been relatively low and the remaining ore had increased zinc feed grades and lowered lead feed grades unlike previous ore, thereby increasing the importance of higher zinc recoveries. Increases in throughput and changes to more difficult ore sources also revealed capacity limitations and increased the necessity for a more efficient plant. A description of the plant before the development phase is given in work by Watsford (1980a). It was demonstrated that the relatively low recoveries of zinc arose from relatively low levels of sphalerite liberation (Young et al. 1997). In summary, the levels of sphalerite liberation were addressed by much finer sizings in the primary and secondary grinding stages and also by creation of new technology to regrind middling streams to product sizings that previously were not considered achievable economically but which were necessary, given the texture of the ore. The liberation of sphalerite (two-dimensional uncorrected values) was improved from 55% to 85% from 1992 to 2000. A new phase of process development for the extremely difficult McArthur River ore (another ore body held by Mount Isa Mines Ltd.) commenced in 1989. Because of the extremely fine-grained nature of this lead-zinc ore, regrinding to 80% passing 7 μm was necessary in the laboratory and pilot plant where modified-Netzsch horizontal stirred-mill technology allowed this regrinding target to be reached. Unfortunately, no industrial-scaleequivalent mill that was modified for mineral industry applications existed for implementation in the planned operation at McArthur River. Hence, prototype and final designs for such a large mill (3,000 L) had to be developed and evaluated on suitable streams in the existing

flotation0.book Page 818 Tuesday, January 2, 2007 7:36 PM

818

FLOTATION PLANT PRACTICE

lead-zinc concentrator located at Mount Isa (Enderle et al. 1997). This IsaMill technology demonstrated the benefits of regrinding key streams for the Mount Isa lead-zinc ore and provided an avenue for further development of that circuit. Although the Mount Isa leadzinc ore was fine grained, it was not as fine as McArthur River ore. However, a portion of each tonne of feed to the lead-zinc concentrator did require unusually fine grinding to maximize liberation and recovery. The development of the lead-zinc concentrator flowsheet culminated with a major upgrade in 1999, which served to implement further relevant technology and to convert the circuit for future production, principally from the George Fisher ore bodies (Young, Pease, and Fisher 2000). The current flowsheet has been reproduced in Figure 1 ( J. Pease, personal communication). The various grinding targets in the circuit are summarized for the original circuit and the new circuit. The technology in each grinding stage is indicated in Table 1, and more details can be obtained in the literature (Pease et al. 2004). The IsaMills provided economic kilowatt-hour-per-tonne values in reaching unusually fine product sizings and, because of the inert media, detrimental side effects from indiscriminant precipitation of iron hydroxides were not incurred. The zinc circuit produced zinc concentrate in three steps (Figure 1): 1. The liberated sphalerite in the zinc rougher concentrate is upgraded and directed to final zinc concentrate by use of columns. 2. The additional sphalerite liberated in the first zinc regrinding step is upgraded and directed to final zinc concentrate by use of closed-circuit conventional cleaners. 3. The additional sphalerite liberated in the second zinc regrinding step is upgraded and directed to final zinc concentrate by use of a second set of closed-circuit conventional cleaners. Progressing through the steps, the actual size range of sphalerite being treated appears to decrease, and the case is made (Pease et al. 2004) that this assists in creating appropriate physical and chemical conditions for flotation of the particles in that size range and avoids using chemical conditions that are only a compromise for a wide range of particle sizes. The sphalerite recovery size data for the three steps are summarized in Figure 1 and show the increasing fineness of the sphalerite in each feed and the high recoveries of sphalerite in the finest fractions (Cyclosizer C6 and C7). The dominant general characteristics of this circuit are recognized as follows: 1. Use of preflotation (without cleaning). 2. Extensive use of Jameson column technology to upgrade and direct liberated galena in the rougher concentrator to the lead concentrate, and conventional column technology to upgrade and direct liberated sphalerite in the zinc rougher concentrate to the zinc final concentrate. 3. Use of IsaMill regrinding technology in the lead circuit and, principally, IsaMill regrinding technology in two zinc circuit locations. 4. The use of three separate cleaning systems in the zinc circuit, with the tailing from the last stage in open circuit (i.e., reporting to final tailing, with the last two cleaning systems containing IsaMill regrinding and closed-circuit conventional cleaners). 5. The use of three stages of closed-circuit lead cleaning with the tailing of the first stage in open circuit reporting to the feed of the zinc circuit.

flotation0.book Page 819 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

50 45 40 35 30 25 20 15 10 5 0 C7 0–4 μm

C6 4–8 μm

C5/C4 C3/C2 C1/38 8–16 μm 16–30 μm 30–53 μm

53

Zn Recovery, %

100 90 80 70 60 50 40 30 20 10 0

75

50 45 40 35 30 25 20 15 10 5 0 C7 0–4 μm

C6 4–8 μm

C5/C4 C3/C2 C1/38 8–16 μm 16–30 μm 30–53 μm

Size Fraction

53

Size Distribution, %

12-µm Zinc Circuit

100 90 80 70 60 50 40 30 20 10 0

Size Distribution, %

Zn Recovery, %

37-µm Zinc Circuit

819

75

Size Fraction 50 45 40 35 30 25 20 15 10 5 0 C7 0–4 μm

C6 4–8 μm

C5/C4 8–16 μm

C3/C1 16–38 μm

38/53

Size Distribution, %

Zn Recovery, %

7-µm Zinc Circuit 100 90 80 70 60 50 40 30 20 10 0 75

Size Fraction

70-μm Primary Grind/Float

37-μm Secondary Grind/Float Prefloat

Rod and Ball Milling

PB-Ro Scavenger

Zn-Ro Scavenger Tailings

Zn-Ro

Ball Milling

Zinc Columns

37% Zinc Recovery, 54% Zinc

Tailings

33% Pb

Zinc Conc.

3 × 1.1 MW Jameson Cell Pb

12-μm Regrind/Float

2x Zinc Conc. 34 % Zinc Recovery, 50% Zinc

Zinc

46% Pb Recovery

1 × 0.52 MW Tower Mill

12-μm Regrind/Float

3 × 1.1 MW 3 stages of closed-circuit conventional cleaning.

Zinc Retreat

Tailings Zinc Zinc Conc. 7-μm Regrind/Float

6% Zinc Recovery, 47% Zinc

Source: J. Pease, personal communication.

