Integrated Membrane Operations in the Food Production 113

Alfredo Cassano, Enrico Drioli Integrated Membrane Operations

Also of interest Lidietta Giorno, Enrico Drioli Membrane Engineering, 2014 ISBN 978-3-11-028140-8, e-ISBN 978-3-11-028139-2

Loredana De Bartolo, Efrem Curcio, Enrico Drioli Membrane Systems: For Bioartificial Organs and Regenerative Medicine, 2014 ISBN 978-3-11-026798-3, e-ISBN 978-3-11-026801-0, Set-ISBN 978-3-11-026802-7

André B. de Haan, Hans Bosch Industrial Separation Processes: Fundamentals, 2013 ISBN 978-3-11-030669-9, e-ISBN 978-3-11-030672-9

Mark Anthony Benvenuto Industrial Chemistry, 2013 ISBN 978-3-11-029589-4, e-ISBN 978-3-11-029590-0

Xiao Dong Chen (Editor-in-Chief) International Journal of Food Engineering ISSN 2194-5764, e-ISSN 1556-3758

www.degruyter.com

Integrated Membrane Operations in the Food Production Edited by Alfredo Cassano, Enrico Drioli

DE GRUYTER

Editors Alfredo Cassano Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected]

Enrico Drioli Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected]

ISBN 978-3-11-028467-6 e-ISBN 978-3-11-028566-6 Set-ISBN 978-3-11-028567-3 Library of Congress Cataloging-in-Publication data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available from http://dnb.dnb.de. © 2014 Walter de Gruyter GmbH, Berlin/Boston Typesetting: Compuscript Ltd., Shannon, Ireland Printing and binding: Hubert & Co. GmbH & Co. KG, Göttingen Cover image: Thinkstock/Hemera ♾ Printed on acid free paper Printed in Germany www.degruyter.com

Preface Process intensification and membranes will play an important role to match the future challenges of agro-food production processes. One of the techniques to intensify processes by target enhancement is the integration of membranes into processing in order to exploit the interesting specific membrane operation properties. This book aims to provide some relevant examples of integrated membrane operations in agro-food productions, highlighting their contribution for an industrial sustainable growth in this area in terms of energy consumption, reduction of environmental impact and product quality. Each chapter reports successful examples of integrated membrane processes in different agro-food sectors, including selected information on basic principles of membrane unit operations, commercial applications and an overview of current research and development. The first chapter (Cuperus and Franken) focuses on ongoing development works based on the use of membrane technology for the production of green products, better and/or natural products. In Chapter 2 (Lutz and Gani) the integration of membrane processes in agro-food production is analyzed according to the process intensification strategy. Integrated membrane operations are reviewed and discussed in different agrofood areas such as fruit juice processing (Chapter 3 – Cassano, Conidi and Drioli), citrus processing (Chapter 4 – Cassano and Jiao), milk processing (Chapter 5 – Mucchetti), whey processing (Chapter 6 – Gésan-Guiziou), winemaking (Chapter 7 – El Rayess and Mietton-Peuchot), brewing and sugar production (Chapter 8 – Lipnizki and Ruby-Figueroa), stevioside purification (Chapter 9 – Mondal and De) and purification of soy extract (Chapter 10 – Mondor). The concentration of polyphenols (Chapter 11 – Tsibranska and Tylkowski) and the recovery of bioactive compounds (Chapter 12 – Brazinha and Crepso) from food processing streams through membrane-based operations are also analyzed. Chapter 13 (Giorno, Mazzei and Piacentini) and Chapter 14 (Charcosset) focus on emerging membrane processes, such as biocatalytic membrane reactors and membrane emulsification, in integrated processes for the production of nutriaceuticals and innovative food formulations. Basic aspects of electrodialysis, as well as its application in integrated processes for food applications, are discussed in detail in the concluding chapter (Chapter 15 – Roux-de Balmann). The editors would like to take also this opportunity to thank all the authors for their expert contribution to this volume.

Enrico Drioli Alfredo Cassano

Contents Preface

 v

Author index 1 1.1 1.2 1.2.1 1.2.2 1.2.3 1.2.3.1 1.2.3.2 1.2.3.3 1.2.3.4 1.2.4 1.2.5 1.3 1.3.1 1.3.2 1.3.3 1.3.4 1.4 1.4.1 1.4.2 1.5 2 2.1 2.1.1 2.1.2 2.2 2.2.1 2.2.2 2.2.3 2.2.4 2.3 2.3.1

 xv

 1 Membrane applications in agro-industry  F. Petrus Cuperus and A.C.M. (Tony) Franken Introduction  1 Membranes in biorefinery  1 What is biorefinery?  1 Mild extraction techniques  2 Use of membranes in biorefinery  4 Crossflow   5 Cross-rotation (CR) filtration  5 Rotating membranes   6 Vibrational membranes   7 Removing minerals from road-side grass   10 Biofuel including microalgae   11 Membranes in vegetable oils and fats   14 Membrane technology applied to vegetable oils  Solvent recovery and reuse   16 Wax removal and/or recovery   17 Goodies in oil   18 Application scale and outlook   20 Application scale   20 Outlook   21 References   21

 14

 25 Process intensification in integrated membrane processes  Philip Lutze and Rafiqul Gani Introduction   25 Background: process intensification   25 Membranes and process intensification   26 Synthesis/design of membrane-assisted PI – overview and concepts   28 Mathematical formulation of the PI synthesis problem   29 PI synthesis based on the decomposition approach   31 Phenomena as building blocks for process synthesis   31 Connection of phenomena   33 Synthesis/design of membrane-assisted PI – workflow   34 Steps of the general workflow   34

viii 

2.3.1.1 2.3.1.2 2.3.1.3 2.3.1.4 2.3.2 2.3.3 2.3.3.1 2.3.3.2 2.3.3.3 2.3.3.4 2.3.4 2.3.4.1 2.3.4.2 2.3.4.3 2.4 2.4.1 2.4.2 2.4.2.1 2.4.2.2 2.4.2.3 2.4.2.4 2.4.2.5 2.4.2.6 2.5 2.5.1 2.5.2 2.5.3 2.5.3.1 2.5.3.2 2.5.3.3 2.5.3.4 2.5.4 2.5.4.1 2.5.4.2 2.5.4.3 2.6 2.7

 Contents

Step 1: Define problem   34 Step A2: Analyze the process   37 B2: Identify and analyze necessary tasks to achieve the process target    37 Step 6: Solve the reduced optimization problem and validate most promising   37 KBS workflow   38 UBS workflow   38 Step U2: Collect PI equipment   38 Step U3: Select and develop models   38 Step U4: Generate feasible flowsheet options   39 Step U5: Fast screening for process constraints   39 PBS workflow   39 Step P3: Identification of desirable phenomena   40 Step P4: Generate feasible operation/flowsheet options   40 Step P5: Fast screening for process constraints   40 Synthesis/design of membrane-assisted PI – sub-algorithms, supporting methods and tools   41 Sub-algorithms   41 Supporting methods and tools   41 Knowledge base tool   42 Model library   42 Method based on thermodynamic insights   42 Driving force method   43 Extended Kremser method   43 Additional tools   43 Conceptual example   45 Step 1: Define problem   45 Step A2: Analyze the process   45 Result of the PBS workflow   46 Step P3: Identification of desirable phenomena   46 Step P4: Generate feasible operation/flowsheet options   48 Step P5: Fast screening for process constraints   49 Step 6: Solve the reduced optimization problem and validate most promising   50 Comparison of solutions obtained from PBS, KBS and UBS   51 Result of the KBS workflow   51 Result of the UBS workflow   53 Comparison of the results   53 Conclusions   55 References   55

Contents 

3

Integrated membrane operations in fruit juice processing   59 Alfredo Cassano, Carmela Conidi and Enrico Drioli 3.1 Introduction   59 3.2 Clarification of fruit juices   59 3.3 Concentration of fruit juices   65 3.3.1 Nanofiltration   65 3.3.2 Reverse osmosis   66 3.3.3 Osmotic distillation   67 3.3.4 Membrane distillation   69 3.4 Integrated membrane operations in fruit juices production   71 3.4.1 Apple juice   71 3.4.2 Red fruit juices   74 3.4.3 Other fruit juices   78 3.4.3.1 Kiwifruit juice   78 3.4.3.2 Cactus pear juice   79 3.4.3.3 Melon juice   81 3.5 Conclusions   81 3.6 References 82 4 4.1 4.2 4.3 4.4 4.4.1 4.4.2 4.5 4.6 4.7 4.8 5

5.1 5.2 5.2.1 5.2.2 5.2.3 5.2.4

 87 Integrated membrane operations in citrus processing  Alfredo Cassano and Bining Jiao Introduction   87 Clarification of citrus juices   89 Debittering of orange juice   92 Concentration of citrus juices   93 Reverse osmosis   93 Membrane distillation and osmotic distillation   95 Recovery of aroma compounds   102 Treatment of citrus by-products   103 Concluding remarks   108 References   109 Integrated membrane and conventional processes applied to milk processing   113 Germano Mucchetti Introduction   113 Fluid milk   114 MF and bacterial removal   114 MF, somatic cells and enzyme removal   118 Membrane reactors for free lactose milk   119 Heat labile ingredients sterilization (MF/UF) and addition to heat-treated milk during packaging   120

 ix

x 

 Contents

5.3 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.4 5.5 6 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 7 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 8 8.1 8.2 8.2.1 8.2.1.1 8.2.1.2 8.2.1.3 8.2.1.4 8.3

Cheese milk   121 Reverse osmosis application to cheese milk   121 Cheese milk concentration   121 Cheese milk medium and high concentration   122 Cheese milk standardization   123 Cream concentration by UF for mascarpone cheese   125 Cheese brine treatment   127 Conclusions   128 References   128  133 Integrated membrane operations in whey processing  Geneviève Gésan-Guiziou Introduction   133 Whey types and composition   133 Concentration and demineralization of whey   135 Concentration of serum proteins   138 Fractionation of individual serum proteins   141 Development of new value-added products from whey   143 Conclusions and challenges   144 References   145  147 Integrated membrane processes in winemaking  Youssef El Rayess and Martine Mietton-Peuchot Introduction   147 Crossflow microfiltration for must, wine and lees clarification  Electrodialysis and bipolar electrodialysis   152 UF and NF for reduction of must sugars   156 RO and NF for sugar must concentration   157 RO, NF and MC for wine dealcoholization   158 Gas control by membrane processes   160 References   161  163 Membrane operations in the sugar and brewing industry  Frank Lipnizki and René Ruby-Figueroa Introduction   163 Beet and cane sugar production   163 Membrane applications on beet sugar production   164 Sugar beet press water and pulp recycling   165 Raw juice purification   167 Demineralization of beet juice   172 Preconcentration of thin juice   173 Membrane application in cane sugar production   175

 148

Contents 

8.3.1 8.3.2 8.3.3 8.3.4 8.4 8.4.1 8.4.1.1 8.4.1.2 8.4.1.3 8.4.1.4 8.5 8.5.1 8.6 9

9.1 9.2 9.3 9.3.1 9.3.2 9.3.3 9.3.4 9.3.5 9.3.6 9.3.7 9.4 9.5 9.5.1 9.5.2 9.5.3 9.5.4 9.5.5 9.6 10

10.1 10.1.1 10.1.2

Raw sugar cane juice purification   176 Concentration of clarified cane juice   181 Molasses treatment   181 Decolorization of remelted raw sugar   182 The brewing industry   183 Membrane applications in the brewing process  Filtration in the lautering process   187 Beer clarification   188 Dealcoholization of beer   190 Beer from tank bottoms   193 Conclusions and outlook   194 Acknowledgements   195 References   195

 xi

 185

Processing of stevioside using membrane-based separation processes    201 Sourav Mondal and Sirshendu De Introduction   201 Physical and biological properties of steviol glycosides   203 Extraction methods of steviol glycosides   205 Ion-exchange   205 Solvent extraction   205 Extraction by chelating agents   206 Adsorption and chromatographic separation   206 Ultrasonic extraction   206 Microwave-assisted extraction   206 Super critical fluid extraction (SCFE)   206 State-of-the-art membrane-based processes   207 Detailed membrane-based clarification processes   208 Hot water extraction   208 Selection of operating conditions and membrane   212 Crossflow ultrafiltration  217 Nanofiltration   220 Diafiltration   222 References   226 Production of value-added soy protein products by membrane-based operations  233 Martin Mondor Introduction   233 Soy as the most important source of plant protein ingredients   233 Production of soy protein isolates by isoelectric precipitation   233

xii 

 Contents

10.1.3 10.2 10.2.1 10.2.1.1 10.2.1.2 10.2.1.3 10.2.2 10.2.2.1 10.2.2.2 10.2.3 10.3 10.3.1 10.3.1.1 10.3.1.2 10.3.1.3 10.3.2 10.3.3 10.4 10.4.1 10.4.2 10.5 10.5.1 10.6

Soy bioactive peptides   234 Membrane technologies in the processing of soy protein products   235 Ultrafiltration   236 Membranes   237 Membrane fouling   237 Operating variables   238 Electrodialysis   239 Conventional electrodialysis   239 Bipolar membrane electrodialysis   241 Integrated electrodialysis-ultrafiltration process   243 Production of soy protein isolates by membrane technologies   244 Ultrafiltration   244 Removal of undesirable components of soy protein extracts   245 Production of soy protein isolate with a high amount of isoflavones   246 Functionality of soy protein isolate produced by ultrafiltration   247 Electrodialysis with bipolar membranes   249 Electrodialysis with bipolar membranes in combination with ultrafiltration-diafiltration   251 Separation of soy peptides by membrane technologies   254 Ultrafiltration   255 Integrated electrodialysis – ultrafiltration approach   258 Concluding remarks and perspectives   261 Acknowledegments   262 References   262

 269 Concentration of polyphenols by integrated membrane operations Iren Tsibranska and Bartosz Tylkowski 11.1 Introduction   269 11.1.1 Beneficial effects of polyphenols   270 11.1.2 Separation/concentration of polyphenols by traditional methods   270 11.1.2.1 Separation of polyphenols at laboratory scale   271 11.1.2.2 Concentration of polyphenols at industrial scale   272 11.2 Concentration of polyphenols by integrated membrane operations   272 11.2.1 Membrane processes for concentration of plant extracts   273 11.2.2 Membrane processes for concentration of juices   281 11.2.3 Membrane processes for recovery/concentration of polyphenols from industrial waste waters (WW)   281 11.3 References   289 11

Contents 

 xiii

12

Valorization of food processing streams for obtaining extracts enriched in biologically active compounds   295 Carla Brazinha and Joao G. Crespo 12.1 Introduction   295 12.2 Market of the natural extracts ingredients    295 12.3 Production of natural extracts – process and final product requirements    297 12.4 Fractionation, concentration and purification of BAC with membrane-processing techniques   300 12.4.1 Fractionation with pervaporation/vapor permeation   300 12.4.2 Extract fractionation and purification by nanofiltration   302 12.5 Concluding remarks   306 12.6 References   306 13 13.1 13.2 13.3 13.3.1 13.3.2 13.3.3 13.3.4 13.3.5 13.3.6 13.4 13.5 14 14.1 14.2 14.2.1 14.2.2 14.2.3 14.3 14.3.1 14.3.2 14.3.3 14.3.4 14.4 14.4.1

Biocatalytic membrane reactors for the production of nutraceuticals  Lidietta Giorno, Rosalinda Mazzei and Emma Piacentini Introduction   311 General aspects   313 Applications   316 Starch sugars   317 Fruit juices processing   317 Production of functional molecules and spices   318 Fats and oils   318 Alcoholic beverages   318 Water purification for food production   319 Conclusions   321 References   322

 311

Membrane emulsification in integrated processes for innovative food  Catherine Charcosset Introduction   323 Membrane emulsification   324 Configurations   324 Membranes   326 Influence of parameters   327 Applications   328 Simple emulsions   328 Multiple emulsions   329 Encapsulation   330 Aerated food gels   331 Integrated processes   331 Beverages   331

 323

xiv 

14.4.2 14.5 14.6

 Contents

Dairy products   332 Conclusions   334 References   334

15

Electrodialysis in integrated processes for food applications Hélène Roux-de Balmann 15.1 Introduction   339 15.2 Principle of electrodialysis   339 15.2.1 Conventional electrodialysis (EDC)    339 15.2.1.1 Membranes and stacks   339 15.2.1.2 Transfer mechanisms and modeling   341 15.2.2 Electrodialysis with bipolar membranes (EDBM)   343 15.2.2.1 Membranes and stacks   343 15.3 Food applications   345 15.3.1 Whey   345 15.3.2 Sugar and beverages industry   346 15.3.3 Wine   348 15.3.4 Organic acids   350 15.4 References   351 Index

353

339

Author index Carla Brazinha Universidade Nova de Lisboa Faculdade de Ciências e Tecnologia (FCT) Department of Chemistry REQUIMTE/CQFB 2829–516 Caparica Portugal [email protected] Chapter 12

Joao G. Crespo Universidade Nova de Lisboa Faculdade de Ciências e Tecnologia (FCT) Department of Chemistry REQUIMTE/CQFB 2829–516 Caparica Portugal [email protected] Chapter 12

Alfredo Cassano Institute on Membrane Technology (ITMCNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapters 3 and 4

F. Petrus Cuperus SolSep BV St Eustatius 65 7333NW Apeldoorn The Netherlands [email protected] Chapter 1

Catherine Charcosset Université de Lyon Laboratoire d’Automatique et de Génie des Procédés UMR CNRS 5007 ESCPE-Lyon 43 Bd du 11 Novembre 1918 69622 Villeurbanne Cedex France [email protected] Chapter 14 Carmela Conidi Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 3

Sirshendu De Indian Institute of Technology Kharagpur Department of Chemical Engineering B-252, IIT Campus Kharagpur 721302 India [email protected] Chapter 9 Enrico Drioli Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 3

xvi 

 Author index

Youssef El Rayess Université de Toulouse (INPT-UPS) Laboratoire de Génie Chimique 1 avenue de l’Agrobiopôle 31326 Castanet-Tolosan France and Centre National de la Recherche Scientifique (CNRS) Laboratoire de Génie Chimique 31432 Toulouse France [email protected] Chapter 7 A.C.M. (Tony) Franken Membrane Application Centre Twente B.V. Gaffelhoek 19 7546MT Enschede The Netherlands [email protected] Chapter 1 Rafiqul Gani Technical University of Denmark Department of Chemical and Biochemical Engineering Computer Aided Process-Product Engineering Center (CAPEC) Soltofts Plads 2800 Kgs. Lyngby Denmark [email protected] Chapter 2 Geneviève Gésan-Guiziou INRA-Agrocampus Ouest UMR1253 Science et Technologie du Lait et de l’Oeuf 65 rue de Saint Brieuc 35000 Rennes [email protected] Chapter 6

Lidietta Giorno Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 13 Bining Jiao Chinese Academy of Agricultural Sciences Citrus Research Institute (CRI-CAAS) Xiema 400712 Beibei, Chongqing PR China [email protected] Chapter 4 Frank Lipnizki Alfa Laval Business Centre Membranes Alfa Laval Nakskov Stavangervej 10 4900 Nakskov Denmark [email protected] Chapter 8 Philip Lutze TU Dortmund University Department of Biochemical and Chemical Engineering Laboratory of Fluid Separations Emil-Figge-Str. 70 44227 Dortmund Germany [email protected] Chapter 2

Author index 

Rosalinda Mazzei Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 13 Martine Mietton-Peuchot Université de Bordeaux Institut des sciences de la vigne et du vin (ISVV) Unité de recherche œnologie, EA 4577 210 chemin de Leysotte 33882 Villenave d’Ornon France and l’Institut national de la recherche agronomique (INRA) Institut des sciences de la vigne et du vin (ISVV) Unité de recherche œnologie, USC 1366 210 chemin de Leysotte 33882 Villenave d’Ornon France martine.mietton-peuchot@u-bordeaux2. fr Chapter 7 Sourav Mondal Indian Institute of Technology Kharagpur Department of Chemical Engineering B-252, IIT Campus Kharagpur 721302 India Chapter 9

 xvii

Martin Mondor Agriculture and Agri-Food Canada Food Research and Development Centre 3600 Casavant Boulevard West Saint-Hyacinthe QC J2S 8E3 Canada [email protected] Chapter 10 Germano Mucchetti University of Parma Food Science Department Parco Area delle Scienze 95/A 43124 Parma Italy [email protected] Chapter 5 Emma Piacentini Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 13 Hélène Roux-de Balmann Université Paul Sabatier Laboratoire de Génie Chimique 31062 Toulouse Cedex 9 France [email protected] Chapter 15

xviii 

 Author index

René Ruby-Figueroa Institute on Membrane Technology (ITM-CNR) c/o Università della Calabria Via P. Bucci, 17/C 87036 Rende CS Italy [email protected] Chapter 8 Irene Tsibranska University of Chemical Technology and Metallurgy Department of Chemical Enineering 8 Kl. Ohridski blvd. 1756 Sofia Bulgaria [email protected] Chapter 11

Bartosz Tylkowski Centre Tecnologico de la Quimica de Catalunya Carrer de Marcelli Domingo s/n Campus Sescelades 43007 Tarragona Spain [email protected] Chapter 11

1 Membrane applications in agro-industry F. Petrus Cuperus and A.C.M. (Tony) Franken 1.1 Introduction Membrane technology is an important part of the engineer’s toolbox. This is especially true for industries that process food and other products with their primary source from nature. However, in many applications these membranes are typically used as end-of-pipe technology, e.g., membranes being used to handle or recycle waste water. In other applications, membranes are used to facilitate production or to improve the products. For example, surface water is purified to grow stainless red tomatoes. Many such applications have been previously described in literature and patent applications. For reference, some of these applications are shown in Table 1.1. This chapter is focused on ongoing development work using membranes. The work is related to agro-business and is driven by the demand for green products, better products and/or natural products. Specifically, this is exemplified by a range of development work on extracting plant compounds for food, cosmetics and wellbeing products. On the other hand, in the more classical “total crop approach”, membranes are thought to have an important role in the future. Very often, the term “biorefinery” is used for all types of cascade that are used for stripping a typical crop-related material. Furthermore, examples of biofuels, especially second and third generation fuels, keep popping up. We will also discuss some research related to vegetable oils and fat processing. These involve new directions for oils such as canola and rapeseed as well as some tropical oils.

1.2 Membranes in biorefinery 1.2.1 What is biorefinery? The American National Renewable Energy Laboratory (NREL) defines a biorefinery as “a facility that integrates biomass conversion processes and equipment Table 1.1: Examples of applications of membrane technology related to agro-food processing Application

Main action/product

Reference

Greenhouses

Softened water provides better products (e.g., stainless tomatoes) RO to concentrate, UF to clarify juice Various Decolorization of red wine Waste water treatment

[1]

Apple juice fabrication Whey processing Winery Potato industry

[2, 3] [4, 5] [6, 7] [8]

2 

 1 Membrane applications in agro-industry

to produce fuels, power and chemicals from biomass. The biorefinery concept is analogous to today’s petroleum refineries that produce multiple fuels and products from petroleum” [9]. Biorefineries stimulated the use of biomass, but because of the oversupply of raw materials in our food chains in the past, there was no strong driver to improve the efficiency in using biomass. However, the increased request in food and non-food applications calls for a change in attitude in the interaction between food and nonfood chains. In this section, some processes will be described that enable both further processing of food products as well as the use of the remaining non-food products (waste).

1.2.2 Mild extraction techniques In many cases, spent biomass in the Netherlands is either burned or composted. However, biomass can also be used for biorefinery because it contains a score of useful components. In biorefinery the biomass is separated into different components that can be used after further processing and separation. To separate the biomass (from plant, field crop, wood, algae, etc.) in different components, the biomass needs to be pretreated in such a way that the functionality of the constituents is not lost. Using destructive processes (such as pyrolysis or thermo-chemical treatment) the constituents are broken down in such a way that they are no longer are fit for highquality applications. Using mild pretreatment processes, the desired components are extracted and remain intact for further processing. Waste streams are minimized in this setup and the yield of the biomass process is maximized. The present state of the art mild extraction techniques for biomass mostly consist of grinding of the biomass to pulp. In this way, the cell structure is ruptured and the cell content is released. This processing method is effective with respect to the release of the desired components from the cells, but it also has the disadvantage that the cellulose-like plant material is reduced and will be present as suspended solids in the solution. These suspended solids cause all sorts of problems in further downstream processing. Another approach to pretreat biomass streams is to consider the extraction and separation processes as one. The pretreatment process consists of a mild extraction step followed by a mild separation technique. In this case, it is of the utmost importance that the extraction and separation steps are geared to one another. From the point of view of the separation techniques, it is important that the extraction techniques are chosen in such a way that the separation can be “simplified”. For example, an extraction technique that minimizes the amount of suspended solids will make the membrane process easier and more effective. In the case of “waste streams” this approach is particularly important.

1.2 Membranes in biorefinery 

 3

Mild extraction techniques that meet the above criterion, amongst others, are: cold aqueous acid extraction, enzymatic extraction, ultrasound, pulsed electric field (PEF) and extraction using CO2. Any technique that does not alter the properties of the desired products can qualify as a mild separation technique. As such, membrane techniques and chromatography are typical mild separation processes. Using this mild extraction and separation process, specialty and/or fine chemicals (e.g., omega-fatty acids, antioxidants, dyes and other bioactive compounds) can be isolated for use in chemistry, pharmacy and the human food industry. The rest stream can be used in application with a lower value, such as animal feed stock and/ or biofuel. Techniques for mild extraction are, among others: 1. Milling and pressing. This is the oldest and most used of the mild extraction techniques. The technique in its most basic form is simple, and consists of chopping up the biomaterials and milling it down to a pulp. In a combined step the material is pressed to gather liquid juice from the plant material. As a result, a concentrated solution is obtained that contains the contents of the plant cells; but cell debris and chlorophyll will be also present in this solution. Although this technique does not fit the criterion that the amount of suspended solids is minimized, the fact that desired materials such as proteins are not denatured means that this process is often referred to as “mild”. In this chapter the technique is listed for comparison reasons. 2. Enzymatic extraction/treatment. Enzymatic treatment of the biomaterial is often used as a pretreatment to aid a further extraction step. In most cases, either an enzyme is added or naturally occurring enzymes are used to weaken the cell structure in order to facilitate the extraction process. This step is often used in combination with milling and pressing. 3. Cold aqueous acid (lactic acid) extraction [10]. In this process a slurry is prepared by dispersing the biomass comprising the naturally occurring microorganisms in an aqueous liquid. In this slurry the conditions have to be chosen in such a way that an aerobic digestion by the microorganisms can take place, in which the naturally occurring microorganisms are capable of converting saccharides into lactic acid. As a result of these conditions, the cells structure is weakened and (part of) the cell content is dissolved into the liquid phase of the slurry. Using this method (part of) the cell content is released into the slurry in a controlled manner, leading to only a clear solution with no or hardly any cell debris. This method is only one of several used for removal of minerals from road-side grass (see section 1.2.4). 4. Ultrasound. Under intense sonication, enzymes or proteins can be released from cells or subcellular organelles as a result of cell disintegration. In order to extract the desired components, the cell membrane must be destructed. Cell disruption is a sensitive process and good control of this is required to avoid an unhindered

4 

5.

6.

 1 Membrane applications in agro-industry

release of all intracellular products, including cell debris and nucleic acids. In addition, product denaturation should be avoided. Ultrasound achieves greater penetration of a solvent into a plant tissue and improves the mass transfer. Ultrasonic waves generating cavitation disrupt cell walls and facilitate the release of matrix components [11, 12]. Pulsed electric field (PEF). PEF is a technology that causes biological cells to be ripped open and perforated. During the process, the biological cells are subjected to an electric field with high field strength, allowing plant and animal cells to be opened up. At higher power settings, microbial inactivation will follow. The high electric field perforates the cell membranes of bacteria and thereby causes their inactivation. By making use of intense but short high-frequency pulses, there is only slight heating of the product itself while the bacterial inactivation effect remains. In order to generate the PEF, both a source and a treatment chamber are required. The treatment chamber consists of at least two electrodes, with an insulating region in between, where the treatment of the product takes place [13]. The PEF process holds promise as a more efficient way of getting useful products out of cell membranes. PEF is particularly well-suited to processing fruit and vegetable juices because the enlargement of the cell pores makes juice extraction easier. PEF may be useful in extracting sugar from sugarbeets and oils from oil bearing plants. PEF may have a use in the developing field of extraction of oil and other products from microorganisms such as algae [14]. Supercritical CO2. Supercritical fluid extraction (SFE) can be used to either remove unwanted material from a product (e.g., decaffeination) or collect a desired product from a solid matrix (e.g., essential oils from herbs). The process relies on the solubility of the extracted compound in supercritical CO2. Process parameters such as pressure and temperature can be altered, allowing for selective extraction. For example, volatile oils can be extracted from plant material at low pressures and both oils and lipids can be extracted using higher pressure [15]. Little is known about the use of supercritical CO2 in extracting valuable components from biomass. Feyecon has carried out successful tests on the extraction of oils from algae, but first cost calculations showed that the process is not viable for large-scale harvesting applications, at least at this moment [16].

1.2.3 Use of membranes in biorefinery The use of membrane technology in downstream processing strongly depends on the way the products are extracted. The traditional way is milling and pressing of biomass, either pretreated or “green”. In both cases, a considerable amount of suspended solids is generated, mainly caused by the cell debris in the solution.

1.2 Membranes in biorefinery 

 5

In this case the membrane technology must be suited to handle a considerable amount of fouling. This means that the used equipment must be able to generate a considerable shear to minimize the fouling and concentration polarization effects.

1.2.3.1 Crossflow Crossflow is one of the oldest methods to avoid membrane fouling. For detailed descriptions of the crossflow principle, the reader is referred to the many textbooks on membrane technology [17–19]. This principle of operation is applied to the difficult filtration of solutions and suspensions, where high shear and good mass transport are necessary to avoid the build-up of particles or macromolecules at the membrane surface. The most common method to realize the high shear is by pumping the feed solution at high speed in relation to a stagnant membrane (module). One of the biggest drawbacks of crossflow membrane filtration, especially when the high shear is realized by pumping the feed around, is the energy consumption. The relation between flow going into the membrane module (thus feed-flow + recirculation flow) and permeate flow can be as high as 50. This means that a large proportion of the energy is not used for filtration, but for moving the feed along the membrane. In contrast, (semi) dead-end operations or reverse osmosis (RO) are operated in a single pass, meaning that all the pumping energy is effectively used for filtration. Several investigations have been conducted to reduce the energy consumption in the filtration process while maintaining a high mass transfer coefficient and to lower the membrane resistance. Some of these methods are as follows: – Module design. Improvements to module design to stimulate the mass transfer involve constructions such as the use of flow diverters or sectioning of the (large) modules in order to get a better flow distribution and avoid channeling. Flow diverters can be used in any type of membrane. – Cross-rotation filtration in which a rotating shaft between the membrane plates is used to create a high crossflow velocity. – The other method of rotation in membrane system is by using rotating discs. – Vibration-enhanced membrane separation, which is the last method to be discussed in this section.

1.2.3.2 Cross-rotation (CR) filtration The principle of CR filtration is shown in Figure 1.1. In a CR filtration system, plates, support layers, membranes and rotors are assembled in a vertical sandwich form. An outer frame and two massive plates at the bottom and the top, along with the plate and rotor stack, comprise a compact unit. A rotating shaft in the middle of the plate stack moves the rotors, creating a velocity > 10 m/s over the membrane surface [20].

6 

 1 Membrane applications in agro-industry

Rotating shaft Membrane Drainage support Filter plate

Concentrate Filtrate Feed Figure 1.1: (A) The principle of cross-rotation filtration and (B) view onto an open plate stack [20]

CR filters are developed for use in “open” ultrafiltration (UF) and microfiltration (MF) applications. Compared to conventional membrane systems, this design allows higher specific filtrate flow rates to be achieved. Pressure drop within the CR filter is minimized because feed and flow over the membranes are independent of one another. Concentration polarization and cake-layer formation on the membrane surface is suppressed by high crossflow conditions. This system is specifically used when a high concentration of the feed is present or required. Typical applications are fermentation broths, pulp bleaching solutions, sludge and polymer solutions [20].

1.2.3.3 Rotating membranes Improvement of the mass transfer can also be achieved by rotation of the membranes. Membrane systems that use a stack of rotating membranes in different configurations have been developed in various configurations. In contrast to the system of CR filtration as described above, here the membrane stack is rotating. Very often, membrane stacks using ceramic disks are used as they provide the necessary stiffness for the membrane stack. In these systems the following module types are used: 1. Single shaft disk filter system (see Figure 1.2 left). In this system, one stack of rotating membranes is used. In most cases a stagnant flow diverter is used inbetween the rotating stack for an improved shear at the membrane surface. Also, the method of supplying the feed to the system influences the shear forces at the membrane surface.

1.2 Membranes in biorefinery 

 7

Figure 1.2: (A) Single shaft disk separator and (B) double shaft disk separator

2.

Double shaft disk filter system (see Figure 1.2, right). In this system two stacks of rotating membranes are used. It is preferable that the two shafts rotate countercurrent to create the maximum shear forces at the membrane surface. This membrane overlapping can increase the permeate flux considerably (the magnitude depends very strongly on the type of feed and the crossflow conditions) [21].

For an overview of dynamic shear-enhanced membrane filtration (a review of rotating disks, rotating membranes and vibrating systems) the reader is referred to the paper by Jaffrin [22]. This paper reviews various systems of dynamic filtration, also called “shear-enhanced filtration”, which consists ofcreating the membrane shear rate necessary to maintain the filtration by a rotating disk, or by rotating or vibrating the membranes. This mode of operation permits very high shear rates, of the order of (1–3) × 105/s and to increase both permeate flux and membrane selectivity [21, 22].

1.2.3.4 Vibrational membranes The traditional method of reducing the effect of fouling in membrane systems is to operate with crossflow of the feed over the membrane. The economical limit to crossflow velocity (mainly caused by limits in module design and energy costs) is given by a shear rate of typically 10,000–15,000/s. As such, the membranes in crossflow operations will still be subject to fouling, because the flow cannot remove solids and particulate retained within the turbulent boundary layer [23]. An alternative method of creating increased shear rates at the membrane surface is to move the membrane itself. The principle of vibratory membrane filtration has been known for more than 20 years. Pall introduced the Pallsep vibrating membrane filter (VMF) that uses an oscillating disc filter stack vibrating at approximately 50 Hz about a vertical axis. With such a system, shear rates in the order of 100,000–150,000/s are generated at the membrane surface (see Figure 1.3 for a comparison of crossflow and vibratory membrane filtration) [23]. The shear developed at the membrane surface is independent of the feedflow rate. This allows independent control of system pressure and shear rate. This

8 

 1 Membrane applications in agro-industry

Crossflow

V SEP

Figure 1.3: Boundary layer resistance in crossflow (left) and V-SEP (right) [22]

operational feature makes a vibratory system well-adapted to handle high viscosity fluids. It also permits operation with high recoveries (high permeate to feed ratio) such as 0.95 vs. less than 0.1 for most crossflow operations [23]. The main use of the Pallsep VMF system is in biotechnological applications (e.g., microfiltration of fermentation broths). An industrial version of the Pall system is developed by New Logic. They introduced the V-SEP (vibratory shear-enhanced processing) [23]. Like the Pall system, V-SEP moves the membrane (leaf) elements in a vibratory motion tangential to the face of the membrane. The feed slurry moves at a low velocity between the parallel membrane leaf elements. The shear waves induced by vibration of the membranes repel solids and foulants from the surface, giving free access for liquid to the membrane pores. A V-SEP system has only two moving parts: the torsion spring (on which the membrane module is mounted) and the bearings. The vibration is induced using a motor with an eccentric weight that is mounted on a metal plate (the seismic mass) supported by a rubber mount. The induced vibration frequency (typically 50 to 60 Hz) is transferred to the membrane module using the torsion spring [23]. The V-SEP resonating drive system is shown in Figure 1.4. The stack of discs is moved at high speed in a torsional oscillation with an amplitude of up to 1.5 inch at 50–60 Hz, thus creating a shear rate of around 150,000/s, which is more than 10 times higher than the maximum shear in crossflow operation [23]. Unlike crossflow filtration, nearly 99% of the total energy utilized is converted to shear at the membrane surface. It should be noted that the magnitude of the flux increase as compared to conventional crossflow strongly depends on the type of membrane process and the type of application. For example, in particle filtration using micro- or ultrafiltration membranes, a flux of more than five times can be achieved. In this case, the increased shear not only lifts the particles but also allows the process to be carried out at increased

1.2 Membranes in biorefinery 

 9

V SEP Resonating drive system

Filter pack drive Torsion spring

Siesmic mass Eccentric weight

Figure 1.4: V-SEP resonating drive system [22]

pressure without the adverse effect of cake-layer formation. The filtration efficiency of typical strong fouling processes like broth filtration is increased by a factor of five to ten. In addition to a better filtration efficiency, another advantage of this process is that it has the ability to concentrate solid content to a much higher end, and also that the slurry can become viscous without blocking the modules. As might be expected, the effect of relative flux increase is less pronounced with processes such as nanofiltration (NF) or RO. In these processes the stagnant boundary layer is less pronounced. However, the other advantages – high concentration and possible high viscosity – remain. If a V-SEP module is compared to spiral-wound elements, it has the advantage that feeds with high viscosity and/or fouling potential can be used without the fear of plugging the feed spacer. Although all the technical equipment shown above is quite impressive, it must be noted that the cost of this equipment and the membrane modules is a manifold of the “standard” capillary or spiral-wound membrane module. Of course, if extraction techniques are used that do not generate large amounts of suspended solids, the process can be simplified considerably. In these cases a “standard” capillary or spiral-wound membrane module can be used. Furthermore, the fluxes and separation properties of the latter systems will be better: not because of the module or system design, but simply because foulants (i.e., suspended solids) are absent. In the next section an application is presented that does not require expensive membrane equipment, because a mild extraction technique is used.

10 

 1 Membrane applications in agro-industry

Table 1.2: Relative costs of different membrane processes Process

Equipment costs

Operational costs

Typical use

Dead-end

Low

Low

Crossflow Cross-rotation Rotating membranes V-SEP

Medium High Very high High

Very high Medium Low Low

Only low fouling aqueous streams Fouling streams High fouling streams High fouling streams High fouling and viscosity streams

1.2.4 Removing minerals from road-side grass Biomass of plant origin may be combusted, directly yielding energy in the form of heat, or it may be converted into convenient energy carriers, for example combustible liquids such as hydrocarbons or alcohols, and combustible gases such as methane. The handling and conversion of biomass of plant origin, however, is difficult because of its physical characteristics, in particular morphology, and because the biomass comprises components that disturb or are harmful in combustion or conversion processes. Undesirable components of biomass of plant origin comprise of carbohydrates, chlorides, alkali metal and alkaline earth metal salts, calcium and magnesium salts, ammonium salts, proteins, ash precursors and water. Unpleasant odors may affect the environment, in particular when drying or when combusting the biomass. Many attempts have been made to treat biomass in order to bring it into a form suitable for the production of energy or energy carriers. Such treatments included the following steps: – Grinding, cutting, milling, or other mechanical treatment aiming at particle size reduction. – Extrusion, pressing or heat treatment aiming at destruction or opening of the biomass cell structure, or removal of water. – Extensive heating or roasting, aiming at melting or pyrolysis, yielding fuels, such as charcoal, tar or gas. – Treatment with strong acid or strong base, or oxidising agents, aimed at conversion of the biomass lignocellulosic components and making them digestible by enzymes. All these methods result in a slurry that on one hand has not reduced the contents of the undesirable components to a level required for combusting the biomass, and on the other hand creates a liquid phase with a large amount of suspended solids.

1.2 Membranes in biorefinery 

 11

The presence of the suspended solids would then require an expensive separation technique, making the process not economically viable. Danvos has described an alternative process for the conversion of biomass into a biomass product that is suitable for use as a fuel [10]. The biomass is of plant origin and comprises microorganisms naturally occurring in the biomass. The process comprises of the following steps: (i) preparing a slurry by dispersing the biomass, including the naturally occurring microorganisms in an aqueous liquid; (ii) maintaining the slurry at conditions suitable for aerobic digestion by the microorganisms to obtain a slurry comprising the biomass product as a dispersed solid phase; and (iii) recovering the biomass product. The recovering process comprises of washing and drying the biomass product. The recovered biomass product can be used in a combustion process after pressing and drying [10]. The resulting liquid phase does not contain any suspended solids. In this way, water can be recycled using the same spiral-wound membranes for nanofiltration and RO as used in water treatment. The concentrate of both steps can be used in bioconversion processes (concentrate of NF) and as inorganic fertilizer (concentrate of reverse RO).

1.2.5 Biofuel including microalgae In the last 10–15  years there has been an increasing interest in the production of chemicals and fuels from renewable resources [24]. Reasons for this trend include growing concerns about global warming and climatic change, volatility of oil supply, increasing instability of crude oil price and existing legislations restricting the use of nonrenewable energy sources. As mentioned previously, several scenarios have been proposed for using new and existing agricultural crops and activities into valid products, but nowadays they are merely included in biorefinery concepts [25–27]. The “first generation biofuels” mainly involve production of carburants from sugar and oil sources – i.e., ethanol and biodiesel – that also could be used for food purposes. To avoid this unhealthy competition, much effort has been put into the “second generation biofuels” that are based on indigestible parts of the food chain or on typical sources that are considered to be waste. These include large parts of sugarbeet and waste from agricultural crops, but also grass, straw etc. Via a combination of processes, liquid or gaseous biocarburant is generated: typically bioalcohol and biogas. “Third generation biofuels”, e.g., from oil algae, are made via alternative “agricultural production schemes” that do not conflict with regular food-oriented agricultural activities. The exploitation of these sources is mainly lab-scale and larger pilot-scale experiments are oriented at learning probable production schemes. It is expected that algae can produce oil and some bioalcohol. There are many types of algae studied and they typically produce a range of products. This is biorefinery at its best, and at this

12 

 1 Membrane applications in agro-industry

moment it is difficult to say which method will come up with winning combinations [28, 29]. However, first generation biofuels are hardly biorefinery products, as the sources are specially grown for making fuels and waste is considered to be just waste. In second generation biofuels, production is aimed towards zero-waste, at least during the generation of fuels, because waste still could be an unwanted by-product. In spite of this, for the main liquid biofuels (bioalcohols) and either first or second generation biofuels, production processes are fairly analogous and basic steps include hydrolysis, fermentation and alcohol refining. Therefore, technical and economic bottlenecks in first and second generation biofuel processing lay in their upstream and downstream processing, with some small differences. One of the main bottlenecks of bioalcohol production is the refinery to almost pure alcohols that can be used as fuel or as fuel additive. The main three biofuels based on alcohols: butanol, ethanol or acetone-butanol-ethanol (ABE) are all produced at low percentages in fermentation (typically < 5 wt%) and have to be rectified. These processes were devised in the 1900–1910s, as excellently reviewed by Garcia [25], but their development has been discontinued by the use of petrochemical oil. Membranes are envisioned to be used in the low-alcohol end (5–40%) by organophilic pervaporation and in overcoming the azeotrope at the high alcohol percentage. The latter can be achieved by either pervaporation or vapor permeation. Overcoming the azeotrope by vapor permeation is now well-practized in many first generation fuel processes in the US that use corn as carbohydrate source. Vapor permeation (VP) is preferred over pervaporation as it integrates more easily with distillation and the fouling problem is hardly an issue. The latter is easily understandable as most severe membrane foulants are non-volatiles and they are not present in a vapor. Usually in these processes the VP unit is integrated with the distillation column that performs the trajectory from the low-end alcohol directly from the fermentor. By clever engineering, the energy costs of the operation are significantly reduced compared to classical processes. Interestingly, typical membranes used in this area can be polymeric (PVAPAN membranes) as well as ceramic membranes (zeolite membranes) [25]. In the downstream part coupled to the fermentation, a lot of research is still being undertaken. This is partly because extraction of the alcohols at low levels keeps the bioproduction at a good level. Also, on many agricultural sites there is no rest-heat to accommodate the first parts of the distillation easily. Moreover, one can understand that if a type of (semi-) continuous fermentation is envisioned, alcohol recovery at low temperature is desired. Using typical organophilic membranes, e.g., poly-di-metyl-siloxane (PDMS) membranes, the principles of such setups have been shown. However, there remain problems that are frequently caused by too-low selectivity or long-term fouling. Typically, organophilic PDMS yield a selectivity (α) of 7 for ethanol and somewhat higher for butanol. Many researchers have screened materials for a more selective membrane. Interesting selectivities have been found using zeolite-filled PDMS (α EtOH~40), PVTMSP (α EtOH~26) and PEBA (α EtOH~25).

1.2 Membranes in biorefinery 

 13

However, in the long-term selectivities tend to lower towards uneconomic values. This is mostly attributed to minor components (butyric acid as one prominent) from the fermentation broth, which are assumed to cause fouling. Using an extra filtration step (NF) to prevent fouling results in too high costs. Using a cleaner substrate helps, but as components as butyric acid are generated in fermentation their presence seems unavoidable [27]. From the third generation fuel production methods the use of microalgae seems the most prominent. In principle, one may regard microalgae systems as sunlight harvesting devices. Using nutrients and sunlight, algae can be pushed to produce carbohydrates or oils but also a number of other compounds that are thought to be useful in cosmetics or healthcare. These include special proteins, PUFAs carbohydrate building blocks [28, 30]. The outcome of a certain process depends on the algae system that is cultivated and the environment during the production. The production of algae oil is often regarded as third generation biodiesel. In its simplest process form it only requires drying and pressing to make biofuel (however, drying can already be costly if not done in sunlight only). Today it is questionable whether microalgae can be economically viable with biofuel income as the only return on investment, hence much research is devoted to the reclaim of more precious components. Worldwide, the interest in microalgae is enormous. In the USA alone more than 500 MUS$ has been raised by different companies to work on algae processing, tackle engineering problems and launch new products. Microalgae were originally grown in water basins but today a large array of vertically hanging polymer bags is considered to be the most effective method. During photosynthesis, the algae absorb CO2 and nutrients and generate carbohydrates. Fully-grown microalgae have to be separated from water and further processed. MF and UF systems [31] were considered to have a role here, but the energy requirements are a major concern. Currently, suction mode UF or MF seems to be the most promising technique [26]. After harvesting the algae, they have to be opened up and extracted to yield their interesting products. Here, membrane filtration is also considered for different routes. In particular, in upstream processes membranes are envisioned, but as these can hardly be seen as real production systems we do not go into detail in this book. Some of these also produce little amounts of ethanol that are also thought to be recovered. To this end, again pervaporation is explored. Of course, in an ideal scheme the microalgae would excrete their beneficial products. This requires microalgae and cultivation routes other than those fostered so far. In an alternative method the mild extraction technologies presented in section 2.2 could help to harvest the oil from the microalgae. For example, OriginOil’s algae single step oil extraction process harvests, concentrates and extracts oil from algae, and separates oil, water and biomass in one step. The process does not use chemicals or heavy machinery and no initial dewatering is required; and it separates the oil, water and biomass in less than 1 hour. The company’s Quantum Fracturing technology combines with electromagnetic pulses and pH modification to break down cell

14 

 1 Membrane applications in agro-industry

Figure 1.5: Pilot plant for recovery of ABE from fermentation using pervaporation (courtesy of Pervatech.com). The same type units may be used for bioethanol and aroma recovery

walls and release oil from the algae cells [32]. Although no details are disclosed, it looks like acid conditions are used to weaken the cell structure, and electromagnetic pulses (PEF) are used to perforate the cell walls. The oil is then released into the aqueous system and is separated by gravity. In particular, the fact that no suspended solids are present improves the separation efficiency. Again it proves that mild extraction is beneficiary in further downstream processing.

1.3 Membranes in vegetable oils and fats 1.3.1  Membrane technology applied to vegetable oils Research in the use of membrane technology for oils and fats has been mostly directed at solvent recovery, processing of the miscellae, degumming, bleaching deacidification, hydrolysis of triglycerides and esterification to obtained structured lipids. The most recent focus is on processing of minority components that could have addedvalue in specialty food, wellbeing products or pharmaceutical-oriented products [24]. Historically, a lot of developmental work has been dedicated to degumming, for various reasons. The first work was dedicated to replace water degumming and partial deacidification [33]. One of the main ideas behind this was that fewer chemicals would be needed and it would yield a product quality improvement and less waste. The process originates from the 1980s and from then on many variations have been described. However, even though considerable scaling-up has been done, in one

1.3 Membranes in vegetable oils and fats 

 15

way or another these efforts never materialized into economic viable process alternatives. One of the main reasons for that is a lack of fouling control as well as the overall costs of the process [34]. Moreover, in a number of production processes, e.g, hexane recycling, considerable lower energy costs were forecasted for the use on membranes. However, as the heat balances of large-scale processes in oil and fat extraction are very much coupled, such energy costs are already low and this hardly improves with the advent of membranes. In addition, process economy of the traditional processes has been improved, and other alternatives have been developed. Hence, less driving force remained for further developments [35–41]. The area where membrane degumming remains interesting is the niche application of fabrication of high-quality lecithin. Using a membrane process reduces heat input – although some hexane still has to be evaporated – and less heat yields a better product. In principle, such a process can be used for various seed oils like rapeseed, sunflower and canola. As the oil and phospholipid concentration may be quite high in such applications, concentration polarization and fouling are important factors. Moreover, for oils like sunflower the plugging of pores by waxes can considerably attribute to flux decline [42, 43]. Several procedures have been tried to counteract such phenomena, but literature hardy mentions the solutions that may have been found. Many concepts of seed oil refining using membrane technology are inspired by the work of Unilever in the 1980s [33]. Many variants of this concept have been published [44, 45]. The setup shown in Figure 1.6 illustrates degumming using water and UF, thus largely avoiding chemicals. The gums (phospholid) concentrate (B) now will also include free fatty acids and some non-trigycerides. In another embodiment, Raw miscella-hexane mix from extraction unit

(A) Washing water

Hexane

(C) Gums hexane

UF

Distillation / stripping

NF Oil-hexane mixture Distillation / stripping Degummed oil

Lecithin production drying/spray drying (C) Lecithin

Figure 1.6: Degumming using membrane technology. Many variants have been proposed, and a major aim is to operate with the minimum of chemicals

Hexane

16 

 1 Membrane applications in agro-industry

water is not added to (A), and (B) largely consists of phospholipids and hexane that is purer than in the first option. The latter setup is mainly aimed at production of “natural lecithin”, hence without the interference of chemicals. Such a process is mainly focused on high-quality lecithin and is therefore destined to be at a relatively small scale. The NF step pictured in Figure 1.6 for the recovery of hexane after extraction is optional. For the current large-scale seed oil extraction processes, it is forecast that NF is not economically viable. From the first processing steps (seed pretreatment and mechanical pre-pressing) so much heat can be recovered that the heat for hexane evaporation is easily provided for. From the work done in the advent of this process (in the 1980s) it is suggested that the type of membrane was not very critical for the final result. As long as it was “tight UF” it worked. However, work after that showed that long-term stability could be an important factor. In addition, tightness of the membrane sometimes seems to have an influence on the final product that is generated. Apart from the fact that the membrane should retain phospholipids, a tighter membrane may also (partially) reject other components such as phytosterols, and /or natural antioxidants like tocopherols [34, 37]. These components have a beneficiary influence on the lecithin produced. Such effects could also be obtained by securing low but not zero water levels in the feed of the membrane. Logically, this is related to the “colloidal” behavior of the feed stream. Water promotes miscellea formation and these in turn can capture several components in their vicinity. It seems also clear that water content has an influence on fouling and concentration polarization in the membrane process, so care has to be taken.

1.3.2 Solvent recovery and reuse The recovery of extracting solvent in vegetable oil has been studied by many researchers. One of the main ideas was that by using membrane, lots of energy could be saved. At first glance, in large-scale application this seems to be the case. However, for many seed oils the desolventizing action is coupled to “toasting” (a process to inactivate enzymes and facilitate oil exit from the seeds). This process inevitably yields rest-heat that can easily be used in the process for solvent recovery. Especially in the case of hexane, this process on a large scale is almost unbeatable as energy is recovered very efficiently. During the 1990s the use of hexane in oil extraction was discussed for health and environmental reasons. Later on the effect on health was proven irrelevant because it is hardly detectable in the refined oil. Nevertheless, these discussions triggered research in to other solvents, such as ethanol, IPA and heptane. Such solvents would require alternative energy housekeeping in large extraction plants and membranes were considered to be a viable tool. Thus, know-how was generated to use other solvents and applied in niche applications in the oil and fats industry. Another solvent used in oil and fats is acetone. In wet fractionation acetone is of particular importance. It has been shown that membranes are easily capable of

1.3 Membranes in vegetable oils and fats 

 17

separation of the acetone and triglycerides. A high rejection of oil is possible and removal of acetone works until a relatively high concentration, not withstanding some osmotic pressure that might occur. The last step still remains a thermal treatment to make the oil acetone free.

1.3.3 Wax removal and/or recovery Wax removal or dewaxing is an issue in many vegetable oils, but especially in sunflower and olive oils [43]. The dewaxing process for oils, vegetable as well as diesel oil, is also known as “winterisation”. Traditionally, the process involves cooling of oil until it is 0–10°C and then settling the oil. Moreover, the onset of crystal formation may also take a long time, and the cleaner the oil the longer it takes. As the viscosity of the oils is fairly high, settling takes a long time. Of course, this can be speeded up by using centrifuges, especially when de-saponification is combined with this process. The use of straightforward filtration has been used for a long time and was followed by using MF and UF in the 1980s. By using membrane filters, the settling process is merely superfluous. These involve filtration of pure oil, hence viscous media, and MF and UF can handle that reasonably well. The use of NF may also be used for wax recovery from low molecular oil. A typical example of this can be found in the recovery of wax from citrus waste [46]. A same process may be used for the winning of waxes from vegetables (and waste). In such cases, (dry) peels or leaves are typically extracted with an organic solvent like ethanol or hexane. A typical process that involves a membrane to recycle solvent is shown in Figure 1.7. These solvents extract oils, including the wax that is either dissolved in the oil or is present on leaves or peel. The typical wax has a C40–C60 backbone and has a small molar mass. Because the amount of wax can be very low ( < 2 wt% peel), there

Solvent recycling

Filter/ centrifuge

Wax Evaporation solidification

Coarse debris

to re-use

NF “polymerics” + solvent

Product: wax

Figure 1.7: Removing wax and recycling solvent from citrus peel oil extraction

18 

 1 Membrane applications in agro-industry

is an urge to reduce the amount of extractant and waste as well as the energy needed for the recovery of the solvent. Here the wax and solvent plus a whole range of different components (cellulosic fragments, some oil, etc.) is extracted and coarsely separated by a centrifuge, a filter or both. Typically waxes are fed to an NF filter together with solvent and some small fragments. This may be partly recycled to the extractor for reuse. The dissolved carbohydrates are largely separated from the solvent and wax that permeate the NF membrane. In the final strip, the solvent is largely evaporated and the wax is separated and solidifies after cooling.

1.3.4 Goodies in oil A number of studies are devoted to upgrading minority compounds of oil. The typical compounds depend on the oil but include tocopherol, carotenoids, phytosterols and many others. The beneficial effects of these “goodies” are often not medically proven and sometimes are only suggested on the basis of history or alternative “pharma belief ”. Omega fatty acids – one of the components of, e.g., fish oil that has proven health effects – originates mainly from fish and related marine source. It is currently not processed via membrane technology and thus it is not discussed here. Most of the components regarded as “goodies” presumably have an antioxidant role in the plant or oil. There has been a time that many of these components were just removed from the oils as they cause yellow color or turbidity. In the traditional chemical refining the components were just washed out as soap or adsorbed to bleach earth. Thus, in general, to acquire a concentrated stream Table 1.3: Minority compounds in oils and their benefits Compound

In oil/plants

“Chemical” action

Presumed health benefit

Tocopherols

Many, especially soy, peanut, rape Many, esp. soy, peanut, rape Many, pumpkin, mustard, rape, palm, fruit, vegetable Rape seed, pine tree, nuts Rice bran All vegetable oils

Antioxidant

Vitamin E related

Antioxidant

Vitamin E related

Antioxidant

Vitamin A related

Antioxidant

Cholesterol control

Antioxidant Cell-wall Antioxidant

Vitamin related Vitamin B/cholesterol control “Anti-cancer”

“Anti-ageing”

Brain

Tocotrienols Carotenoids

Phytosterols Oryzanol Phospholips Polyphenols Omega FFA

Many oils and plants, wine, olive Fish oil

1.3 Membranes in vegetable oils and fats 

 19

of “goodies” one must work with unrefined oils. Quite a body of research agrees that concentrated potions can be made by straightforward NF of oil in a solvent. There are several membranes that have retention for the desired compounds. Interestingly, in the most common solvent for seed oil processing (hexane), retentions are not very high. Using specific solvents such as ethanol or acetone makes manipulation easier, as does the combination with treatments that convert species in more soluble blocks. This is especially true for compounds bonded to cell membranes. In a number of cases pure oils have also been subjected to treatment with membranes. In a number of cases negative retention (−30 to −50%) of tocopherols have been found. However, in these systems permeability of the system is quite low, and even lowered by the negative retention – as the driving force decreases rapidly. The negative retentions are also found when hexane is used as a solvent. Permeabilities are somewhat higher, but would require considerable optimization before becoming a practical reality. Currently there is a revival of interest in the research of recovery processes for polyphenols from olive oil waste but also from grape waste, tea leaves and all kinds of other plants [47–50]. Although polyphenols are a source of trouble in waste water

OH

HO

HO

OH

(a)

OH

(b)

H OH H O HO HO

O

O H H

OH H

O

O O

OH OH

O

(c)

Figure 1.8: Some polyphenols present in olive oils. (A) tyrosol, (B) hydrotyrosol, (C) Oleuropein. Oleuropein are typical tyrosol esters with (cell-wall) carbohydrates. Hence, these compounds are not only present in the olive but also in the leaves. Caffeic acid and verbascoside are other polyphenols. In red wine other polyphenols are present but their structure is very similar

Table 1.4: Specific solvents for recovering “goodies” Target

From

Solvent

Increase in yield (%)

Cartenoids Polyphenols Tocopherols

Palm, algae Olive oil Palm oil

Acetone, ethanol Decolourisation deacidifying [36] Ethanol Removal of oxidised polyphenols Conversion to methyl esters 10 times increase in yield [40]

20 

 1 Membrane applications in agro-industry

Figure 1.9: Membrane unit (in installation) for pilot recovery of “goodies” from extractant solvent. The tank at the top is used for rinsing the membrane with permeate (courtesy of Solsep.com)

treatment, they are considered to have value for nutrition as well as for raw materials for organic synthesis. The first efforts to explore polyphenol waste sources were in the 1990s. A problem is that polyphenols tend to oxidize and polymerize in undefined circumstances. In that state they lose their extractability and their supposed beneficial properties. Polymerized phenols are especially difficult to handle and to break down in a wastewater system. Numerous trials have been and are devoted to extract active polyphenols like e.g., tyrosol (Mw ~138 Da), oleuropein (Mw ~540 Da) and many others. The challenge is to separate the interesting low molecular phenols from the worthless others. As such, large molecules and sediments are removed by physical techniques including centrifugation, sedimentation flocculation and/or micro- and ultrafiltration. The remaining solution is then fractionated by typical low-ultrafiltration and nanofiltration, which resulst in useful solutions of active components [51]. Nevertheless this is a rather complex way to acquire these components. Therefore other chemical and enzymatic treatments on typical olive oil residuals are being researched to release interesting polyphenols that are a little more defined. Subsequently the polyphenols can be extracted and recovered from the liquid.

1.4 Application scale and outlook 1.4.1 Application scale As far as the application scale is concerned, a difference must be made between “membranes in biorefinery” and “membranes in vegetable oil and fat”. For the latter,

1.5 References 

 21

the application scale will be more traditionally organized, which means larger installations at an industrial site for “vegetable oil refinery”. Also the fact the installations must be equipped for non-aqueous solutions (thus water-proof), and the fact that the installations will require professional personnel “24–7”, means that most likely only installations that are integrated into a larger separation processes will be the standard. For the topic “membranes in biorefinery” one tends to look more at the “economy of chain” instead of the more traditional “economy of scale” as described above. In western industry there is a strong tendency towards modularization of food and chemical processing plants. However, this is not the most ideal situation for processing biomass. In fact it makes more sense to process the biomass at the place where it is generated, using fast and flexible processing units that can rapidly be deployed and moved across geographical regions depending on local customer needs. “Factories in containers” inherently have smaller throughputs than traditional large-scale facilities (batch or continuous), but they do have the advantage of smaller transport costs. This setup can also deliver a competitive edge for specialty products in emerging economies, in situations where time-to-market is critical, and/or in a situation where the biomass needs to be processed quickly after harvesting.

1.4.2 Outlook The future for the use of membranes in “biomass refinery” is bright, particularly if the practice of developing installations at the site where the biomass is generated becomes more common. The concept of “factories in containers” can easily be applied to separation using membrane systems. Furthermore, if these separation installations are sensibly coupled (integrated) with a mild extraction process, this type of setup will be seen more often in the future. As far as the use of membranes in “vegetable oil and fat”, much will depend on the way the total refinery process is organized. As mentioned above, it is difficult to replace a part of an existing installation especially if this installation is (heat) integrated with the rest of the plant.

1.5 References 1. Schippers JC. Desalination by reverse osmosis in horticulture. III International Symposium on Water supply, 1981. Available at: actahort.org. 2. Cuperus FP, Bouwer S Th, Boswinkel G, van Gemert RW, Veldsink JW. The upscaling of an enzymatic reactor for the production of apple juice. In: R. Bredesen (editor) Proc. ESF Meeting on Catalytic Membranes. May 30–June 1, 1997 Oslo.

22 

 1 Membrane applications in agro-industry

3. Echavarría AP, Torras C, Pagán J, Ibarz A. Fruit juice processing and membrane technology application. Food Eng Rev 2001;3:136–158. 4. Bird J. The application of membrane systems in the dairy industry. J Soc Dairy Tech 1996;49: 16–23. 5. Bargeman G, Timmer M, van der Horst C. Nanofiltration in the food industry. In: Nanofiltration: Principles and Applications. Schäfer AI, Fane AG, Waite TD (eds). Elsevier Advanced Technology, Oxford; 2005, 305–328. 6. El Rayess Y, Albasi C, Bacchin P, Taillandier P, Raynal J, Mietton-Peuchot M, Devatine A. Crossflow microfiltration applied to oenology: A review. J Memb Sci 2011;382:1–19. 7. Catarino M, Mendes A. Dealcoholizing wine by membrane separation processes. Innov Food Sci Emerg Technol 2011;12:330–337. 8. Rausch KD. Front end to backpipe: membrane technology in the starch processing industry. Starch/Stärke 2002;54:273–284. 9. National Renewable Energy Laboratory (NREL). What is a Biorefinery? http://www.nrel.gov/ biomass/biorefinery.html (Accessed September 2013). 10. Vos DJ, Rustenburg S. A process for the conversion of biomass of plant origin, and a combustion process. Patent application WO 2012023848 A1. 11. Hielscher GmbH. Ultrasound Technology – Ultrasonic Extraction and Preservation. http://www. hielscher.com/ultrasonics/extraction_01.htm (Accessed September 2013). 12. Vinatoru M, Toma M, Radu O, Filip PI, Lazurca D, Mason TJ. The use of ultrasound for the extraction of bioactive principles from plant materials. Ultrason Sonochem 1997;4:135–139. 13. TOP BV. Pulsed Electric Field (description); http://en.topwiki.nl/index.php/Pulsed_Electric_ Field_(PEF) (Accessed September 2013). 14. Pulsed Electric Field Processing; http://en.wikipedia.org/wiki/User:Openman/Pulsed_Electric_ Field_Processing (Accessed September 2013). 15. Supercritical fluid extraction; http://en.wikipedia.org/wiki/Supercritical_fluid_extraction (Accessed September 2013). 16. Lee S, Shah YT. Supercritical fluid extraction of algae oil. In: Biofuels and Bioenergy: Processes and Technologies. CRC Press;2012,37–39. 17. Scott K. Handbook of Industrial Membranes; Elsevier Advanced Technology: Oxford, UK. 18. Mulder M. Basic principles of membrane technology; Kluwer Academic Publishers: Dordrecht. 19. Cross-flow filtration; http://en.wikipedia.org/wiki/Cross-flow_filtration (Accessed September 2013). 20. Driving Forces in Membrane Processes; http://mempro.net/basics/drivingforces.html (Accessed September 2013). 21. Taamneh Y, Ripperger S. Performance of Single and Double Shaft Disk Separators. Physical Separation in Science and Engineering, 2008. Article ID 508617, 5 pages. Hindawi Publishing Corporation. Available from: downloads.hindawi.com/archive/2008/508617.pdf. 22. Jaffrin MY. Dynamic shear-enhanced membrane filtration: A review of rotating disks, rotating membranes and vibrating systems. J Memb Sci 2008;324:7–25. 23. V-SEP, the new way to separate; http://www.vsep.com/pdf/VSEP_Brochure.pdf (Accessed September 2013). 24. Cuperus FP, Ebert K. Nanofiltration in organic media. In: Schäfer A, Fane AG, Waite, TD, eds. Nanofiltration – Principles and Applications. Elsevier: Oxford, UK; 2005, 521. 25. Garcia V, Päkkilä J, Ojamo H, Muurinen E, Kreiski R. Challenges in biobutanol production: how to improve efficiency. Renew Sust Energ Rev 2011;15:964–980. 26. Bilad MR, Vandamme D, Foubert I, Muylaert K, Vankelecom IFJ. Evaluation of submerged microfiltration process to harvest freshwater and marine microalgae. https://lirias.kuleuven.be/ bitstream/123456789/340930/1/Bilad.

1.5 References 

 23

27. Vane LM. A review of pervaporation for product recovery from biomass fermentation process. J Chem Technol Biotechnol 2005;80:603–629. 28. Klok AJ, Martens DE, Wijffels RH, Lamers PP. Simultaneous growth and neutral lipid accumulation in microalgae. Bioresour Technol 2013;134:233–243. 29. Molina Grima E, Belarbia E-H, Acien Fernandeza FG, Robles Medina A, Chisti Y. Recovery of microalgal biomass and metabolites: process options and economics. Biotechnol Adv 2003;20:491–515. 30. Wijffels R. Biorefinery of microalgae. 9th European Congress of Chemical Engineering ECCE9/ ECAB 2. The Hague, April 22–25, 2013. 31. Rossignol N, Vandanjon L, Jaouen P, Quemeneur F. Membrane technology for the continuous separation microalgae: culture medium: compared performances of cross-flow microfiltration and ultrafiltration. Aquacult Eng 1999;20:191–208. 32. Algae oil extraction. http://www.oilgae.com/algae/oil/extract/extract.html (Accessed September 2013). 33. Gupta S. Refining of triglyceride oils. US Patent. 4533501, 1985. 34. Segers JC, den Bieman HACI. Cleaning Method for Membranes. Unilever. European Patent Application EP 1997 702024998, Published, 2001. 35. Degumming of Edible Oil and Fat. Nishin Oil Mills Ltd. Japanese Patent JP 2001 01 7080. Filed January 23, 2001. 36. Lamonica DA. Method for Refining Oil. Rochem Separation Systems. US Patent 5543329. Filed August 13, 1996. 37. Manjula S, Subramanian R. Membrane technology in degumming, dewaxing, deacidifying and decolorizing edible oils. Crit Rev Food Sci Nutr 2006;46:569–592. 38. Jirjis B, Muralidhara HS, Otten, DD (Cargill, Inc). Method for removing phospholipids from vegetable oil miscella, method for conditioning a polymeric microfiltration membrane, and membrane. US Patent 2001;6:207–209. 39. Jirjis B, Muralidhara HS, Otten DD (Cargill, Inc). Method and Apparatus for Processing Vegetable Oil Miscella, Method for Conditioning a Polymeric Microfiltration Membrane, Membrane, and Lecithin Product. US Patent 2004:6833149. 40. Koris A, Vatai G. Dry degumming of vegetable oils by membrane filtration. Desalination 2002;148:149–153. 41. Arora S, Manjula S, Gopala Krishna AG, Subramanian R. Membrane processing of crude palm oil. Desalination 2006;191:454–466. 42. Raman LP, Cheryan M, Rajagopalan N. Deacidification of soybean oil by membrane technology. J Am Oil Chem Soc 1996;73:219–224. 43. Zwijnenberg H, Peinemann K-V, Ebert K, Cuperus FP. Acetone-stable nanofiltration membranes in deacidifying vegetable oil. J Am Oil Chem Soc 1999;76:83–87. 44. Koseoglu SS. Membrane applications and research in the edible oil industry: an assessment. J Am Oil Chem Soc 1990;67:239–249. 45. de Morais Coutihno C, Chih Chiu M, Correa Basso R, Badan Ribeiro AP, Guaraldo LA, Goncalves, Viotto LA. State of art of the applications of membrane technology to vegetable oils. Food Res Int 2006;42:536. 46. Cuperus FP. Membrane used for processing of organic solvents. Presentation at Euromembrane 2012. http://nym14.ce.ic.ac.uk/sites/default/files/Petrus Cuperus NYM14.pdf (Accessed September 2013). 47. Takac S, Karakaya A. Recovery of phenolic antioxidants from olive mill wastewater. Recent Patents Chem Eng 2009;2:230–237. 48. Sarmento LAV, Machado RAF, Petrus JCC, Tamanini TR, Bolzan A. Extraction of polyphenols from cocoa seeds and concentration through polymeric membranes. J Supercrit Fluids 2008;45:64–69.

24 

 1 Membrane applications in agro-industry

49. Liguori L, Russo P, Albanese D, Di Matteo M. Evolution of quality parameters during red wine dealcoholization by osmotic distillation. Food Chem 2013;140:68–75. 50. Jiang H, Zhang J, Jiang Y, Chen P. Environment-friendly preparation method for high-ester catechin tea polyphenol. CN102240343 (A); 2011. 51. Cassano A, Conidi C, Giorno L, Drioli E. Fractionation of olive mill wastewaters by membrane separation techniques. J Hazard Mater 2013;248–249:185–193.

2 Process intensification in integrated membrane processes Philip Lutze and Rafiqul Gani 2.1 Introduction The chemical and biochemical industries produce products that are essential for society. Nowadays, the chemical and bio-based processing industries are facing challenges in order to establish a sustainable production. All of this has to happen under the uncertainty of profit margins because of quicker changing markets, stronger global competition as well as the lower expected lifetime of a product, but at the same time under very high product quality standards [1, 2]. Shifting to more sustainable productions means that existing processes need to increase the efficiency of the raw materials, solvents and energy in the system as well as decreasing the amount of waste. It also means that new raw materials and new catalysts (e.g., biocatalysts) are introduced, which further drives the necessity to totally new production routes and processes. Hence, it is inevitable that the processes which exist and the processes which will be developed to match/face the future challenges require improved designs going beyond those achieved by using the toolbox of conventional process units [3]. Hence, it is believed that one important and even necessary tool, to match these future challenges is process intensification [3]. Food processing technology has always been challenging because of high-quality standards of the product and the process, in some cases a very low life span of raw materials and products, very flexible raw materials in case of agro-based raw materials as well as highly diluted systems, for example, in milk processing. Food production is based on a large use of resources. In a study of the food industry in UK in 2010, it was shown that the food and drink industry takes 11.5% of the total UK energy use and 11% of the total electricity demand [4] and generated during the production around one-third of the total amount of food waste [4].

2.1.1 Background: process intensification A common definition of process intensification (PI) does not exist yet. For some, PI is the reduction of capital costs and volumes [5–7] or the drastic increase in efficiencies [3]; while for others the keypoint is the integrated approach for process and product innovations [8]. In 2009, Van Gerven and Stankiewicz [9] defined four explicit goals of PI, which are achieved within four domains. The goals are: (i) maximize the effectiveness of

26 

 2 Process intensification in integrated membrane processes

intra- and intermolecular events; (ii) optimize the driving forces at every scale and maximize the specific surface area to which these forces apply; (ii) maximize synergistic effects, and, (iv) give each molecule the same processing experience. The four domains are: (i) structure, (ii) energy, (iii) synergy and (iv) time. A more fundamental definition of PI can be derived by concluding that the goals of PI are actually achieved by enhancements of the involved phenomena inside those four domains by: – PI principle A – integration of unit operations; – PI principle B – integration of functions; – PI principle C – integration of phenomena; – PI principle D – targeted enhancements of phenomena in a given operation [10]. PI equipment illustrating this definition is described in the following.

2.1.2 Membranes and process intensification Membranes are widely established in the processing of food products [11]. This is due to some of the key advantages of membranes as compared to conventional separation technologies such as distillation [11]: (i) operation at low-to-moderate temperatures and pressures to circumvent degradation of products; (ii) high selectivity; (iii) easy installation and scale-up (number up). Therefore, membrane-based approaches can provide an essential contribution to the concept of PI, which may lead to even larger performance benefits. In general, several PI membrane operations exist in which the membrane has four functions: (i) mainly as a separator; (ii) as a creator of high surface areas for contacting two phases with each other; (iii) for immobilization of the catalysts; and (iv) as a distributor and controlled provider of reactants into a mixture. An example of such membrane-based PI separators (function i) is an electromembrane filtration, in which the mass transport phenomena are enhanced by adding electrophoretic transport phenomena (PI principle C); section 2.1.1 or a hybrid separation coupling distillation and a hydrophilic pervaporation [12] for the separation of ethanol from water overcoming an existing azeotrope in the system (PI principle A). Membrane distillation [13] or membrane crystallisation [14] are equipment in which the membrane does not primarily act as a separator but as a contactor of phases [function (ii)] and has been applied for desalination. A tubular catalytic membrane reactor on which the catalyst is immobilized on the membrane surface has been investigated for several biocatalytic processes [15] and is an example of function (iii). An example of membrane-assisted PI equipment with function (iv) is a hollow fiber membrane reactor. This equipment has been studied for various uses, for example, for the hydrocarbon partial oxidation in which the oxygen for the reaction is selectively supplied

2.1 Introduction 

 27

through the membrane from an air stream into the hydrocarbon mixture, leading to a higher yield compared to a conventional process [16]. Traditionally, high standards in cleanliness mean that the barriers for new equipment and processes are high in the food sector [17]. Despite this fact, several PI has been reported. As energy transfer is essential in many production routes, new energy forms have been investigated, such as induction-heated mixers, electric fields, radio frequency and microwaves (PI principle C), as well as micro devices (PI principle D), in order to create a more uniformly and quickly targeted heating of a mixture [17]. Also, the application of reactive distillation systems (PI principle A) has been investigated for chemical equilibrium-limited reaction systems and/or for systems in which azeotropes limit the potential separation into pure products (e.g., the production of lactic acid by hydrolysis of methyl lactate [18]). Table 2.1 shows some examples of membrane-assisted PI within the food industry. Table 2.1: Examples of membrane PI equipment in food technology PI Principle Membrane PI equipment function

Case

Improvement

A

i

A

i

Simplification, efficiency, costs Raw material efficiency, costs

A

i, ii

B

i, iv

B

i, ii,

B

i, ii

Dehydration of ethanol Production of lactic acid from sugarcane juice Fermentative production of ethanol Hydrolysis of triolein by lipase Production of n-butyl-oleate Concentration of juices

B

i, ii

Direct contact membrane distillation

B

i, ii

Air sweep membrane distillation

C

i

Twisted tape turbulence promoter microfiltration

Hybrid of distillation and pervaporation Microfiltration fermentor Coupling of fermentation and membrane distillation Hollow fiber membrane reactor Ultrafiltration membrane reactor Direct contact membrane distillation

Concentration of sugar solutions Concentration of juices Separation of milk proteins

Reference

[19] [20]

Increase in productivity

[21]

Not evaluated

[22]

Not evaluated

[23]

Energy efficiency, moderate temperatures Energy efficiency, moderate temperatures Energy efficiency, moderate temperatures Reduced fouling, capacity increase

[24–26]

[26]

[27]

[28]

28 

 2 Process intensification in integrated membrane processes

2.2 Synthesis/design of membrane-assisted PI – overview and concepts The identification and/or development of PI options, and membrane-assisted PI options specifically, are not simple. One of the reasons is that a large number of process options is available, and a number of decision criteria (operational constraints, performance criteria) need to be matched. Therefore, from the process systems engineering (PSE) toolbox, process synthesis tries to provide solutions, tools and methods for flowsheet synthesis/design. Process synthesis is the identification of the optimal path to reach a desired product of the desired quality and quantity from a given set of raw materials and utilities within a set of specified boundaries on the process (Figure 2.1). In general, existing process synthesis methodologies can be classified into: heuristic approaches [29–31]; approaches using thermodynamic insights [32 ]; approaches based on mathematical programming such as superstructure optimization [33, 34]; or a combination of those [35, 36]. An overview of specific synthesis methods developed to intensify processes can be found in Lutze et al. [10]. Mostly, unit operations have been used as building block for process synthesis. However, other scales/concepts apart from unit operations have also been used to synthesize processes. Such examples of other building blocks are: tasks [30]; mass and heat building blocks [37]; elementary process functions [38]; reactor/ mass exchanger building blocks [39]; as well as attempts based on phenomena [40 –42 ]. Until now, owing to the complexity of PI, the need for quantitative measures and the necessity to apply PI to the whole process rather than specific parts, all those synthesis/design methods have not been applied fully for PI [10]. But their ideas/concepts need to be integrated as supporting tools into a general PI synthesis/design methodology. Within this chapter, a general synthesis/design methodology for PI is presented. It combines mathematical formulation with methods and tools based

Sustainable production

Quick changing markets

Raw materials

Products

Utilities

Waste Process synthesis problem

Global competition

Innovative products

Figure 2.1: Simplified scheme of the general synthesis/design problem

2.2 Synthesis/design of membrane-assisted PI – overview and concepts  

 29

Process level

Unit-operation level

Phenomena level

Figure 2.2: Dependency of the levels of aggregation in chemical processes

on thermodynamic insights as well as knowledge. The mathematical formulation of PI synthesis problem is introduced in section 2.2.1. The process synthesis problem is handled using three types of building blocks of different scales: the process, unit operation and phenomena level. Processes consist of a set of unit operations, which are connected to achieve the process target. The behavior of each unit operation depends on the interaction of the involved phenomena within the operation (see Figure 2.2). At the process level, a knowledge base synthesis (KBS) approach is followed in which, based on analysis of the system under investigation, a PI knowledge base tool is searched for identification of a suitable complete process solution. Further building blocks are PI unit-based synthesis (UBS) and phenomena-based synthesis (PBS) operations. While the complexity and the time spent to obtain a solution arises from KBS to PBS, the novelty and potentially the achieved process improvement also increases. However, independent of the selected building block for synthesis/design, a large number of process options may be generated to intensify a process. Therefore, an efficient solution procedure is necessary. Here, the decomposition-based solution approach (section 2.2.2 ) is used [43]. PI is evaluated using a set of performance metrics, which can be selected according to Lutze et al. [10 ] and Criscuoli and Drioli [44]. In addition to the mathematical problem formulation and the solution procedure, the concept behind phenomena as a building block (section 2.2.3) as well as the connection of phenomena (section 2.2.4) is explained here in more detail.