FIGURE 1 Flowsheet for George Fisher (Australia) zinc-lead-silver ore showing three stages in the zinc circuit

TABLE 1

Summary of product sizing for each grinding and regrinding step Product Sizing, 80% passing size, µm

Circuit Original New

Primary Grinding Circuit 150* 70*

Secondary Grinding Circuit 75* 37*

* Closed-circuit ball milling. † Open-circuit IsaMills. ‡ Closed-circuit tower mill followed by open-circuit IsaMills. § Open-circuit IsaMills processing cyclone underflow from feed. NA = Not available.

Lead Regrind NA 12†

Zinc Regrind 1 35* 12‡

Zinc Regrind 2 NA 7§

flotation0.book Page 820 Tuesday, January 2, 2007 7:36 PM

820

FLOTATION PLANT PRACTICE

R E D D O G C O N C E N T R AT O R

The Red Dog concentrator commenced production in 1989 and is presently the largest producer of zinc concentrate in the world. The phases in the development and simplification of the current circuit (Figure 2) have been documented by Lacouture and Hope (2003). The ore is considered to be of moderate difficulty in its processing properties. The reason for introducing major changes in the flowsheet in Figure 2 in 2001 was to ensure that adequate installed flotation capacity and, therefore, residence time existed in each part of the circuit along with the quoted flowsheet changes for the zinc circuit (Lacouture and Hope 2003). The changes were as follows: • “Bypass part of the zinc rougher concentrate as sphalerite liberation in the first few cells was in excess of 80% and laboratory and plant testwork confirmed it to have no impact on zinc final concentrate grade and recovery. • Increase the number of cleaning steps from two to three. • Make the circulating loads more logical and easier to control by operators. • Minimize entrainment by allowing reduced pulp density with sufficient residence time. • Have two separate circuits to enable a different chemical and regrind environment to suit the complex particles.” The dominant general characteristics of this circuit are: 1. Use of preflotation (without cleaning). 2. Extensive use of column cleaning technology in lead and zinc cleaning and in the zinc retreatment section.

Feed Preflotation

Lead Roughers

Zinc Roughers

Zinc Cleaners Lead Scavengers

Zinc Retreat

Lead Conc.

FIGURE 2

Zinc Conc.

Tailings

Red Dog value-improvement-program circuit (United States)

flotation0.book Page 821 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

821

3. Use of tower mill regrinding technology in the lead circuit and zinc circuit (two separate locations). 4. Use of two separate closed-circuit cleaning systems in the zinc cleaners with the tailing of the second stage in open circuit, and the zinc first cleaner tailing in the main cleaning system being directed to a zinc retreatment bank whose tailing became the exit point from the cleaning system. 5. The tailing of the lead cleaning system was in closed circuit, being recycled to the middle of the lead roughers. Extensive use was reported of reduced tonnage trials and flotation simulation to establish appropriate installed capacities throughout the circuit. Plans were to reduce the percent solid values in the pulp from 41% in the zinc roughers and 20%–30% in the zinc cleaners to 35% and 10%–20%, respectively, to lower entrainment of gangue. Hence, capacity increases were required to accommodate the larger water additions in the zinc roughers and cleaners, as well as to respond to throughput increases and existing bottlenecks. Lower percent solid values were sought because the mechanisms for recovery of unwanted minerals had been determined, and the large contribution from entrainment of liberated nonsulfide gangue was understood (Gorain and Stradling 2003). C E N T U R Y Z I N C C O N C E N T R AT O R

The Century concentrator commenced production in 1999 and is presently the second largest producer of zinc concentrate in the world. The startup of the Century plant has been described by Burgess et al. (2003). The ore is considered to be of extreme difficulty in its processing properties with many similarities to the McArthur River ore body. The flowsheet is reproduced in Figure 3 (Burgess et al. 2003).

Cyclones (25 cm, 2 clusters × 24)

Primary Sizer

Lead Rougher Tank Cells (6)

Run-of-Mine Hopper or ey nv Co

Water

SAG Mill (11.0 m × 4.9 m)

Con

r veyo

Flotation Screen (2.4 m × 6.1 m) Ball Mill

(6.1 m × 9.8 m)

Oversize to Waste Dump

Stockpile

Tailings Thickener

Zinc Rougher Zinc Scavenger Tank Cells (7) Tank Cells (11)

Tailings Dam Prefloat Lead Cleaner 1 Rougher Tank Cells (3) Tank Cells (7)

Lead Thickener Conditioner

Zinc Cleaner 1 Tank Cells (6)

To Pipeline Lead Cleaner 2 Tank Cells (2)

Cyclones (5 cm, 20 Clusters × 16)

Sand Grinders Conditioners (15) (2)

Conditioner Zinc Regrind (6)

Zinc Cleaner 3 Tank Cells (8)

Lead Storage Tanks (1) Zinc Cleaner 5 Tank Cells (6)

To Pipeline

Zinc Cleaner 5 Tank Cells (6)

Zinc Cleaner 4 Tank Cells (8)

Source: Burgess, et al. 2003.

FIGURE 3

Flowsheet of Century zinc concentrator in 2003

Zinc Thickener

Zinc Storage Tanks (3)

flotation0.book Page 822 Tuesday, January 2, 2007 7:36 PM

822

FLOTATION PLANT PRACTICE

The extremely difficult processing properties arise principally from the very fine grained nature of the ore, particularly with respect to the minerals reporting to the sphalerite concentrate. These difficulties existed even though the sphalerite is of unusually high purity (typically 0.6 mm are not fed to the primary float. The large mesh size of these screens means that very little oversize returns to the mill. The primary flotation rougher cells consist of 20-m3 forced-air tank cells. The primary rougher tailings are fed to an open circuit secondary ball mill where they are ground to their final grind size of about 75% passing 75 μm. A densifying cyclone is used to ensure that the correct density is fed to the mill. From the secondary mill, the pulp is fed to the secondary rougher circuit, which is a mirror image of the primary rougher circuit. The secondary rougher tailings proceed to final tailings. The concentrate from both roughers is fed to the primary cleaner circuit. This circuit has three stages of cleaning and produces a high-grade concentrate from the easily floated material. The primary cleaner tailings are fed to the secondary cleaning circuit, which produces a low-grade concentrate from the slower-floating material. These two concentrates are combined and then sent to the smelter after thickening in a high-rate vertical thickener. The secondary cleaner tailings are combined with the secondary rougher tailings as final tailings in a conventional thickener and are sent to the tailings dam. The cleaner circuits are made up of a combination of 10-m3 and 5-m3 naturally aspirated tank cells. The flowsheet for Lonmin Platinum is shown in Figure 1. Flotation Circuit Operation

The reagents used in the flotation circuit are as follows: • Copper sulfate is added to each of the milling stages as an activator. • Xanthate (sodium isobutyl xanthate, or SIBX) is added to the two rougher stages as a sulfide and metallic collector.

flotation0.book Page 834 Tuesday, January 2, 2007 7:36 PM

834

FLOTATION PLANT PRACTICE

Shaft

High-Rate Thickening

Bulk Ore Handling, Primary Crushing and Coarse Ore Storage

High-Grade Cleaners, 3 Stages

Primary ROM Milling and Screening

Secondary Open-Circuit Milling

Primary Rougher Flotation

Secondary Rougher Flotation

Low-Grade Cleaners, 2 Stages

Conventional Thickening

Tailings Dam Concentrate Dispatch

FIGURE 1

Smelter

Flowsheet of Lonmin Platinum operation (South Africa)

Depressant (carboxy methyl cellulose, or CMC) is added to the roughers and the cleaners for talc control. • Frother (Dowfroth 200) is added to the roughers and the cleaners. • Flocculant (anionic) is added to the concentrate thickeners to achieve the desired density. Reagent additions are controlled using peristaltic pumps and flow meters based on the feed rate of ore to the plant. The float is conducted at the natural pH of the ore, which is about 8.5, and the water used in the float is a combination of recycled water and municipal water for makeup. The float cells are controlled to predetermined levels by Mintek’s FloatStar, and the final concentrate mass pull is controlled using a mass-pull-control algorithm based on concentrate flow rate and density. The chromite grade in the concentrate is monitored continuously using an Outokumpu Courier system. •

S T I L LWAT E R P L AT I N U M - PA L L A D I U M M I N E

The Stillwater platinum-palladium mine is located in the Beartooth Mountains, 130 km southwest of Billings, Montana, United States. In 2003, Stillwater produced totals of 450,000 oz Pd and 134,000 oz Pt. Associated by-product metals include rhodium, gold, silver, nickel, and copper. The mine has been in production since 1987, and in 2003, a new flotation concentrator was built at the East Boulder mine site to increase production (Major et al. 2003). The PGMs are contained in the narrow J-M reef, and the PGMs are associated with pentlandite, pyrrhotite, and chalcopyrite. More than 100 platinum and palladium species are said to be present, including the occasional grains of braggite (Kennedy 1987). Ore sent to the new East Boulder concentrator is treated in a 6.7-m-diameter × 2.29-m SAG mill driven by a 1.5-MW motor. Discharge from the SAG mill and ball mill are combined before pumping to four cyclones. A portion of the cyclone underflow is treated in a

flotation0.book Page 835 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

835

SAG Discharge Rougher Cells

Unit Cell

Scavenger Cells

Rougher Conditioner

Flotation Tails

Cleaner-Scavenger Cells

Cleaner Conditioner

Cleaner Cells

Cleaner Column Concentrate Thickener

Flotation Conc.

FIGURE 2

Flowsheet of Stillwater operation (United States)

10-m3 Svedala unit flotation cell. The remaining portion is combined with the unit flotation cell tailings and directed to a 3.8-m-diameter × 4.88-m ball mill. The unit flotation concentrate reports to the final concentrate thickener. Approximately 45% to 50% of PGMs in the feed are recovered in the unit cell concentrate, and it is estimated that treatment of the cyclone underflow by flotation improves overall PGM recovery by 1% to 2% ( J. Sargent, personal communication). The cyclone combined overflow products at a P80 of 120 μm into two banks of five rougher flotation cells, five middling cells, and five scavenger cells. All the cells are of 8.5 m3 capacity. A three-stage cleaning circuit is used consisting of two 2.8-m3 mechanical cells for the first cleaner and four 1.4-m3 mechanical cells for the second cleaner. The final stage cleaning is performed in two 0.91-m-diameter × 9.3-m column cells. The first cleaner tailings are pumped to a regrind mill, and the ground product is treated in a cleaner-scavenger bank consisting of six 2.8-m3 mechanical cells. Tailings from the scavenger circuit are recycled back to the head of the rougher circuit, and the tailings from the second-stage cleaning are recycled back to the first-stage cleaners. Column tailings are recycled back to the head of the second-stage cleaners. Flotation residence time is nominally 60 minutes in the rougherscavenger circuit and 30 minutes in the cleaner circuit. Targeted platinum and palladium recoveries are 95% and 91%, respectively. The flotation reagent regime consists of a CMC talc depressant and Cytec reagents Aero 350 and Aero 3477 as collectors. The flotation flowsheet is shown in Figure 2. L A C D E S I L L E S PA L L A D I U M M I N E

The Lac des Iles mine is located 115 km north of Thunder Bay, Canada. The mine commenced operation in 1993, and a new concentrator was built in 2001 to increase the ore treatment rate from 2,400 t/d to 15,000 t/d (Martins et al. 2003).