2.2.1 Mathematical formulation of the PI synthesis problem The mathematical formulation of the general PI synthesis problem is given by Eqs. 2.1–2.5. Z = min FObj(Y, X, d, q ) Subject to Y, X, d, q with Y∈{0,1}, X, d, q ∈R and,

(2.1)

30 

 2 Process intensification in integrated membrane processes

Logical constraints: gLogical (Y)  ≤  0

(2.2)

gStructural (Y)  ≤  0

(2.3)

gOperational(Y, X, d, q )  ≤  0

(2.4)

Structural constraints:

Operational constraints:

The complete process model based on unit operations or phenomena: ∂X ___       = hP( Y, X, d, q ) ∂t

(2.5)

Design (optimization) variables X are, for example, temperatures T, pressures P and/or compositions x. Binary decision variables Y describe the existence of units U and streams F in a solution. If Y = 1 then a stream/unit exists in the solution, otherwise it does not. Equipment parameters are included in the vector d. Product and process specification are included in the vector q. Logical constraints (Eq. 2.2) are rules for process/operation synthesis. One example is the check whether an inlet flow into a unit operation has the correct state. The connectivity between unit operations is defined within structural constraints (Eq. 2.3). If the integration of two units/phenomena is not advantageous compared to the corresponding sequential connection, they cannot be combined. Process models hp are the set of equations to describe the behavior of the process. These can be built through models at unit operation level or at phenomena level. Dependent on the process scenario, those models are steady-state, dynamic or distributed. A feasible process option must satisfy all constraints. The evaluation of different process options is done using the objective function (Eq. 2.1), which can also be substituted and/or supported by performance criteria (Eq. 2.6). Ψ(Y, X1...v, q ) – ΨTarget (Y, X1...v, q )  ≥  0

(2.6)

Process models (Eq. 2.5) may be replaced by short-cut models (see Eq. 2.7), which do not describe the full in-depth behavior of the process, covering only a subset v of the set of process variables X. ∂X ___       = hP,simple( Y, X1..v, d, q  ) ∂t

(2.7)

2.2 Synthesis/design of membrane-assisted PI – overview and concepts  

 31

2.2.2 PI synthesis based on the decomposition approach The formulated general synthesis problem (Eqs. 2.1–2.5) is a mixed-integer non-linear problem with likely very complex process models. Dependent on the size and complexity of the synthesis problem, the identification of the global optimal solution by solving all equations simultaneously is not simple and may in some cases even impossible. Therefore, instead of solving the whole synthesis problem (Eqs. 2.1–2.5) simultaneously, the problem is decomposed into a manageable set of sub-problems [43]. This decomposition-based solution procedure is highlighted in Figure 2.3. In a stepwise manner, logical and structural constraints (Eqs. 2.2–2.3) are solved, reducing the number of process options remaining in the initial search space. Thus, the next set of equations is decomposed into subsets of fixed binary variables which mean subsets of existing process options. For each process option the corresponding process model (Eq. 2.5 and/or Eq. 2.7) together with operational constraints (Eq. 2.7) are solved to identify whether a feasible solution may exist. Subsequently, performance criteria (Eq. 2.6) are applied to identify the most promising process options from the set of feasible solutions. This small number of the top-ranked feasible alternatives is further optimized (solving Eqs. 2.1–2.6) to identify the best solution. With this approach, global optimality cannot be guaranteed. However, the obtained solution most likely would be the best or as near-optimal as is possible to obtain.

2.2.3 Phenomena as building blocks for process synthesis In this approach, the processes are aggregated at different levels (see Figure 2.2). The lowest level is the phenomena level. Phenomena as building blocks consist of mass, Search space Logical constraints (Eq. 2.2) Structural constraints (Eq. 2.3) Phenomena/unit-operation model (Eq. 2.5) Operational constraints (Eq. 2.4) Performance criteria (Eq. 2.6)

Best option (Eq. 2.1 – 2.6) Figure 2.3: Process synthesis/design via a decomposition approach

32 

 2 Process intensification in integrated membrane processes

component, energy and momentum balances as well as constraint equations describing the phenomenon as well as the inlet and outlet stream conditions. Overall mass balance (with c for coordinate axis): ∂ρ __      = ∂t

∂(ρw )  in

c ∑ ______       |       ∂c out

(2.8)

Component balance for each component i: ∂___ ρi      = ∂t

∂(ρiwc)        + r ∑______        ∂c |out i  in

(2.9)

Energy balance: ∂(ρu) _____       = ∂t

∂(ė )  in

c            + q ∑ ____ ∂c |out generated

(2.10)

Momentum balance (with index f as the number of forces): ∂(ρw) _____       = ∂t

∑      ff

(2.11)

f

All equations in the building blocks have to be linked to additional constitutive equations, such as equations representing thermodynamic properties or reaction kinetics. In general, phenomena can be classified into eight different classes: 1. Mixing. Mixing phenomena describe the mixing within one phase or between several existing phases. 2. Stream dividing. A stream dividing phenomenon divides a stream into two or more streams. Temperature, pressure and concentrations remain unchanged. 3. Phase contact. Phase contact phenomena describe the contact (and the resistances) directly at the phase boundary of two (or more) phases. 4. Phase transition. These phenomena describe the mass transfer of components between at least two phases. An example of a phase transition phenomenon is the vapor-liquid equilibrium relationship. 5. Phase change. These phenomena describe the change of state of a complete stream. Concentrations within the phase do not change. An example is full evaporation of a liquid stream. 6. Phase separation. Phase separation phenomena describe the degree of separation of the two phases. This may be ideal, meaning that the outlet streams are pure with respect to the occurring phases within each outlet stream or not. 7. Reaction. Within a reaction phenomenon, the mass of one (or more than one) component is changed between inlet and outlet streams. 8. Energy transfer phenomena. Energy supply/removal between energy sources and sinks are described with energy transfer phenomena. The classes do not overlap and form a full set of classes to represent chemical processing.

2.2 Synthesis/design of membrane-assisted PI – overview and concepts  

 33

2.2.4 Connection of phenomena Each class of phenomena (except of the stream dividing) can be further sub-classified. For example, for phase transition phenomena the subclasses are dependent on the involved phases (e.g., V-L, L-L). For each class of phenomena the number of inlet and outlet streams are defined, which is important for connectivity between phenomena of different classes. In general, phenomena can be connected in two ways. Phenomena can be connected into a simultaneous phenomena building block (SPB) when they are occurring at the same time, at the same position and having a combined operating window. This connection is called “interconnection”. All others can be connected sequentially when the necessary phenomena for the state change in the second SPB is provided to match input/output constraints of SPBs to be connected. The dividing phenomenon is always an SPB itself. Similar to the SMILES notation for molecules, a notation for phenomena-based flowsheets is defined throughout this document: – – is used for a sequential connection of phenomena – = is used for an interconnection of phenomena SPBs with at least two phases are established as co-current-flow, crossflow as well as countercurrent flow (see Figure 2.4). One or more SPBs form an unit operation. One or several operations form a process.

(a) Co-current Phase 1

Phase 1

Phase 1

Phase 2

Phase 2

Phase 2

Phase 1

Phase 1

Phase 1

Phase 2

Phase 2

Phase 2

Phase 1

Phase 1

Phase 1

Phase 2

Phase 2

Phase 2

(b) Cross

(c) Counter-current

Figure 2.4: Connectivity options between two-phase SPBs

34 

 2 Process intensification in integrated membrane processes

2.3 Synthesis/design of membrane-assisted PI – workflow The workflow of the PI synthesis methodology decomposes the problem into three branches depending on the scale to be used for PI: KBS for identification of processes; unit operation-based synthesis (UBS); and phenomena-based synthesis (PBS). Dependent on the branch, a certain number of steps have to be performed. In each step, the user needs to take certain decisions supported by algorithms and tools to proceed to the next step. The link between the different steps and tools and algorithms is presented in Figure 2.5. MBS, model-based search; LBSA, limitation/bottleneck analysis; APCP, analysis of pure component properties; AMP, analysis of mixture properties; AR, analysis of reactions; OPW, operating process window; DS, development of superstructure; SoP, selection of phenomena; AKM, apply the extended Kremser method; KS, knowledge base search. The decision regarding which workflow is to be followed depends on the maturity and the maximum development time of the PI processes to be developed. Based on analysis with respect to those criteria in the first steps of the general framework (section 2.3.1), for novel designs the PBS methodology (section 2.3.4) is followed, while for all others the unit operation-based (section 2.3.3) is followed. In cases were not only the single equipment needs to be mature, but rather the whole process, the KBS workflow (section 2.3.2) is entered. Necessary sub-algorithms, methods and tools are explained in sections 2.4.

2.3.1 Steps of the general workflow Here, the steps that need to be performed for all workflows – KBS, UBS and PBS – are explained in detail.

2.3.1.1 Step 1: Define problem The starting point of the methodology is either a base-case design of an existing or a conceptual process or input/output specification of the process. In step 1, the synthesis/design problem with respect to PI is defined, including the definition of the objective function (Eq. 2.1), the process/operation scenario and the constraints that the options need to match. In addition, the metrics for evaluation of generated options need to be selected from the available set of metrics as well as information about desired simplification of the flowsheet (complexity) and maturity of later considered equipment. – Input: System and reaction description, process and operation scenario, specifications, target, performance criteria (base-case design). – Output: Translated input into mathematical formulated synthesis/design problems.

• Literature • ICAS databases • ProPred, ProCAMD • ICAS, Pro II • SustainPro • KS, MBS, LBSA, APCP, AMP, AR, OPW

Yes

Step A2: Analyze the process

• Literature • PI knowledge base

Process data/Base-case available?

Step 1: Define problem

No

Step B2: Identify and analyze necessary tasks to achieve the process targets

Process data of existing/conceptual process OR Inlet/Outlet specifications, reactions system and a list of utilities

• OF-method • Phenomena library • ICAS databases • ProPred, ProCAMD • APCP, AMP, AR, OPW, SoP

2.3 Synthesis/design of membrane-assisted PI – workflow 

 35

Step U4: Generate feasible flowsheet options

• Model library • DS

Novel

Optimal intensified, feasible process option

Step 6: Solve the reduced optimization problem and validate promising

Step K2: Collect (PI) process(es)

PBS

• MINLP Solver/e. g. GAMS • ICAS, ICAS-MOT, Pro\II • ProCAMD, ProPred • SoP

Step P5: Fast screening for process constraints

Step P4: Generate feasible operation/ flowsheet options

Step P2: Identification of desirable phenomena

KBS High: whole process

Maturity?

• ICAS-MOT • ICAS, Pro\II • PI knowledge-base

• OF-method • AKM (extendet Kremser method)

• PI knowledge base • OF-method • Phenomena library • ICAS databases • ProPred, ProCAMD • KS, APCP, AMP AR, OPW, SoP

Figure 2.5: Workflow of the methodology MBS, model-based search; LBSA, limitation/bottleneck analysis; APCP, analysis of pure component properties; AMP, analysis of mixture properties; AR, analysis of reactions; OPW, operating process window; DS, development of superstructure; SoP, selection of phenomena; AKM, apply the extended Kremser method; KS, knowledge base search.

Tools Sub-Algorithms

Workflow

Step U5: Fast screening for process constraints

Step U3: Select and develop models

• Systematic modelling tool • Model library • ICAS MOT

• ICAS-MOT • ICAS, Pro\II • ProCAMD, ProPred • SoP

Step U2: Collect PI equipment

• PI knowledge base • KS, APCP, SoP

UBS

Low/medium/ high mature

36   2 Process intensification in integrated membrane processes

2.3 Synthesis/design of membrane-assisted PI – workflow 

 37

2.3.1.2 Step A2: Analyze the process The objective of step 2 is to collect all data about the process, necessary to gain full understanding of the process. This understanding is required to identify bottlenecks and/or limitations for improvement in order to search for already existing PI strategies/equipment in a knowledge base. Firstly, the process is translated into a task-based flowsheet as well as phenomena-based flowsheet in order to identify limitations/bottlenecks of involved phenomena in the design. For the selection for PI based on input/output specification, a base-case design is built using algorithms and rules based on identification of tasks and selection of the phenomena to fulfill the tasks. Available methods to identify bottlenecks/ limitations are based on knowledge stored in a PI knowledge base tool as well as on model-based algorithms evaluating the performance of the process. If sustainability is used as measure, the method by Carvalho et al. [45] based on mass and energy indicator is used. – Input: Mathematical formulated synthesis/design problem. Base-case design. – Output: Rule based translation of unit operation-based flowsheet into tasks and phenomena. List of limitations/bottlenecks in the process. List of phenomena causing the limitation/bottleneck.

2.3.1.3 B2: Identify and analyze necessary tasks to achieve the process target This step is analogue to step A2 but is used where a base-case design may not exist. Therefore, a simple task-based flowsheet has to be constructed first. For this, the means-ends-analysis [30] has been adapted. The output of this step is the same as in step A2. – Input: Mathematical formulated synthesis/design problem. – Output: List of limitations/bottlenecks in the process. List of phenomena causing the limitation/bottleneck.

2.3.1.4 Step 6: Solve the reduced optimization problem and validate most promising In step 6, the objective function (Eq. 2.1) for all remaining process options is calculated to identify the best option. The best among these can be found through optimization (fine tuning the optimal solution). Depending on the size of the search space, the process options are optimized separately (NLP) or through applying complex superstructure optimization methods (MINLP). – Input: Results from KBS, UBS, PBS. Mathematical formulated synthesis/design problem. All models. – Output: Best PI process option.

38 

 2 Process intensification in integrated membrane processes

2.3.2 KBS workflow This workflow consists of one-step K2 in which the KS-algorithm is used to search in a PI knowledge base tool (section 2.4.2.1) for suitable PI or non-PI processes using the system description (components/reactions, component/reaction classes) as keywords. – Input: System description. – Output: Whole process(es) to fulfill the need of (PI) process (from step 1).

2.3.3 UBS workflow Within the UBS workflow, PI equipment is identified to target the desired improvement. Based on this a large number of process options are generated, which are then screened using the decomposition approach. The initial search space is identified in step U2 and screened through stepwise applying constraints at different steps until the best option with respect to the objective function is identified. The detailed workflow for UBS can be found in Lutze et al. [46]. Examples highlighting this workflow have been the intensification of the production of N-acetyl-D-neuraminic acid [46], the production of 5-hydroxymethylfurfural and the production of hydrogen peroxide via anthraquinone route [47].

2.3.3.1 Step U2: Collect PI equipment With knowledge about limitations/bottlenecks of the involved phenomena of the process, a PI knowledge base which stores/retrieves available information (see section 2.4.2.1), is consulted retrieving possible PI principles (predictive approach) and already developed PI equipment to overcome the obtained limitations/bottlenecks for process improvement. The obtained PI solutions are pre-screened already with respect to feasibility, for example, process conditions, maturity and scale-up ability. – Input: List of limitations/bottlenecks in the process. List of phenomena causing the limitation/bottleneck. – Output: List of potentially feasible PI equipment.

2.3.3.2 Step U3: Select and develop models The objective of step U3 is to provide the process/operational mathematical models needed for the subsequent calculation/evaluation steps. The necessary models may be selected from a model library, or, if the model is not available, developed through a modeling tool using a systematic procedure proposed by Cameron and Gani [48]. Models have an important role to play, but for these to be reliable, validation is

2.3 Synthesis/design of membrane-assisted PI – workflow 

 39

necessary and will require carefully collected and analyzed experimental data. All process options based on unreliable process models are removed from the search space. Models are saved for later retrieval into a model library. – Input: List of PI equipment. Experimental/literature data. – Output: Models for PI equipment.

2.3.3.3 Step U4: Generate feasible flowsheet options In step U4, the objective is to generate all feasible intensified options through synthesis rules (logical constraints, Eq. 2.2) fixing sets of binary variables in a superstructure to give the set of feasible process options. Subsequently, structural constraints, again fixing sets of binary variables in a superstructure, are employed to get the set of structural promising process options. Redundant options are removed. – Input: Flowsheet. List of PI equipment. Models. Superstructures. Logical and structural constraints from step 1. – Output: List of feasible and structural promising process options.

2.3.3.4 Step U5: Fast screening for process constraints In step U5, all remaining PI options are fast screened using simple models (shortcut models, Eq. 2.7) or partly rigorous simulation for matching operation constraints (Eq. 2.4). The remaining process options are screened for additional process constraints based on performance metrics. – Input: List of feasible and structural promising process options. – Output: List of potential most promising PI process options. The next step is step 6 (see section 2.3.1.4).

2.3.4 PBS workflow In analogue to the UBS workflow, in the PBS workflow phenomena are identified to target the desired improvement. Based on these a large number of process options are generated, which are screened using the decomposition approach. The initial search space is identified in step 3 and stepwise reduced through applying constraints at different steps until the best option with respect to the objective function is identified. In general, the unit operation-based search space is part of the phenomena-based search space. The reason for this is that from the phenomena all different types of units can be potentially synthesized. An additional example highlighting this workflow has been the intensification of the separation of hydrogen peroxide from water, leading to an internally heat integrated setup [47].

40 

 2 Process intensification in integrated membrane processes

2.3.4.1 Step P3: Identification of desirable phenomena In step P3, the analysis of the process phenomena and the identified limitation/ bottleneck are used for identification of desirable phenomena or desirable tasks to be integrated in order to overcome the limitation/bottleneck. For each PI possibility, the best phenomena fulfilling the desired PI is identified and selected. Output of this step is the initial search space of necessary and desirable phenomena within the process. – Input: List of limitations/bottlenecks in the process. List of phenomena causing the limitation/bottleneck. List of performance criteria (from step 1). – Output: List of phenomena to fulfill the need of the desired PI process.

2.3.4.2 Step P4: Generate feasible operation/flowsheet options The inlet to this step is the initial search space of identified phenomena (nP,tot). In step P4, phenomena are connected to SPBs. The theoretical maximum number of SPBs can be expressed by giving the total number of phenomena in the search space nP,tot and the maximum number of phenomena within an SPB nP,max (from step P4.1.1) using Eq. 2.12: nP, max

NSPBmax =

1)! ∑  (____________  (   n  (n− k −− 1)!k!     )+1 P,tot

k = 1

(2.12)

P,tot

Subsequently, SPBs are connected to form unit operations and unit operations are combined to form process flowsheets. To connect SPBs to operations, the number of stages to achieve a process target is identified by the extended Kremser method (see section 2.4.2.5). With the knowledge of the number of necessary stages, the SPBs are connected using generic superstructures retrieved from a model library. Between each scale, screening steps using logical and structural constraints are performed to identify the feasible subset. – Input: List of phenomena. List of logical and structural constraints. – Output: Feasible phenomena-based process options.

2.3.4.3 Step P5: Fast screening for process constraints The remaining phenomena-based process options are screened using operational constraints and performance. After this, the most promising options are transformed to unit operations using a set of rules. These options are additionally screened by operational constraints and performance on the unit operation level. – Input: List of feasible and structural promising phenomena-based process options. – Output: List of most promising unit operation-based processes. The next step is step 6 (see section 2.3.1.4).

2.4 Synthesis/design of membrane-assisted PI 

 41

2.4 Synthesis/design of membrane-assisted PI – sub-algorithms, supporting methods and tools In several steps of the different workflows, the same information must be be processed based on the input given. Therefore, these can be outsourced into so-called sub-algorithms (2.4.1). Additionally, the tools and methods needed to create knowledge to take decisions and proceed to the next step are explained here (section 2.4.2).

2.4.1 Sub-algorithms Sub-algorithms are used at different steps of the methodology within different workflows. They are described in detail in Lutze et al. [46, 47] – MBS (model-based search): Determination of limitations/bottlenecks through a model-based search. – LBSA (limitation/bottleneck sensitivity analysis): Quick determination of the limitations/bottlenecks that have the potential to improve the process performance most when this task is enhanced. – APCP (analysis of pure component properties): Collection, generation and analysis of pure component properties based on thermodynamic insights (see also section 2.4.2). – AMP (analysis of mixture properties): Collection, generation and analysis of mixture properties. – AR (analysis of reactions): Analysis of the involved reactions. – OPW (operating process window): Determination of the operating window of each phenomenon and of each unit operation. – DS (development of a superstructure): Generation of the superstructure based on the items in the search space Ω. – SoP (selection of phenomena): Selection of the initial search space of phenomena and screening for potentially most promising phenomena to enhance a specific task. – AKM (apply the extended Kremser method): Identify of the configuration of SPBs (co-, counter, cross-current) and the number of stages necessary to achieve a defined task in the process (see also section 2.4.2). – KS (knowledge base search): Retrieval of existing knowledge about PI from a knowledge base tool (see also section 2.4.2). Data in the knowledge base can be searched through different keywords simultaneously in a forward or a reverse manner or integrating both.

2.4.2 Supporting methods and tools The developed methodology depends on a number of supporting methods and tools (Figure 2.5). Those are presented briefly in this section. Deatailed information can be found elsewhere [10, 46, 47].

42 

 2 Process intensification in integrated membrane processes

2.4.2.1 Knowledge base tool At different steps of the developed workflows knowledge about existing PI processes (KBS), unit operations (UBS) and PI principles (PBS) must be provided. Therefore, the purpose of the knowledge base tool is to store and provide this necessary knowledge obtained from scientific literature in an efficient way. In order to structure the information in a simple, efficient and flexible way, an ontology for PI knowledge representation has been developed [10, 46]. The following main items are stored in the knowledge base tool: – Existing PI processes and equipment and their operating windows. – Knowledge for identification of PI principles. – Translated performance metric into logical, structural and operational constraints at different levels of abstraction: unit operation, phenomena. – Rules for the transformation of phenomena into a unit operation (step P5). Currently, the knowledge base [46] comprises around 12000 information items for 135 PI equipment and internals in around 200 different process configurations. In total, 17 different pieces of PI membrane equipment are stored.

2.4.2.2 Model library The developed methodology is based on mathematical models to evaluate PI options on a quantitative basis. Therefore, different models are necessary, derived and validated with known data, in different modeling depths as well as superstructures for the unit operation-based workflow and the phenomena-based workflow. Hence, a model library for storage as well as retrieval of models has been developed which contains the following: – superstructure library for connection of units – models for PI equipment, reactions – superstructure library for connection of phenomena – models for phenomena.

2.4.2.3 Method based on thermodynamic insights In step U2 of the unit operation-based workflow and step P3 of the phenomenabased workflow, depending on the case, suitable PI separators or phase transition phenomena may need to be identified to improve the process. This needs a general and quick check whether the underlying phenomena are capable of separating the components. As a large number of options may exist, the method should be quick and efficient using only as little data as possible. The method developed by Jaksland et al. [32] serves this purpose. It allows the identification of feasible and potentially most promising separation systems based on pure component properties. First, for all existing mixtures in the system, all binary

2.4 Synthesis/design of membrane-assisted PI 

 43

ratios of pure component properties are calculated. The difference of these binary ratios determines the suitability of a separation based on a set of rules giving necessary minimum ratios. For example, a large difference in the boiling point (binary ratio > > 1.05) of two components identifies a distillation phase transition by vaporliquid contacting as a potential easy separation of these two.

2.4.2.4 Driving force method In the phenomena-based approach, a potentially large number of different phase transition phenomena may be used to fulfill a separation task. Here, the driving force (DF) method is applied [49]. The DF between a binary pair to be separated Dij is defined by: Dij = yi – xi

(2.13)

The DF is a measure of the ability to separate components from each other. Hence, it allows a quick comparison of different phase transition phenomena or the same phase transition phenomenon at different process condition (pressures, temperatures). If the DF is zero, for example, at the azeotrope point, a separation by this phase transition phenomenon is not possible. By plotting the DF of different phase transition phenomena for one task, it is possible to identify the ease of a separation (larger DF) by different techniques in different sections of the composition space.

2.4.2.5 Extended Kremser method In step P3, it is necessary to identify the minimum number of stages for different flow arrangements to synthesize feasible processes and to quickly identify the most promising options. For this, the Kremser method, originally developed for absorption columns [50], has been extended. It links the number of stages to the achieved/ desired outlet-concentration/conversion. The Kremser method is based on material balances around a column/flow arrangement with feed inlet and outlet in the top and bottom. A column is transformed into a series of countercurrent blocks between inlet and outlet streams. Columns with feeds or outlets within the column can also be handled as they can be transformed into two blocks of countercurrent flow arrangements, as illustrated for a distillation column in Figure 2.6.

2.4.2.6 Additional tools Several additional tools are used following the unit operation-based workflow as well as the phenomena-based workflow. Tools are necessary for: – providing pure component property data from a database (CAPEC-Database);

44 

 2 Process intensification in integrated membrane processes

V0 , y0

P = 1.3 bar N=?

Distillate x D, A = 0.95 Feed x A = 0.25 P = 1.3 bar N=?

Feed x A = 0.25

Bottom x B, W = 0.98

L 0, x0

VN+1, yN+1

L N, xN

V0 , y0

L 0, x0

P = 1.3 bar N=? VN+1, yN+1

L N, xN

Figure 2.6: Transformation of a distillation column into two blocks of countercurrent flow arrangement

– – – –

– – –

providing mixture data based on thermodynamic models from a database (CAPEC-Database); prediction of pure compound properties in case a compound and/or a property is not found in the database (ProPred); selection of solvents (ProCAMD); analysis of the base-case design in order to identify the most sensitive limitation within an operation for achieving the highest improvement in the overall process (SustainPro); analysis of physical boundaries within non-reactive and reactive separations (PDS); model derivation and analysis, parameter fitting (MOT); model simulation, optimization and validation (MOT, ICASSim).

All those tools are available in ICAS, an integrated computer aided system, developed in CAPEC, a research center at the Technical University of Denmark, DTU: – CAPEC-Database – ProPred – ProCAMD – ICASSim – PDS [51] – SustainPro [45] – ICAS-MOT [52]

2.5 Conceptual example 

 45

2.5 Conceptual example The methodology is applied to a case study in order to highlight the method and tools. Even though all workflows KBS, UBS, PBS are applied, only the PBS workflow is highlighted in more detail. However, all obtained results are later compared against the non-membrane-assisted results of the UBS and KBS workflow. As a case study, the production of isopropyl acetate (IPAc), whose main use is as a solvent in chemical processes [53] but also in the food industry [54]. Similar esters are also used as food flavoring [55]. IPAc is produced by the reaction (Eq. 2.14) of isopropanol (IPOH) and acetic acid (HOAc): CH3COOH+C3H7OH ↔ C5H10O2+H2O.

(2.14)

The reaction takes place in the liquid phase and is heterogeneously catalyzed by Amberlyst 15.

2.5.1 Step 1: Define problem In step 1, the scenario is restricted to design of a reactor incorporating a membrane (functions i–iii; see section 2.1.2) for the continuous production of IPAc. The objective is to identify one piece of apparatus, with the lowest operational costs at lowest capital costs, that achieves a conversion of 0.99 and a purity of 98 mol.% of IPAc. The production capacity of IPAc is defined as 50000 t/year. The feed is an equimolar mixture of the reactants. Energy, simplification, volume and waste are selected as PI performance criteria. Based on these, logical and structural constraints are retrieved from a knowledge base and operational constraints are set. The complete problem definition is presented in Table 2.2.

2.5.2 Step A2: Analyze the process The base-case design, a simple isothermal continuous stirred-tank reactor (CSTR), is transformed into a task-based flowsheet (necessary to identify the necessary phenomena for the process) and a phenomena-based flowsheet. The task is the conversion of IPOH and HOAc to IPAc. The involved phenomena are ideal liquid phase mixing, reaction and convective cooling for isothermal operation. Based on the task description, the reaction phenomenon is classified as necessary. The reaction analysis based on kinetics given by Lai et al. [53] confirmed the equilibrium limitation

46 

 2 Process intensification in integrated membrane processes

Table 2.2: Problem definition of the case study Objective (Fobj) Design scenario

Process scenario Logical constraints by problem formulation Structural constraints for efficiency

Structural constraints for energy

Structural constraints for simplification

Operational constraints

Operational constraints PI screening criterion for step P5

Yield, volume and pieces of equipment Design of PI equipment for the reaction task incorporating a membrane for production of 50000 t/year IPAc Continuous L1: Reaction 1 necessary L2: Reaction 1 has to be in the first stage S1: Do not integrate units which inhibit each other’s performance S2: Add phenomena and stages to the position in the flowsheet in which they have the highest efficiency S3: Always end the flowsheet with the phenomena giving the highest yield last S4: Do not provide energy to streams without purpose S5: Do not connect units with alternating heat addition and heat removal S6: Do task in units of the same kind S7: Do not use pre-reactors S8: Do not use recycle streams if not necessary (when efficiency can be reached) S9: Remove options in which stages are redundant Raw materials are pure Use raw materials according to the stoichiometry of the reaction(s) (Waste) Yield = 0.99 (phenomena level) Thermal energy consumption (phenomena level)

of the reaction (K  > 1). At a temperature of 330 K the molar-based conversion of an equimolar feed is 0.65.

2.5.3 Result of the PBS workflow 2.5.3.1 Step P3: Identification of desirable phenomena The knowledge base is contacted to identify possibilities for PI to overcome the identified limitations. As this case has an unfavorable equilibrium, the integration of the reaction task with a second reaction (of IPAc and/or H2O) or a separation task (to

2.5 Conceptual example 

 47

Table 2.3: Decision table regarding PI solutions to limitations in a process

Limitation

Necessary task:

Reaction

Reaction

Reaction

Desirable Task:

Separation

Heat supply Mixing

Reaction Reaction

Contact problems of reactants Product reacts further/is intermediate Activation problems Degradation by T Energy management Slow reaction Limiting equilibrium

this particular intensified option is reported to overcome this limitation; the activation of an option through a knowledge search.

remove IPAc and/or H2O) using suitable phenomena are options obtained from the database (Table 2.3). Applying the method based on thermodynamic insights (section 2.4.2.3) and a reaction screening, 21 different phase transition phenomena serving the identified (six are membranes) and two reactions have been identified. For example, phase transition by relative volatility has been identified by occurring differences in boiling points for the removal of H2O and IPAc [PT(VL)] as well as phase transition by pervaporation identified by differences in radius of gyration to remove H2O [PT(PVL)]. However, not all of them are feasible because the operating window does not match the liquid phase reaction phenomenon and/or potential promising because their DF is too low. Once H2O by pervaporation is removed, two membranes found in the database, one of them is a zeolite [56], the other one is polymeric [57]. The pervaporation has been modeled using correlations for selectivity and flux of a membrane based on experimental data of a zeolite membrane from Van Hoof et al. [56] and for a polymeric membrane from Sanz and Gmehling [57]. The zeolite membrane also has higher fluxes and selectivities for small water concentrations. Hence, the polymer membrane is removed from the search space. Table 2.4 gives an overview of the screening. Only two-phase transitions are kept in the search space: the zeolite pervaporation [PT(PVL)] and relative volatility [PT(VL)]. Subsequently, additional phenomena are in addition to the necessary and desirable phenomena. Those are selected from the knowledge base using a set of rules: – mixing (liquid (L): ideal (Mid), tubular flow (Mtub), rectangular flow (Mrec) – vapor (Mv): ideal – two-phase V-L: ideal (2 phM) – stream dividing (D) – convective heating (H) and cooling (C)

48 

 2 Process intensification in integrated membrane processes

Table 2.4: Identified phenomena for each desirable task using the algorithm APCP and AMP with pure component properties, and result of the screening of phenomena for each desirable task using the algorithm SoP Task

Identified phenomena

Method of determination

Separation H2O/Rest

Pervaporation Vapor permeation

Rg, VM, VVdW Rg, VM, VVdW

Nanofiltration

Rg, VM, VVdW, δSP

Pervaporation

Rg, VM, VVdW

Vapor permeation

Rg, VM, VVdW

Nanofiltration

Rg, VM, VVdW, δSP

Relative volatility

TB, PLV

Separation IPac/Rest

Separation IPAc and H2O/Rest Reaction with H2O Reaction with IPAc

Reaction phenomenon Reaction phenomenon

Screened out

Not matching operating window with necessary task Suitable membrane not in database Suitable membrane not in database Not matching operating window with necessary task Suitable membrane not in database Note: Possible until the ternary azeotrope No additional reaction found Not desired

TM, melting point; TB, boiling point; PLV, vapor pressure; Rg, radius of gyration; VM, molar volume; VVdW, Van der Waals volume; δSP, solubility parameter). The remaining phenomena and tasks kept in the search space are written in bold.

– – –

heterogeneous reaction (R) ideal phase contact of V-L (PC) ideal phase separation of V and L (PS).

Overall, nP,tot = 13 phenomena are in the search space.

2.5.3.2 Step P4: Generate feasible operation/flowsheet options Firstly, all 13 phenomena are interconnected to SPBs. The number of competing phenomena (Eq. 2.15) that cannot be present within one SPB is nP,compete = 4 because of competing: heating or cooling (-1); liquid flow mixing phenomena (-2); dividing is by definition one SPB alone (-1). Based on nP,tot and the maximum number of phenomena allowed within an SPB nP,max (Eq. 3.2), a total number of 4019 SPBs (NSPBmax, Eq. 2.16) are generated.

NSPBmax =

nP,max = nP,tot – nP,compete = 13 – 4 = 9

(2.15)

∑k = 1  ( nP,tot,w/oD!/(nP,tot,w/oD − k)!k!) ) + 1 = 4019

(2.16)

nP, max

2.5 Conceptual example 

 49

Subsequently, the generated SPBs are screened for feasibility using connectivity rules and the information of the operating window of each phenomenon (not shown here). For example, the operating window with respect to temperature/pressure for reaction phenomenon is restricted to have a liquid present, and by the deactivation energy of the catalyst while the maximum allowable temperature/pressure of the phase transition phenomenon by pervaporation is limited to the membrane stability of Pmax= 2bar and Tmax= 300K. In total, 58 SPBs are feasible in terms of conditions of the operating windows of the integrated phenomena (not shown here). Subsequently, the performance by using the conversion as criterion of each SPB is checked by using the extension of the Kremser method to identify the number of necessary stages and their connection. For example, when the reaction phenomenon is coupled with phase transition by pervaporation a conversion of 0.99 may be achieved within one stage. In other SPBs, such as reaction with phase transition by V–L, a conversion of 0.1 can achieved in 13 countercurrent stages. In order to allow the integration of different SPBs using different phase transition phenomena but still aiming at a simple operation, it is ideal to use a crossflow-current SPB arrangement and the use of betweenone and three connected stages. The simplified superstructure for this form is retrieved from the model library (Figure 2.7). The introduction of a dividing phenomenon into one stage of the three-stage superstructure enables up to six different recycles of the liquid while for configurations with two stages, three recycles are possible. That means that based on the 58 feasible SPBs in the search space 218,892 different process options are generated (Eq. 2.17). NOOmax = 583 + 582 + 581 + 6·582 + 3·581 = 218892

(2.17)

Subsequently, all options are screened by logical constraints (Table 2.5), fixing the binary variables Y for feasible processes by ensuring a matching operating window of inlet/outlet streams of the SPBs and by ensuring the formation of the product. That gives 24,142 remaining options. By subsequent structural screening (Eq. 2.3; see Table 2.5), for example, by removing energy redundant options, 506 structural promising processes are identified (see Table 2.5).

2.5.3.3 Step P5: Fast screening for process constraints The remaining 506 feasible and structural most promising options are screened by solving operational constraints and process models (Eqs. 2.4–2.5) which lead to 118

5 1

6 2

Stage 1

3 Stage 2

4 Stage 3

7 Figure 2.7: Superstructure of maximum three stages in crossflow. Only the liquid streams are shown

50 

 2 Process intensification in integrated membrane processes

Table 2.5: Results of the screening in steps P4–P5 Constraint Generated Formally feasible operation Logical: Product formed (L1) Logical: Reaction before or simultaneously with phase transition (L2) Logical: Process within operating window Structural: Energy redundant operations are removed (S5) Structural: Efficiency redundant stages are removed (S1, S2, S4, S9) Structural: SPBs with highest effect on improving the yield are last (S3) Structural: Process in 1 unit possible (S6, S7) Structural: Process in 1 unit operation possible and no external recycle (S8) Operational: Yield Performance screening: Energy Operational: Feasibility at unit operational level

Number of options 218,892 121,610 102,424 64,179 24,142 12,244 11,153 7,619 518 506 118 22 2

remaining options (Table. 2.5). Subsequently, all options are ranked by its energy consumption as PI screening criterion (see Table 2.2) and 22 most promising options are selected to remain in the search space (Table 2.5). All those 22 phenomena-based options are transformed into unit operations using a knowledge base and a set of rules. For example, because the material of the membrane is stiff (and not flexible as for polymers), the membrane is arranged either tubular or as plate-and-frame module. Additionally, boundaries for the unit operational level are retrieved from the knowledge base, leading to a subsequent screening at the unit operation level. In total, only two options remain in the search space that have been identified as achieving the lowest operational costs (represented by raw material and energy consumption). For example, an integrated CSTR type pervaporator system has been removed because necessary reactor sizes are too large. The two remaining options are: – a tubular (#1; D- MFl,tub = R = H = 2phM = Mv = PC = PT(PVL) = PS(VL)MFl,tub = R = H = 2phM = Mv = PC = PT(PVL) = PS(VL)) –

or a rectangular

(#2; D- MFl,rec = R = H = 2phM = Mv = PC = PT(PVL) = PS(VL)MFl,rec = R = H = 2phM = Mv = PC = PT(PVL) = PS(VL)) flow reactor-pervaporator with integrated simultaneous heating (see Figure 2.8).

2.5.3.4 Step 6: Solve the reduced optimization problem and validate most promising In step 6, a reduced NLP optimization problem involving the remaining two options is solved in which the diameter (and/or volume) of the tube/channel is optimized to

2.5 Conceptual example 

 51

Vapor flow Membrane

Liquid flow

Rigid wall

Feed

Heating liquid flow Width W

Liquid flow Height H

...

Figure 2.8: Detailed design of the rectangular plate-frame-flow reactor-pervaporator

achieve a conversion of 0.99, a purity of 98 mol.% IPAc in the outlet and a pressure drop Δp < 1 bar. It is assumed that 100 kg/m3 catalysts can be put into the channel of a static mixer type of packing. Additional operating constraints are fluid conditions on the membrane (Re  > 20) and that the number of parallel arrangements should not exceed 100. The kinetic model can be found in Lai et al. [53] while the experimental data for the correlations of the membrane can be found in Van Hoof et al. [56]. The results of the optimization are given in Table 2.6. The best option is the plate-andframe heat exchanger-reactor-pervaporator with a rectangular channel (option #2; see Table 2.6, Figure 2.8).

2.5.4 Comparison of solutions obtained from PBS, KBS and UBS The new reactor design using the PBS workflow (even though it is a result of targeting the intensification only in the reactor part) is benchmarked against the results from the KBS und UBS workflows.

2.5.4.1 Result of the KBS workflow By using the KBS workflow and searching in a knowledge base for the keywords of “production of isopropyl acetate” and “esterifications”, a process proposed by Corrigan and Stichweh [58] is found. It consists of one reactor, six distillation systems, one extractor using water as a solvent and one decanter (see Figure 2.9).

Table 2.6: Simulation results for both remaining process option #1 and #2 Option

T (K)

#1 #2

373 373

Parallel Rectangular channel units (–) W/H H (m) W (m) 100 20

1000* 0.0021

W, width; H, height; *at boundary constraint.

2.1

Amembrane/ V (m)

Vsingle (m3)

100 476

1.46 0.75

r or rh(m) Dp (bar)

0.01 0.0021

1.0 1.0

HAc H2O

Distillation

HAc

Distillation

Decanter

Figure 2.9: Base-case design for the production of IPAc [58]

IPOH

Reactor HAc + IPOH IPAc + H2O HAc

IPAc IPOH H2O Extractor

H2O

Distillation

H2O

H2O IPOH IPAc

IPAc IPOH H2O

IPAc IPOH H2O

IPAc H2O

H2O IPOH

IPAc IPOH H2O

IPAc IPOH H2O

H2O

Distillation

IPOH H2O

Distillation

IPAc H2O

IPAc

Distillation

52   2 Process intensification in integrated membrane processes

2.5 Conceptual example 

 53

2.5.4.2 Result of the UBS workflow Starting from the information that reaction is equilibrium-limited, the PI knowledge base tool is used to identify potential candidate PI equipment to be used for this process. This limitation can be overcome by integrating either a second reaction or a separation step (see Table 2.3). Hence, the knowledge base is again contacted to identify suitable equipment for reactive separators. In the PI knowledge base tool, in total of 22 reactive separators are stored. Among them, reactor coupled with pervaporation, VP and nanofiltration are available. However, only suitable membranes for pervaporation have been found. After screening, the best process option is a reactive distillation system [53], which consists of one reactive distillation column with a decanter at the top of the column and an external stripping column (Figure 2.10). The product is obtained as the bottom product from the stripping column. The mass and energy data for their process is given in Table 2.7. Inlet to the process is raw materials at industrial grade.