flotation0.book Page 836 Tuesday, January 2, 2007 7:36 PM

836

FLOTATION PLANT PRACTICE

The most common PGM minerals are kotulskite and palladoarsenide, and the sulfide minerals include pyrite, chalcopyrite, pentlandite, and pyrrhotite. The talc content of the ore varies between 2% and 4%. Ore from the mine is fed to a 9.1-m-diameter × 4.3-m SAG mill driven by a 6.3-MW motor. Secondary grinding is performed in two 6.1-m-diameter × 10.4-m ball mills that produce a flotation feed product having a P80 of 75 μm. The flotation circuit consists of two parallel trains of rougher-scavenger cells. Two Outokumpu 50-m3 TankCells as rougher cells are followed by fourteen Outokumpu 130-m3 tank scavenger flotation cells. The rougher concentrates are combined and reground to a P80 of 20 μm before being cleaned. In the original cleaning circuit, the reground rougher concentrate was cleaned in a single 1.7-m-diameter × 11-m column cell. The scavenger concentrates were also reground (P80 of 20 μm) prior to cleaning in a bank of nine Outokumpu 38-m3 flotation cells. The tailings from the cleaner cells were combined with the rougher-scavenger tailings prior to disposal. The cleaner concentrate was also cleaned in a bank of Denver D-100 cells, and the concentrate from these cells was pumped to two 1.3-m-diameter × 11-m column cells in parallel. The tailings from the columns reported back to the feed of the Denver cleaner cells, whereas the tailings from the second stage of cleaning were recycled back to the Outokumpu cleaner cells. The column cell concentrates were combined with the rougher column cell concentrate to form the final concentrate. The flotation reagents consist of a CMC depressant (750 g/t of mill feed), two collectors (potassium amyl xanthate and di-isobutyl dithiophosphate), and MIBC as frother. From startup, poor froth stability problems were encountered and the addition of Dowfroth 200 did not eliminate the problem. Decreasing the addition rate of the CMC provided a more workable solution by allowing some of the talc to stabilize the froth. However, the more stable froths created pumping problems around the regrind and cleaning circuits, and the CMC addition requires control in a narrow operating range (0–200 g/t) to prevent problems with either over- or underfrothing. In addition to the frothing problems, it soon became apparent that the recovery performance at 68% was well below the target of 82%. In addition, the MgO content in the final concentrate exceeded the levels acceptable to the smelter (9.9% versus a target of 7%). An extensive flotation survey and mineralogical program followed, and this identified that both problems were related to insufficient cleaning capacity. The first change to the cleaning circuit consisted of changing the rougher-cleaner circuit to three stages by including a bank of mechanical cells after the regrinding stage. The concentrate from these cells was then cleaned in the two small column cells that formed part of the scavenger cleaner circuit before being pumped to the large column cell. For the scavenger cleaning circuit, the three-stage cleaning approach was retained, but all columns were replaced by conventional mechanical cells. This circuit provided substantial concentrate quality benefits as well as improvements to overall recovery. Despite these issues, additional plant surveys and mineralogical work identified parts of the cleaning circuits that were overloaded. Therefore, the cleaning circuit was modified as shown in Figure 3, and recovery and MgO content improved to 74% and 7%, respectively. Work continues to improve the overall recovery process, and finer grinding of the concentrate has been shown at laboratory scale to be beneficial to improving recovery. Ultrafine losses of the PGMs are also the subject of further studies.

flotation0.book Page 837 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

Flotation Feed

837

Rougher Cells Scavenger Cells

Flotation Tails Regrind Mill

Scavenger Cleaner Cells

Cleaner Tailings

Cleaner Cells Recleaner Scavenger Recleaner Column

Regrind Mill

Final Flotation Conc.

Column Cells

FIGURE 3

Flowsheet of Lac des Iles operation (Canada)

REFERENCES

Abramov, A.A., and K.S.E. Forssberg. 2005. Chemistry and optimal conditions for copper minerals flotation: Theory and practice. Miner. Process. Extr. Metall. Rev. 26(2):77–143. Ackerman, P.K., G.H. Harris, R.R. Klimpel, and F.F. Aplan. 1984. Effect of alkyl substituents on performance of thionocarbamates as copper sulfide and pyrite collectors. Pages 69–78 in Reagents for the Mineral Industry. Edited by M.J. Jones and R. Oblatt. London: The Institution of Mining and Metallurgy. ———. 1987a. Evaluation of flotation collectors for copper sulfides and pyrite. I. Common sulfhydryl collectors. Int. J. Miner. Process. 21(1–2):105–127. ———. 1987b. Evaluation of flotation collectors for copper sulfides and pyrite. III. Effect of xanthate chain length and branching. Int. J. Miner. Process. 21(1–2):141–156. ———. 1999. Use of chelating agents as collectors in the flotation of copper sulfides and pyrite. Miner. Metall. Process. 16(1):27–35. Adam, K., K.A. Natarajan, and L. Iwasaki. 1984. Grinding media wear and its effect on the flotation of sulfide minerals. Int. J. Miner. Process. 12(1):39–54. Ahn, J.H., and J.E. Gebhardt. 1991. Effect of grinding media-chalcopyrite interaction on the selfinduced flotation of chalcopyrite. Int. J. Miner. Process. 33(1–4):243–262. Allan, G.C., and J.T. Woodcock. 2001. A review of the flotation of native gold and electrum. Miner. Eng. 14(9):931–962. Barzyk, W., K. Malysa, and A. Pomianowski. 1981. The influence of surface oxidation of chalcocite on its floatability and ethyl xanthate sorption. Int. J. Miner. Process. 8(1):17–29. Boldt, J.R., Jr. 1967. The Winning of Nickel. Edited by P. Queneau. London: Methuen. Bowen, P. 1999. The operational and metallurgical development of the McArthur River concentrator. In Flotation Colloquium—April 28, 1999. Australia, University of Queensland: Sir James Foots Institute of Mineral Resources. Bradburn, R.G. 1978. Brenda mines concentrator. Pages 88–91 in Milling Practice in Canada. Edited by D.E. Pickett, W.S. Hall, and G.W. Smith. Montreal: Canadian Institute of Mining and Metallurgy. Bruce, R.W. 1976. The effect of grinding media on the selective flotation of copper-lead-zinc ores. Paper presented at the SME-AIME Fall Meeting, Denver, Colorado. Salt Lake City: SME.