2.5.4.3 Comparison of the results With the phenomena-based design incorporating membranes retrieved from applying the PBS workflow, a conversion of 99% based on IPOH is achieved compared to around 94% for conventional as well as 93% for the reactive distillation system (Figure 2.11). The thermal heat requirement of the novel design is only 10% of the thermal heat requirement of the conventional process and around 20% of the reactive

Decanter Aqueous product

Reactive distillation

HAc + IPOH

HAc

Stripping column

IPAc + H2O

IPOH

IPAc, 99mol%

Figure 2.10: Reactive-distillation-stripper configuration for the production of IPAc [53]. Colorized is the reactive zone in the first column containing the catalyst Amberlyst-15

54 

 2 Process intensification in integrated membrane processes

Table 2.7: Mass and energy data of the reactive distillation configuration [53] Inlet

Flowrates [kmol/h] HOAc IPOH IPAc H2O Total [kmol/h] Conversion (IPOH) Conversion (HOAc) Heat [MW] Catalyst Cat density [kg/m3] Cat costs [US-$] Price [US-$/kg] Amount [kg cat]

Outlet

50.7 50.0 26.0 126.7 93.0 91.6 5.57

Aqueous

Bottom

0.00008 2.3 0.67 76.9 79.7

0.0047 0.46 46.5 0.0019 47.0

770 74200 7.7162 9616.1

100 99 98 97 96 95 94 93 92 91 90

Base-Case-Design Reactive distillation Catalytic static mixer flow reactor-pervaporator

Conversion (IPOH) [%] 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0

0.25 0.20 0.15 0.10 0.05 0 Heat [kW/t product per year]

Amount of catalyst [kg cat/t product per year]

Figure 2.11: Comparison of the design for the production of 46.5 kmol/h IPAC for the criteria conversion, heat and amount of catalyst

distillation system. The amount of catalyst in the novel design can be reduced to a value of 10% compared to the reactive distillation system.

2.6 Conclusions 

 55

2.6 Conclusions PI and membranes will play an important role in matching the future challenges of production processes. One of the techniques to intensify processes by target enhancement of limiting process phenomena is the integration of membranes into processing in order to exploit solely specific membrane properties. In order to provide systematic designs as well as quantitative reasoning, systematic synthesis/methods are necessary. Here, one PI synthesis/design method is proposed, which offers, depending on the time and resources to be spent for the solution, different levels as building blocks that are complete processes at the knowledge-based level, unit operations and phenomena. While the novelty of the solution and the potential performance improvement increases from processes to phenomena, time and resources also increase while maturity decreases.

2.7 References 1. Stankiewicz A, Moulijn J. Process intensification: Transforming chemical engineering. Chem Eng Prog 2000;96:22–34. 2. European Roadmap for Process Intensification. Creative Energy – Energy Transition, 2009. http:// www.efce.info/index.php?id=111785&site=efce&lang=en (accessed September 28th, 2013). 3. Moulijn J, Stankiewicz A, Grievink J, Gorak A. Process intensification and process systems engineering: A friendly symbiosis. Comput Chem Eng 2008;32:3–11. 4. Hall GM, Howe J. Energy from waste and the food processing industry. Process Saf Environ 2012;90:203–212. 5. Ramshaw C. “HIGEE” Distillation – An Example of Process Intensification. Chem Eng 1983;13–14. 6. Tsouris C, Porcelli JV. Process Intensification – Has its time finally come? Chem Eng Prog 2003;99:50–55. 7. Cross WT, Ramshaw C. Process Intensification – laminar-flow heat-transfer. Chem Eng Res Des 2000;64:293–301. 8. Becht S, Franke R, Geißelmann A, Hahn H. An industrial view of process intensification. Chem Eng Process 2009;48:329–332. 9. Van Gerven T, Stankiewicz A. Structure, Energy, Synergy, Time – The Fundamentals of Process Intensification. Ind Eng Chem Res 2009;48:2465–2474. 10. Lutze P, Gani R, Woodley JM. Process intensification: A perspective on process synthesis. Chem Eng Process 2010;49:547–558. 11. Mulder M. Basic Principles of Membrane Technology. Kluwer Academic Publishers: Dordrecht, The Netherlands; 1991. 12. Lipnizki F, Field RW, Ten PK. Pervaporation-based hybrid process: A review of process design, applications and economics. J Membrane Sci 1999;153:183–210. 13. Curcio E, Drioli E. Membrane distillation and related operations – A review. Sep Purif Rev 2005;34:35–85. 14. Curcio R, Criscuoli A, Drioli E. Membrane crystallizers. Ind Eng Chem Res 2001;20:2679–2684. 15. Jochems P, Satyawali Y, Diels L, Dejonghe W. Enzyme immobilization on/in polymeric membranes: status, challenges and perspectives in biocatalytic membrane reactors (BMRs). Green Chem 2011;13:1609–1623. 16. Wang HH, Tablet C, Schiestel T, Werth S, Caro J. Partial oxidation of methane to syngas in a perovskite hollow fiber membrane reactor. Catal Commun 2006;7:907–912.

56 

 2 Process intensification in integrated membrane processes

17. Reay D, Ramshaw C, Harvey H. Process Intensification – Engineering for Efficiency, Sustainability and Flexibility. Elsevier; Oxford, UK: 2008. 18. Mujtaba IM, Edreder EA, Emtir M. Significant thermal energy reduction in lactic acid production processes. Appl Energ 2012;89:74–80. 19. Stürken K. Organophile Pervaporation: Ein Membranverfahren zur Aufarbeitung verdünnter wäßrig-organischer Lösungen. Dissertation. Technische Universität Hamburg – Harburg, 1994. 20. Sikder J, Roy M, Dey O, Pal P. Techno-economic analysis of a membrane-integrated bioreactor system for the production of lactic acid from sugarcane huice. Biochem Eng J 2012;63:81–87. 21. Gryta M. The fermentation process integrated with membrane distillation. Sep Purif Technol 2001;24:283–296. 22. Goto M, Goto M, Nakashio F, Yoshizuka K, InoueK. Hydrolysis of triolein by lipase in a hollow fiber reactor. J Membrane Sci 1992;74:207–214. 23. Habulin M, Knez Z. Enzymatic synthesis of n-butyl-oleate in a hollow fiber membrane reactor. J Membrane Sci 1991;61:315–324. 24. Nene S, Kaur K, Sumod K, Joshi B, Raghavarao KSMS. Membrane distillation for the concentration of raw cane-sugar syrup and membrane clarified sugarcane juice. Desalination 2002;147:157–160. 25. Calabro V, Jiao BL, Drioli E. Theoretical and experimental study on membrane distillation in the concentration of orange juice. Ind Eng Chem Res 1994;33:1803–1808. 26. Alves VD, Coelhoso IM. Orange juice concentration by osmotic evaporation and membrane distillation: A comparative study. J Food Eng 2006;74:125–133. 27. Deshmukk SK, Sapkal VS, Sapkal RS. Evaluation of direct contact membrane distillation for concentration of orange juice. Int J Chem Res 2011;1;39–48. 28. Popovic SS, Joviecevic DZ, Duric MS, Milanovic SD, Tekic MN. Influence of twisted tape turbulence promoter on fouling reduction in microfiltration of milk proteins. Hem Ind 2011;65:233–239. 29. Douglas JM. A hierarchical decision procedure for process synthesis. AIChE J 1985;31:353–362. 30. Siirola JJ, Rudd DF. Computer aided synthesis of chemical process designs. Ind Eng Chem Fundam 1971;10:353–362. 31. Barnicki SD, Fair JR. Separation Systems Synthesis: A Knowledge-Based Approach. 1. Liquid Mixture Separations. Ind Eng Chem Res 1990;29:421–432. 32. Jaksland CA, Gani R, Lien KM. Separation process design and synthesis based on thermodynamic insights. Chem Eng Sci 1995;50:511–530. 33. Brueggemann S, Oldenburg J, Zhang P, Marquardt W. Robust dynamic simulation of three phase reactive batch distillation columns. Ind Eng Chem Res 2004;43:3672–3684. 34. Grossmann IE, Aguirre PA, Barttfeld M. Optimal synthesis of complex distillation columns using rigorous models. Comp Chem Eng 2005;29:1203–1215. 35. D’Anterroches L, Gani R. Group contribution based process flowsheet synthesis, design and modeling. Fluid Phase Equilibr 2005;228–229:141–146. 36. Lutze P, Gani R, Woodley JM. Phenomena-based synthesis and design to achieve process intensification. Comput Aided Chem Eng 2011;29(C):221–225. 37. Papalexandri KP, Pistikopoulos EN. Generalized modular representation framework for processs synthesis. AIChE J 1996;42:1010–1032. 38. Peschel A, Freund H, Sundmacher K. Methodology for the design of optimal chemical reactors based on the concept of elementary process functions. Ind Eng Chem Res 2010;49:10535–10548. 39. Linke P, Kokossis A. Attainable designs for reaction and separation processes from a superstructure-based approach. AIChE J 2003;49:1451–1470. 40. Rong BG, Kolehmainen E, Turunen I. Methodology of conceptual process synthesis for process intensification. 18th European Symposium on Computer Aided Process Engineering – ESCAPE 18. Lyon, France, 1–4 June, 2008.

2.7 References 

 57

41. Arizmendi-Sánchez J, Sharratt P. Phenomena-based modularization of chemical process models to approach intensive options. Chem Eng J 2008;135:83–94. 42. Lutze P, Babi DK, Woodley JM, Gani R. Phenomena-based Synthesis and Design to achieve Process Intensification. Proceedings of 11th International Symposium on Process Systems Engineering. Elsevier BV, 2012, pp. 1697–1701. 43. Karunanithi AT, Achenie LEK, Gani R. A new decomposition-based computer-aided molecular/ mixture design methodology for the design of optimal solvents and solvent mixtures. Ind Eng Chem Res 2005;44:4785–4797. 44. Criscuoli A, Drioli E. New metrics for evaluating the performance of membrane operations in the logic of process intensification. Ind Eng Chem Res 2007;46:2268–2271. 45. Carvalho A, Gani R, Matos H. Design of sustainable chemical processes: Systematic retrofit analysis generation and evaluation of options. Process Saf Environ 2008;86:328–346. 46. Lutze P, Roman–Martinez A, Woodley JM, Gani R. A systematic synthesis and design methodology to achieve process intensification in (bio)chemical processes. Comput Aided Chem Eng 2010;28(C):241–246. 47. Lutze P. An Innovative Synthesis Methodology for Process Intensification. J&R Fredenberg: Copenhagen, Denmark; 2012. 48. Cameron IT, Gani R. Product and Process Modelling: A Case Study Approach. 1st ed. Elsevier: Amsterdam, The Netherlands; 2011. 49. Bek-Pedersen E, Gani R. Design and synthesis of distillation systems using a driving-force-based approach. Chem Eng Process 2004;43:251–262. 50. Seader JD, Henley EJ. Separation Process Principles. 1st edition. John Wiley & Sons Inc.: New York, USA; 1998. 51. Gani R, Hytoft G, Jaksland C, Jensen AK. An integrated computer aided system for integrated design of chemical processes. Comput Chem Eng 1997;21:1135–1146. 52. Heitzig M, Sin G, Sales Cruz M, Glarborg P, Gani R. Computer-aided model framework for efficient model development, analysis and identification: Combustion and reactor modeling. Ind Eng Chem Res 2011;50:5253–5265. 53. Lai IK, Hung SB, Hung WJ, Yu CC, Lee MJ, Huang HP. Design and control of reactive distillation or ethyl and isopropyl acetates production with azeotropic feeds. Chem Eng Sci 2007;62:878–898. 54. US Food and Drug Administration (FDA). Appendix 6. Toxicological Data For Class 3 Solvents. http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ ucm073403.pdf (Accessed 4 September, 2012). 55. Monument Chemicals Inc., 2012: Safety Date Sheet. http://www.monumentchemical.com/ documents/IPAc_Data_and_Safety_Sheet_US_EU_1.pdf (Accessed 4September, 2012). 56. Van Hoof V, Dotremont C, Buekenhoudt A. Performance of Mitsui NaA type zeolite membranes for the dehydrogenation of organic solvents in comparison with commercial polymeric pervaporation membranes. Sep Purif Technol 2006;48:304–309. 57. Sanz MT, Gmehling J. Esterification of acetic acid with isopropanol coupled with pervaporation: Part I: Kinetics and pervaporation studies. Chem Eng J 2006;123:1–8. 58. Corrigan TE, Stichweh LA. Esterification process development. Chem Eng Sci 1968;23:991.

3 Integrated membrane operations in fruit juice processing Alfredo Cassano, Carmela Conidi and Enrico Drioli 3.1 Introduction Fruit juices are the biggest product of a number of fruits and are recognized as important components of the human diet. They provide a range of key nutrients as well as many non-nutrient phytochemicals, which are important for their role in preventing chronic diseases such as cancer, cardiovascular and neurological disorders. The consumption of fruit juice has significantly increased in recent years. The growing demand for juices of high sensory and nutritional quality has led to the investigation of new improved food processing technologies. With a large number of advantages such as high-efficiency, simple equipment, convenient operations and low-operating consumption, membrane technology is today one of the most important separation techniques to support the production and marketing of innovative juices designed to exploit the sensory characteristics and nutritional peculiarities of fresh fruits [1]. Fruit juice clarification, stabilization, depectinization and concentration are typical steps where membrane processes such as microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), reverse osmosis (RO), osmotic distillation (OD) and membrane distillation (MD), have been successfully utilized as alternative technologies to the traditional fruit juice production [2]. Membrane operations represent a valid alternative to thermal evaporation processes, which cause the deterioration of heat-sensitive compounds leading to a remarkable qualitative decline of the final product. Conversely, the current filtration of a wide variety of juices is performed by using fining agents such as gelatine, diatomaceous earth, bentonite and silica sol, which cause environmental problems upon their disposal. The possibility to realize integrated membrane systems in which all the steps of the fruit juice production are based on molecular membrane separations, or in many cases in combination with other conventional separation units, often allows better performance in terms of product quality, plant compactness, environmental impact and energetic aspects [3, 4].

3.2 Clarification of fruit juices Fruit juices are naturally cloudy, but to different degrees, because of the presence of polysaccharides (pectin, cellulose, hemicelluloses, lignin and starch), proteins,

60 

 3 Integrated membrane operations in fruit juice processing

tannins and metals [5]. As the juice’s clear appearance is a determinant factor for consumers, the fruit juice industry has been investing in methods that optimize this feature. Conventional methods of producing clarified juice involve many steps, such as enzymatic treatment (depectinization), cooling, flocculation (gelatin, silica sol, bentonite and diatomaceous earth), decantation, centrifugation and filtration. These processes are generally slow; in particular, the clarification step with gelatine and silica sol may take a minimum of 6–18 h to accomplish the necessary sedimentation of the colloidal particles. The incomplete sedimentation of the colloidal material results in prolonged processing times and significant juice loss. The clarified supernatant juice fraction remaining after the settling of flocs may be centrifuged prior to subsequent filtration treatments, but the voluminous, viscous precipitate requires the use of sludge frame filters or high-vacuum rotary filtration systems with diatomaceous earth as a filtering aid to recover some of the juice captured in the colloid sediment solution [6]. In addition, the use of fining agents is characterized by different drawbacks, such as the risks of dust inhalation with consequent health problems caused by handling and disposal, environmental problems and significant costs. UF and MF processes represent a valid alternative to the use of traditional fining agents and filter aids. They are typical pressure-driven membrane processes capable of separating particles in the approximate size range of 1–100 μm and 0.05–10 μm, respectively. Basically, large species such as microorganisms, lipids, proteins and colloids are retained, while small solutes such as, for example, vitamins, salts and sugars flow together with water. Advantages of UF and MF processes over conventional fruit juice processing are: increased juice yield; the possibility of operating in a single step to reduce working times; the possibility of avoiding the use of gelatines, adsorbents and other filtration aids; the reduction in enzyme utilization; easy cleaning and maintenance of the equipment; reduction of waste products; the elimination of the need for pasteurization [7]. In addition, the low temperatures used during the process preserve the fruit juice freshness, aroma and nutritional value. In these processes the juice is separated into a fibrous concentrated pulp (retentate) and a clarified fraction free of spoilage microorganisms (permeate), improving the microbiological quality of the clarified juice. Polysulfone (PS) membranes have been used extensively for juice UF; in comparison with cellulose acetate membranes they can withstand short exposures to hypochlorites during periodic cleaning cycles. Polyvinylidene fluoride (PVDF), polyamide (PA) and polypropylene (PP) membranes are also largely used in some juice UF systems because they are inexpensive if compared with ceramic membranes. However, the major drawback of polymeric membranes is their low stability in drastic conditions of pH and, consequently, limited shelf-life for juice processing applications. Ceramic membranes have greater resistance to chemical degradation and a much longer shelf-life; however, one of the significant limitations of ceramic membranes is their higher cost in comparison with the polymeric ones. The higher cost of ceramic membranes is due to the utilization of expensive inorganic precursors (alumina and

3.2 Clarification of fruit juices 

 61

zirconia) and higher sintering temperature during membrane fabrication. Therefore, the development of low-cost ceramic membranes is performed by utilizing low-cost inorganic precursors and lower sintering temperatures. Nandi et al. [8] studied the clarification process of mosambi juice using low-cost inorganic membranes prepared with different low-cost inorganic precursors, such as kaolin, quartz, feldspar, sodium carbonate, boric acid and sodium metasilicate. Different physico-chemical parameters such as color, clarity, pH, citric acid content, total soluble solids (TSS) were measured before and after the clarification process to study the effect of membrane processing on the properties of the juice. The prepared membranes showed high-efficiency in preserving the properties (pH, TSS, acidity, density) of both centrifuged and enzyme-treated centrifuged juices; however, a significant decrease in color, alcohol insoluble solids and viscosity and an increase in clarity were observed because of the removal of pectin materials. The most used configurations for the clarification of fruit juices at industrial level are tubular, capillary, hollow fiber and plate-and-frame membrane modules [9–11]. For pulpy juice, with high solids content and viscosity, large bore-tubular modules or plate-and-frame modules with large spacers are preferred. However, the tubular configuration is associated with low packing density and high membrane replacement costs. Conversely, hollow fiber membranes present the advantage of a high membrane area per volume unit of module, low manufacturing costs and a simple handling in comparison with other membrane configurations. Permeate fluxes and quality of the clarified juices in both MF and UF processes are strongly affected by operating conditions, such as transmembrane pressure, crossflow velocity, temperature and volume reduction factor (VRF), nature of the membrane and nature of the feed solution. The clarification of fruit juices by MF and UF has been extensively studied in recent years by various authors. Most of these works were devoted to the preservation of sugars, vitamins and other constituents in order to maintain the characteristics of the fresh product: fundamental requirements for the viability of these processes and the consumer acceptance. The effect of operating conditions, pore size and nominal molecular weight cut-off (NMWCO) on the permeate flux in the clarification of pineapple juice was investigated by Laorko et al. [12] using both MF and UF membranes. In particular, MF membranes with pore sizes of 0.1 μm and 0.2 μm and UF membranes with NMWCO of 30 and 100 kDa were employed. According to the obtained results, the 0.2 μm MF membrane was considered to be the most suitable for the clarification of pineapple juice as it showed the highest recovery of phytochemical compounds and the highest permeate flux. Carneiro et al. [13] successfully clarified pineapple juice with a tubular polyethersulfone (PES) 0.3 μm pore size MF membrane associated with an enzymatic treatment. The clarification process produced a great reduction of haze and viscosity without modifying the acidity and the soluble solid content of the juice. Additionally,

62 

 3 Integrated membrane operations in fruit juice processing

the microbiological parameters of the microfiltered juice fell within the standards required by the Brazilian legislation for juices and drinks. The retention of sugars in the clarification of pineapple juice by MF and UF is affected by the membrane pore size and MWCO as well as by the geometry of the membrane module [14]. The clarification of melon juice by a crossflow MF process was studied by Vaillant et al. [15]. Results showed that the clarified juice was highly similar to the initial juice except for the absence of suspended solids and carotenoids. Similar results were obtained with the clarification of pomegranate juice using two PVDF membranes with pore sizes of 0.22 and 0.45 μm. The clarified juice did not show significant changes in its chemical properties when compared to enzymatic methods. In addition, it presented a more desiderable color if compared with the fresh juice, thus improving the marketability of the product [16]. Cassano et  al. [17] studied the influence of the UF process on the recovery of bioactive compounds from kiwifruit juice by using a 30 kDa cellulose acetate membrane. Most bioactive compounds of the depectinised kiwifruit juice were recovered in the clarified fraction of the UF process. The rejection of the UF membrane towards total phenolics was 13.5%. The recovery of glutamic, folic, ascorbic and citric acids, in the clarified juice, with respect to the initial feed, was dependent on the final VRF of the process: an increase of the VRF determined an increase of these compounds in the clarified juice. The rejections of the UF membrane towards these compounds were in the range 0–4.3%. The pretreatment of the juice is another important aspect that affects the performance of MF and UF membranes in terms of juice quality and permeate flux. Fruit juices contain a high concentration of pectin, cellulose, hemicelluloses and proteins, which make the juice highly viscous and difficult to treat in the following clarification step. A preliminary enzymatic treatment of the juice with pectinases determines a hydrolysis of pectins, leading to an improvement of the permeate flux caused by the reduction of the viscosity of the juice and the removal of pectinious materials, which tend to form a deposited foulant layer on the membrane surface [18]. Rai et al. [19] evaluated the effect of different pretreatment methods (centrifugation, fining by gelatin, fining by bentonite, fining by bentonite followed by gelatin, enzymatic treatment, enzymatic treatment followed by centrifugation and enzymatic treatment followed by fining with bentonite) on permeate flux and quality during the UF of mosambi juice. The enzymatic treatment followed by adsorption with bentonite produced the highest permeation flux and a clarified juice with more than 93% clarity without deterioration of the juice quality. The juice composition plays a very important role in the fouling of MF and UF membranes. Membrane fouling leads to a decline in permeate flux through the membrane, a more frequent membrane cleaning and replacement, and, consequently, an increasing in the operating costs.

3.2 Clarification of fruit juices 

 63

The quantification of flux decline in the clarification of fruit juice by MF and UF processes has been analyzed by several authors with different methods. The flux decline in the unstirred UF of enzymatically treated mosambi juice was analyzed by Rai et al. [20] through the gel filtration theory. The obtained results show that the gel porosity decreased with the operating pressure, indicating the formation of a compressed gel layer over the membrane surface. The membrane resistance was found to be dominant in the first few seconds of the process (4.5 s), and beyond 143–361 s the filtration was found to be entirely controlled by the gel layer. Jiraratananon and Chanachai [21] analyzed the flux decline in the UF of passion fruit juice according to the resistance-in-series model. According to this model the permeate flux (Jp) for UF is usually written in terms of transmembrane pressure difference and total resistance: Jp = ___  ΔP   mRt

(3.1)

where Jp is the permeation flux (m/s), ΔP the transmembrane pressure difference, Rt the total resistance (m–1) and μ is the viscosity of the solution (Pa s). Rt is defined as: Rt = Rm + Rp,re + Rp,ir + Rf

(3.2)

where Rm is the resistance of the membrane; Rp,re is the resistance of the reversible polarized layer consisting of the concentration polarization layer plus a precipitated gel resulting from the limit of solubility of macromolecules (it can be removed by rinsing with water at low flow rate); Rp,ir is a semi-reversible polarized layer loosely bound to the fouling layer (it can be removed by rinsing with water at high flow rate); Rf is the fouling resistance caused by an irreversible adsorbed layer that can be removed only by chemical cleaning. All these resistances can be calculated by measuring the water flux through the membrane after cleaning with water and chemical susbstances. In Figure 3.1 a schematic representation of the resistance-in-series model is shown. Experimental results indicated that all resistances increased with the operating pressure and juice concentration and decreased with the feed flow rate. An increase in temperature determined a reduction of Rp,re and Rp,ir and an increase in Rf. At high temperatures (50°C) the reversible polarized layer changed to a cross-linked gel, significantly increasing Rf. According to the results obtained by Tasselli et al. [22] in the UF of kiwifruit juice with modified poly(ether ether ketone) (PEEK WC) hollow fiber membranes, Rm controlled the permeate flux at ΔP values lower than 30 kPa, while at higher ΔP the permeate flux was controlled by Rf. Rf was also the major resistance to the permeate flux over the whole range of flow rate investigated. Constela and Lozano [23] analyzed the flux decay in the UF of apple juice with hollow fiber membranes using different approaches. They found that at constant VRF

64 

 3 Integrated membrane operations in fruit juice processing

Rp,ir Rm

Rp, re cg

Bulk

Rf

cb Permeate Membrane Figure 3.1: Schematic representation of the resistance-in-series model (Rm, membrane resistance; Rf, fouling resistance; Rp,re, resistance of the reversible polarized layer; Rp,ir, resistance of the semi-reversible polarized layer; cg, gel concentration; cb, bulk concentration)

the membrane fouling can be described adequately through the exponential model of the following equation: Jp = J0 − B ln(VRF)

(3.3)

where J0 is the initial permeate flux and B is a constant that depends on the system, operating conditions and juice properties. Additionally, the fouling phenomena under increasing VRF can be adequately described by using classical flow-through filtration models. Cassano et al. [24] analyzed the fouling phenomena in the UF of blood orange juice with tubular PVDF membranes according to the mathematical model presented by Field et al. [25] in which the permeate flux decline with time is described by the following equation: dJ − __      = k ⋅ (J − J lim ) ⋅ J2−n dt

(3.4)

where Jlim represents the limit value of the permeate flux obtained in steady-state conditions; k and n are a phenomenological coefficient and a general index, respectively, depending on fouling mechanism. The model permits the establishment of the fouling mechanism involved in the process according to the estimated value for n as follows: – Complete pore blocking (n = 2), which occurs when particles are larger than pore size and a complete pore obstruction is obtained. – Partial pore blocking (n = 1), a dynamic situation in which particles may bridge a pore by obstructing the entrance but not completely blocking it. – Cake filtration (n = 0), in this case particles that do not enter the pores form a cake on the membrane surface. – Internal pore blocking (n = 1.5; jlim = 0), which occurs when particles enter the pores, reducing the pore volume.

3.3 Concentration of fruit juices 

 65

Analysis of the results revealed that, in fixed operating conditions of ΔP and temperature, the fouling mechanism evolved from a partial to a complete pore blocking condition in dependence on the axial velocity. A similar model applied to the clarification of passion fruit by MF [26] and of pineapple juice by UF [27] indicated that internal pore blocking dominated in the case of ceramic tubular membranes while cake filtration was dominant in the case of hollow fiber membranes.

3.3 Concentration of fruit juices The production of concentrated fruit juices is of interest at industrial level, as they can be used as ingredients in many products, such as ice creams, fruit syrup, jellies and fruit juice beverages. In addition, the concentration of fruit juices includes a series of advantages such as weight and volume reduction, with a consequent reduction of packaging, transport, handling and storage costs. The product stability is also enhanced by the reduction of the water activity. Finally, the concentration step allows a better product preparation for a final drying treatment. The industrial concentration of fruit juices is usually performed by multi-stage vacuum evaporation processes, in which water is removed at high temperatures followed by the recovery and concentration of volatile flavors and their addition back into the concentrated product [28]. However, the main drawbacks of the traditional evaporation processes are high energy consumption, off-flavor formation, color change and reduction of nutritional values caused by thermal effects. An alternative technique to thermal evaporation is cryoconcentration, in which water is removed as ice and not as vapor. This technology preserves the juice quality but the achievable concentration is lower (about 50°Brix) if compared to thermal evaporation; in addition, the cryoconcentration is characterized by a significant energy consumption [29–31]. In comparison with conventional technologies, membrane operations meet most requirements of the modern food industry. They are environmentally friendly with high effectiveness and low energy consumption; in addition, the possibility to operate at low temperatures allows for the preservation of the sensory and nutritional qualities of the fresh juice. In the following section some unit operations employed in fruit juice concentration are described and discussed.

3.3.1 Nanofiltration NF is a relatively new pressure-driven membrane process situated between the separation capabilities of UF and RO, which can be used to separate low molecular weight solutes at low pressure based on steric, Donnan and dielectric exclusion effects. The

66 

 3 Integrated membrane operations in fruit juice processing

molecular weights are in the range of 200–400 Da and divalent ions are rejected more than monovalent ions in mixed electrolyte solutions. The NF process offers higher fluxes and lower energy consumption than RO and better retention than UF for lower molar mass molecules such as sugars, natural organic matters and ions [32 ]. As a result, NF represents a promising process in the food and beverage industry for reducing dissolved contaminants. It has been successfully employed in the concentration of grape must [33], milk demineralization [34] and the treatment of spent process water from the food and beverage industries [35]. Interesting applications have also been developed in the field of fruit juice concentration. Warczok et al. [36] studied the concentration of apple and pear juice at low pressures (8–12 bar) by using different NF membranes (AFC80, MPT-34, Desal-5DK) in tubular and flat-sheet configurations. The results indicated that in membrane selection both retention and permeation values should be considered, and that irreversible fouling of fruit juices is relatively low (30 ± 8%). Bánvölgyi et  al. [37] studied the NF process for the concentration of blackcurrant juice using a flat-sheet membrane with a salt rejection of 78.11%. NF was performed at selected values of temperature (30°C), pressure (20 bar) and feed-flow rate (400 l/h). At VRF of 2.23 the average permeate flux was of approximately 18 l/m2h and the retention of total extract content was 96.72%. NF has also been successfully employed for concentrating bioactive compounds extracted from vegetables and fruit juices. For example, an integrated process UF-NF was studied for the separation and concentration of polyphenols from bergamot juice [38]. The depectinised juice, after a preliminary clarification by UF, was treated with NF membranes with different MWCO (450, 750 and 1000 Da) in order to evaluate their selectivity towards sugars, organic acids and polyphenols. The results showed that the use of an integrated process based on the use of the 450 Da NF membrane was the best procedure for the separation of polyphenols from sugars. The proposed process resulted in a clear solution enriched in sugars and organic acids (permeate) and a fraction enriched of phenolic compounds with high antioxidant activity.

3.3.2 Reverse osmosis Fruit juice concentration by RO has been of interest to the fruit processing industry for about 30 years. RO separates mainly water from the juice and obtains a high-quality product because of low operating temperatures and retention of nutritional, aroma and flavor compounds, with low energy consumption and low capital costs. The concentration of a variety of fruit juices by RO has been studied [39, 40]. These works were mainly devoted to evaluating the effect of different operating conditions on permeate fluxes and retention of juice compounds. Matta et al. [41] evaluated

3.3 Concentration of fruit juices 

 67

the quality of the acerola juice concentrated by RO for sensory, microbiological and nutritional values. The acerola juice, after an enzymatic treatment, was clarified by MF and then concentrated by a thin-film composite RO membrane in plate-and-frame configuration with a NaCl rejection of about 95%. The MF process resulted in a good level of clarification regarding all the substances that cause haze, and in a clear juice free of pulp and suspended substances. The RO process concentrated the clarified juice from 7°Brix to 29.2°Brix: the concentrated juice presented a content of ascorbic acid of 5229 mg/100 g, 4.2 times higher than the initial value. In spite of the high selectivity and solute retention capacity of RO membranes, this process has a significant drawback. The high osmotic pressure of fruit juices precludes their concentration at the required level of TSS. For cellulosic and noncellulosic membranes the most efficient flux and solute recovery were obtained at a concentration lower than 30°Brix [40]. This suggests the implementation of integrated processes in which RO is used as a preconcentration step before a final concentration with other technologies (freeze concentration, thermal evaporation, MD and OD).

3.3.3 Osmotic distillation Osmotic distillation (OD), also known as osmotic evaporation, membrane evaporation and isothermal MD, has attracted considerable interest in the concentration of thermo-sensitive solutions, such as fruit juices, because it works under atmospheric pressure and room temperature, thus avoiding thermal and mechanical damage of the solutes [42]. In comparison with pressure-driven membrane processes it does not suffer from strong limitations when high osmotic pressures are involved [43]. The process involves the use of a macroporous hydrophobic membrane to separate two circulating aqueous solutions at different solute concentrations: a dilute solution and a hypertonic salt solution. The difference in solute concentrations, and consequently in water activity of both solutions, generates a vapor pressure difference at the vapor-liquid interface, causing a vapor transfer from the dilute solution towards the stripping solution (Figure 3.2). The water transport through the membrane can be summarized in three steps: (i) evaporation of water at the dilute vapor-liquid interface; (ii) diffusional or convective vapor transport through the membrane pore; (iii) condensation of water vapor at the membrane-brine interface [44, 45]. The typical OD process involves the use of concentrated brine as a stripping solution at the downstream side of the membrane. A number of salts such as MgSO4, CaCl2, K2HPO4 are suitable. The most important parameters that affect the water flux in OD are mainly feed and brine flow rate and brine concentration. The flow rates directly affect the thickness of the boundary layer at the membrane surface that presents a resistance to the mass transfer, while the concentration of brine affects the vapor pressure gradient through

68 

 3 Integrated membrane operations in fruit juice processing

High water vapor pressure

Low water vapor pressure Vapor flux

Feed diluted aqueous solution

Air

Stripping solution (concentrated salt solution)

Macroporous hydrophobic membrane Figure 3.2: Schematic representation of the osmotic distillation mechanism

the membrane, which is directly related with magnitude of the driving force. According to the results obtained by Ravindra Babu et al. [46] in the OD concentration of pineapple juice, the contribution of concentration polarization on transmembrane flux is more prominent when compared to that of temperature polarization. Additionally, at low TSS of the feed juice the flux decay is more attributable to the dilution of the stripping solution; at higher TSS values it mainly depends on juice viscosity (viscous polarization) and, consequently, on juice concentration and temperature [47, 48]. In recent years different applications of OD in the concentration of fruit juices have been reported. Hongvaleerat et al. [49] evaluated the potential of using OD to concentrate both single-strength and clarified pineapple juice. The OD experiments were performed in a laboratory-scale system consisting of two different circuits, one for the juice and the other one for the brine solution. As an extraction phase a saturated calcium chloride solution was employed at a concentration in the range of 5.5–0.6 mol/l. A 0.2 μm flat-sheet hydrophobic membrane was used, composed of a polytetrafluoroethylene (PTFE) layer supported by a polypropylene (PP) porous support. Concentrated juices with a TSS content of at least 56°Brix were obtained. The results showed that an increase in the temperature of the juice from 20°C to 35°C approximately doubled the evaporation flux, whereas the increase in the circulation velocity of the salt solution increased the evaporation flux of about 7%. The increase of flux with temperature can be attributed to the increase of water partial pressure at the liquid– gas interface on the juice side, which increases the driving force for water transfer. Higher evaporation fluxes were obtained for the clarified juice (8.5 kg/m2h) in comparison with the single-strength juice (6.1 kg/m2h) indicating a clear effect of pulp on the performance of the OD process. Additionally, the characterization of the juice showed no significant changes in the color and other main quality criteria of the juice. In particular, the titrable acidity and the phenolic content of both clarified and singlestrength juices increased proportionally to the TSS concentration factor.

3.3 Concentration of fruit juices 

 69

Cassano et al. [50] evaluated the potential of the OD process in the concentration of clarified kiwifruit juice and the impact of the OD process on the product quality in terms of ascorbic acid content and total antioxidant activity (TAA). The clarified kiwifruit juice, with an initial TSS content of 9.4°Brix, was concentrated up to a final value of 66°Brix by using a laboratory bench plant equipped with an OD membrane module (Liqui-Cell Extra-Flow 2.5 × 8″ membrane contactor, Membrana, Charlotte, NC, USA) realized with PP hollow fiber membranes. A 60 w/w% calcium chloride dihydrate was used as a stripping solution. The clarified juice was pumped through the shell side of the membrane module, while the stripping solution was pumped through the fiber lumens (tube side). The analytical measurements showed that the OD process has no influence on the vitamin C content, independently on the concentration degree achieved, while in the juice concentrated by thermal evaporation at 66.6°Brix, a reduction of 87% of vitamin C was observed in comparison to the clarified juice. Also, the total antioxidant activity (TAA) of the clarified juice was preserved during the OD process, while a 50% reduction of TAA was measured in the thermally concentrated juice. The efficiency of the OD process in maintaining the nutritional quality of the fresh juice was also observed in the concentration of passion fruit [48], camu-camu [51], noni [52], orange [53, 54] and apple [55] juices. By referring to the sensorial quality of the juice, some losses of aroma compounds were observed during the initial hours of concentration of clarified orange juice by OD. However, these losses can be drastically limited by preconditioning the OD membrane with the clarified juice from the clarification unit and by avoiding thermal regeneration of brine during concentration [53]. The athermal concentration of pasteurized pineapple juice by OD produced a 51°Brix concentrate, retaining an average of 62% of the volatile compounds present in the initial juice [56]. A higher retention of citral and ethyl butyrate, two aroma compounds relevant in the orange juice aroma, was observed during the OD process of the juice when compared to the MD process [57]. A strict correlation between the degree of retention of organic volatile flavor components and membrane surface pore size was demonstrated by Barbe et al. [58]. In particular, membranes with relatively large sizes at the surface showed higher organic volatiles retention per unit water removal than those with smaller surface openings. This phenomenon was attributed to a greater intrusion of the liquid feed and brine streams with a resulting increase in the thickness and resistance of the boundary layer entrance of larger pores.

3.3.4 Membrane distillation Membrane distillation is an emerging and promising technology that is attracting interest for various applications in the food industry. Similarly to OD, in MD two aqueous solutions at different temperatures are separated by a macroporous hydrophobic membrane. Due to the hydrophobicity of the membrane material, the liquid

70 

 3 Integrated membrane operations in fruit juice processing

water cannot enter the pores and a liquid interface is formed on either side of the membrane pores [59]. In these conditions a net pure water flux occurs from the warm side to the cold side. The process takes place at atmospheric pressure and at a temperature that may be much lower than the boiling point of the solutions. The driving force is the vapor pressure difference between the two solution-membrane interfaces because of the existing temperature gradient. The phenomenon can be described as a three phase sequence: (i) formation of a vapor gap at the warm solution-membrane interface; (ii) transport of the vapor phase through the macroporous system; (iii) its condensation at the cold side membrane-solution interface [60]. Depending on the mechanism exploited to obtain the required driving force, MD processes can be divided into four different categories: (i) direct contact membrane distillation (DCMD); (ii) air gap membrane distillation (AGMD); (iii) sweep gas membrane distillation (SGMD); (iv) vacuum membrane distillation (VMD). – In DCMD, water with a lower temperature than liquid in the feed side is used as a condensing fluid in the permeate side. In this configuration, the liquid on both sides of the membrane is in direct contact with the hydrophobic macroporous membrane [61]. – In AGMD, water vapor is condensed on a cold surface that is separated from the membrane via an air gap. The heat losses are reduced in this configuration by addition of a stagnant air gap between membrane and condensation surface. – In SGMD, a cold inert gas is used in the permeate side for sweeping and carrying the vapor molecules outside the membrane module where the condensation takes place [62, 63]. – In VMD, the driving force is maintained by applying vacuum at the permeate side. The applied vacuum pressure is lower than the equilibrium vapor pressure. Therefore, condensation takes place outside the membrane module [64]. The MD process is proposed as an innovative technology for the concentration of fruit juice, allowing the drawbacks of conventional thermal evaporation to be overcome. In particular, DCMD offers some key advantages due to its suitability for applications in which the volatile component is water [65]. The advantages of MD compared to other traditional technologies for the concentration of fruit juice are: high-quality of concentrates; low operating temperatures (24–48°C); the possibility to achieve high contents of dry substances (60–70%) and low energy costs. Theoretical and experimental studies on MD in the concentration of orange juice were performed by Calabrò et al. [66] using a commercial plate PVDF membrane with a nominal pore size of 0.1 μm made by Millipore Corp. (Billerica, MA, USA). The used membrane showed a very good retention of orange juice compounds, such as soluble solids, sugars and organic acids, with a 100% rejection of sugars and organic acids. An increase in permeate flux was observed by pretreating the juice by UF. A concentrated apple juice with a TSS content of 64°Brix was produced using a PP hollow fiber module (Microdyn ENKA MD-020 –2N-CP) with tube and shell configuration [67]. Transmembrane fluxes were of approximately 1 kg/m2h; which

3.4 Integrated membrane operations in fruit juices production 

 71

was increased significantly (up to 20%) by thermal OD using CaCl2 as a stripping solution. Temperature polarization was found to be more important than concentration polarization and was located mainly on the feed side. In addition, the presence of osmotic effect on the juice stream generated flux inversion. However a small ΔT of 4°C was sufficient to compensate this osmotic effect. Higher evaporation fluxes were obtained in the concentration of ultrafiltered and depectinised apple juice using a PVDF membrane with a mean pore size of 0.45 mm and a porosity of 80–85% (MKKK-3 type, NPO Polymersyntes, Russian Federation) [68]. Permeate fluxes were of the order of 9 l/m2h at a TSS content of 50°Brix. Further concentration of the juice up to 60–65°Brix resulted in a reduced productivity (3.8–3.0 l/m2h). The influence of the apple juice pretreatment before concentration by MD was also studied by Lukanin et  al. [69]. In particular, after a fermentation process, the apple juice was submitted to an enzymatic treatment with protease, followed by a clarification step by UF. The pretreatment method permitted a removal of biopolymers and proteins, which potentially play a detrimental role in the MD process. Results showed that an increase in the biopolymer removal enhances the transmembrane flux because of a reduction in juice viscosity. VMD was studied for the recovery of volatile aroma compounds from blackcurrant [70] and pear juice [71]. Aroma enrichment factors up to 15 were experimentally obtained.