flotation0.book Page 838 Tuesday, January 2, 2007 7:36 PM

838

FLOTATION PLANT PRACTICE

Buckley, A.N., and R. Woods. 1992. The interaction of diethyl dithiophosphate with freshly exposed galena and chalcocite surfaces. Electrochemistry in Mineral and Metallurgical Processing III. Pages 269–285 in Proceedings of the 3rd International Symposium on Electrochemistry in Mineral and Metallurgical Processing. Volume 92-17. Edited by R. Woods and P.E. Richardson. Pennington, NJ: The Electrochemical Society. Burgess, F., L. Reemeyer, M. Spagnolo, M. Ashley, and D. Brennan. 2003. Ramp up of the Pasminco Century concentrator to 500 000 tpa zinc metal production in concentrate. Pages 153–163 in Proceedings of the Eighth Mill Operators’ Conference, Melbourne: Australasian Institute of Mining and Metallurgy. Burns, J., editor. 1987. Leslie Bradford Golden Jubilee Oration. Melbourne: Australasian Institute of Mining and Metallurgy. Carter, J.M., editor. 1957. The Milling of Canadian Ores. Sixth Mining and Metallurgical Congress. Montreal: Canadian Institute of Mining and Metallurgy. Chryssoulis, S., and L. Cabri. 1990. Significance of gold mineralogical balances in mineral processing. Trans. Inst. Min. Metall. Sect. C 99:C1–C9. Crozier, R.D. 1992. Flotation: Theory, Reagents, and Ore Testing. Oxford: Pergamon Press. Damjanovi´c, B., and J.R. Goode, editors. 2000. Pages 95–155 in Canadian Milling Practice. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Dana, E.S., and W.E. Ford. 1945. A Textbook of Mineralogy. 4th edition. New York: John Wiley & Sons. Djingheuzian, L.E. 1957. History of milling in Canada. Pages 3–36 in The Milling of Canadian Ores. Edited by J.M. Carter. Montreal: Canadian Institute of Mining and Metallurgy. Dunkin, H.H., editor. 1953. Ore Dressing Methods in Australia and Adjacent Territories. Melbourne: The Australasian Institute of Mining and Metallurgy. Dunne, R., R. Chittenden, G. Lane, and S. Morrell. 1999. The Cadia gold copper project— exploration to start-up. Paper presented at SME Annual Conference, Denver, March. Enderle, U., P. Woodall, M. Duffy, and N. Johnson. 1997. Stirred mill technology for regrinding McArthur River and Mount Isa zinc-lead ores. Pages 71–78 in Proceedings of the XX International Mineral Processing Congress, Volume 2. Germany: GDMB, Clausthal-Zellerfeld. Fahlström, P.H. 1974. Autogenous grinding of base metal ores at Boliden Aktiebolag. CIM Bull. 67(743):128–141. Fahlström, P.H., O. Fägremo, and A.S. Gjerdrum. 1975a. Autogenous grinding at Boliden’s Aitik plant, Part I. World Min. 28(3):42–47. ———. 1975b. Autogenous grinding at Boliden’s Aitik plant, Part II. World Min. 28(4):42–46. Fairthorne, G., D. Fornasiero, and J. Ralston. 1997. Interaction of thionocarbamate and thiourea collectors with sulfide minerals: A flotation and adsorption study. Int. J. Miner. Process. 50(4):227–242. Forssberg, E., and T.V. Subrahmanyam. 1993. Grinding, pulp chemistry and particle floatability. Volume 1, Comminution and classification. Pages 1–6 in Proceedings of the XVIII International Mineral Processing Congress, Sydney. Volume 1. Melbourne: Australasian Institute of Mining and Metallurgy. Forssberg, E., S. Sundberg, and Z. Hongxin. 1988. Influence of different grinding methods on floatability. Int. J. Miner. Process. 22(1–4):183–192. Fuerstenau, M.C., J.L. Huiatt, and M.C. Kuhn. 1971. Dithiophosphate versus xanthate flotation of chalcocite and pyrite. Trans. SME-AIME 250(3):227–231. Gleisner, A.G., and J.C. Vega. 1992. Froth Flotation of Sulfide Ores Using Dithiazine in Collectors. Chile: Establecimientos Industrales Quimicos Oxiquim. Gorain, B., and A. Stradling. 2003. The Red Dog VIP mill optimisation project. Part B: An integrated approach to modelling and design of the flotation plant. CIM Bull. 96(1069):100–106. Grano, S., P.L.M. Wong, W. Skinner, N.W. Johnson, and J. Ralston. 1994. The effect of autogenous and ball mill grinding on the chemical environment and flotation of the copper ore of Mount Isa Mines Limited, Australia. Pages 351–388 in Flotation—a Volume in Memory of Alexander Sutulov. Volume 2. Edited by S. Castro and J. Alvarez. Concepción, Chile: Andros Ltd.

flotation0.book Page 839 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

839

Hamilton, J.K., and J.T. Woodcock. 1993a. Copper ore treatment at Bougainville Copper Limited, Panguna, Papua New Guinea. Pages 639–645 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: The Australasian Institute of Mining and Metallurgy. ———. 1993b. Zinc-copper ore treatment by Murchison Zinc Company Pty. Ltd. at the Scuddles mine, Golden Grove, WA. Pages 541–543 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: The Australasian Institute of Mining and Metallurgy. Heyes, G.W., and J.M. Phelan. 1988. The application of separate conditioning to improve zinc metallurgy at Woodlawn Mines. Pages 85–89 in Proceedings of the Third Mill Operators’ Conference. Melbourne: The Australasian Institute of Mining and Metallurgy. Heyes, G.W., and W.J. Trahar. 1977. The natural floatability of chalcopyrite. Int. J. Miner. Process. 4(4):317–344. ———. 1979. Oxidation-reduction effects in the flotation of chalcocite and cuprite. Int. J. Miner. Process. 6(3):229–252. Holmes, W., D. MacNamara, G. Marrs, P. Wells, and P. Fragomeni. 2000. Falconbridge LimitedStrathcona mill. Pages 95–97 in Canadian Milling Practice. Edited by B. Damjanovi´c and J.R. Goode. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Iwasaki, I., K.J. Reid, H.A. Lex, and K.A. Smith. 1983. Effect of autogenous and ball mill grinding on sulfide flotation. Min. Eng. 35(8):1184–1190. Johnson, N., and P. Munro. 2002. Overview of flotation technology and plant practice for complex sulphide ores. Pages 1097–1123 in Mineral Processing Plant Design, Practice, and Control Proceedings. Volume 1. Edited by A.L. Mular, D.N. Halbe, and D.J. Barratt. Littleton, CO: SME. Johnston, D.L., R. Simkus, and B. Caines. 2000. Highland Valley copper. Pages 84–88 in Canadian Milling Practice. Edited by B. Damjanovi´c and J.R. Goode. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Jones, M.H., and J.T. Woodcock. 1984. Applications of pulp chemistry to regulation of chemical environment in sulphide mineral flotation. Pages 147–183 in Principles of Mineral Flotation. Edited by M.H. Jones and J.T. Woodcock. Melbourne: The Australasian Institute of Mining and Metallurgy. Kelebek, S. 1993. The effect of oxidation on the flotation behaviour of nickel-copper ores. Flotation II and Miscellaneous. Pages 999–1005 in Proceedings of the XVIII International Mineral Processing Congress. Volume 4. Melbourne: The Australasian Institute of Mining and Metallurgy. Kennedy, A. 1987. Stillwater platinum palladium mine. Min. Mag. 157(11):418–427. Kerr, A. 2002. An overview of recent developments in flotation technology and plant practice for nickel ores. Pages 1142–1158 in Mineral Processing Plant Design, Practice, and Control Proceedings. Volume 1. Edited by A.L. Mular, D.N. Halbe, and D.J. Barratt. Littleton, CO: SME. Kimble, J.B., and R.M. Parlman. 1982. Trithiocarbonates and their use in flotation. Belgium Patent BE 890,634. Assigned to Phillips Petroleum Co. ———. 1987. Recovering metal sulfides by flotation using mercaptoalcohols. U.S. Patent 4,643,823. Assigned to Phillips Petroleum Co. Kozyrev, S.M., M.Z. Komarova, L.N. Emelina, O.I. Oleshkevich, O.A. Yakovleva, D.V. Lyalinov, and V.I. Maximov. 2002. The mineralogy and behaviour of PGM during processing of the Noril’skTalnakh PGE-Cu-Ni ores. Pages 757–775 in The Geology, Geochemistry, Mineralogy and Mineral Beneficiation of Platinum-Group Elements. CIM Special Volume 54. Edited by L.J. Cabri. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Krishnan, N.I. 1993. Lead-zinc ore concentrator practice by Pasminco Mining–Rosebery, Tas. Pages 519–522 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: The Australasian Institute of Mining and Metallurgy. Lacouture, B., and G. Hope. 2003. The Red Dog VIP mill optimisation project. Part A: Rectifying the shortage of flotation capacity. CIM Bull. 96(1068):64–68. Laplante, A., and R. Dunne. 2002. The gravity recoverable gold test and flash flotation. Paper No 7 in Proceedings of the 34th Annual Meeting of the Canadian Mineral Processors. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum.

flotation0.book Page 840 Tuesday, January 2, 2007 7:36 PM

840

FLOTATION PLANT PRACTICE

Large, R.R., and W.G. Mumme. 1975. Junoite, wittite, and related seleniferous bismuth sulfosalts from Juno mine, Northern Territory, Australia. Econ. Geol. 70(2):369–383. Lawson, B., and D. Carr. 2003. The flotation capacity project to increase the recovery in the Mount Isa Mines copper concentrator. Pages 93–99 in Proceedings of the Eighth Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Leppinen, J.O. 1990. FTIR and flotation investigation of the adsorption of ethyl xanthate on activated and non-activated sulfide minerals. Int. J. Miner. Process. 30(3–4):245–263. Leppinen, J.O., C.I. Basilio, and R.H. Yoon. 1989. In-situ FTIR study of ethyl xanthate adsorption on sulfide minerals under conditions of controlled potential. Int. J. Miner. Process. 26(3–4):259–274. Li, X., and I. Iwasaki. 1992. Effect of cathodic polarization on the floatability of chalcopyrite in the absence of oxygen. Miner. Metall. Process. 9(1):1–6. Liechti, D., M. Mayhew, A. Kerr, and J. Truskoski. 2000. INCO Limited, Ontario DivisionClarabelle mill. Pages 97–101 in Canadian Milling Practice. Edited by B. Damjanovi´c and J.R. Goode. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Lin, Q., X. Yang, and D. Wang. 1999. Chelate collector alpha-oximophosphonate preparation and structure-property relationship. Sel. Pap. Eng. Chem. Metall. (China) 81–88. Lumsdaine, I., and S.M. O’Hare. 1993. Copper concentrator practice at Mount Isa Mines Limited, Mount Isa, Qld. Pages 649–651 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: Australasian Institute of Mining and Metallurgy. Ma, J., L. Yu, and K. Sun. 1984. Novel froth-collector for flotation of sulfide minerals—CEED. Pages 287–290 in Reagents for the Mineral Industry. Edited by M.J. Jones and R. Oblatt. London: Institution of Mining and Metallurgy. Maier, G.S., X. Qiu, and B. Dobias. 1997. New collectors in the flotation of sulfide minerals: A study of their electrokinetic, calorimetric, and flotation properties of sphalerite, galena, and chalcocite. Colloids Surf. A 122(1–3):207–225. Major, K., P. Lefebvre, J. Sargent, and B. Darnton. 2003. Design, construction and start-up of Stillwater Mining’s East Boulder concentrator. Pages 51–69 in Proceedings of the 35th Annual Meeting of Canadian Mineral Processors. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Martin, C.J., R.E. McIvor, J.A. Finch, and S.R. Rao. 1991. Review of the effect of grinding media on flotation of sulfide minerals. Miner. Eng. 4(2):121–132. Martins, C., T. Ohrling, T. Olsen, and P. Taggart. 2003. From tribulation to triumph—flotation improvements at the new Lac des Iles concentrator. Pages 13–27 in Proceedings of the 35th Annual Meeting of Canadian Mineral Processors. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. McLaglen, V. 1993. Copper concentrator practice at the Warrego plant of Poseidon Gold Limited, Tennant Creek, NT. Pages 652–653 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: Australasian Institute of Mining and Metallurgy. Michell, F.B. 1950. The Practice of Mineral Dressing. London: Electrical Press. Mielczarski, J.A., E. Mielczarski, and J.M. Cases. 1998. Influence of chain length on adsorption of xanthates on chalcopyrite. Int. J. Miner. Process. 52(4):215–231. Mill, J.H.A., B.A. Clifford, R.J. Dudley, and P.A. Ruxton. 1990. Scuddles zinc-copper deposit at Golden Grove. Pages 583–590 in Geology of the Mineral Deposits of Australia and Papua New Guinea. Edited by F.E. Hughes. Melbourne: Australasian Institute of Mining and Metallurgy. Moyses, D.R. 1965. Mount Isa Mines Ltd. copper. Pages 109–118 in The Australian Mining, Metallurgical, and Mineral Industry. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Murray, G., and H.H. Bein. 1951. Rebuilding Lepanto—The Far East’s largest producer of copper. Eng. Min. J. 152(March): 72–75. Murray, R.M. 1953. Treatment of copper ores. Pages 173–179 in Ore Dressing Methods in Australia and Adjacent Territories. Edited by H.H. Dunkin. Melbourne: Australasian Institute of Mining and Metallurgy.