3.4 Integrated membrane operations in fruit juices production Several studies have confirmed the efficiency of the membrane technology in substituting conventional unit operations involved in different steps of the fruit juices production (such as stabilization, clarification, fractionation, concentration and recovery of aroma compounds). Today, the possibility to redesign the industrial transformation cycle of the fruit juice production through the introduction of membrane technologies appears a valid approach for a sustainable industrial growth within the PI strategy. The aim of this strategy is to introduce into the productive cycles new technologies characterized by low hindrance volume, advanced levels of automation capacity, modularity, remote control, reduced energy consumption or waste production [72]. In the following section some examples of integrated membrane systems in the treatment of apple, red fruits and other juices are reviewed and discussed.

3.4.1 Apple juice Apple juice is a popular beverage worldwide, and is perceived as a wholesome and nutritious product. Overall quality of apple juice is an important factor to consider in processing, because some attributes, such as aroma, color and flavor, are greatly appreciated by the final consumer, and are associated with freshness and authenticity.

72 

 3 Integrated membrane operations in fruit juice processing

Apple juice has been traditionally pasteurized by thermal means using continuous pasteurization, which may be carried out by passage through plate heat exchangers, and by tunnel pasteurizers. Thermal processing inactivates spoiling microorganisms efficiently, but may also degrade taste, color, flavor and nutritional quality of the juice: consequently, conventional apple juice production results in a juice depleted in important bioactive compounds, such as flavonoids, and TAA [73, 74]. Different alternative processes have been proposed in recent years for the treatment of apple juice and for the production of apple juice with the properties of the initial fresh fruits. A process design related to the clarification and concentration of apple juice based on the use of membrane and conventional separation systems was proposed by Alvarez et al. [75]. The process involves the clarification of raw apple juice using an enzymatic membrane reactor, the preconcentration of the clarified juice up to 25°Brix by RO, the aroma compounds recovery and concentration by pervaporation (PV) and a final concentration up to 72°Brix using conventional evaporation (Figure 3.3). An economic evaluation of the integrated membrane system indicated a reduction of the total capital investment of 14% and an increasing in process yield of 5% when compared with the conventional process. Total manufacturing costs decreased by 8% because less energy was required to concentrate the juice. Membrane replacement accounted only for 2% of operating costs and membrane life was estimated to be 2, 3 and 2 years for UF, RO and PV membranes, respectively. An integrated process for the production of high-quality apple juice concentrate was proposed by Aguiar et al. [76]. The enzymatically treated juice was clarified by MF, preconcentrated by RO (up to 29°Brix) and then concentrated by OD (up to 53°Brix). Analytical results in the different fraction of the integrated membrane process, showed a 18% reduction of phenolic compounds and a loss of more volatile compounds during the concentration step. However, sensory evaluations showed that the reconstituted concentrated juice was excellent in terms of odor and flavor, with high acceptance percentages by consumers. Recently, Onsekizoglu et al. [77] evaluated the impact on the product quality of different integrated membrane processes for the clarification and concentration of apple juice. The fresh apple juice, with an initial TSS content of 12°Brix, was previously clarified by a combined application of fining agents (gelatine and bentonite) and UF. The clarified juice was then concentrated using different membrane processes such as MD, OD, and a coupling of MD and OD and by conventional thermal evaporation up to 65°Brix. The different membrane-based concentration techniques were very efficient, as the concentrated juice presented nutritional and sensorial qualities similar to that of the original juice, especially regarding the retention of the bright color and pleasant aroma, which are lost during thermal evaporation. Furthermore, among all the concentration treatments applied, only thermally evaporated samples resulted in the formation of 5-hydroximethilfurfural, an indicator of the Maillard reaction. The content of phenolic and organic acids and sugars was preserved during the different

PV, 20 °C

Milling Enzymes

Aroma concentrate

RO, 20–25°C

Preconcentrated juice 25 °Brix

Depectinization tank

Condenser Vacuum pump

Pressing

Water (apples washing, etc.)

UF, 20°C

Clarified juice

Figure 3.3: Integrated membrane process for the production of apple juice concentrate and apple juice aroma (adapted from Alvarez et al. [75])

MEE 60 – 80 °C

Concentrated juice 70 –72 °Brix

Apples

3.4 Integrated membrane operations in fruit juices production 

 73

74 

 3 Integrated membrane operations in fruit juice processing

concentration processes while the aroma profile, studied in terms of trans-2-exenal, was remarkably lost in apple juice concentrates produced by thermal evaporation. A higher retention of trans-2-exenal was observed in the OD process in comparison to MD. The coupled operations of MD and OD retained the aroma profile of the initial juice and so were seen as the most promising alternative to the conventional thermal evaporation technique.

3.4.2 Red fruit juices Red fruits are among the most important dietary source of polyphenols, such as anthocianins, flavonols, flavan-3-ols, benzoic and hydroxicinnamic acid derivates. Numerous in vitro studies have reported their high antiradical activity and capacity to inhibit the human low-density lipoprotein and liposome oxidation. Biochemical and pharmacological activities have been attributed to free radical scavenging, effects on immune and inflammatory cell functions, anti-carcinogenic and antitumor properties [78]. Within the group of red fruits, pomegranate (Punica granatum), redcurrant (Ribes rubrum L.), blood orange (Citrus sinensis L.), blackcurrant (Ribes nigrum L.) and cherry (Prunus avium L.) juices are among the richest in anthocyanins, which are responsible of the bright red color and the strong antioxidant capacity, gaining huge interest as ingredients in the design of functional juices. In order to better preserve the properties of red fresh fruits, several new “mild” technological processes have been proposed in the last years. A multi-step membrane process on a laboratory- and large-scale was proposed by Kozak et al. [79, 80] for the treatment of blackcurrant juice. The integrated system consisted of a MF step to clarify the juice, a RO unit to pre-concentrate the juice up to 26°Brix and a final OD process to concentrate the juice up to 63–72°Brix. Experiments were performed, at first, at a laboratory scale to determine the optimal operating parameters. The large-scale measurements were carried out on the basis of the laboratory results. In large-scale experiments the depectinised juice was prefiltered through a 100 μm bag filter and then preconcentrated through a RO flat-sheet membrane module (MFTKöln) at an operating pressure of 51 bar and a temperature of 24°C. The concentration was carried out by using a PP hollow fiber membrane module (MD 150 CS 2N, Microdyn) with an average pore size of 0.2 μm. In the concentrated juice the anthocyanin content was three times higher than the raw juice. The sensory analysis showed a little loss of aroma compounds in the reconstituted juice when compared to the raw juice, while the color intensity and the acidic flavor intensity remained unchanged. An interesting process design based on the use of integrated membrane systems for the blackcurrant juice concentration was proposed by Sotoft et al. [81] to replace traditional multiple step evaporators and aroma recovery. The processes consisted of a VMD unit for aroma recovery and water removal by a combination of NF, RO and DCMD. In particular, a preliminary concentration of the raw juice up to 45°Brix is

3.4 Integrated membrane operations in fruit juices production 

 75

obtained using a combined NF/RO process: combining these two techniques it was possible to utilize the high rejection of RO membranes and the high concentration factor of NF membranes in order to overcome the high osmotic pressure limitations typically encountered in RO. The raw juice was first treated by RO with a dense membrane that had a high degree of sugar retention (99.7%). The microbial quality of the water in the permeate stream was very high, suggesting a potential reuse as a source of drinking water or process water in food production. The RO retentate was processed by NF. The NF permeate was recirculated back to the RO unit, while the retentate stream was submitted to the final concentration DCMD step, producing a concentrated juice with a TSS content of 65–70° Brix (Figure 3.4). The production of the proposed system was fixed at 17,283 ton of 66°Brix concentrated juice/year with a production price of 0.40 €/kg assuming a membrane lifetime of 1 year. The estimated operation cost is lower than the price of a traditional process by about 43%: therefore the economical potential of the process is very promising in order to replace conventional evaporators. Galaverna et al. [82] investigated two membrane-based configurations for the production of highly concentrated blood orange juices. The process included an initial clarification of the freshly squeezed juice by UF, in order to separate the liquid serum from the pulp. The clarified juice was successively concentrated using two consecutive processes: (i) first RO, as a preconcentration step up to 25–30°Brix, (ii) then OD to obtain a final concentration of 60°Brix. Alternatively, the clarified juice was directly concentrated by OD up to 60°Brix. The proposed system was very efficient in preserving the TAA of the juice, even at high concentrations (60°Brix). Among the different antioxidant components a slight decrease in the OD retentate was observed for ascorbic acid (15%) and anthocyanins (23%), whereas flavanones and hydroxycinnamic acids were very stable. The final TAA value obtained in both configurations was not significantly different from that observed in the traditional thermal treatment. The concentrated juice retained its bright red color and its pleasant aroma, which was, on the contrary, completely lost during thermal concentration. A similar UF/OD integrated membrane process was proposed and investigated for the clarification and concentration of pomegranate juice [83]. The raw juice, with an initial TSS content of 16.2°Brix, was clarified by hollow fiber UF membranes and then concentrated by using a Liqui-Cell Extra-Flow 2.5 × 8″ membrane contactor (Membrana, Charlotte, NC, USA) up to a TSS value of 52°Brix. The analytical measurement performed on clarified and concentrated samples showed that sugars, organic acids (malic, ascorbic and citric acids), polyphenols and anthocyanins were well-preserved during the process independently on the TSS content (Table 3.1). The evaluation of the TAA in the OD samples confirmed the validity of the proposed process in preserving juice bioactive compounds. In particular, the concentrated juice at 52°Brix showed only a 10% reduction of the TAA when compared with the clarified juice. The integrated process for the clarification and concentration of pomegranate juice is depicted in Figure 3.5.

1

Permeate

Permeate in

Aroma concentrate

Concentrated juice

Distillation

VMD

VMD retentate

Permeate out

DCMD

4

NF

RO permeate

RO

3

2

Figure 3.4: Integrated membrane process for aroma recovery and blackcurrant juice concentration. (1) aroma recovery by vacuum membrane distillation (VMD) followed by distillation; (2) preconcentration by reverse osmosis (RO); (3) preconcentration by nanofiltration (NF); (4) final concentration by direct contact membrane distillation (DCMD) (adapted from Sotoft et al. [81])

Vacuum

Condenser

Filtered macerated juice

76   3 Integrated membrane operations in fruit juice processing

3.4 Integrated membrane operations in fruit juices production 

 77

Table 3.1: General composition of pomegranate juice clarified and concentrated by integrated membrane process

Suspended solids, %(w/w) pH Total acidity (g/l) Total soluble solids (°Brix) Total antioxidant activity (mM Trolox) Ascorbic acid (mg/l) Malic acid (g/l) Citric acid, (g/l) Total polyphenols (g catechin/l) Total anthocyanins (mg/l)

Fresh juice

Permeate UF

4.8 3.75 0.41 16.2 12.9

0 3.78 0.35 16.2 10.6

5.3 3.74 0.44 – 14.1

52.0 10.1*

68.0 1.90 1.47 1.57

47.0 1.82 1.45 1.31

71.0 2.01 1.24 1.70

44.0* 1.80* 1.26* 1.22*

102.8

Retentate UF Retentate OD

90.7

100.6

– – –

75.85*

UF, ultrafiltration; OD, osmotic distillation *value referred to a TSS content of 16.2°Brix.

Pomegranate fruits Arils juice Washing

Peeling

Incubation with pectinase (1%, room temperature, 4 h) Sieving (200 mm)

Arils

Ultrafiltration Permeate Retentate Squeezing

Osmotic distillation

Seeds

Concentrated juice

Pressing

Washing

Figure 3.5: Integrated membrane process for the production of pomegranate juice concentrate

78 

 3 Integrated membrane operations in fruit juice processing

The possibility to concentrate different red fruits juices, by using a coupled operation of MD and OD, referred to as “membrane osmotic distillation” (MOD), was evaluated by Koroknai et al. [84]. In the proposed system, three different red fruits juices (chokeberry, redcurrant and cherry), were firstly clarified by UF and then concentrated by MOD. The clarification step improved the efficiency of the MOD process, providing a less viscous feed stream with significantly lower fouling behavior during the concentration, at the same time excluding the possibility of microbiological contamination in the further concentration process. During the concentration process, the integration of OD and MD processes permitted an increase of the driving force, which resulted in enhanced water flux during the operation. The resulting process was more effective than MD or OD alone [85]. The obtained evaporation fluxes were in the range of 4.51–5 kg/m2h for all the investigated juices. An excellent preservation of the valuable antioxidant capacity ( > 97%) was observed, indicating the validity of the process in preserving the original nutritional value of the fresh fruits.

3.4.3 Other fruit juices 3.4.3.1 Kiwifruit juice Kiwifruit (Actinidia sp.) is one of the most nutrient-dense fruits and is characterized by significant amounts of biologically active compounds, including ascorbic acid [86, 87]. In particular, it contains more ascorbic acid than the average amounts found in fruits such as grapefruit, oranges, strawberries and lemons. It also has an impressive antioxidant capacity, containing a wealth of phytonutrients, including carotenoids, lutein, phenolics, flavonoids and chlorophyll. Conventional processing of kiwifruit into juice is affected by a number of factors, including excessive browning, formation of haze/precipitates and flavor change [88]. Membrane-based separation technologies have been studied and proposed as a valid method for the preservation of nutritional properties of the kiwifruit juice. An integrated membrane process for the production of high-quality kiwifruit juices was proposed by Cassano et al. [89]. The fresh depectinised kiwifruit juice was clarified by UF. After the clarification step, the kiwifruit juice, with an initial TSS of 12.5°Brix, was concentrated by OD up to a value higher than 60°Brix. The effect of UF and OD processes on the TAA and other analytical parameters of the kiwifruit juice was also studied. During the integrated membrane process, the vitamin C content was very well-preserved and only a small reduction of TAA in the concentrated juice was measured. The possibility to introduce a PV step devoted to the recovery of aroma compounds into the integrated system was also investigated [90]. For most part of aroma compounds, the enrichment factor in the permeate of the fresh juice was higher than in both the clarified and concentrated juice, suggesting the use of PV

3.4 Integrated membrane operations in fruit juices production 

 79

Clarified juice (10 –11 °Brix)

Pulp

UF Fruit juice (10 –11 °Brix)

Diluted brine

Concentrated brine

OD

PV Aromatic compounds

Concentrated juice (64–65 °Brix)

Figure 3.6: Integrated membrane process for the production of kiwifruit juice concentrate (UF, ultrafiltration; PV, pervaporation)

directly on the fresh juice before the clarification step. The recovered aroma can be added to the final concentrated juice for the production of beverages with high nutritional values (Figure 3.6).

3.4.3.2 Cactus pear juice Cactus pear juice has attracted great attention in recent years for its nutraceutical and functional importance. Concentrated juices (up to 63–67°Brix) can be obtained by centrifuge vacuum evaporator at approximately 40°C. However, the thermal treatment determines a color damage and a herbaceous aroma appears after the concentration process [91]. An integrated membrane process for the production of concentrated cactus pear juice was proposed by Cassano et al. [92]. This was based on a preliminary clarification of the depectinised juice by UF followed by a concentration of the UF permeate by OD (Figure 3.7). The final retentate of the OD process, with a TSS concentration of 58°Brix, presented a content of betalains similar to that of the fresh juice. A good preservation of vitamin C and TAA was also evaluated (Table 3.2 ). The retentate of the OD process was considered a good source of antioxidants with potential applications for food and nutritional supplements. The high betalain concentration achieved in the OD retentate (227 mg/l for betaxanthins and 54 mg/l for betacyanins) was

80 

 3 Integrated membrane operations in fruit juice processing

Cactus pear fruits Washing Peeling Peel

Peeled fruits

Scrap

Squeezing

Washing

Puree

Washing liquid Mixing

Depectinization Sieving (200 mm)

Retentate

Cactus pear juice UF

Clarified juice

Concentrated juice

OD

Diluted brine

Concentrated brine

Figure 3.7: Integrated membrane process for the production of cactus pear juice concentrate (UF, ultrafiltration; OD, osmotic distillation) Table 3.2: General composition of cactus pear juice clarified and concentrated by integrated membrane process Fresh juice Betaxanthins (mg/l) Betacyanins (mg/l) Total soluble solids (°Brix) Total antioxidant activity (mM Trolox) Ascorbic acid (mg/l) Citric acid (g/l) Glutammic acid, (g/l)

Permeate UF

Retentate UF

Retentate OD

57.1 19.9 13.4

53.4 12.7 13.0

61.6 32.8 14.1

52.5* 12.4* 58.0

5.0

4.8

4.9

4.6*

39.3 416.0 2.06

37.3 395.0 2.05

43.0 427.4 1.95

36.0* 375.0* 2.05*

UF, ultrafiltration; OD, osmotic distillation. *value referred to a TSS content of 13.0°Brix.

considered to be a good source of natural colorants with potential applications in the food industry.

3.5 Conclusions 

 81

Melons Hand-peeling Cutting and seeds removal

Retentate

Milling/grinding MF

Clarified juice

OD

Concentrated juice

Finishing (1.6 mm) Fresh juice

Diluted brine

Concentrated brine

Enzymatic treatment Figure 3.8: Integrated membrane process for the production of melon juice concentrate (MF, microfiltration; OD osmotic distillation) (adapted from Vaillant et al. [93])

3.4.3.3 Melon juice The effects of a combined MF-OD process on the physico-chemical, nutritional and microbiological qualities of melon juice were evaluated by Vaillant et  al. [93]. The juice was clarified by a ceramic multichannel membrane (Membralox® 1P10-40, Pall Exekia, Bazet, France) with an average pore diameter of 0.2 μm and then concentrated by OD by using a module containing PP hollow fibers (10 m2 membrane surface). Calcium chloride, used as a stripping solution, was circulated outside the fibers. The clarified juice presented physico-chemical and nutritional properties similar to those of the fresh melon juice, except for the absence of suspended solids and carotenoids, which remained totally concentrated in the MF retentate. In particular, β-carotene was retained by the MF membrane, probably because it is strongly associated with the membrane and wall structures of the cell fragments. Microbiological analyses showed that MF can ensure microbiological stability of the juice in a single step. The concentrated juice at 55°Brix preserved the main physico-chemical and nutritional properties of the fresh juice. The integrated membrane process (Figure 3.8) was proposed as an innovative way of treating melon juice, as it allowed high-value products to be obtained from fruits discarded by the fresh market.

3.5 Conclusions The potential advantages of membranes operations over conventional methodologies in fruit juice clarification and concentration have been successfully demonstrated.

82 

 3 Integrated membrane operations in fruit juice processing

In addition technological inputs and economical benefits can be achieved through the integration of different unit membrane operations between themselves or with other conventional technologies. In this view, the development of hybrid processes can give a significant contribution to redesign traditional flow sheets in fruit juice processing with consequent advantages in terms of the improvement of final products, recovery of by-products and high added-value compounds, control of environmental impact, and a reduction of energy and water consumption. The possibility to develop integrated membrane operations in agro-food productions, as well as in other industrial areas, can be considered a valid approach for a sustainable industrial growth within the PI strategy.

3.6 References 1. Jing L, Howard AC. Applications of membrane techniques for purification of natural products. Biotechnol Lett 2010;32:601–608. 2. Jiao B, Cassano A, Drioli E. Recent advances on membrane processes for the concentration of fruit juices: a review. J Food Eng 2004;63:303–324. 3. Drioli E, Romano M. Progress and new perspectives of integrated membrane operations for sustainable industrial growth. Ind Eng Chem Res 2001;40:1277–1300. 4. Drioli E, Fontananova E. Membrane technology and sustainable growth. Chem Eng Res Des 2004;82:1557–1562. 5. Vaillant F, Millan A, Dornier M, Decloux M, Reynes M. Strategy for economical optimization of the clarification of pulpy fruit juices using crossflow microfiltration. J Food Eng 2001;48:83–90. 6. Meyer AS, Koser C, Adler-Nissen J. Efficiency of enzymatic and other alternative clarification and finig treatments on turbidity and haze in cherry juice. J Agr Food Chem 2001;49:3644–3650. 7. Echavarria AP, Torras C, Pagan J, Ibarz A. Fruit juice processing and membrane technology application. Food Eng Rev 2011;3:136–158. 8. Nandi BK, Das B, Uppaluri R, Purkait MK. Microfiltration of mosambi juice using low cost ceramic membrane. J Food Eng 2009;95:597–605. 9. Alvarez V, Andres LJ, Riera FA, Alvarez R. Microfiltration of apple juice using inorganic membranes: process optimization and juice stability. Can J Chem Eng 1996;74:156–162. 10. Cassano A, Donato L, Drioli, E. Ultrafiltration of kiwifruit juice. Operating parameters, juice quality and membrane fouling. J Food Eng 2007;79:613–621. 11. Cassano A, Tasselli F, Conidi C, Drioli E. Ultrafiltration of clementine mandarine juice by hollow fiber membranes. Desalination 2009;241:302–308. 12. Laorko A, Li Z, Tongchitpkdee S, Chantachum S, Youravong W. Effect of membrane property and operating conditions on phytochemical properties and permeate flux during clarification of pineapple juice. J Food Eng 2010;100:514–521. 13. Carneiro L, Sa IDS, Gomes FDS, Matta VM, Cabral LMC. Cold sterilization and clarification of pineapple juice by tangential microfiltration. Desalination 2002;148:93–98. 14. Carvalho LMJ, Castro IM, Silva CAV. A study of retention of sugars in the process of clarified pineapple juice (Ananas comosus, L. Merril) by micro- and ultra-filtration. J Food Eng 2008;87: 447–454. 15. Vaillant V, Cisse M, Chaverri M, Perez A, Dornier M, Viquez F, Claudie Dhuique-Mayer C. Clarification and concentration of melon juice using membrane processes. Innov Food Sci Emerg 2005;6:213–220.

3.6 References 

 83

16. Mirsaeedghazi H, Eman-Djomeh Z, Mohammad Mousavi S, Aroujalian A, Navidbakhsh M. Clarification of pomegranate juice by microfiltration with PVDF membranes. Desalination 2010;264:243–248. 17. Cassano A, Donato L, Conidi C, Drioli E. Recovery of bioactive compounds in kiwifruit juice by ultrafiltration. Innov Food Sci Emerg 2008;9:556–562. 18. Alvarez S, Alvarez R, Riera FA, Coca J. Influence of depectinization on apple juice ultrafiltration. Colloid Surface A 1998;138:377–382. 19. Rai P, Majumdar GC, Das Gupta S, De S. Effect of various pretreatment methods on permeate flux and quality during ultrafiltration of mosambi juice. J Food Eng 2007;78:561–568. 20. Rai P, Majumdar GC, Das Gupta S, De S. Quantification of flux decline of depectinized mosambi (Citrus sinensis (L.) Osbeck) juice using unstirred batch ultrafiltration. J Food Process Eng 2005;28:359–377. 21. Jiraratananon R, Chanachai A. A study of fouling in the ultrafiltration of passion fruit juice. J Membrane Sci 1996;111:39–48. 22. Tasselli A, Cassano A, Drioli E. Ultrafiltration of kiwifruit juice using modified poly(ether ether ketone) hollow fiber membranes. Sep Purif Technol 2007;57:94–102. 23. Constela DT, Lozano JE. Hollow fiber ultrafiltration of apple juice: macroscopic approach. LWT-Food Sci Technol 1997;30:373–378. 24. Cassano A, Marchio M, Drioli E. Clarification of blood orange juice by ultrafiltration: analyses of operating parameters, membrane fouling and juice quality. Desalination 2007;212: 15–27. 25. Field RW, Wu D, Howell JA, Gupta BB. Critical flux concept for microfiltration fouling. J Membrane Sci 1995;100:259–272. 26. Oliveira RC, Docê RC, Barros STD. Clarification of passion fruit juice by microfiltration: analyses of operating parameters, study of membrane fouling and juice quality. J Food Eng 2012;111: 432–439. 27. Barros STD, Andrade CMG, Mendes ES, Peres L. Study of fouling mechanism in pineapple juice clarification by ultrafiltration. J Membrane Sci 2003;215:213–224. 28. Barbe AM, Bartley JP, Jacobs AL, Johnson RA. Retention of volatile organic flavor/fragrance components in the concentration of liquid foods by osmotic distillation. J Membrane Sci 1998;145:67–75. 29. Aider M, de Halleux D. Cryoconcentration technology in the bio-food industry: Principles and applications. LWT-Food Sci Technol 2009;42:679–685. 30. Jariel O, Reynes M, Courel M, Durand N, Dornier M, Deblay P. Comparison of some fruit juice concentration techniques. Fruits 1996;51:437–450. 31. Köseoglu SS, Lawhon, JT, Lusas EW. Use of membranes in citrus juice processing. Food Technol 1990;44:90–97. 32. Conidi C, Cassano A, Drioli E. Recovery of phenolic compounds from orange press liquor by nanofiltration. Food Bioprod Process 2012;90:867–874. 33. Massot A, Mietton-Peuchot M, Peuchot C, Milisic V. Nanofiltration and reverse osmosis in winemaking. Desalination 2008;231:283–289. 34. Daufin G, Escudier JP, Carrère H, Bérot S, Fillaudeau L, Decloux M. Recent and emerging applications of membrane processes in the food and dairy industry. Trans IChemE 2001;79: 89–102. 35. Noronha M, Britz T, Mavrov V, Jarnke HD, Chmiel H. Treatment of spent process water from a fruit juice company for purposes of reuse: hybrid process concept and on-site test operation of a pilot plant. Desalination 2002;143:183–196. 36. Warczok J, Ferrando M, Lopez F, Guell C. Concentration of apple and pear juices by nanofiltration at low pressure. J Food Eng 2004;63:63–70.

84 

 3 Integrated membrane operations in fruit juice processing

37. Bánvölgyi S, Horváth S, Békássy-Molnár E,Vatai G. Concentration of blackcurrant (Ribes nigrum L.) juice with nanofiltration. Desalination 2006;200:535–536. 38. Conidi C, Cassano A, Drioli E. A membrane-based study for the recovery of polyphenols from bergamot juice. J Membrane Sci 2011;375:182–190. 39. Alvarez V, Alvarez S, Riera FA, Alvarez R. Permeate flux prediction in apple juice concentration by reverse osmosis, J Membrane Sci 1997;127;25–34. 40. Medina BG, Garcia A. Concentration of orange juice by reverse osmosis. J Food Process Eng 1998;10:217–230. 41. Matta M, Moretti R, Cabral L. Microfiltration and reverse osmosis for clarification and concentration of acerola juice. J Food Eng 2004;61:477–482. 42. Mengual JI, Zárate JMO, Peña L, Velázques A. Osmotic distillation through porous hydrophobic membranes. J Membrane Sci 1983;82:129–140. 43. Kostantinos BP, Harris NL. Osmotic concentration of liquid foods. J Food Eng 2001;49:201–206. 44. Kunz W, Benhabiles A, Ben-Aim R. Osmotic evaporation through macroporous hydrophobic membranes: a survey of current research and applications. J Membrane Sci 1996;121:25–36. 45. Hogan PA, Canning RP, Peterson PA, Johnson RA, Michaels AS. A new option: osmotic distillation. Chem Eng Prog 1998;94:49–61. 46. Ravindra Babu B, Rastogi NK, Raghavarao KSMS. Concentration and temperature polarization effects during osmotic membrane distillation. J Membrane Sci 2008;322:146–153. 47. Courel M, Dornier M, Henry JM, Rios GM, Reynes M. Effect of operating conditions on water transport during the concentration of sucrose solutions by osmotic distillation. J Membrane Sci 2000;170:281–289. 48. Vaillant F, Jeanton E, Dornier M, O’Brien GM, Reynes M, Decloux M. Concentration of passion fruit juice on industrial pilot scale using osmotic evaporation. J Food Eng 2001;47:195–202. 49. Hongvaleerat C, Cabral LMC, Dornier M, Reynes M, Ningsanond S. Concentration of pineapple juice by osmotic evaporation. J Food Eng 2008;88:548–552. 50. Cassano A, Drioli E. Concentration of clarified kiwifruit juice by osmotic distillation. J Food Eng 2007;79:1397–1404. 51. Rodrigues RB, Menez HC, Cabral LMC, Dornier M, Rios GM, Reynes M. Evaluation of reverse osmosis and osmotic evaporation to concentrate camu-camu juice (Myciaria dubia). J Food Eng 2004;63:97–102. 52. Valdés H, Romero J, Saavedra A, Plaza A, Bubnovich V. Concentration of noni juice by means of osmotic distillation. J Membrane Sci 2009;330:205–213. 53. Cissé M, Vaillant F, Perez A, Dornier M, Reynes M. The quality of orange juice by coupling crossflow microfiltration and osmotic evaporation. Int J Food Sci Tech 2005;40:105–116. 54. Cassano A, Drioli E, Galaverna G, Marchelli R, Di Silvestro G, Cagnasso P. Clarification and concentration of citrus and carrot juices by integrated membrane processes, J Food Eng 2003;57:153–163. 55. Cissé M, Vaillant F, Bouquet S, Pallet D, Lutin F, Reynes M, Dornier M. Athermal concentration by osmotic evaporation of roselle extract, apple and grape juices and impact on quality. Innov Food Sci Emerg 2011;12:352–360. 56. Shaw PE, Lebrun M, Ducamp MN, Jordan MJ, Goodner KL. Pineapple juice concentrated by osmotic evaporation. J Food Quality 2002;25:39–49. 57. Alves VD, Coelhoso IM. Orange juice concentration by osmotic evaporation and membrane distillation: a comparative study. J Food Eng 2006;74:125–133. 58. Barbe AM, Bartley JP, Jacobs AL, Johnson RA. Retention of volatile organic flavor/fragrance components in the concentration of liquid foods by osmotic distillation. J Membrane Sci 1998;145:67–75. 59. Schofiel RW, Fane AG, Fell CGD. Heat and mass transport in membrane distillation. J Membrane Sci 1987;33:299–313.

3.6 References 

 85

60. Khayet M. Membrane and theoretical modeling of membrane distillation: A review. Adv Colloid Interfac 2011;164:56–88. 61. Khayet M, Mengual JI, Matsura T. Porous hydrophobic/hydrophilic composite membranes. Application in desalination using direct contact membrane distillation. J Membrane Sci 2005;252:101–113. 62. Khayet M, Godino P. Mengual JI. Nature of flow on sweeping gas membrane distillation. J Membrane Sci 2000;170:243–255. 63. Khayet M, Mengual JI, Zakrewska-Trznadel G. Theoretical and experimental studies on desalination using the sweeping gas membrane distillation. Desalination 2003;157:297–305. 64. Bandini S, Gostoli C, Sarti C. Separation efficiency in vacuum membrane distillation. J Membrane Sci 1992;73:217–229. 65. Lawson KW, Loyd DR. Membrane distillation II. Direct contact MD. J Membrane Sci 1996;120: 123–133. 66. Calabrò V, Jiao B, Drioli E. Theoretical and experimental study on membrane distillation in the concentration of orange juice. Ind Eng Chem Res 1994;33:1803–1808. 67. Laganà F, Barbieri G, Drioli E. Direct contact membrane distillation: modelling and concentration experiments. J Membrane Sci 2000;166:1–11. 68. Gunko S, Verbych S, Bryk M, Hilal N. Concentration of apple juice using direct contact membrane distillation. Desalination 2006;190:17–124. 69. Lukanin OE, Gunko SM, Bryt MT, Nigmatullin RR. The effect of content of apple juice byopolymers on the concentration by membrane distillation. J Food Eng 2003;60:275–280. 70. Bagger-Jørgensen R, Meyer AS, Varming C, Jonsson G. Recovery of volatile aroma compounds from black currant juice by vacuum membrane distillation. J Food Eng 2004;64:23–31. 71. Diban N, Voinea OC, Urtiaga A, Ortiz I. Vacuum membrane distillation of the main pear aroma compound: experimental study and mass transfer modelling. J Membrane Sci 2009;326:64–75. 72. Stankiewicz AI, Moulijn AJ. Process intensification: transforming chemical engineering. Chem Eng Prog 2000;96:22–33. 73. Charles-Rodríguez AV, Nevárez-Moorillón GV, Zhang QH, Ortega-Rivas E. Comparison of thermal processing and pulsed electric fields treatment in pasteurization of apple juice. IChemE 2007;85:93–97. 74. Van Der Sluis AA, Dekke M, Skrede G, Jongen WMF. Activity and concentration of polyphenolic antioxidants in apple juice. 2. Effect of novel production methods. J Agric Food Chem 2004;52:2840–2848. 75. Alvarez S, Riera FA, Alvarez R et al. A new integrated membrane process for producing clarified apple juice and apple juice aroma concentrate. J Food Eng 2000;46:109–125. 76. Aguiar BI, Miranda NGM, Gomes FS et al. Physicochemical and sensory properties of apple juice concentrated by reverse osmosis and osmotic evaporation. Innov Food Sci Emerg 2012;16: 137–142. 77. Onsekizoglu P, Savas Bahceci K, Jale Acar M. Clarification and concentration of apple juice using membrane processes: A comparative quality assessment. J Membrane Sci 2010;352:160–165. 78. Bermudez-Soto MJ, Tomas-Barberan FA. Evaluetation of commercial red fruit juice concentrates as ingredients for antioxidant functional juices. Eur Food Res Technol 2004;219:133–141. 79. Kozák Á, Bánvölgyi Sz, Vincze I, Kiss I, Békássy Molnár E, Vatai G. Comparison of integrated large-scale and laboratory scale membrane processes for the production of black currant juice concentrate. Chem Eng Proc 2008;47:1171–1177. 80. Kozák A, Békássy-Molnár E, Vatai G. Production of black-currant juice concentrate by using membrane distillation. Desalination 2009;241:309–314. 81. Sotoft LF, Christensen KV, Andrénsen R, Norddahl B. Full scale plant with membrane based concentration of blackcurrant juice on the basis of laboratory and pilot scale tests. Chem Eng Proc 2012;54:12–21.

86 

 3 Integrated membrane operations in fruit juice processing

82. Galaverna G, Di Silvestro G, Cassano A, Sforza S, Dossena A, Drioli E, Marchelli R. A new integrated membrane process for the production of concentrated blood orange juice: Effect on bioactive compounds and antioxidant activity. Food Chem 2008;106:1021–1030. 83. Cassano A, Conidi C, Drioli E. Clarification and concentration of pomegranate juice (Punica granatum L.) using membrane processes. J Food Eng 2011;107:366–373. 84. Koroknai B, Csanádi Z, Gubicza L, Bèlafi-Bakó K. Preservation of antioxidant capacity and flux enhancement in concentration of red fruit juices by membrane processes. Desalination 2008;228:295–301. 85. Bèlafi-Bakó K, Koroknai B. Enhanced water flux in fruit juice concentration: Coupled operation of osmotic evaporation and membrane distillation. J Membrane Sci 2006;269:187–193. 86. Hunter Denise C, Denis H, Parlane NA, Buddle BM, Stevenson LM, Skinner MA. Feeding ZESPRItm GOLD Kiwifruit puree to mice enhances serum immunoglobulins specific for ovalbumin and stimulates ovalbumin-specific mesenteric lymphonode cell proliferation in response to orally administered ovalbumin. Nutr Res 2008;28:251–257. 87. Kvesitadze GI, Kalandya AG, Papunidze SG, Vanidze MR. Identification and quantification of ascorbic acid in kiwifruit by high-performance liquid chromatography. Appl Bioch Micr 2001;37:215–218. 88. Cano Pilar M. HPLC separation of chlorophyll and carotenoid pigments for four kiwifruit cultivars. J Agric Food Chem 1991;40:594–598. 89. Cassano A, Jiao B, Drioli E. Production of concentrated kiwifruit juice by integrated membrane processes. Food Res Int 2004;37:139–148. 90. Cassano A, Figoli A, Tagarelli A, Sindona G, Drioli E. Integrated membrane process for the production of Highly Nutritional Kiwifruit Juice. Desalination 2006;189:21–30. 91. Mobhammer MR, Stintzing FC, Carle R. Evaluation of different methods for the production of juice concentrated and fruit powders from cactus pear. Innov Food Sci Emerg 2006;7:275–287. 92. Cassano A, Conidi C, Timpone R, D’Avella M, Drioli E. A membrane-based process for the clarification and the concentration of the cactus pear juice. J Food Eng 2007;80:914–921. 93. Vaillant F, Cisse M, Chaverri M, Perez A, Dornier M, Viquez F, Dhuique-Mayer C. Clarification and concentration of melon juice using membrane processes. Innov Food Sci Emerg 2005;6:213–220.

4 Integrated membrane operations in citrus processing Alfredo Cassano and Bining Jiao 4.1 Introduction Citrus is one of the world’s major fruit crops, largely developed in tropical and sub-tropical countries. The annual global production of citrus fruits was estimated to be over 115 million metric tons in 2011, with oranges contributing more than half of the worldwide citrus production. According to 2012 data from the United Nations Food and Agriculture Organization (FAO), China, Brazil, the USA, India, Mexico, and Spain are the world’s leading citrus fruit-producing countries, representing close to 60% of the global production [1]. The most well-known examples of citrus fruits with commercial importance are sweet oranges, lemons, limes, grapefruit and mandarins (also known as tangerines). Of the total citrus production, close to 60% is consumed in the fresh market and approximately 40% is utilized after processing. Orange juice production accounts for nearly 85% of total processed consumption. Among fruits and vegetables, citrus fruits have been recognized as an important food and integrated into our daily diet, representing a very rich source of “healthpromoting substances” [2]. Citrus fruits are a good source of carbohydrates: the sugar content ranges from 4% to 7% depending on the specific fruit and cultivar. Sucrose is predominant in orange juice while fructose is predominant in lemon juice [3]. Fresh citrus fruits are also a good source of dietary fiber, which is associated with gastrointestinal disease prevention and lowered circulating cholesterol. In addition, they contain many B vitamins (thiamin, pyridoxine, niacin, folate, riboflavin and pantothenic acid), minerals and biologically active phytochemicals such as carotenoids, flavonoids and limonoids [4]. These biological constituents are of vital importance in human health improvement because of their antioxidant properties [5, 6]. The major categories of orange juice present on the market include fresh squeezed juice, frozen concentrated orange juice (FCOJ), not-from-concentrate (NFC) orange juice and refrigerated orange juice from concentrate (RECON). – The freshly squeezed juice, obtained from fresh fruits without being pasteurized, is characterized by a shelf-life of only a few days. – FCOJ is generally obtained by removing water from the juice in a vapor form through thermally accelerated short-time evaporators (TASTEs) and then stored at −6.6°C or lower until it is sold or packaged for sale. A typical citrus fruit processing plant layout, illustrating the steps involved in the production of citrus juice and by-products, is depicted in Figure 4.1.