flotation0.book Page 841 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

841

Musgrove, N.H. 1980. Bismuth production at Peko Mines, Tennant Creek, N.T. Pages 609–610 in Mining and Metallurgical Practices in Australasia. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Nickel, E.H., and M.C. Nichols. 1991. Mineral Reference Manual. New York: Van Nostrand Reinhold. Nihill, D., C. Stewart, and P. Bowen. 1998. The McArthur River mine—the first years of operation. Pages 73–82 in AusIMM Annual Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Pease, J.D., M.F. Young, D. Curry, and N.W. Johnson. 2004. Improving fines recovery by grinding finer! Pages 65–78 in Metallurgical Plant Design and Operating Strategies [electronic resource]. Melbourne, Australasian Institute of Mining and Metallurgy. Persson, P., and B. Malmensten. 1991. Interactions between sulfide minerals and alkylxanthate ions. 5. A vibration spectroscopic study of the interactions between chalcocite, synthetic copper(I) sulfide, acanthite, and synthetic silver(I) sulfide, and ethylxanthate ions in aqueous and acetone solutions. Colloids Surf. 59(1):279–292. Pickett, D.E., W.S. Hall, and G.W. Smith, editors. 1978a. Milling of nickel-copper ores. Pages 133– 151 in Milling Practice in Canada. Montreal: Canadian Institute of Mining and Metallurgy. ———. 1978b. Milling Practice in Canada. Montreal: Canadian Institute of Mining and Metallurgy. Ragozzini, R.J., and G.J. Sparrow. 1987. Selective dissolution of uranium from Olympic Dam copper concentrates. Hydrometallurgy 18(3):367–382. Rao, M.K.Y., and K.A. Natarajan. 1989. Effect of electrochemical interactions among sulfide minerals and grinding medium on chalcopyrite flotation. Miner. Metall. Process. 6(3):1146–1151. ———. 1990. Effect of electrochemical interactions among sulfide minerals and grinding media on the flotation of sphalerite and galena. Int. J. Miner. Process. 29(2):175–194. Richardson, P.E., and G.W. Walker. 1985. The flotation of chalcocite, bornite, chalcopyrite, and pyrite in an electrochemical flotation cell. Pages 198–210 in Congress International el Mineralurgie. Volume 2. Compte Rendu 15th. St. Etienne, France: GEDIM. Robinson, R.J. 1980. Tin-copper ore treatment at Cleveland Tin Limited, Luina, Tas. Pages 418–422 in Mining and Metallurgical Practices in Australasia. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Roos, J.R., J.P. Celis, and A.S. Sudarsono. 1990. Electrochemical control of chalcocite and covellitexanthate flotation. Int. J. Miner. Process. 29(1–2):17–30. Senior, G.D., and R. Ahveninen. 2001. Beneficiation of sulfide ores with the flotation of pyritic fraction. U.S. Patent 6,170,669. Senior, G.D., and W.J. Trahar. 1991. The influence of metal hydroxides and collector on the flotation of chalcopyrite. Int. J. Miner. Process. 33(1–4):321–341. Shaw, D.R. 1981. Dodecyl mercaptan: A superior collector for sulfide ores. Min. Eng. 33(6):686–692. Shellshear, W. 1916. Flotation of gold and copper ores at Mount Morgan, Queensland. Proc. Aus. Inst. Min. Eng. 22:51–74. Shirley, J.F. 1981. By-product molybdenite plant design. Pages 26–31 in Proceedings of the 13th Annual Meeting of Canadian Mineral Processors. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum. Simmons, G.L., J.N. Orlich, J.C. Lenz, and J.A. Cole. 1999. Implementation and start-up of N2TEC flotation at the Lone Tree mine. Pages 183–195 in Advances in Flotation Technology. Edited by B.K. Parekh and J.D. Miller. Littleton, CO: SME. Smart, R.St.C. 1991. Surface layers in base metal sulfide flotation. Miner. Eng. 4(7–11):891–909. Stainlay, R.L. 1980. Copper ore concentration at Mount Gunson Mines Pty. Ltd., Mount Gunson, S.A. Pages 316–317 in Mining and Metallurgical Practices in Australasia. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Sutherland, K.L., and I.W. Wark. 1955. Principles of Flotation. Melbourne: Australasian Institute of Mining and Metallurgy. Suttill, K.R. 1991. A technical buyer’s guide to mining chemicals. Eng. Min. J. 192(8):23–33. Thornton, E. 1973. The effect of grinding media on flotation selectivity. Pages 224–239 in Proceedings of the 5th Annual Meeting of Canadian Mineral Processors. Montreal: Canadian Institute of Mining, Metallurgy, and Petroleum.

flotation0.book Page 842 Tuesday, January 2, 2007 7:36 PM

842

FLOTATION PLANT PRACTICE

Torrisi, C., and R. Smith. 2003. Bringing split flotation to reality—the Cannington experience. Pages 73–80 in Proceedings of the Eighth Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Trahar, W.J. 1981. A rational interpretation of the role of particle size in flotation. Int. J. Miner. Process. 8(4):289–327. ———. 1984. The influence of pulp potential in sulfide flotation. Pages 117–135 in Principles of Mineral Flotation. Edited by M.H. Jones and J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Tveter, E.C., and F.W. McQuiston Jr. 1962. Plant practice in sulfide mineral flotation. Pages 382–426 in Froth Flotation 50th Anniversary Volume. Edited by D.W. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Valli, M., B. Malmensten, and I. Persson. 1994. A vibration spectroscopic study of the interaction between some sulfide minerals and O,O-diethyl dithiophosphate ions in aqueous solution. Colloids Surf. A 83(3):227–236. Valli, M., and I. Persson. 1994a. Interactions between sulfide minerals and alkylxanthates, 7. A vibration and X-ray photoelectron spectroscopic study of the interaction between covellite and alkylxanthate ions and the non-sulfide mineral malachite and ethylxanthate ions in aqueous solution. Colloids Surf. A 83(3):199–206. ———. 1994b. Interactions between sulfide minerals and alkylxanthates, 8. A vibration and X-ray photoelectron spectroscopic study of the interaction between chalcopyrite, marcasite, pentlandite, pyrrhotite, and troilite, and ethylxanthate and decylxanthate ions in aqueous solution. Colloids Surf. A 83(3):207–217. Van Olst, M., N. Brown, P. Bourke, and S. Ronkainen. 2000. Improving flotation plant performance at Cadia by controlling and optimising the rate of froth recovery using Outokumpu Frothmaster. Pages 127–135 in Proceedings of the Seventh Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Volkov, V.I., N.I. Geonya, V.V. Rybas, G.P. Ponomaryov, and M.I. Mantsevich. 1991. Creation of the technology of copper-nickel-iron bearing ore beneficiation of Talnakh deposit on the basis of flotation with the use of inert gas. Pages 335–340 in Proceedings of the Copper 91—Cobre 91 International Symposium. Volume II. New York: Pergamon Press. Vonk, A.S.M. 1993. Copper concentrator practice at the Olympic Dam Joint Venture, Roxby Downs, S.A. Pages 656–658 in Australasian Mining and Metallurgy. Edited by J.T. Woodcock and J.K. Hamilton. Melbourne: Australasian Institute of Mining and Metallurgy. Waltho, A., S. Alnutt, and A. Radojkovic. 1993. Geology of the Century zinc deposit, Northwest Queensland. Pages 111–129 in Proceedings of the International Symposium on Zinc. Melbourne: Australasian Institute of Mining and Metallurgy. Waterhouse, L.V. 1920. Flotation practice at Mount Lyell. Proc. Aus. Inst. Min. Metall. 38:105–123. Watsford, R.M.S. 1980a. Lead-zinc-silver ore concentration by Mount Isa Mines Ltd., Mount Isa, Qld. In Mining and Metallurgical Practices in Australasia, the Sir Maurice Mawby Memorial Volume. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. ———. 1980b. Copper ore concentration at Mount Isa Mines Ltd., Mount Isa, Qld. Pages 317–321 in Mining and Metallurgical Practices in Australasia. Edited by J.T. Woodcock. Melbourne: Australasian Institute of Mining and Metallurgy. Winkers, A. 2002. Overview of recent developments in flotation technology and plant practice for copper gold ores. Pages 1124–1141 in Mineral Processing Plant Design, Practice, and Control Proceedings. Volume 1. Edited by A.L. Mular, D.N. Halbe, and D.J. Barratt. Littleton, CO: SME. Woodcock, J.T., editor. 1965. The Australian Mining, Metallurgical, and Mineral Industry. Melbourne: Australasian Institute of Mining and Metallurgy. ———. 1968. Ore dressing developments in Australia, 1967. Aus. Min. 60(7):46–91. ———. 1980. Mining and Metallurgical Practices in Australasia. Melbourne: Australasian Institute of Mining and Melbourne. Woodcock, J.T., and J.K. Hamilton, editors. 1993. Australasian Mining and Metallurgy. Melbourne: Australasian Institute of Mining and Metallurgy.