Molasses 50 °Brix

Dry cattle feed

Drying

Pressing

Hydrolysis

Peel and rag

Juice

Finishing

Pulpy juice

Pasteurization

Aseptic storage

Not from concentrate

Evaporation

Pulp wash 65 °Brix

Deareation/deoling

Centrifugation

Washing

Pasteurization

Pulp wash

Defect removal

Bin storage

Figure 4.1: Schematic of citrus fruit processing

d-Limonene

Evaporation

Press liquor

Fruit unloading

Frozen concentrate

Frozen storage

Cooling

Evaporation

Pasteurization

Extraction

Grading

Aqueous aroma

Condensation

Essential oil

Winterization

Centrifugation

Oil emulsion

88   4 Integrated membrane operations in citrus processing

4.2 Clarification of citrus juices 

–

–

 89

NFC orange juice is processed and pasteurised by flash heating immediately after squeezing the fruit, without removing the water content from the juice. It can be stored freezed or chilled for at least a year. RECON is a juice that has been processed to obtain the frozen concentrate and then reconstituted by adding back the water that had been originally removed. Reconstituted single-strength juice is normally reconditioned by the packager and sold as a ready-to-serve product, either chilled or in an aseptic form without the need of refrigeration.

Traditional technologies involved in the production of all these juices are characterized by some drawbacks concerning the quality of the product, the energetic consumption and the environmental impact. It is known that the thermal treatment by pasteurization and/or thermal concentration produces a severe loss of the volatile organic flavor/fragrance components as well as a partial degradation of ascorbic acid and natural antioxidants, accompanied by a certain discoloration and a consequent qualitative decline [7, 8]. Membrane technologies, as “mild technologies”, represent very efficient systems to preserve the nutritional and organoleptic properties of citrus fresh products because of the possibility of operating at room temperature with low energy consumption. In addition, they offer interesting alternatives to the traditional techniques for the recovery of high added value compounds from citrus by-products. A full exploitation of the potential of these techniques may be achieved through the integration of different processes. This chapter will provide an overview of membrane operations that can substitute traditional operations in the clarification, concentration and aroma recovery of citrus juices as well as in the recovery of bioactive compounds from by-products of citrus juice production. Integrated membrane processes, which can contribute to redesign the traditional industrial transformation of citrus fruits within the PI strategy, are also presented and discussed.

4.2 Clarification of citrus juices The traditional methods of citrus juice clarification are based on the use of different technologies, including centrifugation, depectinisation, fining agents (such as bentonite and gelatin) and filtration by diatomaceous earth. The use of microfiltration (MF) and ultrafiltration (UF) membranes presents many advantages over conventional clarification, including the possibility to operate at room temperature (thus avoiding pasteurization and sterilization), increased juice yields, reduced labor and capital costs, elimination of filter aids, reduction of waste products, easy maintenance of the equipment [9].

90 

 4 Integrated membrane operations in citrus processing

Permeate fluxes and permeate quality are the most important aspects for the selection of a proper membrane. The build-up of macromolecular or colloidal species at the upstream interface of the membrane such as pectins, proteins, tannins and fibers determines a rapid permeate flux decay followed by a long and gradual decline towards a steady-state limit value. This phenomenon is known as concentration polarization. Fouling mechanisms, such as adsorption of particles on the membrane pore walls and pore plugging, are additional phenomena. Pretreatment methods can reduce the particulate matter in the juice, leading to a remarkable improvement of permeate fluxes and the attainment of higher concentration factors. Rai et al. [10, 11] studied the effect of seven different pretreatment methods on the performance of a 50 kDa thin-film composite polyamide membrane in the clarification of mosambi juice including centrifugation, fining by gelatin, fining by bentonite, fining by bentonite followed by gelatin, enzymatic treatment with pectinase, enzymatic treatment followed by centrifugation and enzymatic treatment followed by fining with bentonite. Among the various pretreatment methods the enzymatic treatment followed by bentonite was found the best to maximize the permeate flux. Permeate fluxes in MF and UF processes depend strongly on operating and fluid dynamic conditions and on the nature of the membrane and feed solutions. Cassano et  al. [12] evaluated the effect of operating parameters on membrane fouling and juice quality in the clarification of depectinised blood orange juice by using a polyvinylidene fluoride (PVDF) UF tubular membrane module with a NMWCO of 15 kDa (Koch Membrane Systems Inc., Wilmington, MA, USA). Permeate fluxes increased with transmembrane pressure (TMP) up to a limiting value (TMPlim) depending on the physical properties of the juice and axial velocity. An increase in the feed flow rate produced a linear increase of the permeate flux caused by the effect of the shear stress at the membrane surface, which enhanced the rate of removal of deposited particles reducing the polarized layer thickness. The increase in the operating temperature produced a reduction of juice viscosity, together with an increase of the diffusion coefficient of macromolecules with a consequent enhancing of the permeate flux. In optimized operating conditions (TMP 0.85 bar, feed flow rate 800 l/h and temperature 25°C) the initial permeate flux of 19 l/m2h decreased at a steady-state value of about 11 l/m2h when the VRF reached a value of 3 and remained constant up to the final VRF value of 6. In the fixed operating conditions of TMP and temperature, the fouling mechanism evolved from a partial to a complete pore blocking condition in dependence of the axial feed velocity. Ascorbic acid, anthocyanins, narirutin and hesperidin, which contribute to the TAA of the juice were recovered in the clarified juice, while suspended solids were completely removed by the UF membrane. Thus, the flux decline during the UF process was attributed to the formation of fouling layers through a combination of suspended particles and adsorbed macromolecular impurities. A mass balance of the UF process is depicted in Figure 4.2.

4.2 Clarification of citrus juices 

 91

Retentate (1.57 I)

Depectinized orange juice (9.33 I)

UF

Suspended solids

933 g

Soluble solids

1119.6 g

Ascorbic acid

6.54 g

Anthocyanins

564.0 mg

Narirutin

436.08 mg

Hesperidin

310.96 mg

Suspended solids

933 g

Soluble solids

211.95 g

Ascorbic acid

1.00 g

Anthocyanins

95.0 mg

Narirutin

73.41 mg

Hesperidin

48.88 mg

Clarified juice (7.76 I)

Suspended solids

0g

Soluble solids

869.12 g

Ascorbic acid

4.98 g

Anthocyanins

424.0 mg

Narirutin

363.09 mg

Hesperidin

261.27 mg

Figure 4.2: Mass balance of the UF process in the clarification of blood orange juice with tubular PVDF membrane

An improvement of color and clarity of mandarin juice through the removal of suspended solids was also achieved by using modified poly(ether ether ketone) (PEEKWC) and polysulfone (PS) hollow fiber (HF) membranes prepared through the phase inversion process [13]. PEEKWC membranes showed a lower rejection towards TSSs, total phenolics and TAA in comparison with the PS membranes in agreement with the lower rejection observed for PEEKWC membranes towards dextrans with specific molecular weight. The analysis of membrane fouling in the clarification of orange juice with PVDF MF membranes (Tri-Cor 2-MFK, Koch Membrane Systems Inc.) with a pore size of 0.3  μm revealed that the separation process is controlled by cake filtration mechanisms at relatively low velocity (i.e., Reynolds number  = 5000) and low TMPs. At higher Reynolds numbers an increase of applied TMP allows an increase in the limit permeate flux by a factor of about 4. In these conditions the filtration process is controlled by a complete pore blocking mechanism and flux decay is negligible [14]. The use of electric fields in UF to reduce fouling and concentration polarization has been also investigated in the treatment of citrus juices. Pectin is negatively charged and its charge density depends on pH and degree of methoxylation. Therefore, the

92 

 4 Integrated membrane operations in citrus processing

application of an external field with appropriate polarity can reduce the pectinous gel layer thickness due to the electrophoretic migration of pectin molecules away from the membrane surface. Sarkar et  al. [15] evaluated the effect of the electric field, applied in both constant and pulse modes, on the permeate flux in the UF of mosambi [Citrus sinensis (L.) Osbeck] juice with a 50 kDa polyethersulfone (PES) membrane. The application of the electric field resulted in a significant improvement of the permeate flux, achieving a 22% reduction of total energy consumption per unit volume of permeate. In addition, PEF was more advantageous in terms of permeate flux improvement and energy consumption if compared with constant electric fields. MF and UF membranes allow a complete removal of pulp and water soluble pectins from orange juice. In particular, the use of HF PS membranes with a MWCO of 50 kDa permits a complete separation of suspended solids from freshly squeezed orange juice: most of pectin materials are retained by membranes and the viscosity of the juice is appreciably reduced [16]. Oxygenated aroma compounds, such as alcohols, esters and aldehydes, flow freely through the membrane while less polar aroma compounds like limonene and valencene tend to be associated with the retained pulp. Todisco et al. [17] found that more water soluble compounds, such as aldehydes, esters and alcohols, pass through PVDF UF membranes while hydrocarbons and less polar aroma compounds like limonene are retained with the pulp fraction. Conversely, some esters such as methyl acetate, ethyl acetate, ethyl butyrate and methyl butyrate, which contribute to the “top-note” of citrus flavors are recovered in the clarified juice. Consequently, even if UF membranes remove some volatile aroma compounds, the greatest contributors of orange flavors can be preserved, allowing the production of orange juice with an improved appearance. The clarification of lemon juice was investigated using a 0.2 μm MF membrane in a flat-sheet configuration [18]. The clarified juice presented titrable acidity, pH and TSS values comparable with those of untreated fresh lemon juice. Optimal performances were obtained at a TMP of 0.6 bar and a feed flow rate of 1 m/s.

4.3 Debittering of orange juice Bitterness in citrus fruits is attributed to the presence of limonin, a 22-carbon triterpenoid dilactone, and naringin, a 15-carbon glycosylated flavonoid [19]. In particular, the soluble fruit fraction contains a non-bitter precursor of limonin, “limonoate A-ring lattone” (LARL), which diffuses into the juice during fruit processing and is converted in limonin under acidic conditions. This conversion is accelerated by enzymes such as the limonin D-ring lactone hydrolase present in the juice. Different approaches have been used until now to solve the problem of limonin formation in orange juice, including the enzymatic treatment of the juice, the exposure of the fruit to ethylene, the adsorption of limonin on cellulose acetate and the use of agents, such

4.4 Concentration of citrus juices 

 93

as β-ciclodextrin, that able to form complexes with limonin. However, until now none of these systems can be considered completely satisfactory for preventing bitterness in orange juice. UF can be used to separate the suspended pulp from the juice, where the non-bitter precursor is located, minimizing the contact time between pulp and serum and thus allowing the control of non-bitter precursor to limonin conversion. Todisco et al. [17] clarified freshly squeezed orange juice by using tubular PVDF UF membranes. They found that the limonin concentration in the permeate was reduced by about 65% in comparison with the fresh juice, although the molecular weight of limonin is 500 (remarkably lower than the MWCO of the membrane). This result was attributed to the continuous and faster removal of serum from pulp if compared with the diffusion of LARL and its conversion in bitter limonin. If compared with other technologies, UF does not require a chemical modification of the juice, allowing a better-quality final product with a concentration of limonin ( < 5–6 ppm) too small to cause a bitter taste.

4.4 Concentration of citrus juices 4.4.1 Reverse osmosis Traditional processes of citrus juice concentration are based on the use of multi-stage vacuum evaporation, which involves the use of water evaporation at high temperatures followed by recovery and concentration of volatile flavors and their addition back to the concentrated product. These processes lead to a significant deterioration of juice quality and a partial loss of fresh juice flavors, accompanied by juice discoloration and the appearance of a typical cooked taste caused by the thermal effects [20]. In addition the evaporation process is characterized by significant energy consumption (tripleeffect evaporators require up to 660 Btu of energy per kg of water removed). Freeze concentration and sublimation concentration techniques require less energy and preserve juice quality but these methods can be costly and limited in the degree of concentration achieved [21]. Fruit juice concentration by RO offers different advantages over conventional concentration processes in terms of low thermal damage to product, reduction in energy consumption and lower capital investments [22] as the process is carried out at low temperatures and it does not involve phase change for water removal. The retention of juice constituents, especially flavors, and the permeate flux, regarding the RO performance, are two major factors, which are related to the type of membranes and the operating conditions used during the process. Oil-soluble aroma compounds of orange juice are easily retained by cellulose acetate membranes in comparison with water soluble aromas [23]. High recoveries of sugars (higher than 98%), acids (up to 85%) and flavor-volatile compounds were

94 

 4 Integrated membrane operations in citrus processing

obtained using spiral-wound polyamide membranes [24]. The different rejection of the RO membrane towards sugars and organic acids produced an increase in the TSS/ acid ratio with a consequent reduction of juice bitterness. Braddock et al. [3] reported that volatile compounds, except for methanol, ethanol and traces of limonene, were not detected in measurable quantities in the permeate, during the concentration of orange and lemon juices in the range of 22–25°Brix at pulp contents of 7–10%, using a composite tubular RO membrane (ZF99, Patterson Candy International Ltd., 99% NaCl rejection, surface area 0.9 m2). However, a 17–30% loss of volatile peel oil (measured as limonene) was found if the membrane system was not totally closed during the recirculation of the process stream. The enzymatic pretreatment of orange juice with pectinase increases the permeate flux of RO membranes without affecting the solute recovery. A further improvement of permeate flux can be attained by using clarified juice: RO of clarified satsuma mandarin juice produced higher permeate fluxes than those obtained with cloudy juice (3% v/v) [25]. The first commercial RO plant for the concentration of orange juice was described by Gadea (1987). It was based on the use of thin-film polyamide composite membranes (AFC99, PCI Membrane Systems) in tubular configuration. Operating at a feed flow rate between 4.2 and 9.7 m3h, the water removal rate was of approximately 2 m3h. RO membranes showed an excellent retention of juice constituents as shown in Table 4.1. Köseoglu et al. [9] proposed a cold process for separating orange juice into three fractions: (i) a pulp fraction; (ii) a heat-sensitive solution containing small molecules such as flavors, acids and sugars; (iii) a heat-insensitive solution containing color, proteins, other molecules and microbes. In this process the depulped orange juice is pumped to a UF membrane system (constituted by Romicon HF membranes with a MWCO of 50 kDa, Koch Membrane Systems Inc.) producing a clarified juice (the heat-sensitive fraction), which is concentrated by RO. A tubular RO membrane system containing ZF-99 non-cellulosic membranes (Patterson Candy International Ltd., Witchurch, Hampshire, UK) was used for this purpose. The RO permeate is directed back into the UF feed acting as diafiltration water to improve the removal of sugar and aroma compounds through the UF membrane. The RO permeate is also collected to reconstitute orange juice. The pulp and the heat-insensitive fraction (UF retentate) can be mixed and submitted to a short-time heat treatment. Hence they can be aseptically combined and filled with the cold-concentrated heat-sensitive fractions to produce a concentrated juice. An optional deacidification procedure by using weakly basic anion-exchange resins can be considered to remove the desired amount of acid. The whole process flow diagram is depicted in Figure 4.3. As also reported in Chapter 3, the concentration of fruit juices by RO is limited by the high operating pressures needed to reach high concentrations of soluble solids because of the high osmotic pressure of the juice (the osmotic pressure of a 42°Brix pulpy orange juice is greater than 90 bar). For cellulosic and non-cellulosic membranes

4.4 Concentration of citrus juices 

 95

Table 4.1: Retention of orange juice constituents in RO concentration (adapted from [26]) Component TSSs (°Brix) Glucose (g/l) Fructose (g/l) Saccharose (g/l) Citric acid (%, w/w) Vitamin C (ppm) Ash (%) Sodium (ppm) Potassium (ppm) Phosphorous (ppm) Aspartic acid (mg/100 ml) Asparagine (mg/100 ml) Proline (mg/100 ml) Glycine (mg/100 ml) Alanine (mg/100 ml) Valine (mg/100 ml) Isoleucine (mg/100 ml) Leucine (mg/100 ml) Phenlylalanine (mg/100 ml) γ-Aminobutirric (mg/100 ml) Histidine (mg/100 ml) Ornitine (mg/100 ml) Lysine (mg/100 ml) Arginine (mg/100 ml) Total nitrogen (g/l)

Juice stength

Permeate

Rejection (%)

11.2 26.6 28.4 35.1 0.85 480 4.1 59 1427 200 28 45 80 1.5 9.9 1.8 0.0 0.5 3 24 1.1 0.8 2.9 70 1363

0.0 0.1359 0.1552 0.0815 0.01 6.0 0.11 2 47 2 0.35 1.0 2.4 Traces 1.5 0.034 Traces Traces 0.04 0.46 0.062 0.32 0.14 0.42 11

100 99.4 99.4 99.7 98.8 98.7 97.3 96.6 96.7 99.0 98.7 97.7 97.0 – 84.8 98.1 – – 98.6 98.1 94.3 60.0 95.1 99.4 99.2

the most efficient flux and solute recovery are at a concentration lower than 30°Brix. This suggests the use of RO as a preconcentration step in combination with other concentration techniques like freeze concentration or evaporation in order to reduce energy consumptions and to increase the production capacity [26, 27].

4.4.2 Membrane distillation and osmotic distillation Membrane distillation (MD) and osmotic distillation (OD) can be used to selectively extract water from aqueous solutions under atmospheric pressure and at room temperature, also at high osmotic pressures [28–30]. Therefore they are suitable for the concentration of heat-sensitive products like fruit juices, including citrus juices. Both processes are based on the use of a macroporous hydrophobic membrane separating two aqueous solutions. The driving force for the water vapor transport through the membrane is the vapor pressure difference between the two solution-membrane interfaces that is caused bythe existing temperature gradient in MD and concentration gradient in OD.

UF

RO permeate recycle

Permeate (flavors, aromas, sugars, aminoacids, minerals)

Permeate (water, trace flavors)

RO

Deacidification (optional)

Reduced acid RO retentate

Sterilizing filter

RO permeate discard

Retentate (flavors, aromas, sugars, aminoacids, minerals)

Orange juice concentrate

Figure 4.3: Flow diagram of integrated process for orange juice concentration (adapted from [9]). UF, ultrafiltration; RO, reverse osmosis

Finished orange juice

Retentate (enzymes, pectins, proteins, pulp, oil)

Evaporation Sterilization

Aseptic packaging

96   4 Integrated membrane operations in citrus processing

4.4 Concentration of citrus juices 

 97

The concentration of orange juice by MD was investigated by Drioli et al. [31] by using a commercial PVDF membrane (Millipore Corp.) with a nominal pore size of 0.22 m and a laminated membrane (G0712) with a pore size of 0.2 μm (Gelman Science Technology, Ltd., Ann Arbor, MI, USA). The evaporation flux was remarkably higher for the PVDF membrane. It decreases with an increase of feed juice concentration caused by the decrease of the vapor pressure of the juice and to the exponential increase of its viscosity. At high concentration ratios, permeate fluxes were higher in MD than in RO. An increase of the MD flux was observed by increasing the temperature difference between orange juice and cooling water. Similarly, an increase of the evaporation flux was observed by increasing the feed flow rate because of the generated shearing forces, which reduced the accumulation of particulates, such as pectin and cellulose on the membrane surface. Conversely, a lower crossflow velocity hindered the heattransfer from the bulk of the solution to and from the membrane surface, leading to a more severe temperature polarization. The PVDF membrane showed a good retention of orange juice compounds, such as TSS, sugars and organic acids. A 42.1% decrease in vitamin C was attributed to the use of high temperatures and oxidation. For this purpose Drioli et al. [31] suggested to maintain the operating temperature in MD as lower as possible. Similar results were obtained by Calabrò et al. [32] in the concentration of orange juice by MD in which a commercial plate PVDF membrane (Millipore Corp.) with a nominal pore size of 0.11 μm was used. It was found that UF of the single-strength juice resulted in an increase of evaporation fluxes and that the flux remained essentially constant during an approximately two-fold concentration. The MD flux of the unfiltered juice, by contrast, decreased steadily over the same concentration range. The improvement in MD flux after UF was attributed to a reduction in juice viscosity as a result of pulp and pectin removal. Integrated membrane processes involving MF and OD for the clarification and concentration of orange juice, respectively, have been implemented on pilot and semi-industrial scales. In the process investigated by Shaw et al. [33] the orange juice was clarified by crossflow MF by using Membralox IP19–40 membranes of 0.2 μm average pore diameter (SCT, Bazet, France) and then concentrated by using a pilot scale osmotic evaporator containing polypropylene (PP) HF membranes with average pore diameter of 0.2 μm. Calcium chloride solution (4.6 M) was used as brine. The clarified juice was concentrated 3-fold up to 35°Brix. Headspace gas chromatographic analyses showed a loss of about 32% of the volatile compounds. No significant differences were found between the initial juice and the reconstituted concentrate in a panel test evaluating specific flavor characteristics. In particular, the average flavor score for each characteristic was slightly lower in juice from concentrate that in the initial juice. Concentrated orange juices at 62°Brix were obtained in an integrated process investigated by Cissé et al. [34] on a semi-industrial scale. The single-strength juice

98 

 4 Integrated membrane operations in citrus processing

was clarified using a MF unit equipped with the ceramic membrane Membralox IP19–40. The clarified juice was concentrated by using an OD plant equipped with PP HF membranes and calcium chloride as a stripping solution. The clarified juice presented a composition very similar to that of the fresh pulpy juice except for the carotenoids, which were completely retained by the membrane, and some aroma compounds, such as terpenic hydrocarbons, which were partially retained because of their apolar properties and association with the insoluble solids of the retentate fraction. The quality of concentrate was also very similar to that of the clarified juice in terms of organic acids and sugar content (Table 4.2). A small loss of vitamin C at the beginning of the OD process was attributed to oxidation phenomena. Authors reported that losses of aroma compounds could be limited by preconditioning the OD membrane with the clarified juice and by avoiding thermal regeneration of brine during concentration. In the selected operating conditions the evaporation fluxes in OD decreased from 0.7 l/m2h to 0.67 l/m2h when TSS reached 45°Brix and to 0.59 l/m2h when TSS reached 62°Brix. The effect of an integrated membrane process on bioactive compounds and antioxidant activity (TAA) of blood orange juice was investigated by Galaverna et al. [35].

Table 4.2: Main characteristics of orange juice clarified and concentrated by MF/OD integrated process (adapted from [34]) Component pH (20°C) Water activity (25°C) Viscosity (25°C, mPa S) Density (kg/m3) Total soluble solids (°Brix) Suspended insoluble solids (g/kg) Titrable acidity (g citric acid/kg TSS) Glucose (g/kg TSS) Fructose (g/kg TSS) Sucrose (g/kg TSS) Carotenoids (g/kg TSS) Vitamin C (g/kg TSS) Color (L°) Hue angle (H°) Color purity (C°) aafter

dilution to 11.5°Brix.

Single-strength juice

Clarified juice

3.62 0.98 1.1 1032 11.8

3.58 0.99 1.2 1028 11.5

3.52 0.77 28.2 1290 62.0

80

0

0

68

61

62

186 220 491 0.38 3.7 52 88 30

185 220 489 < 0.02 3.5 62 88.3 17

Concentrated juice

187a 221a 491a < 0.02a 3.3a 61a 88.3a 17a

4.4 Concentration of citrus juices 

 99

The process was based on a preliminary clarification of freshly squeezed juice by UF followed by a preconcentration step (up to 25–30°Brix) performed by RO and a final concentration by OD up to about 60°Brix. The slight decrease (ca. 15%) of TAA during the concentration process was attributed to the partial degradation of ascorbic acid (ca. 15%) and anthocyanins (ca. 20%). However, this degradation was lower than that observed in thermally concentrated juices where the reduction of TAA, ascorbic acid and anthocyanins were of the order of 26%, 30% and 36%, respectively (Figure 4.4). On the contrary, no significant variations were observed for hydroxycinnamic acids and flavanones, which were well-preserved during the integrated membrane process (Figure 4.5). Similar results were obtained in a two-step UF/OD process in which TAA variations were particularly attributed to variations of the ascorbic acid content. According to these results, some authors proposed a process scheme in which the preconcentration step via RO effects time saving and efficiency increases without affecting the quality of the final product (Figure 4.6). A picture of samples obtained in the three-step membrane process is reported in Figure 4.7. Recently, Destani et al. [36] implemented an integrated process on a laboratory scale to obtain formulations of interest for food and/or pharmaceutical industry starting from the blood orange juice produced in southern Italy. The freshly squeezed juice, after a depectinisation step, was submitted to a UF process in order to recover natural antioxidants, such as hydroxycinnamic acids, hydroxybenzoic acids, flavanones,

800 Ascorbic acid Total anthocyanins

Concentration [mg/l]

700 600 500 400 300 200 100 0 Fresh juice TSS (°Brix)

UF RO OD TE permeate retentate retentate retentate

12.6

12.4

21.4

60.6

56.3

TAA (mM Trolox) 8.65

8.21

7.47

7.33

6.40

Figure 4.4: Variation of TAA, ascorbic acid and total anthocyanins in blood orange juice concentrated by integrated membrane process and thermal evaporation UF, ultrafiltration; RO, reverse osmosis; OD, osmotic distillation; TE, thermal evaporation; TAA, total antioxidant activity; TSS, total soluble solids.

100 

 4 Integrated membrane operations in citrus processing

60 Sinapic acid Ferulic acid p-Cumaric acid Narirutin Hesperidin

Concentration [mg/l]

50

40

30

20

10

0 Fresh juice (12.6 °Brix)

UF permeate RO retentate OD retentate (12.4 °Brix) (21.4 °Brix) (60.6 °Brix)

TE retentate (56.3 °Brix)

Figure 4.5: Variation of hydroxycinnamic acids and flavanones in blood orange juice concentrated by integrated membrane process and thermal evaporation UF, ultrafiltration; RO, reverse osmosis; OD, osmotic distillation; TE, thermal evaporation Pulp

Raw fruit juice

Condenser UF

Condensate to waste Diluted brine

Clarified juice

Evaporator

Concentrated brine Pasteurization

Water

OD

RO Water

Concentrated juice

Pre-concentrated juice Reconstituted juice Figure 4.6: Proposed flow diagram for the clarification and concentration of blood orange juice UF, ultrafiltration; RO, reverse osmosis; OD, osmotic distillation.

flavan-3-ols, and anthocyanins. The UF permeate, with an initial TSS content of 10.5°Brix, was concentrated by OD up to a final concentration of 61.4°Brix. Suspended solids were completely removed by UF, producing a clear juice in which phenolic

4.4 Concentration of citrus juices 

(a)

(b)

(c)

 101

(d)

Figure 4.7: Samples of blood orange juice clarified and concentrated by integrated membrane process: (A) depectinised juice (12.6°Brix); (B) clarified juice by UF (12.4°Brix); (C) preconcentrated juice by RO (21.4°Brix); (D) concentrated juice by OD (60.6°Brix)

compounds were well-preserved in comparison to the fresh juice (the rejection of the UF membrane towards these compounds was in the range 0.4–6.9%). The recovery of phenolic compounds in the concentrated juice in comparison with the unclarified juice, was of 95–100% for hydroxycinnamic acids and 100% for the other investigated compounds (phenolic acids, flavanones and flavan-3-ols). A well-known OD module designed for laboratory applications is the Liqui-Cell Extra-Flow 2.5 × 8″ membrane contactor (Membrana), which contains microporous PP HFs (external and internal diameters of 300 μm and 220 μm, respectively) with a mean pore diameter of 0.2 μm and a total membrane surface area of 1.4 m2. Modules containing nominal membrane areas of 19.2 and 135  m2 are also commercially available [37]. The membrane contactor provides a shell-and-tube configuration: the clarified juice to be concentrated enters the shell side of the module while the stripping solution is recirculated in the lumen side with its flow countercurrent to the feed. The brine pressure drop required to supply adequate brine velocity for the concentration process is within the burst limit of the fibers and below the pressure level for liquid intrusion into the membrane pores (the intrusion pressure of water into the pores is well in excess of 100 psig). Successfull applications related to the use of Liqui-Cel® membrane contactors for the concentration of fruit and vegetable juices were realized on pilot plant facilities located in Mildura and Melbourne, Australia. The Melbourne facility, designed by Zenon Environmental (Burlington, Ontario, Canada), was a hybrid plant consisting of UF and RO pretreatment stages and an OD section containing two 19.2 m2 LiquiCel® membrane modules. Fresh fruit juices were concentrated up to 65–70°Brix at an average throughput of 50 l/h [30]. The Mildura plant, designed by Vineland Concentrates and Celgard LLC, contained 22 Liqui-Cel® membrane modules (4  × 28″ type) for a total interfacial area of 425 m2. It was used for the concentration of grape juices to

102 

 4 Integrated membrane operations in citrus processing

make wine from reconstituted concentrate. The installation produced approximately 20–25 l/h of 68°Brix concentrate [38]. The surface tension of citrus juices is reduced by peel oils and highly lipophilic flavor components, which can promote wetting of hydrophobic surfaces such as those of PP membranes. The use of HF membranes from more hydrophobic polymers such as polytetrafluoroethylene (PTFE) or PVDF or laminate membranes that prevent liquid intrusion without impeding vapor transport are some possible solutions to overcome this drawback [30, 39]. A strict correlation between the degree of retention of organic volatile flavor/ fragrance components and membrane pore size was observed in the concentration of Valencia orange juice by OD [40]. In particular, membranes having a relatively large pore size at the surface exhibited a higher organic volatiles retention per unit water removal than those with smaller surface openings. This was attributed to a greater intrusion of the juice and stripping solutions in membranes with large pores with a resulting increase in the thickness and resistance of the boundary layer at the pore entrance. A comparative study between OD and MD in terms of water flux and aroma retention in the concentration of orange juice was performed by Alves and Coelhoso [41]. In particular, the transport of two relevant aroma compounds of the orange juice aroma, citral and ethyl butyrate, in both OD and MD processes was evaluated. Experimental results revealed a higher retention for both compounds and higher water fluxes in the OD process. The presence of suspended solids and macromolecules in the orange juice was considered as the main cause of mass transfer resistance during OD.

4.5 Recovery of aroma compounds The aroma profile of orange juice comprises a large number of volatile organic compounds including alcohols, hydrocarbons, esters and aldehydes. These compounds are susceptible to chemical changes or complete degradation when the juice is submitted to thermal processes such as pasteurization or thermal evaporation. The use of separation techniques such as distillation, SFE, adsorption and pervaporation (PV), finalized to recover these compounds, is a possible way to minimize these problems. PV presents different advantages over traditional techniques in terms of low energy consumption, no heat damage of heat-sensitive aromas, minimum loss of aroma compounds and no additional separation treatments for added solvents. In this process a liquid mixture is partially vaporized through a permselective membrane which can be either a non-porous polymeric or a nanoporous inorganic (ceramic/ zeolite) membrane. The vaporous permeate is subsequently condensed to obtain a liquid product. The DF for mass transfer across the membrane is generally accomplished by applying a gradient in partial vapor pressure between the liquid feed and the vaporous permeate [42–44].

4.6 Treatment of citrus by-products 

 103

The pervaporative recovery process of volatile aroma compounds from orange juice was studied by Aroujalian and Raisi [45] using a commercial polydimethylsiloxane (PDMS) membrane (GKSS Forscungszentrum, Geestacht, Germany). The authors found that the feed flow rate had no significant effect on the performance of the process while an increase in the feed temperature produced a higher flux and enrichment factor. Additionally, an increase in permeate pressure produced a slight decrease in the enrichment factor of some aroma compounds such as limonene, linalool and α-terpineol. Conversely, for some aroma compounds such as ethyl acetate, ethyl butirate and hexanal, the enrichment factor increased by increasing the permeate pressure. Well-spaced longitudinal outflow HF modules containing dense PDMS fibers were also showed to be a feasible option for aromas recovery from water phase orange juice stream by PV [46]. Results suggested that the water phase aromas can be enriched up to 8% w/w in the PV permeate stream. As phase separations were observed in the permeate streams, operational temperatures must necessarily consider the possibility of enriched aroma recycling.

4.6 Treatment of citrus by-products In the orange juice production, only around the half of the fresh oranges’ weight is transformed into juice; the residual waste, containing water (in average 82%), peels, seeds, orange leaves and whole orange fruits, does not meet the quality requirement. This waste is traditionally spread on soil, producing dried peel by natural evaporation that can be used as swine or cattle feed. This method of handling presents environmental and health problems because of uncontrolled fermentation, and produces leachates containing high concentrations of organic matter, which can contaminate surface and ground waters. An alternative handling option involves the treatment of citrus waste with lime followed by milling and pressing. The resulting press liquor and press cake can be used as animal feed [47]. The press liquor containing in average a TSS content of 10°Brix can be concentrated up to 65–70°Brix by multiple effect evaporation to obtain citrus molasses, which can be used in the production of beverage alcohol and as cattle feed. As the evaporative concentration is characterized by high energy consumption, efforts are needed to find alternative dehydration techniques. RO can be used as a preconcentration system of citrus press liquors in order to produce a permeate stream (which can be reused in the juice production process depending on its quality) and a concentrated stream (which can be submitted to an evaporation step to produce citrus molasses). The performance of a spiral-wound RO membrane (Filmtec SW30–2540, Dow Chemical, Midland, MI, USA) in the preconcentration of model solutions of sucrose (10, 20 and 30°Brix) at different pressures, temperatures and flow rate was evaluated by Garcia et al. [48]. An empirical function was

104 

 4 Integrated membrane operations in citrus processing

developed to predict permeate fluxes for some specific operating conditions. Results indicated that in the range of conditions investigated, concentration and effective pressure were the most important factors affecting the RO preconcentration process. In a design of 24 membrane elements it was found that the RO system had 7.7 times lower energy consumption when compared to a preconcentration system with a multiple effect evaporation [49]. The RO preconcentration process was also investigated using an aromatic polyamide spiral-wound membrane (SWC2-2540, Hydranautics, Oceanside, CA, USA) on two synthetic feed solutions prepared with and without addition of pectin in order to simulate a complete depectinisation step before the RO treatment [50]. Preconcentration of synthetic liquor with pectin was only possible up to a VRF 1.2 at the maximum tested TMP (50 bar) because of the high solution viscosity and membrane fouling. Starting from an initial feed concentration of 8.5°Brix, the highest concentration achieved in selected operating conditions (TMP 50 bar, temperature 20°C) was 11°Brix. In addition, the presence of pectin led to a low-quality permeate. Conversely, press liquors without pectin were well preconcentrated for all tested conditions: in general, increments in TMP led to higher solute concentration factors. For these liquors a complete pore blocking was identified as a predominant mechanism in the earlier stages of the treatment while cake filtration was considered dominant for later stages. The maximum concentration obtained with RO is still far from the value (72°Brix) reached by evaporation: consequently, RO is unsuitable for obtaining citrus molasses directly. Forward osmosis (FO) has been recently investigated as an alternative method for dewatering orange press liquor [51]. In this approach, concentrated draw solutions of NaCl (2M and 4M) were used to remove water from synthetic press liquors through a flat-sheet cellulose acetate membrane (Hydration Technologies Inc., Albany, OR) with a NaCl rejection of 95–99%. Concentration factors up to 3.7 were obtained when using a 4M NaCl solution and a synthetic press liquor without pectin. As in the previous studies of preconcentration by RO, pectin was found to be the main compound responsible in membrane fouling. In particular, the combination of pectin and calcium led to severe flux decays, although citric acid competes with pectin in complexing Ca2+ ions partially mitigating fouling phenomena. Fouling also affected negatively the concentration factor, as a maximum concentration factor of 1.44 was reached when solutions containing pectin were used as feed. Although pectin and its derivatives form a gel-like structure over membrane surfaces reducing the permeate flux, this behavior can be exploited to concentrate and purify pectin solutions by using MF or UF membranes. Pectin is widely used in the food and cosmetic industry as gel-forming agent, stabilizer and emulsifier [52]. Some interesting pharmacological activities (cholesterol decreasing, anti-metastasis, antiulcer) have been also reported [53–55]. Currently, industrial processes for pectin production from citrus peel are based on the use of large amounts of ethanol resulting in high operating costs.

4.6 Treatment of citrus by-products 

 105

Lianwu et al. [56] evaluated the performance of a tubular UF ceramic membrane (ZrO2, 30,000 MW) in the treatment of a pectin-containing solution extracted from citrus peel. They observed a more than 90% retention rate of macromolecular pectin in comparison with pigments and other components for which the retention rate was less than 20%. According to these results the authors concluded that the decolorization, separation and purification of pectin preparations can be achieved simultaneously through the use of UF ceramic membranes. A crossflow MF system based on the use of a regenerated cellulose membrane with a nominal pore size of 0.2 μm (Sartorius, Götingen, Germany) was also investigated to concentrate and purify soluble pectin extracted from mandarin peels [57]. The MF system effectively concentrated pectin extracts (the galacturonic acid content increased about 4.2% at a VRF of 4), saving 75% of ethanol consumption required for the precipitation of pectin. A further purification of pectin was achieved through a diafiltration step that removed undesirable impurities, such as polyphenols and carotenoids, from concentrated pectin extracts. As previously reported, citrus by-products are enriched in bioactive compounds (i.e., flavonoids and phenolic acids) recognized for their beneficial implications in human health because of their antioxidant activity and free radical scavenging ability [58–60]. The recovery of these compounds offers new opportunities for the formulation of products of interest in the food industry (dietary supplements and functional foods production), pharmaceuticals (products with antibacterial, antiviral, anti-inflammatory, antiallergic and vasodilatory action) and the cosmetic industry [2]. In recent years, different studies have proposed the recovery of flavonoids from by-products of orange juice processing based on the use of organic solvents [61], resins [62], heat treatment [63], γ-irradiation [64] and enzymes [65]. However, the proposed methodologies are characterized by some drawbacks. For example: the extraction with organic solvents presents safety problems (some of them are believed to be toxic); low efficiency and highly time-consuming; heat treatment results in pyrolysis; γ-irradiation assisted extraction is still unknown in terms of safety. Recently, membrane technology has attracted attention as an alternative molecular separation technology to conventional systems for the recovery of bioactive compounds from vegetable sources. In particular, a large number of potential applications involving the use of NF membranes have been proposed for the fractionation and concentration of solutes from complex solutions [66–68]. Integrated membrane processes for the recovery of bioactive compounds from orange press liquors can be properly designed to obtain formulations of food or pharmaceutical interest. In these hybrid processes UF is a valid approach to remove from the liquor macromolecules, such as pectins and proteins, ensuring the production of a clarified solution containing health benefit compounds [69]. The optimization of operating conditions (TMP, feed flow rate and temperature) to improve permeate flux and to reduce the fouling index in the UF of orange press

106 

 4 Integrated membrane operations in citrus processing

liquor has been recently investigated by Ruby-Figueroa et al. [70]. The performance of UF HF membranes (PS, 100 kDa, China Blue Star Membrane Technology Co. Ltd., Beijing, China) was analyzed using the response surface methodology approach in order to evaluate the effects of multiple factors and their interactions [71]. Maximum permeate fluxes of 23.7 kg/m2h and a minimum fouling index of 48% were estimated in optimized TMP, temperature and feed flow rate values of 1.4 bar, 15°C and 167 l/h, respectively. By using a similar approach, other authors also evaluated the effect of operating conditions on the membrane rejection towards polyphenols and the recovery of antioxidant compounds in the permeate stream [72]. Optimization of multiple responses established operating parameters that gave maximum recovery of TAA in the permeate and minimum polyphenols rejection, simultaneously. The obtained results indicated a minimum polyphenols rejection of the UF membrane (28.4%) under operating conditions of minimal concentration polarization and fouling (feed flow rate, 244.64 l/h; TMP, 0.2 bar). The potential for NF membranes in the separation and concentration of bioactive compounds from orange press liquors obtained by pigmented orange peels was investigated by Conidi et al. [73]. Spiral-wound NF membranes with different MWCO (180, 300, 400 and 1,000 Da) and polymeric material (PA, polypiperazine amide and PES) were evaluated for their rejection towards anthocyanins, flavonoids and sugars in order to identify a suitable membrane to separate phenolic compounds from sugars. A strong reduction of the average rejection towards sugar compounds was observed by increasing the MWCO of selected membranes while for anthocyanins the rejections were higher than 89% independent of the pore size. In particular, the NF PES 10 membrane with a MWCO of 1,000 Da showed the lowest rejection towards sugar compounds (22.8%) and high rejection towards anthocyanins (89.2%) and flavonoids (69.3%) (Table 4.3). Recently a membrane-based study for the recovery of polyphenols from bergamot juice was investigated by Conidi et al. [74]. Bergamot is a citrus hybrid fruit derived from bitter orange and lemon, exploited for the production of its essential oil widely

Table 4.3: Nanofiltration of clarified orange press liquor. Rejections (R) of NF membranes towards sugars, flavonoids and anthocyanins Membrane Manufacturer type

Membrane material

NF 70 NF 200

Polyamide Polypiperazine-amide

180 300

Polyetehrsulphone Polyetehrsulphone

N30F NF PES10

Dow/Filmtec Dow/Filmtec Microdyn Nadir Microdyn Nadir

Ranthocyanins (%)

Rsugars (%)

95.4 88.4

95.9 94.2

93.4 69.8

400

82.5

93.5

42.8

1000

69.3

89.2

22.8

MWCO Rflavonoids (%) (Da)

4.6 Treatment of citrus by-products 

 107

used for pharmaceutical, cosmetical and food applications. Conversely, the juice has not found a real use in the food industry because of its bitter taste: therefore it is considered a waste of the essential oil production. However, natural phenols of the juice, and especially flavonoids, have a great potential as active principles in the pharmaceutical industry and as antioxidant compounds in the food industry [75–78]. The extraction of polyphenols from vegetable materials with organic solvents, although commonly used in many industrial processes, involves high capital costs and is considered unsafe for food aims because of the presence of solvent traces in the final extract. In addition, polyphenols can be denatured by high temperatures required to increase the extraction rate. In the process investigated by Conidi et al. [74] the bergamot juice was clarified by UF and then submitted to a treatment with UF and NF membranes with different MWCO (450, 750 and 1,000 Da) in order to evaluate their selectivity towards sugars, organic acids and polyphenols. According to the experimental results, an integrated process based on the preliminary UF of the depectinised juice, followed by a NF step with a 450 Da membrane, was proposed. The UF pretreatment produced a removal of suspended solids, reducing fouling phenomena in the following NF step. Flavonoids were recovered in the NF retentate while more than 50% of sugars were recovered in the NF permeate according to the highest difference in the observed rejection towards these compounds for the NF 450 Da membrane (Table 4.4). Recently, Cassano et al. [79] also evaluated the potential of an integrated membrane process for the clarification and concentration of bergamot juice in order to produce a concentrated product that can be used for food or pharmaceutical formulations. The process is based on a preliminary clarification of the depectinised juice with HF UF membranes (PS, 100 kDa, China Blue Star Membrane Technology, Beijing, China) followed by the concentration of the clarified juice by OD with a Liqui-Cell Extra-Flow 2.5 × 8” membrane contactor (Membrana) up to 54°Brix. Suspended solids were completely removed in the UF process. Flavonoids and ascorbic acid were recovered in the UF permeate and well-preserved during the subsequent concentration process. The evaluation of the TAA in clarified and concentrated samples confirmed the validity of the process in producing a concentrated juice without modifying the main quality criteria of the fresh juice.

Table 4.4: Effect of MWCO on the rejection of UF and NF membranes towards sugars and polyphenols in the treatment of clarified bergamot juice Membrane type

Manufacturer

Membrane material

Inopor®nano Inopor®nano Etna 01PP

Inopor Inopor Alfa Laval

TiO2 TiO2 Fluoropolymer

MWCO (Da)

450 750 1000

Rsugars (%) 48.7 30.3 2.1

Rflavonoids (%) 95.4 53.4 3.25

108 

 4 Integrated membrane operations in citrus processing

4.7 Concluding remarks The possibility to realize integrated membrane systems in which all the steps of the productive cycle are based on molecular membrane separations is considered today to be a valid approach for a sustainable industrial growth within the PI strategy. The integration of different membrane operations, or in combination with traditional separation units, offers significant advantages in terms of product quality, energy consumption, plant compactness, environmental impact, recovery of water and high added-value compounds. In this chapter the combination of different membrane operations in citrus production has been presented in order to illustrate its effect on the juice quality and the recovery of high added-value compounds from citrus by-products. Figure 4.8 shows how the traditional flow sheet of the blood orange juice processing can be redesigned through the implementation of an integrated membrane process. The proposed process for the production of highly nutritional concentrated juice is based on the preliminary clarification of the squeezed juice by UF followed by a concentration

Blood oranges

Raw fruit juice

Juice extraction essential oil extraction

Pulp UF

Peels

Essential oils

Pasteurization Clarified juice

Pressing Clarified juice

Press liquor

Diluted brine

Concentrated brine

UF Suspended solids

OD Sugars, minerals NF

Pasteurized pulp, water

Concentrated juice

Enriched polyphenols solution Industrial colorants nutraceuticals pharmaceuticals

Reconstituted juice

Figure 4.8: Integrated membrane process in the industrial transformation of blood oranges

4.8 References 

 109

step by OD. The fractionation of the orange press liquor through an integrated UF/NF process leads to a solution enriched in phenolic compounds thatcan be employed as an industrial colorant or as formulations of pharmaceutical and nutraceutical interest. Advantages of membrane clarification and concentration processes over conventional techniques have been successfully demonstrated. Although today fruit juice concentration by membranes may be more expensive than evaporation, with the enlargement of the world’s fruit juice market and the demand for product quality, commercial applications of membrane processes in concentrated citrus juice processing will expand in the near future. In addition, the utilization of low-cost raw materials, such as citrus wastes, combined with mild technologies, including membrane operations, is expected to offer significant economical and environmental advantages. Research efforts related to the preparation of new membranes both highly selective and permeable, or robust and stable in long-term applications as well as improvements of process engineering including module and process design, are expected to fuel the growth of this technology in citrus juice processing.

4.8 References 1. Food and Agriculture Organization of the United Nations. Citrus fruit. Fresh and processed. Annual statistics. CCP:CI/ST/2012 http://www.fao.org/fileadmin/templates/est/COMM_ MARKETS_MONITORING/Citrus/Documents/CITRUS_BULLETIN_2012.pdf. United Nations: Rome, Italy; 2012. (Accessed May 27, 2013). 2. Benavente-Garcia O, Castillo J, Marin FR, Ortuno A, Del Rio JA. Use and properties of citrus flavonoids. J Agr Food Chem 1997;45:4505–4515. 3. Braddock RJ, Nikdel S, Nagy S. Composition of some organic and inorganic compounds in reverse osmosis–concentrated citrus juices. J Food Sci 1988;53:508–512. 4. Liu Y, Heying E, Tanumihardjo SA. History, global distribution and nutritional importance of Citrus fruits. Compr Rev Food Sci F 2012;11:530–545.  5. Dhuique-Mayer C, Caris-Veyrat C, Ollitrault P, Curk F, Amiot MJ. Varietal and interspecific influence on micronutrient contents in citrus from the Mediterranean area. J Agr Food Chem 2005;53: 2140–2145. 6. Yao LH, Jiang YM, Shi J, Tomás-Barberán FA, Datta N, Singanusong R, Chen SS. Flavonoids in food and their health benefits. Plant Food Hum Nutr 2004;59:113–122. 7. Arena E, Fallico B, Maccarone E. Evaluation of antioxidant capacity of blood orange juices as influenced by constituents, concentration process and storage. Food Chem 2001;74:423–427. 8. Maccarone E, Campisi S, Cataldi Lupo MC, Fallico B, Nicolosi Asmundo C. Thermal treatments effects on the red orange juice constituents. Ind Bevande 1996;25:335–341. 9. Köseoglu SS, Lawhon JT, Lusas EW. Use of membranes in citrus juice processing. Food Technol 1990;44:90–97. 10. Rai P, Majumdar GC, Jayanti VK, Das Gupta S, De S. Alternative pretereatment methods to enzymatic treatment for clarification of mosambi juice using ultrafiltration. J Food Process Eng 2006;29:202–218. 11. Rai P, Majumdar GC, Das Gupta S, De S. Effect of various pretreatment methods on permeate flux and quality during ultrafiltration of mosambi juice. J Food Eng 2007;78:561–568.

110 

 4 Integrated membrane operations in citrus processing

12. Cassano A, Marchio M, Drioli E. Clarification of blood orange juice by ultrafiltration: analyses of operating parameters, membrane fouling and juice quality. Desalination 2007;212:15–27. 13. Cassano A, Tasselli F, Conidi C, Drioli E. Ultrafiltration of clementine mandarin juice by hollow fibre membranes. Desalination 2009;241:302–308. 14. Todisco S, Peña L, Drioli E, Tallarico P. Analysis of the fouling mechanism in microfiltration of orange juice. J Food Process Pres 1996;20:453–466. 15. Sarkar B, DasGupta S, De S. Flux decline electric field-assited crossflow ultrafiltration of mosambi (Citrus sinensis (L.) Osbeck) juice. J Membrane Sci 2009;331:75–83. 16. Hernandez E, Chen CS, Shaw PE, Carter RD, Barros S. Ultrafiltration of orange juice: effect on soluble solids, suspended solids, and aroma. J Agr Food Chem 1992;40:986–988. 17. Todisco S, Tallarico P, Drioli E. Modelling and analysis of the effects of ultrafiltration on the quality of freshly squeezed orange juice. Ital Food Bev Technol 1998;12:3–8. 18. Espamer L, Pagliero C, Ochoa A, Marchese J. Clarification of lemon juice using membrane process. Desalination 2006;200:565–567. 19. Boylston TD. Fruit Juices. In: Hui YH, ed. Handbook of Food Products Manufacturing. John Wiley & Sons, Inc.: Hoboken, NJ; 2007;847–866. 20. Alvarez S, Riera FA, Alvarez R, et al. A new integrated membrane process for producing clarified apple juice and apple juice aroma concentrate. J Food Eng 2000;46:109–125. 21. Chen CS, Shaw PE, Parish ME. Orange and tangerine juices. In: Nagy S, Chen CS and Shaw PE, eds. Fruit Juice Processing Technology. AgScience: Auburndale, FL; 1993;110–165. 22. Merson RL, Paredes G, Hosaka DB. Concentrating fruit juices by reverse osmosis. In: Cooper AR, ed. Ultrafiltration Membranes and Applications. Plenum Press: New York, USA; 1980;405–413. 23. Merson RL, Morgan AI. Juice concentration by reverse osmosis. Food Technol 1968;22:631–634. 24. Medina BG, Garcia A. Concentration of orange juice by reverse osmosis. J Food Process Eng 1988;10:217–230. 25. Fukutani K, Ogawa H. A comparison of membrane’s suitability and effect of operating pressure for juice concentration by reverse osmosis. Nippon Shokuhin Kogyo Gakk 1983;30:636–641. 26. Gadea A. Reverse osmosis of orange juice. Proceedings of the International Fruit Juice Congress. Orlando, FL, USA, 1987. 27. Sheu MJ, Wiley RC. Preconcentration of apple juice by reverse osmosis. J Food Sci 1983;48: 422–429. 28. Kunz W, Benhabiles A, Ben-Aim R. Osmotic evaporation through macroporous hydrophobic membranes: a survey of current research and applications. J Membrane Sci 1996;121:25–36. 29. Lawson KW, Lloyd DR. Membrane distillation. J Membrane Sci 1997;124:1–25. 30. Hogan PA, Canning RP, Peterson P, Johnson RA, Michaels AS. A new option: osmotic distillation. Chem Eng Prog 1998;7:49–61. 31. Drioli E, Jiao B, Calabrò V. The preliminary study on the concentration of orange juice by membrane distillation. Proc Int Soc Citriculture 1992;3:1140–1144. 32. Calabrò V, Jiao B, Drioli E. Theoretical and experimental study on membrane distillation in the concentration of orange juice. Ind Eng Chem Res 1994;33:1803–1808. 33. Shaw PE, Lebrun M, Dornier M, Ducamp MN, Courel M, Reynes M. Evaluation of concentrated orange and passionfruit juices prepared by osmotic evaporation. LWT Food Sci Technol 2001;34: 60–65. 34. Cissé M, Vaillant F, Perez A, Dornier M, Reynes M. The quality of orange juice processed by coupling crossflow microfiltration and osmotic evaporation. Int J Food Sci Tech 2005;40: 105–116. 35. Galaverna G, Di Silvestro G, Cassano A, Sforza S, Dossena A, Drioli E, Marchelli R. A new integrated membrane process for the production of concentrated blood orange juice: Effect on bioactive compounds and antioxidant activity. Food Chem 2008;106:1021–1030. 36. Destani F, Cassano A, Fazio A, Vincken JP, Bartolo G. Recovery and concentration of phenolic compounds in blood orange juice by membrane operations. J Food Eng 2013;117:263–271.

4.8 References 

 111

37. Liqui-Cell. Data sheets of Liqui-Cel Extra-Flow products. http://www.liqui-cel.com/productinformation/data-sheets.cfm. (Accessed May 27, 2013) 38. Thompson D. The application of osmotic distillation for the wine industry. Aust N Z Grapegrow Winemak 1991;11:11–14. 39. Michaels AS. Osmotic Distillation Process Using a Membrane Laminate. US Patent 1999;5938928. 40. Barbe AM, Bartley JP, Jacobs AL, Johnson RA. Retention of volatile organic flavor/fragrance components in the concentration of liquid foods by osmotic distillation. J Membrane Sci 1998; 145:67–75. 41. Alves VD, Coelhoo IM. Orange juice concentration by osmotic evaporation and membrane distillation. J Food Eng 2006;74:125–133. 42. Nagai K. Fundamentals and perspectives for pervaporation. In: Drioli E, Giorno L, eds. Comprehensive Membrane Science and Engineering, Volume 2: Membrane Applications in Molecular Separations. Elsevier: Oxford, UK; 2010;243–271. 43. Lipnizki F, Olsson J, Trägårdh G. Scale-up of pervaporation for the recovery of natural aroma compounds in the food industry. Part 1: simulation and performance. J Food Eng 2002;54: 183–195. 44. Sahin S. Principles of pervaporation for the recovery of aroma compounds and applications in the food and beverage industries. In: Rizvi SSH, ed. Separation, Extraction and Concentration Processes in the Food, Beverage and Nutraceutical Industries. Woodhead Publishing: Cambridge, UK; 2010;219–243. 45. Aroujalian A, Raisi A. Recovery of volatile aroma components from orange juice by pervaporation. J Membrane Sci 2007;303:154–161. 46. Shepherd A, Habert AC, Borges CP. Hollow fiber modules for orange juice aroma recovery using pervaporation. Desalination 2002;148:111–114. 47. Braddock RJ. Handbook of citrus by-products and processing technology. John Wiley & Sons Inc.: New York, NY; 1999. 48. Garcia E, Gozálves JM, Lora J. Use of reverse osmosis as a preconcentration system of waste leaching liquid from the citric juice production industry. Desalination 2002;148:137–142. 49. Garcia-Castello EM, Lora-Garcia J, Garcia-Garrido J, Rodriguez-Lopez AD. Energetic comparison for leaching waste liquid from the citric juice production using both reverse osmosis and multiple-effect evaporation. Desalination 2006;191:178–185. 50. Garcia-Castello EM, Mayor L, Chorques S, Arguelles A, Vidal-Brotons D, Gras ML. Reverse osmosis concentration of press liquor from orange juice solid wastes: Flux decline mechanisms. J Food Eng 2011;106:199–205. 51. Garcia-Castello EM, McCutcheon JR. Dewatering press liquor derived from orange production by forward osmosis. J Membrane Sci 2011;372:97–101. 52. Thakur BR, Singh RK, Handa AK. Chemistry and uses of pectin. Crit Rev Food Sci Nutr 1997;37:47–73. 53. Platt D, Raz A. Modulation of the lung colonization of B16–F1 melanoma cells by citrus pectin. J Natl Cancer I 84;438–442. 54. Kiyohara H, Hirano M, Wen XG, Matsumoto T, Sun XB, Yamada H. Characterization of an anti-ulcer pectic polysaccharide from leaves of Panax ginseng C.A. meyer. Carbohyd Res 1994;263:89–101. 55. Ismail MF, Gad MZ, Hamdy MA. Study of the hypolipidemic properties of pectin, garlic and ginseng in hypercholesterolemic rabbits. Pharmacol Res 1999;39:157–166. 56. Lianwu X, Xiang L, Yaping G. Ultrafiltration behaviours of pectin–containing solution extracted from citrus peel on a ZrO2 ceramic membrane pilot unit. Korean J Chem Eng 2008;25:149–153. 57. Cho CW, Lee DY, Kim CW. Concentration and purification of soluble pectin from mandarin peels using crossflow microfiltration system. Carbohyd Polym 2003;54:21–26. 58. Anagnostopoulou MA, Kefalas P, Papageorgiou VP, Assimopoulou AN, Boskou D. Radical scavenging activity of various extracts and fractions of sweet orange peel (Citrus sinensis). Food Chem 2006;94:19–25.

112 

 4 Integrated membrane operations in citrus processing

59. Bocco A, Cuvelier ME, Richard H, Berset C. Antioxidant activity and phenolic composition of citrus peel and seed extract. J Agr Food Chem 1998;46:2123–2129. 60. Yu J, Wang L, Walzem RL, Miller EG, Pike LM, Patil BS. Antioxidant activity of citrus limonoids, flavonoids, and coumarins. J Agr Food Chem 2005;53:2009–2014. 61. Li BB, Smith B, Hossain MdM. Extraction of phenolics from citrus peels I. Solvent extraction method. Sep Purif Technol 2006;48:182–188. 62. Di Mauro A, Fallico B, Passerini A, Maccarone E. Waste water from Citrus processing as a source of hesperidin by concentration on styrene-divinylbenzene resin. J Agr Food Chem 2000;48: 2291–2295. 63. Xu GH, Ye XQ, Chen JC, Liu DH. Effect of heat treatment on the phenolic compounds and antioxidant capacity of citrus peel extract. J Agr Food Chem 2007;55:330–335. 64. Oufedjikh H, Mahrouz M, Amiot MJ, Lacroix M. Effect of γ-irradiation on phenolic compounds and phenylalanine ammonia-lyase activity during storage in relation to peel injury from peel of Citrus clementina Hort. Ex. Tanaka. J Agr Food Chem 2000;48:559–565. 65. Li BB, Smith B, Hossain MdM. Extraction of phenolics from citrus peels II. Enzyme-assisted extraction method. Sep Purif Technol 2006;48:189–196. 66. Mello BCBS, Petrus JCC, Hubinger MD. Concentration of flavonoids and phenolic compounds in aqueous and ethanolic propolis extracts through nanofiltration. J Food Eng 2010;96:533–539. 67. Cissé M, Vaillant F, Pallet D, Dornier M. Selecting ultrafitration and nanofiltration membranes to concentrate anthocyanins from roselle extract (Hibiscus sabdariffa L.). Food Res Int 2011;44: 2607–2614. 68. Tylkowski B, Tsibranska I, Kochanov R, Peev G, Giamberini M. Concentration of biologically active compounds extracted from Sideritis ssp. L. by nanofiltration. Food Bioprod Process 2011;89:307–314. 69. Pap N, Mahosenaho M, Pongrácz E, et al. Effect of ultrafiltration on anthocyanin and flavonol content of black currant juice (Ribes nigrum L.). Food Bioprocess Tech 2012;5:921–928. 70. Ruby Figueroa R, Cassano A, Drioli E. Ultrafiltration of orange press liquor: optimization for permeate flux and fouling index by response surface methodology. Sep Purif Technol 2011;80:1–10. 71. Anjum MF, Tasadduq I, Al-Sultan K. Response surface methodology: a neutral network approach. Eur J Oper Res 1997;101:65–73. 72. Ruby-Figueroa R, Cassano A, Drioli E. Ultrafiltration of orange press liquor: optimization of operating conditions for the recovery of antioxidant compounds by response surface methodology. Sep Purif Technol 2012;98:255–261. 73. Conidi C, Cassano A, Drioli E. Recovery of phenolic compounds from orange press liquor by nanofiltration. Food Bioprod Process 2012;90:867–874. 74. Conidi C, Cassano A, Drioli E. A membrane-based study for the recovery of polyphenols from bergamot juice. J Membrane Sci 2011;375:182–190. 75. Mollace V, Sacco I, Janda E, Malara C, Ventrice D, Colica C, Visalli V, Muscoli S, Ragusa S, Muscoli C, Rotiroti D, Romeo F. Hypolipemic and hypoglycaemic activity of bergamot polyphenols: from animal models to human studies. Fitoterapia 2011;82:309–316. 76. Pernice R, Borriello G, Ferracane R, Borrelli R, Cennamo F, Ritieni A. Bergamot: a source of natural antioxidants for functionalised fruit juices. Food Chem 2009;112:545–550. 77. Di Donna L, De Luca G, Mazzotti F, Napoli A, Salerno R, Taverna D, Sindona G. Statin-like principles of Bergamot fruit (Citrus bergamia): Isolation of 3-hydroxymethylglutaryl flavonoid glycosides. J Nat Prod 2009;72:1352–1354. 78. Miceli N, Mondello MR, Monforte MT, Sdrafkakis V, Dugo P, Crupi ML, Taviano MF, De Pasquale R, Trovato A. Hypolipidemic effect of Citrus bergamia Risso et Poiteau juice in rats fed a hypercholesterolemic diet. J Agr Food Chem 2007;55:10671–10677. 79. Cassano A, Conidi C, Drioli E. A membrane-based process for the valorization of the bergamot juice. Sep Sci Technol 2013;48:537–546.

5 Integrated membrane and conventional processes applied to milk processing Germano Mucchetti 5.1 Introduction In 2012 more than 474 × 106 tons of milk were produced worldwide [1]. The production of fluid milk, whole dried milk, skimmed dried milk and cheese accounted for 36.8%, 7.4%, 9.5% and 32.0% of total milk, respectively. The remaining 14.3% of milk was transformed into other products (butter, proteins, fermented milk, etc). Processing each of these milk products is traditionally characterized by different operations of separation. The oldest operation, common to all the three milk product categories, is the standardization of milk fat content or milk skimming by centrifugal separation. Fluid milk stabilization for quality and safety until the expiry date was proposed by Appert in 1831 coupling water partial removal by means of heat and mass transfer operations with in container heat sterilization of concentrated milk [2]. More than one century occurred before fluid milk heat pasteurization became common practice and the first standard (61°C for 30 min) was proposed in the USA [3]. Around 65  years occurred before crossflow tangential microfiltration (MF) was industrially applied in France, where the Laiterie de Villefranche sur Saone started to produce the milk called “Marguerite” by mixing raw microfiltered skim milk with heat pasteurized cream. Cheese may be considered as the typical result of the selective concentration of some milk components (casein and fat) by separation from the others (whey proteins, lactose and water). Separation is traditionally obtained by casein modification caused by enzymatic hydrolysis or acid addition followed by drainage of the whey fraction through a pierced mould or a cloth. Whey separation can be alternatively performed by ultrafiltration (UF), or prevented by milk preconcentration by UF, depending on cheese processes. The application of membrane operation to milk processing improved the separation options that are industrially applied today, enhancing the separation specificity (e.g., microbial spores, somatic cells or native phosphocasein) and efficacy when compared to traditional technologies. The aim of this chapter is to describe the most relevant applications of membrane separation to milk processing, as integrated in the actual manufacturing of some important milk products (fluid milk and cheese) and to compare them to the conventional processes.

114 

 5 Integrated membrane and conventional processes applied to milk processing

5.2 Fluid milk The shelf-life of fluid milk mainly depends on microbial count and enzymatic activity residual to heat and/or other physical treatment, and on the efficacy of the milk packaging conditions (clean, ultraclean or aseptic filling) in preventing post-contamination [4, 5]. Pasteurized and extended shelf-life (ESL) milk present a heat-resistant microbial count, and for this reason both pasteurized and ESL milk have to be refrigerated, while UHT and in-bottle sterilized milk are sterile products that can be stored at room temperature [5].

5.2.1 MF and bacterial removal To extend milk shelf-life without increasing the time and/or the temperature of heat treatment, which is responsible for higher losses of nutritional value and generally unwanted milk changes, an efficient strategy is the reduction of raw milk heatresistant microbial counts by non-thermal treatments. With the same number of decimal reductions or lethal effect, a lower initial microbial count leads to a lower residual count and to a potentially longer shelf-life of the heat-treated milk. Centrifugal separation (e.g., bactofugation™) and MF are two competing or synergic operations that are able to reduce the number of microbial cells contaminating raw milk. Bacterial removal from milk by separators, mainly heat-resistant spores, is a relatively old technology [6], applied firstly to fluid milk and then to cheese milk (hardcooked Italian cheeses), to prevent a late blowing defect [7, 8]. Separation is based on the difference of size and specific gravity (SG) of the milk particles (fat globules, casein micelles, bacterial and somatic cells) (Table 5.1) and the continuous phase of milk, as defined by Stoke’s law. Distribution of bacteria sizes is rather large and the smallest bacterium is Brevundimonas diminuta, usually, but not always, retained by a 0.2 μm rated membrane filter [9]. Spoilage and pathogenic microorganisms present in raw milk are usually a larger size [10, 11]. SG of bacterial spores (1,120 to 1,380  kg/m3 according to species and measurement method [12 ], and SG of vegetative bacterial cells (1020 –1094 kg/m3) [13] are higher than SG of milk (1028–1032 kg/m3). As the difference of SG between spores and milk higher is than that between vegetative cells and milk, centrifugal separation removes spores more efficiently than vegetative cells from milk. Centrifugation can reduce the total bacterial count by 80 –90%, the number of anaerobic sporeforming microorganisms such as Clostridium by 98–99.5% and that of aerobic sporeformers such as Bacillus by 95% [14 ].

5.2 Fluid milk 

 115

Table 5.1: Composition of milk and characteristics of its particles (dispersion, size and options of membrane separation) Component

Form of the dispersion of the component

Approximate content

Size (nm)

Separation method

Microorganism Somatic cells Fat Casein

Suspension Suspension Emulsion Colloidal Soluble Soluble Colloidal Soluble Soluble

< 100.000 CFU/ml < 300.000 cells/l ~3.8 g/100 g ~2.6 g/100 g

~200–5000 ~ 6000–15,000 ~500–10,000 ~40–300 ~10 ~3–6

MF

Whey proteins Ash Non-protein nitrogen Lactose Water

Soluble

~0.6 g/100 g ~0.54 g/100 g ~0.26 g/100 g ~0.20 g/100 g ~4.8 g/100 g ~87.2 g/100 g

UF

< 0.5 90%

Sweet whey (6% dry matter) Concentrated whey (18–26% dry matter)

Nanofiltration

Resins + Nanofiltration

Electrodialysis

Electrodialysis

Electrodialysis + Resins + Nanofiltration Resins + Electrodialysis

138 

 6 Integrated membrane operations in whey processing

a simple process that has the advantage of simultaneously concentrating the liquid (20–22% of dry matter at a VRR of ~4), which is often desired, and demineralizing (35–50% and even 70% with diafiltration). As whey in most instances has to pass through a concentration stage prior to further processing, the NF option is very attractive because the demineralization is obtained without further cost. In the past, the first concentration was often carried out using RO. However, nanofiltration is more appropriate for the concentration of whey because it simultaneously demineralizes the whey proteins: because of the low osmotic pressure difference between retentate and permeate compared to RO (attributed to the transfer of monovalent ions), the transmembrane pressure is lower and the operation is generally more cost-effective. Nanofiltration offers low investment costs and simple installations, which are easy to run. Moreover the amount of effluent is greatly reduced in comparison with the other demineralization processes. Electrodialysis and ion-exchange actually lead to high investment and running costs, mainly due to membrane, spacers and electrodes replacement as well as wastewater treatment for electrodialysis and high consumption of regeneration chemicals for ion-exchange. It has been demonstrated that the running cost of these demineralization techniques are 25–55% higher than NF. In addition, the effluents generated by nanofiltration have a lower BOD compared to other demineralization processes, but still require further treatment (classically RO) before sending to the purification treatment plant. The loss of lactose, non-protein nitrogen (N) and protein in nanofiltrate are today lower than those found in electrodialysis or ion-exchange, making retentate more valuable and leading to a permeate with lower BOD. Urea does leak quite extensively. Also, organic acids like lactic and acetic acid can pass through the membrane to a large extent, presenting the possibilities of desacidification of acid whey. For large demineralization installations (those treating more than 400 m3/day), and depending on the proportion of salts to be removed, investment in combining technologies may be of interest (Table 6.2). Today many modern demineralization plants are combinations of classical ion-exchange and/or electrodialysis with NF. By doing this, the ionic load of the ion exchangers is reduced in combination with lower volumes to treat, resulting in principle in a reduction in the size of the columns. Other applications of NF in whey processing include the concentration and partial demineralization of whey UF permeates prior to the manufacture of lactose and lactose derivatives.

6.4 Concentration of serum proteins The development of UF membrane processes in the 1970s has offered new possibilities to fully exploit the nutritional, biological and functional properties of the whey proteins. Whey is nowadays considered as a valuable by-product, leading to

6.4 Concentration of serum proteins 

 139

high-value-added products. However until the 1970s, the interesting properties of whey proteins could not be fully exploited because of the damaging effect of heat/ acid treatments used for their separations from whey. It has actually long been practice to fully obtain whey proteins in a denatured form, by heating acidified whey (temperature > 90°C and pH < 6) according to the technology used for making whey cheeses, such as ricotta. Today most whey protein concentrates are produced using UF. By a combination of UF and diafiltration it is possible to produce whey protein concentrates (WPC) with 35–80% protein (expressed in nitrogen N × 6.38 over dry matter), compared with whey powder with only 12–15% protein. The upper limit of 80% is usually dictated by the residual fat content of the starting whey – fat that is preferentially concentrated along with proteins. Most WPCs on the market contain either 35% or ~80% protein and are important sources of protein for a large variety of food products, ranging from processed meat and sausages, to health foods, beverages and confectionery. WPCs containing ~35% protein for example are used in the manufacture of processed cheese, yogurt and infant formula, and in various bakery applications. The concentrates are also marketed for use in stews and sauces because of their thickening properties, and meat patties. Ultrafiltration membranes usually have a cut-off, ranging from 10 to 20 kg/mol in order to remove both the lactose and ions, and retain proteins. The obtained retentate can be then further processed by diafiltration (for protein purification) followed by evaporation and spray-drying. The protein content of the final whey protein product depends on the degree of concentration during UF: a VRR of 4.5–7.0 is required for 35% WPC, and it should reach 13–20 for 50–60% WPC. Combined with a diafiltration, which removes minerals and lactose from the retentate, whey UF (VRR 30–35) can lead to WPC purity of 75–80%. The fouling of UF membrane treating whey was largely studied in the 1980s and 90s and it was demonstrated that during UF, membrane fouling is mainly attributed to three different species: (i) presence of residual lipids coming from the cheese manufacture and still present even after a previous centrifugation of the whey to be treated; (ii) accumulation of proteins at the membrane surface, more pronounced at pH closed to their isolectric point (pH ~ 5.0–5.5) and (iii) precipitation of calcium phosphate enhanced under neutral pH (7.0–7.5) and high temperature (55°C). Over the past years several whey pre-treatments have been proposed [7]. Some pre-treatments increase the purity of the final concentrates, especially by reducing the residual lipids content, which impairs the functionalities of whey proteins (emulsifying, foaming and gelling characteristics) and promote, development of off-flavors. Some other pre-treatments improve UF performance, especially by limiting calcium phosphate precipitation and protein accumulation [8]. Among them, some use membrane operations. MF was included as a first step for removal of bacteria, casein fines and fat, thus contributing to an improved quality of the final product. More recently, an optimized pretreatment, based on a “thermocalcic aggregation process” of the

140 

 6 Integrated membrane operations in whey processing

residual fat initially suggested by Pearce [9] and Maubois’s group [10] was proposed and consists of several steps: 1. Whey is first concentrated by UF until a concentration of 4–5. 2. The pH of the retentate is increased and adjusted to 7.5. 3. The temperature is maintained at 55°C for 8 min (in order to favor the aggregation of the lipoproteins-Ca). 4. The formed aggregates as well as the small fat globules and bacteria are separated using a 0.1 μm membrane MF. Owing to its high content of phospholipids, whey MF retentate that represents a volume of no more than 2% of the initial volume of whey [11] has potential as an effective emulsification agent for food applications or cosmetics. The absence of fat in the “clarified” whey (permeate) and high pH strongly reduce the fouling of subsequent UF, resulting in longer running time and higher permeation flux. Such a pretreatment of whey accompanied by the introduction of a diafiltration in the UF process allows whey protein isolates (WPI) with 90% protein in the total solids and high foaming and gelling properties. Although high performances are obtained with this pretreatment, higher results are observed for WPI issued from the “ideal” whey. Because of its composition and because of the native state of its proteins, the microfiltrate of skimmed milk MF (0.1 μm) is today the most appropriate fluid to produce efficiently WPI up to 95% protein in the total solids with very high functional and nutritional properties, in particular in comparison with WPC obtained from cheese whey. In industrial processes, whey UF is performed mainly in multi-stage SW systems with polyethersulfone membranes (Figure 6.2). Processes are currently operated at temperatures either around 50°C or 10°C. Operating at temperatures of 50°C requires a

Figure 6.2: UF plant for whey protein concentration (Courtesy of SPX Flow Technology SAS, France)

6.5 Fractionation of individual serum proteins 

 141

pretreatment in order to avoid severe fouling during operation, mainly due to calcium phosphate in these conditions. Flux is about twice as high as flux at 10°C, which is a major incentive for operation at such a high temperature, and thermal denaturation of proteins is minimized. However, because of the relatively low membrane prices, most of the manufacturers of protein concentrates prefer to operate at a lower temperature (10–12°C) in spite of a lower flux. In these conditions, a much lower growth of thermoduric bacteria in the spiral-wound filtration equipments is observed, which results in a better microbiological quality of the end product. In addition membrane fouling was reduced because of the increase of solubility of calcium phosphate.

6.5 Fractionation of individual serum proteins To exploit the particular properties of individual proteins, which are known to exert a wide range of nutritional, functional and biological activities (Table 6.3), fractionation of whey protein mixture for the isolation of one or a group of proteins is useful. Some putative activities of serum proteins are digestive function (β-lactoglobulin), anti-carcinogenic (α-lactalbumin), antimicrobial (lactoferrin and lactoperoxidase) and passive immunity (immunoglobulins). Considerable progresses have been made over the last 20 years in technologies aimed at separation, fractionation and isolation in a purified form of many interesting proteins occurring in both bovine colostrum and milk. Most of them however are based on gel filtration and chromatographic techniques. Some industrial-scale methods have been developed using membrane technologies for isolation of some proteins but their large-scale manufacture is still limited [13]. WPI, either obtained from defatted whey or microfiltrate of skimmed milk MF are both excellent starting materials for the industrial production for further fractionation of concentrates into individual proteins. Table 6.3: Some biological functionalities of serum proteins (adapted from [12]) Biological functions β-lactoglobulin α-lactalbumin Immunoglobulins Bovine serum albumin Lactoferrin

Lactoperoxydase

Retinol carrier, potential antioxidant, precursor for bioactive peptides, binds fatty acids Lactose synthezis in mammary gland, calcium carrier, immunomodulator, precursor for bioactive peptide Specific immune protection (antibodies and complement system), potential precursor for bioactive peptides Precursors of bioactive peptides Antimicrobial, antioxidative, anti-carcinogenic, anti-inflammotory, iron transport, call growth, regulation, precursor for bioactive peptides, immunomodulator Antimicrobial, synergetic effect with immunoglobulins and lactoferrin

142 

 6 Integrated membrane operations in whey processing

Some studies have led to the industrial manufacture of enriched protein fractions or separate proteins such α-lactalbumin, β-lactoglobulin and immunoglobulins using membrane technologies. The cost of extracting these proteins is high but often justified when recognizing the great value-added benefits when incorporated in hygiene products, functional foods and nutraceutical products. Among the commercially interesting proteins, the two main serum proteins, α-lactalbumin and β-lactoglobulin, can be produced in enriched fractions using membrane or centrifugation processes: α-lactalbumin has a great potential market because of its high content in tryptophan (4 residues per mole) and in infant milk formula. α-lactalbumin is the main soluble protein of the human milk, and is therefore required in concentrated forms for the manufacture of adapted formula for infant use. The main utilizations of β-lactoglobulin appear to be in gel and foam-type products and in the manufacture of protein hydrolysates for food ingredients. As initially proposed by Pearce [9] a heat treatment combined with pH adjustment process has been developed to fractionate α-lactalbumin and β-lactoglobulin from whey. This involves the reverse aggregation of α-lactalbumin by heating at 55°C for 30 min at a low pH and separation of the resulting aggregates consisting of α-lactalbumin and whey proteins others than β-lactoglobulin by MF or centrifugation. α-lactalbumin is actually a calcium metalloprotein that it contains one mole of calcium per mole of protein. This protein loses its bounded calcium and its stability when at ~55°C (30 min) pH is lowered to 3.8. At this pH calcium is released in the solution and α-lactalbumin unfolds and precipitates at temperatures of 50–65°C. Such physicochemical conditions involve the reversible polymerization of the protein that precipitates, together with immunoglobulins and bovine serum albumin [14]. The permeate or supernatant contains the β-lactoglobulin fraction that can be then processed by UF in combination with diafiltration to yield purified β-lactoglobulin (95% purity). Starting from the permeate of milk microfiltration performed at temperatures lower than 45°C, this principle can be used to produce high purity non-lactosylated β-lactoglobulin [15]. If highly purified β-lactoglobulin is obtained through this process, there are still problems, to our knowledge, related to the purity of the industrially recovered α-lactalbumin, that can be recovered from the retentate/sediment after solubilization at neutral pH, followed by UF. The purity of α-lactalbumin does not exceed 60–70% because of the presence of some denaturated immunoglobulins, contamination by β-lactoglobulin and bovine serum albumin. Further work is required for better knowledge of the structural conformation of this protein and of its interactions with the other protein components present in whey. UF membrane with a cut-off approximately 100 kg/mol or more can also be used to isolate immunoglobulins (Igs) from whey, but whey is a poor source of Igs compared to colostrum or milk produced by hyperimmunized cows. Cow colostrum contains substantially higher concentrations of immunoglobulins than mature milk (20–200 g/l vs. 0.15–0.8 g/l) and can then be used as an appropriate starting fluid for a two-step immunoglobulins extraction procedure. Korhonen et al. [16] used various

6.6 Development of new value-added products from whey 

 143

filtration techniques (RO, UF, MF) and cation-exchange resins as a molecular sieve, to concentrate immunoglobulins from colostral whey. The Igs level of the final freezedried concentrates varied from 45 to 75%. Piot et al. [17] obtained enriched Igs fraction using MF and UF techniques: the colostrum is first microfiltered using a 0.1 μm pore membrane to obtain a permeate (named “serocolostrum”) that is crystal clear, free of blood and somatic cells as well as fat globules and casein micelles. The permeate that contains 80% of the initial immunoglobulins can then be further concentrated using ultrafiltration (100 kg/mol). Commercial immunoglobulins products are mostly used in veterinary medicine or neonatal ruminants and pigs. Because ruminants are born without blood antibodies, they are very susceptible to infection, and it is highly desirable that they receive protection either by suckling colostrum for at least 1 week or by ingesting an immunoglobulin concentrate.