flotation0.book Page 843 Tuesday, January 2, 2007 7:36 PM

PLANT PRACTICE: SULFIDE MINERALS AND PRECIOUS METALS

843

Woodcock, J.T., and M.H. Jones. 1969. Oxygen concentrations, redox potentials, xanthate residuals, and other parameters in flotation plant pulps. Pages 439–468 in Mineral Processing and Extractive Metallurgy. Edited by M.J. Jones. London: Institution of Mining and Metallurgy. ———. 1970. Chemical environment in Australian lead-zinc flotation plant pulps, I. pH, redox potentials, and oxygen concentrations. Proc. Aus. Inst. Min. Metall. 235:45–60. Woods, R. 1976. Electrochemistry in sulfide flotation. Pages 298–333 in Flotation—A.M. Gaudin Memorial Volume. Edited by M.C. Fuerstenau. New York: American Institute of Mining, Metallurgical, and Petroleum Engineers. Woods, R., C.A. Young, and R.H. Yoon. 1990. Ethyl xanthate chemisorption isotherms and Eh-pH diagrams for the copper/water/xanthate and chalcocite/water/xanthate systems. Int. J. Miner. Process. 30(1–2):17–33. Xu, M., P.F. Wells, and G.P. Wong. 2003. Fundamental studies of INCO matte separation process. Page 278 in Abstracts of the XXII International Mineral Processing Congress. Edited by L. Lorenzen and D.J. Bradshaw. Johannesburg: South African Institute of Mining and Metallurgy. Young, C.A., C.I. Basilio, and R.H. Yoon. 1991. Thermodynamics of chalcocite-xanthate interactions. Int. J. Miner. Process. 31(3–4):265–279. Young, M., J. Pease, and K. Fisher. 2000. The George Fisher project to increase recovery in the Mount Isa lead/zinc concentrator. Pages 157–163 in Proceedings of the Seventh Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Young, M., J. Pease, N. Johnson, and P. Munro. 1997. Developments in milling practice at the lead/ zinc concentrator of Mount Isa Mines Limited from 1990. Pages 3–12 in Proceedings of the Sixth Mill Operators’ Conference. Melbourne: Australasian Institute of Mining and Metallurgy. Zachwieja, J.B., G.W. Walker, and P.E. Richardson. 1987. Electrochemical flotation of sulfides: The bornite-ethylxanthate system. Miner. Metall. Process. 4(3):146–151.

flotation0.book Page 844 Tuesday, January 2, 2007 7:36 PM

BLANK LEFT PAGE

flotation0.book Page 845 Tuesday, January 2, 2007 7:36 PM

Plant Practice: Nonsulfide Minerals A.E.C. Peres, A.C. Araujo, H. El-Shall, P. Zhang, and N.A. Abdel-Khalek

INTRODUCTION

Industrial minerals serve as raw material for many applications, of which phosphate and potash for fertilizer production are by far the largest. Basically, the fertilizers can be divided into two groups: phosphate- and potash-based fertilizers. The concentration method of phosphate and potash as raw materials differs significantly, as phosphate rock is insoluble and potash is soluble in water. Industrial minerals can be floated directly or indirectly to remove impurities such as silica, calcite, or heavy minerals, and so forth. Plant practice of flotation separation of some important nonsulfide minerals such as phosphate, potash, iron ore, feldspar, and so on is discussed in this chapter. P H O S P H AT E O R E F L O TAT I O N

The major producers of phosphate rock are the United States, Morocco, Tunisia, China, Russia, and Brazil. About 138 Mt of phosphate rock were produced worldwide in 2003. The term phosphate ore includes all phosphate-bearing minerals and rocks, but only two are of industrial interest: apatite and sedimentary phosphate (e.g., collophane). Igneous Phosphate Ores

Apatite, Ca5(PO4)3(F, Cl, OH), is of igneous origin and can be formed only in the molten state and upgraded only by flotation. Crystalline apatite has been formed in all stages of magmatic solidification and, thus, is found in nearly any magmatic rock. Beautifully colored crystals are formed in pegmatite and in alpine clefts. Apatite can easily be mixed up with other minerals (e.g., beryl). Therefore, the name apatite is derived from apateon, which means “delusive.” Typical gangue minerals are quartz, carbonates, mica, and clay. Apatite may also occur as a gangue mineral in iron ores. Upgrading of apatite is carried out industrially only by flotation. Deposits can be exploited only if the phosphate (expressed as P2O5) content is at least 7%, which is about 17% apatite. Because of the good crystalline character of the apatite, a high concentration factor of between 5 and 6 can be reached. It is possible to obtain concentrates by flotation with a grade of 38%–42% P2O5 or 87%–97% apatite, and these values can be achieved with apatite recoveries as high as 80%–90%. Prior to flotation, the apatite minerals must be liberated by grinding the rock, typically to 100% –250 μm. In some cases, the subsequent discarding of slimes (