6.6 Development of new value-added products from whey Membrane operations are now viewed as efficient tools for the development of new added-value products, making it possible to separate minor compounds such as bioactive peptides, or growth factors from whey or fermented whey-based products. Milk and whey are rich sources of bioactive peptides and it is now established that pressure-driven membrane-based processes can be used to fractionate peptide mixtures and amino acids. Bioactive peptides have a positive impact on body functions and conditions and may ultimately influence health (regulation of weight; mood, memory and stress control; immune defence, improvement of heart, bone, dental digestive health, etc.) [18]. Their production is therefore of considerable commercial interest for applications in the food sector. The market of bioactive peptides is increasing because the possibility of designing new dairy products with health-promoting benefits looks promising and offers a perspective for consumers and producers. Over the past few decades a number of methods have been developed for their purification, and to date some casein-derived peptides have been manufactured at an industrial scale. Among the different methods used to fractionate and enrich peptides (precipitation with salts or solvents, filtration or chromatography), NF allows the separation of peptides based both on the size and charge of molecules. A significant advantage of membrane processes such as UF or NF is that membranes can be added to the production process of peptides (through enzymatic hydrolysis of precursor proteins, using gastrointestinal enzymes, usually pepsin and trypsin or fermentation) and the product can be separated continuously in the “bioreactors”. The application of UF reactor for continuous extraction of permeates enriched with bioactive peptides has been described for the production of several peptides [12]. For the preparation of phosphopeptides from casein, Brulé et al. [19] proposed, for example, the use of an ultrafiltration membrane for processing permeate after the digestion of caseinate in solution with a proteolytic enzyme. The separation of

144 

 6 Integrated membrane operations in whey processing

the phosphopeptides present in the permeate was performed by ultrafiltration of the peptides solution after addition of a bivalent cation salt (calcium chloride) so as to cause aggregation of phosphopeptides. The non-phosphorylated peptides pass the membrane and diafiltration against water, used to purify the phosphopeptides in the retentate, results in a preparation which is rich ( > 90% w/w) in the desired phosphopeptides. The glycomagropeptide (C-terminal part of the κ-casein release in whey by the action of chymosin) was shown to be separated from sodium caseinate using UF membrane. This peptide has numerous uses (action on satiety, inhibition of Escherichia coli cells adhesion to intestinal walls, etc) and in particular it contains no Phe, which makes it suitable for use as a nutritional protein supplement for patients suffering from phenylketonuria, who did not digest protein with phenylalanine owing to their lack in the appropriate degrading enzyme. The occurrence of many bioactive peptides in bovine milk is now well established [18], but at present the industrial-scale production of such peptides is limited by a lack of suitable separation technologies. Among them, membrane techniques, such as NF or UF are used industrially to produce ingredients that contain bioactive peptides based on casein or whey protein hydrolysates and seem to be the best technology available for the enrichment of bioactive peptides. Growth factors such as transforming growth factor β (TGF-β) and insulin-like growth factor I and II have been classically separated from whey by means of cationexchange chromatography [20]. However, some recent developments of membrane applications have enabled the recovery of growth factors from whey [21]. Some attention has already been paid to the recovery of the growth factor from bovine colostrum, which typically contains 10–15 times the amount of milk in terms of growth factors. MF and UF separations have then been proposed for the extraction of these compounds [17]. Industrial applications of such separations do not exist yet, but providing some solid scientific evidences on their bioactivity in humans are developed, a sustainable market can be expected.

6.7 Conclusions and challenges Membrane operations have revolutionized the field of whey processing in many aspects, and have been part of profound changes in the dairy industry worldwide. In the last 40 years, the approach to whey processing and utilization has changed from considering whey as a simple waste to capitalizing on the opportunities that the whey offers for product innovations. Nowadays, industrial processing of whey is a highly specialized, technologically advanced segment of the dairy industry, requiring up-to-date knowledge and focused attention. The ultrafiltration processes, largely used worldwide for the production of protein concentrates and isolates, partially solve the problems of dealing with large volumes of whey or whey permeate in traditional cheese

6.8 References 

 145

manufactures. Nanofiltration and in a lesser extent microfiltration enabled the development of whey-based ingredients. Despite the impressive growth of membrane applications in this sector, membrane technologies have still not been fully exploited for the development of added-value ingredients and products, mainly because of commercial constraints such as immature market demand for value-added ingredients, occurrence of alternative technologies or introduction of non-dairy equivalents of dairy ingredients. Although a wide variety of fascinating new applications of membrane technology in dairy processes are expected in the next decade, the field of membrane science is now facing important challenges worldwide, such as shortage of drinking water supplies, global warning and a potential global energy crisis. With the reduction of available water and the constant rise of energy costs, it is imperative that control of water and energy uses will be a key factor in the development and growth of membrane-based processes in the future. In that context some recent work of Omont et  al [22] carried out on the fractionation of whey proteins discussed the potentialities of membrane processes compared to chromatographic techniques. Control of fouling and improvements of cleaning steps will undoubtedly remain as a highpriority research domain, whereas development of new module designs, coupling of separation operations, optimization of the overall fractionation processes might offer interesting alternatives.

6.8 References 1. Marshall KR. Industrial fractionation of milk proteins: serum proteins. In: Fox PF, ed. Developments in Dairy Chemistry–1. Applied Science Publishers: New York; 1982. 2. Maubois JL. New applications of membrane technology in the dairy industry. Aus J Dairy Technol 1991;46:91–95. 3. Gésan-Guiziou G, Boyaval E, Daufin G. Critical stability conditions in crossflow microfiltration of skimmed milk: transition to irreversible deposition. J Membrane Sci 1999;158:211–222. 4. Jeantet R, Schuck P, Famelart MH, Maubois JL. Nanofiltration benefit for production of spray-dried demineralized whey powder. Lait 1996;76:283–301. 5. Gernigon G, Schuck P, Jeantet R. Demineralization. In: Fuquay JW, Fox PF, McSweeney PLH, eds. Encyclopedia of Dairy Science, 2nd edn, Vol. 4. Elsevier: London; 2011,738–743. 6. Largeteau D (Eurodia). Electrodialysis in food processing. Aarhus seminar. 2009. 7. Pouliot Y, Jelen P. Pretreatments of dairy fluids to minimize long-term membrane fouling. In: IDF ed. Fouling and cleaning in membrane pressure driven membrane processes. Bulletin of the IDF Special Issue 9504;1995:80–93. 8. Maubois JL, Ollivier G. Extraction of milk proteins. In: Damodaran S, Paraf A, eds. Foods Proteins and Their Applications. Marcel Dekker Inc.: New York; 1997,579–595. 9. Pearce RJ. Thermal separation of beta-lactoglobulin and alpha-lactalbumin in bovine cheddar cheese whey. Aust J Dairy Technol 1983;38:144–148. 10. Maubois JL, Pierre A, Fauquant J, Piot M. Industrial fractionation of main whey proteins. IDF Bulletin 1987;212:154–159. 11. Baumy JJ, Gestin L, Fauquant J, Boyaval E, Maubois JL. Technologies de purification des phospholipides du lactosérum. Process 1990;1047:29–33.

146 

 6 Integrated membrane operations in whey processing

12. Korhonen H, Pihlanto A. Technological options for the production of health-promoting proteins and peptides devived from milk and colostrums. Curr Pharm Des 2007;13:829–843. 13. Bonnaillie L, Tomasula PM. Whey protein fractionation. In: Onwulata CI, Huth PJ, eds. Whey Processing, Functionality and Health Benefits. Wiley Blackwell: Singapore; 2008;15–39. 14. Bramaud C, Aimar P, Daufin G. Whey protein fractionation: Isoelectric precipitation of α–lactalbumin under gentle heat treatment. Biotechnol Bioeng 1997;56:391–397. 15. Maubois JL, Fauquant J, Famelart MH, Caussin F. Milk microfiltrate, a convenient starting material for fractionation of whey proteins and derivates. In Proc of the 3rd Int Whey Conf, Munich Sept 12–14, Behr. 16. Korhonen H, Syväoja EL, Vasara E, Kosunen T, Marnila P. Pharmaceutical composition, comprising complement proteins, for the treatment of Helicabacter infections and a method for the preparation of the composition. PCT Patent Application WO98/00150, 1998. 17. Piot M, Fauquant J, Madec MN, Maubois JL. Preparation of “serocolostrum” by membrane microfiltration. Lait 2004;84:333–342. 18. Korhonen H. Milk-derived bioactive peptides: from science to applications. J Funct Foods 2009;I:177–187. 19. Brulé G, Roger L, Fauquant J, Piot M. Phosphopeptides from casein-based material. US patent 4358465. 1981. 20. Smithers GW. Isolation of growth factors from whey and their application in food and biotechnology industries – A brief review. In Advances in Fractionation and Separation Processes for Novel Dairy Applications. Brussels, Belgium. Int dairy Fed 2004;16–19. 21. Gauthier SF, Poulit Y, Maubois JL. Growth factors from bovine milk and colostrum: composition, extraction and biological activities. Lait 2006;86:99–125. 22. Omont S, Froelich D, Gésan-Guiziou G, Rabiller-Baudry M, Thueux F, Beudon D, Tregret L, Buson C, Auffret D. Comparison of milk protein separation processes by life cycle analysis: chromatography vs filtration processes. Proc Eng 2012;44:1825–1827.

7 Integrated membrane processes in winemaking Youssef El Rayess and Martine Mietton-Peuchot 7.1 Introduction Membrane filtration has been applied to wine for a long time. At present, the clarification cartridges are integrated in bottling units. Subsequently, in a crossflow filtration mode, microfiltration membranes were the first to be applied for wine clarification. Today, crossflow microfiltration (CFMF) is largely used in oenology for must, lees and wine filtration at different membrane cut-offs, from 0.2 to 1.2 μm. The development of RO application in must concentration was almost done in parallel with that of microfiltration (MF) in clarification [1]. This chapter is an overview of the application of membrane processes in winemaking. The aim is to present both the application of membrane processes in winemaking and a general philosophy of their development from a process engineering point of view. Several examples illustrate this approach; in particular, applications of nanofiltration (NF) and RO membranes. Reduction of alcohol content is studied with different techniques [NF+ evaporation, NF+ membrane contactor (MC), decrease of sugar content, etc.]. Reduction of sugar content of the musts [ultrafiltration (UF) + NF] could be an alternative process to reduce the alcohol content of the wine and to improve its quality. The volatile acidity or malic acid reduction could also be done by coupling two stages of RO. As the free acids are poorly retained by the membrane, the permeate after the first stage filtration contains free acids, salts, esters and other small molecules. Once the permeate is neutralized with the pH of the targeted acid, it will be retained by the second stage membrane in a salty form. The other components passing through are re-injected in the initial must or wine. The potassium hydroxide is used for neutralization. The proposed processes integrate different steps: two membrane techniques, a membrane process and a chemical reaction or new developments (rotating membranes, bipolar membranes, etc.). The principal condition for further development of membrane processes in winemaking is a good understanding of membrane techniques, separation techniques and characterization of the membrane itself and the product (must or wine) to be filtered. The second constraint, given the complexity and variability of must and wine composition, is not trivial and needs a considerable effort at both industrial and research levels. For the same reasons, the coupling of the membrane and other physical-chemical treatments appears to be a promising research domain.

148 

 7 Integrated membrane processes in winemaking

7.2 Crossflow microfiltration for must, wine and lees clarification During red or white winemaking, the filtration process is usually involved in several steps of product elaboration. Filtration in winemaking is used to accomplish two main objectives: clarification and microbial stabilization. – In clarification, large particles that affect the visual appearance of wine are removed. – In microbial stabilization, bacteria or yeasts are removed with the aim of reducing the probability of re-fermentation or spoilage. The limpidity and the microbiological stabilization of wine are two essential parameters that could affect wine organoleptic quality. The compounds removed by filtration can be classified in three groups according to their size: (i) solutes ( < 1 nm), (ii) colloids (between 1 nm and 1 μm) and (iii) particles ( > 1 μm) [2]. Depending on the objectives of the filtration and wine characteristics, the winemakers choose the most adapted technique. Wineries can perform filtration according to two main different technologies, by using precoat or membrane filters. In order to obtain the required wine quality, different types of filtration equipments are available: drum filtration, plate-and-frame filtration, cartridges and crossflow filtration [3]. A filtration process must be efficient in terms of retention and produce adequate flow rates without prejudice to the quality of the wine. These criteria can be difficult to reconcile because of the fouling by the filtering solution over time. The fouling modifies the flow rate and the retention characteristics. CFMF is well implemented in wine cellars. The first trials of CFMF have been conducted in oenology at the beginning of the 1980s with unsatisfactory results in terms of wine quality because the membranes used (UF membranes) were not specific to wine filtration. This technique is an attractive process to the wine industry for one-step clarification and microbiological stabilization compared to traditional techniques. In order to have a limpid wine, the wine makers implement successive solid-liquid separations using traditional technologies such as centrifugation, filtration on plates, diatomaceous earth filtration and the use of exogenic additives. The traditional techniques showed quickly their limits in terms of wine quality, wine loss and its implementation especially in cellars dealing with huge volumes of wines. In addition, the filtration additives have a negative effect on the environment. Their disposal must be done in special waste treatment sites. In addition to a great simplification of the wine processing line, CFMF offers a number of additional advantages, such as elimination of earth use and its associated environmental problems as well as the combination of clarification, stabilization and sterile filtration in one single continuous operation. For a long time, the development of the CFMF suffered from the significant fouling of the membranes by wine compounds. The consequence of this is a reduction in permeation rates, affecting the economic viability of the process, and a risk of excessive

7.2 Crossflow microfiltration for must, wine and lees clarification  

 149

retention of some components, which may affect the product quality. Crude wine after fermentations is a very complex medium with solute molecules as ions, organic acids and sugars ( < 1 nm), molecules with colloidal behavior as polyphenols, polysaccharides and proteins (between 1 nm and 1 μm) and particles as microorganisms, tartaric crystals and organic precipitates ( > 1 μm). Several studies have been conducted in order to identify the wine molecules responsible for membrane fouling, the fouling mechanisms and the methods to limit or control membrane fouling. Studies were initiated in the mid −980s and were focused on the identification of the most suitable membrane pore size for wine filtration. Studies [4, 5] showed that 0.2 μm as average pore size presented the best results in term of permeate flux and wine quality. After this, studies were oriented in order to identify the wine compounds responsible for membrane fouling. This was mainly reported as polysaccharides and polyphenols. Vernhet et al. [6] studied the effect of wine polysaccharides on an organic (PES) membrane fouling. They showed that the effects of polysaccharides on fouling are not similar, owing to the nature of polysaccharides fraction involved in the fouling. It was shown that the pectic polysaccharides of low molecular weight [rhamnogalacturonan type II(RG-II)] have no noticeable effect on the permeation flux, whereas mannoproteins play a crucial role in reducing the fluxes. The researchers noticed that the membrane fouling by a given wine is not directly related to its total polysaccharides content but rather to the composition, structure of these polysaccharides and the balance between different groups of polysaccharides [7]. In 2009, Ulbricht et al. [8] demonstrated that different membrane materials exhibit different levels of polysaccharides adsorption. They showed that larger amounts of polysaccharides were adsorbed on hydrophilic membranes than on hydrophobic membranes. In 2011, El Rayess et al. [9] showed that polysaccharides formed a compact gel layer on the membrane surface which is dependant of transmembrane pressure. Phenolic compounds have a much more important affinity for membranes than the polysaccharides and there are both quantitative and qualitative differences between the different materials tested. It is worth noticing that polyphenols are amphipathic molecules with hydrophobic aromatic rings and hydrophilic phenolic hydroxyl groups. So their adsorption involves both hydrophobic effects and the formation of hydrogen bonds. The preferential adsorption of phenolic compounds with low polarity suggests the predominance of hydrophobic interactions [2]. It was shown that an increase in polyphenol concentration in wine lead to a decrease of membrane permeability and thus an increase of membrane fouling [10]. Suspended particles yeast, bacteria and cell debris play also a role in membrane fouling. Boissier et al. [11] proved that the increase in the total resistance related to yeast deposition is caused by the compaction of the cake layer on the surface of the membrane. They found also that fouling is governed by fines particles (lactic bacteria and colloidal aggregates) more than yeast. In 2011, El Rayess [12] showed that yeasts may protect the membrane from colloids fouling whether by forming a secondary membrane or by disturbing the pectic gel layer to be uniformly installed on the membrane surface.

150 

 7 Integrated membrane processes in winemaking

To control membrane fouling, it is important to adjust the operating conditions of wine CFMF. Critical flux can be a key parameter of the control of membrane fouling as it is depending at the same time on the hydrodynamics and physico-chemistry (membrane/solutes interactions). El Rayess et al. [9] studied the critical flux for irreversibility (Jci) during wine CFMF. The critical flux for irreversibility (Jci) is defined as the permeate flux above which an irreversible fouling appears on the membrane surface. The method used to determine Jci is the square wave barovelocimetry (SWB) developed by Espinasse et al. [13]; this method also enables the distinction between reversible and irreversible fouling with the determination of reversible (Rrf ) and irreversible (Rif ) resistance, respectively. If the permeate flux is the same between these two steps, the fouling associated is considered as totally reversible. If it is not the case, the fouling associated is considered as partly irreversible and the critical flux for irreversibility is determined at the last step where fouling is totally reversible. For all tested macromolecules (tannins, pectins and mannoproteins) and associate concentrations and in the range of tested pressures (200–1000 mbar), it was found impossible to determine a value of the critical flux for irreversibility Jci. This was not the case for filtered wine where Jci was found beyond 1000 mbar (Jci   ≥   1.4  ×  10–4 m/s). To improve the efficiency of the filtration process and maintain the state of cleanliness of the membrane filtering surfaces at an acceptable level during the filtration process, most filtration devices are equipped with a reverse filtering system. Backflushing, backwashing and back-pulsing are all methods of operation in which the transmembrane pressure is periodically inverted by the use of a secondary pump, so that permeate flows back into the feed, lifting the fouling layer from the surface of the membrane. The main difference in the methods is mainly the time-frame in which the process operates. Technically, the back-pulsing process is very similar to backflushing or backwashing that is widely used for commercial applications. However, the fundamental difference between a back-pulse and a back-flush is the force and time used to lift accumulated deposits off the membrane. Generally, in back-flushing flow reversal occurs for a few seconds once every several minutes, while back-pulsing occurs at a higher frequency and the pulses are applied for a very short time ( < 1 s). When a flow reversal system is applied, the residues of particles separated from the wine are eliminated from the membrane surface and the filtrate flow rate is increased. As the restoration of flux lasts only for a relatively short period, the back-flush should be repeated at frequent intervals for maximum effect. The main advantages are: (i) a faster filtration process; (ii) more filtration capacity and overall performance increases of up to 60%; (iii) less filtration stress; (iv) less temperature pick-up; (v) minimum impact on the wine quality. New developments in CFMF in the wine industry are oriented to filter high charged (suspended solids), viscous and abrasive fluids such as tank bottoms, wine lees, fining agents, retentates from crossflow filters and centrifuged sludges. Therefore, the Bucher Vaslin and Velo group (2011) introduced new membranes to filter very charged fluids from the wine industry. These membranes are made from sintered stainless steel

7.2 Crossflow microfiltration for must, wine and lees clarification  

 151

Figure 7.1: Sintered stainless steel tubular membranes

powder. In fact, the porous stainless steel is used to support the ceramic filtration layer (generally made by titanium oxide TiO2 or zirconium oxide ZrO2). These tubular membranes (Figure 7.1) have been used for years, but their introduction into the wine industry is very recent. These membranes (when developed) were supposed to be applied to food filtration in the future as an alternative to the generally used organic and ceramic membranes. This is because the organic ones are easily fouled by food suspensions and need to be frequently cleaned by chemical solutions, which results in a short lifespan; and the ceramic ones are brittle that they are likely to break during service. Another new development in wine filtration of high charged fluids is the introduction of dynamic filtration. Dynamic or shear-enhanced filtration consists of creating the shear rate at the membrane by a moving part such as a rotating membrane, or a disk rotating near a fixed circular membrane, or by vibrating the membrane either longitudinally or torsionally around a perpendicular axis [14]. This mode of filtration constitutes an alternative to crossflow filtration, which not only increases substantially the permeate flux, but has a favorable effect on membrane selectivity. It also permits to decouple from membrane shear rate, the inlet flow rate into the module, which can be varied independently and does not need to be much larger than the filtration rate. The drawbacks are the complexity and limitations in membrane area for some systems, such as cylindrical rotating membranes or multi-compartment rotating disk systems, which raise the equipment cost. TMCI Padovan introduced the first rotating dynamic crossflow filter (Dynamos®, Figure 7.2A) into the wine industry. This filter is also equipped with a back-pulse system and is covered by an international patent (WO2011/033537) [15]. This system is known in the literature as a multi-shafts systems with rotating ceramic membranes and was developed and commercialized many years ago but without the back-pulse system (Rotostream®, Buss-SMS-Canzler GmbH; MSD, Westfalia-separator). In this system, two stacks of rotating membranes are used. Preferably the two shafts rotate countercurrent to create the maximum shear forces at the membrane surface. This membrane overlapping (Figure 7.2B) can increase the permeate flux considerably (the magnitude depends very strongly on the type of feed and the crossflow conditions).

152 

 7 Integrated membrane processes in winemaking

Membrane hollow shafts

Overlapping zone Back-pulse system

Retentate

Pump

Permeate

(a)

(b)

Figure 7.2: (A) Dynamos® system configuration; (B) membrane overlapping

Table 7.1: Analytical results of wine lees after different types of filtration [15] Cabernet Lees Cabernet after rotary drum Alcohol (%v/v) Total SO2 (ppm) Anthocyanins (ppm) Total phenolic index Suspendid solids (%v/v) Color intensity Turbidity (NTU) Temperature intake (°C)

12.44 82 713 1516 30 0.766 n.d. 0

12.2 52 650 1402 0.5 0.7 50 2

Cabernet after tubular crossflow

Cabernet after dynamic filtration

12.41 75 675 1358 0 0.683 0.5 10

12.4 70 691 1456 0 0.735 0.5 1–3

Table 7.1 gives the results of analyses of wine lees after comparison of three different methods of filtration (rotary drum filtration, crossflow filtration and dynamic filtration). The results showed little differences between the three types of filtration with a preference to dynamic filtration.

7.3 Electrodialysis and bipolar electrodialysis Tartrates are naturally occurring crystals that often form in wines and may be considered undesirable to most customers. The crystals are formed from potassium hydrogenotartrate (KHT), a naturally occurring organic acid, and their solubility in wines decrease with the presence of ethanol. This unstable state can lead to the occurrence of KHT crystals in bottles with dramatic consequences on the final aspect of the wine. Wine stabilization as part of the winemaking process reduces the concentration of KHT in wine. Traditionally, wineries lower the potassium tartrate solubility in wine by chilling the wine to approximately −4°C. There are different types of

7.3 Electrodialysis and bipolar electrodialysis 

Brine

Anode effluent

Treated wine

T– –

Cathode effluent

T– –

K+ TH –

TH –

Cathode

Ca++

Anode

Ca++

K+

TH –

 153

TH –

Anode feed

Cathode feed

Untreated wine

Water

Cation exchange membrane Anion exchange membrane

Figure 7.3: Electrodialysis principle for tartrate stabilization in wines

cold stabilization enhancement that may expedite the process, particularly seeding the wine with potassium tartrate crystals [16]. Wines are typically maintained at this temperature for a period of 1.5–3 weeks, depending on how easy it is to crystallize the potassium bitartrate, i.e., how “stable” the wine is. The cold stabilization is time- and energy-consuming and difficult to control. The electrodialysis (ED) process (Figure 7.3), as an alternative to traditional methods, removes the tartaric acids from the wine by passing it through an electric field and collecting ions (potassium K+ and calcium Ca++) and negatively charged tartaric acids on anionic and cationic membranes. The Eurodia Company and the French National Agronomic Research Institute (INRA) developed ED in the 1990s. The principle of this technology consists of the application of an electric field, because the presence of electrodes on both sides of the cell will lead to ion migration: cations (K+ and Ca++) will migrate towards the negative electrode (the cathode), whereas anions (TH– and T–) will move towards the positive electrode (the anode). The cations in the compartment “untreated wine” will be able to cross the cationic membrane and thus be eliminated from this compartment. However, they will not be able to leave the compartment “water” as they would find an anionic membrane on their way. The anions can be exported from compartment “untreated wine” because they will not be able to cross the next (cationic) membrane. In principle, all cations and anions can be affected during ED. However, not all ions show the same behavior and their extraction depends on various factors. In theory, the transfer kinetics of ions depends especially on the speed constant (ϖ). This constant is characteristic for every ion in the solution. In practice, this constant

154 

 7 Integrated membrane processes in winemaking

depends essentially on the ion mobility and dimension, as well as on the nature of the membrane. Therefore, the choice and development of ED membranes was conducted on the considerations for the regulatory agencies, the technological characteristics of the available membranes and on enological criteria. The electrodialyser for continuous wine tartrate stabilization is composed of a stack of ED membranes of two compartments filled with inert gas, feeding pumps and two tanks allowing the return of the fluids (wine and concentrate) into their respective compartments. The wine return-tank is equipped with level sensors and the two circuits are both provided with a conductimeter. With help of the conductivity probes, an automatic device conducts the circulation sequences and the concentration level of the discharge. The wine is recirculated in the electrodialyser until achieving the reduction level of ions concentration. The deionization rate or the degree of deionization (DD) is the percentage reduction of electrical conductivity of the wine during the ED process. This DD must be determined with accuracy before the wine treatment. It is calculated as follow: DD =

wine initial conductivity – wine final conductivity × 100% wine initial conductivity

(7.1)

If the predicted DD is underestimated, the wine must be treated again by ED in a second step, which increases the operation time and complexity of the process and adversely affects the reliability of the technology [17]. The necessary DD can be determined by applying the conventional cold treatment to a sample of wine as this process is a very time-consuming; the DD is predicted using the minicontact test, which simulates at a bench-scale the cold treatment with seeding. In 2003, Benitez et al. [18] compared the cold treatment and ED for tartrate stabilization of three sherry wines on an industrial scale with the objective of checking the efficacy of these techniques. The analytical results of the common enological parameters are shown in Table 7.2. The influence of ED treatment on the alcohol content, color intensity, pH and volatile acidity was small or negligible, while the effect on the titratable acidity was small and in direct relationship with the DD. The cold treatment produced a pH decrease of 0.2 units in the Fino wines, while the maximum decrease produced by ED in these wines was only 0.05 units. With regard to potassium, the ED produced an appreciable reduction in its concentration while the effect of cold treatment on potassium reduction depended on the wine type. Cold treatment produced a larger decrease in the tartrate content than ED. In 2010, the Australian Wine Research Institute (AWRI) published a report comparing the cold treatment and ED for tartrate stabilization in wines [19]. The report showed that ED offers significant advantage in the power consumption and wine losses while wastewater volume and labor requirements are higher for ED than the cold treatment. This report concluded that based on the obtained results, ED appears to offer viable alternative method to tartrate stabilization in wines. In the past two decades, it has been observed that wines are suffering from higher ethanol levels and pH values that are higher and higher. This phenomenon has

DD (%) Alcohol content (% vol.) pH Color Intensity Titratable acidity (g/l) Volatile acidity (g/l) Saturation temperature (°C) Minicontact test (μS/cm) K (mg/l) Ca (mg/l) Tartaric acid (g/l) Malic acid (g/l) Acetic acid (g/l)

15.1 3.18 0.02 4.23 0.16 23.9 85 695 102 2.43 0.177 0.14

15.1 2.98 0.019 3.84 0.13 13 1 520 90 1.43 0.161 0.125

CT 20.8 15.3 15.2 3.54 3.5 0.077 0.08 4.12 4.11 0.6 0.62 24.4 18.8 67 9 1470 825 97 97 1.55 1.41 0.2 0.206 0.59 0.6

Control

30.4 14.9 3.13 0.036 3.9 0.13 16.6 6 450 71 2.1 0.153 0.125

ED

Control 19.6 15 3.16 0.017 4.01 0.14 19.7 61 535 72 2.33 0.189 0.136

Medium

Fino

30.1 15.2 3.51 0.079 4.03 0.69 16.6 5 620 92 1.33 0.192 0.675

ED

Table 7.2: Analytical results of common enological parameters during electrodialysis and cold treatment [18]

15.3 3.45 0.064 4.02 0.68 19.5 9 820 120 1.13 0.215 0.653

CT

15.5 3.43 0.104 4.09 0.65 21 9 705 95 1.59 0.132 0.59

Control

Cream ED 18.3 15.3 3.37 0.095 3.94 0.63 16.6 5 520 70 1.41 0.148 0.575

15.5 3.43 0.071 3.88 0.67 19.1 8 695 120 1.35 0.158 0.58

CT

7.3 Electrodialysis and bipolar electrodialysis 

 155

156 

 7 Integrated membrane processes in winemaking

been attributed to global warming and to enological practices. The lack of acidity in wines leads to the development of undesirable microorganisms and a disequilibrium in the organoleptic qualities of the wines. According to the European Community, only the addition of tartaric acid is allowed to correct the wines pH. An alternative to the addition of tartaric acid, which is difficult to control because tartaric precipitation, the European Commission and the International Organization of Vine and Wine (361–2010 resolution) in 2010 accepted the method of acidification by bipolar ED. The principle of wine acidification by bipolar-membrane ED is similar to ED, but the anion-exchange membrane is replaced by a bipolar-membrane. In its simplest form, a bipolar-membrane is a cation-exchange membrane laminated together with an anion-exchange membrane, through an intermediate layer (the “junction” layer). In the junction layer, water is split into hydroxide (OH–) ions and protons (H+) by a disproportionate reaction. The hydroxide ions and protons produced are separated by migration in the respective membrane layer out of the membrane. When the electric current is applied, the potassium ions (K+) contained in the wine are attracted towards the cathode, they pass through the cationic membrane and are stopped by the bipolar-membrane. The OH– ions migrate towards the positive pole (anode) into the brine (concentrate), whereas the H+ ions migrate towards the negative pole (cathode) and replace the potassium ions that are extracted from the wine in order to conserve the ion equilibrium. This operation causes acidification (lowering the pH) by decreasing the potassium content and thus the salified form of organic acids in the wines. For a lowering of pH values there is a concomitant increase in titratable acidity. Acidification by bipolar ED can correct wine pH with a precision of 0.05 units. The target value of the treatment is determined following tasting with the producer. The maximum treatment value is 0.3 units of pH. Deacidification using an electromembrane process (ED with bipolar membranes) of musts (resolution OIV-OENO 483-2012) or wines (resolution OIV-OENO 484-2012) is accepted in 2012 by the International Organization of Vine and Wine. The principle of wine deacidification by bipolar-membrane ED is similar to the acidification principle, but the anions are affected in this process because the cation-exchange membranes are replaced by anion-exchange membranes. The application of the electric current drives the anions (TH– and M–) towards the anode. They pass through the anionic membrane and are stopped by the bipolar-membrane. The anions are transferred from the wine compartment to the brine compartment where they are associated with H+ cations losing their ionic form. The wine is impoverished in organic acids, reducing the titratable acidity and thus the wine is de-acidified.

7.4 UF and NF for reduction of must sugars The process for reduction of must sugars comprises two membrane processes: UF and NF. This process is patented by the Bucher Vaslin company and is at present marketed

7.5 RO and NF for sugar must concentration 

 157

under the name of REDUX®. In 2012, a specific application on the reduction of sugar content in musts through membrane coupling was adopted (Resolution OIV-OENO 450B-2012). The UF and NF are carried out according to the principle of tangential flow filtration: mass transfer across the membrane is caused by a pressure gradient that may vary from 5 to 80 bar according to the methods. In the REDUX® process, UF is used to separate macromolecules (especially polysaccharides and polyphenols) from musts before sugar concentration by NF. Osmotic pressure of the ultrafiltered must is therefore lowered. The sugar concentration by NF may be greater. The higher sugar concentration leads to a decrease in volume loss (elimination of a semi-concentrated must approximately 400 g/l sugar). The retentate of UF by macromolecules is reintroduced into the original must. Plus, the permeate of NF is also reincorporated in the original must. The treated wine contains less alcohol but is richer in solids. NF is preferred to RO because it gives greater flow rates. NF membranes presenting higher cut-off thresholds allow the transfer of organic acids and potassium into the permeate.

7.5 RO and NF for sugar must concentration The climate and the local growing conditions make each year’s harvest unique. In some years, especially during poor growing conditions, grape musts do not have sufficient potential alcohol content. So, it is necessary to increase sugar concentration to obtain an increase in alcohol content in order to have a well-balanced wine. Winemakers use chaptalization techniques (adding sugars), must concentrate (MC) or rectified must concentrate (RMC) to compensate for poor growing seasons, or if they are located in areas of the world that experience cooler climates. The additive techniques increase wine volumes and sometimes lead to unbalanced wines. To avoid these problems there is a growing interest in subtractive techniques such as pervaporation, cryoconcentration, RO and NF. RO is used to eliminate a certain volume of water of grape must by applying hydrostatic pressure (higher than osmotic pressure), leading to an increase in must sugar concentration. RO is considered an adequate technology to concentrate grape musts, with the advantages of: – lower energy consumption – minimal heat damage of the quality properties – maintenance of sensory characteristics and nutritional value of the products – absence of caramelization reactions – compact and easy to operate facilities – higher purity of the permeate compared to other membrane processes. In 2002, Mietton-Peuchot et al. [1] showed that RO could be an alternative to chaptalization and vacuum evaporation. They showed that low temperature (about 10°C) and

158 

 7 Integrated membrane processes in winemaking

high applied pressure (about 75 bar) are conditions that prevent the must components from crossing the membrane. Conversely, the high pressure and the concentration provoked tartaric acid precipitation at the membrane surface, which increased the membrane fouling. Kiss et al. [20] showed that RO (60 bar, 20°C) coupled to high pressure NF (70 bar, 40°C) may produce must concentrate with high sugar concentration (around 45°Brix ≈450 g/l). They showed that the cost estimations of the membrane process are much cheaper than evaporation. In 2011, Santana et al. [21] evaluated the concentration of grape must by RO. The experiments were carried out in two steps. First, three transmembrane pressures (40, 50 and 60 bar) and four temperatures (20, 30, 40 and 50°C) were evaluated without juice concentrations. Under the evaluated conditions, the process conducted at 50°C and 60 bar presented the higher permeate flux and the maintenance of all the physical and chemical parameters of the product. In the second step, the grape juice was concentrated at three temperatures, 20, 30 and 50°C, always at 60 bar of transmembrane pressure. The temperature of 30°C resulted in an adequate value for the permeate flux in addition to maintaining the physicalchemical and nutritional characteristics of the concentrated product when compared to the single-strength juice. NF is another membrane process tested as an alternative to RO. This process uses a pressure gradient (up to 40 bar) to transport must or wine through the membrane. Versari et al. [22] and Santos et al. [23] tested several NF membranes for must concentration. They found that NF membranes provided rejection coefficients ranging from 80–95% for sugars. In the literature, little has been published about this subject and there are possibilities to develop research activities in this sector.

7.6 RO, NF and MC for wine dealcoholization Production techniques for manufacturing low or reduced alcoholic strength beverages have been developed over the last 15 to 20 years in order to satisfy a consumer demand for healthier alcoholic products. Decreasing alcohol consumption is a worldwide trend and lower alcohol consumption rates are associated with certain positive health benefits. Several technologies and strategies have been applied for removing alcohol, and among them the membrane processes after alcoholic fermentation. Overall, RO is probably the most widely used technique at present for reducing the alcohol content in wine. It involves pressure filtration of the wine through a fine porous membrane that is permeable to alcohol and water, but not to most of the wine components. In order to reduce the alcohol content and to limit the volume loss, water from the RO filtrate must be recuperated. The most used process is RO coupled to column distillation (Figure 7.4). In the patent (WO 2004/113489), Gonçalves and De Pinho [24] propose a process of ethanol removal from wine based on the use of NF membranes coupled to a distillation operation. The NF membranes provide higher alcohol flow rates together with greater permeation rates than RO.

7.6 RO, NF and MC for wine dealcoholization 

Water

Water + alcohol

Concentrate

Wine de-alcoholized

 159

Alcohol

Distillation column

Wine

Figure 7.4: Wine de-alcoholisation process using membrane processes coupled to distillation

Another process tested for wine dealcoholization is OD. OD is a MC technique also known as osmotic evaporation, membrane evaporation, isothermal MD or gas membrane extraction. OD is a membrane transport process in which the wine containing one or more volatile components is allowed to contact one surface of a microporous membrane whose pores are not wetted by the wine, while the opposing surface is in contact with a second non-wetting liquid phase (usually water) in which the volatile components are soluble or miscible. The membrane thereby functions as a vapor gap between the two liquid phases, across which any volatile component is free to migrate by either convection or diffusion. The DF for such transport is the difference in vapor pressure of each component over each of the contacting liquid phases. OD of a high alcohol content wine, at a temperature of 10–20°C and using water as the stripping liquid, can rapidly reduce the alcohol content to levels down to 6%. The mechanism of the process takes advantage of three factors: (i) ethanol is the most volatile component in the wine and the most rapidly diffusing species across the hydrophobic membrane, (ii) the vapor pressure of the flavour/fragrance components is low and, thus, so is their OD flux, and iii) the solubilities of the flavor/fragrance components in alcohol/water solutions are substantially higher (and their vapor pressures correspondingly lower) than they are in water. The main advantage of OD lies in its ability to work at low temperatures and pressures, thus avoiding mechanical damage and thermal degradation of the components and aroma of wines. Diban et  al. [25] investigated the feasibility of applying hydrophobic HF MCs to decrease the alcohol content of a synthetic wine solution. They showed that the ethanol content can be reduced by the same amount independent of the initial ethanol concentration present in wine in the range studied of 10–13% (v/v) of alcohol. They showed that a partial dealcoholization (reduction of 2% vol.) of Merlot wine gave acceptable aroma losses without a perceptible depletion of the product quality. Liguori et al. [26] showed that the optimal conditions for ethanol removal

160 

 7 Integrated membrane processes in winemaking

from model solutions at 10% vol. were obtained working in laminar conditions for both feed and stripping streams. They also observed a decrease in ethanol flux, while increasing the ethanol content of the solutions explained by the saturation phenomena. An increase in temperature accelerates the dealcoholization process. No significant differences in chemical analyses between crude and dealcoholised wine were found. MC can be used on its own for wine dealcoholization or coupled to other membrane process. For example, the Australian process marketed by the company Memstar consists of alcohol reduction by a two-stage process of RO followed by a membrane module known as a MC (Liqui-Cel®). Wine to be treated is first separated by RO into concentrate and permeates streams. The wine concentrate contains all of the wine characters. The alcohol-rich permeate is passed through the MC, on the other side of which is a counter-flow of treated strip water. Alcohol passes through the membrane from the permeate into the water. The dealcoholised permeate is then cooled and recombined with the wine, lowering the alcohol of the final wine.

7.7 Gas control by membrane processes All through vinification and ageing, as well as during ageing of the bottled wine, oxygen is a major actor in wine transformation. It has a beneficial role in many steps of the wine process (increase of the yeast population, color stabilization, etc.), but oxygen may also be detrimental when present during specific steps (oxidation, growth of microorganisms, etc.) So, micro-oxygenation techniques, allowing mastered oxygen input, are nowadays well known for their stabilizing effect. The basic principle of a micro-oxygenation technique is to continuously deliver to the wine, stored in tanks, oxygen quantities always smaller than the chemical or biological demand, to avoid any accumulation of dissolved oxygen. This oxygen input is done by microbubbling of pure gaseous oxygen into the wine with a small microfiltration membrane and this technique is now accepted practice in wine manufacturing. The major difficulty of micro-oxygenation with hydrophilic microporous membranes is the control of the dimension of the bubbles and of the yield of transfer. All the microporous systems (stainless steel, ceramic, organic membranes) have been characterized by gas-liquid porometry to obtain the pore size distribution. Depending on the media, the pore size distribution varies from 11 μm to 0.3 μm. MC could be used to control the gas transfer in wines (carbon dioxide and oxygen) before bottling. An important developed application of MC is the carbonation of sparkling beverages and the use of MC in oenology belongs to this latter type of application [25]. Indeed, it concerns the control of dissolved gases concentrations (CO2 and O2) in the case of wine carbonation/de-oxygenation or wine de-carbonation/de-oxygenation or wine de-oxygenation at the packaging step [27].

7.8 References 

 161

7.8 References  1. Mietton-Peuchot M, Milisic V, Noilet P. Grape must concentration by using reverse osmosis. Comparison with chaptalization. Desalination 2002;148:125–129. 2. El Rayess Y, Albasi C, Bacchin P, Taillandier P, Mietton-Peuchot M, Devatine A. Crossflow microfiltration applied to oenology: a review. J Memb Sci 2011;382:1–19. 3. Ribéreau-Gayon P, Glories Y, Maujean A, Dubourdieu D. Handbook of enology. Volume 2: The Chemistry of Wine, Stabilization and Treatments. Dunod, Paris, France, 2006. 4. Poirier D, Bennasar M, Tarodo de la Fuente B, Gillot J, Garcera D. Clarification et stabilisations des vins par ultrafiltration tangentielle sur membranes minérales. Lait 1984;64:141–142.  5. Peri C, Riva M, Decio P. Crossflow membrane filtration of wines: comparison of performance of ultrafiltration, microfiltration and intermediate cut-off membranes. Am J Enol Vitic 1988;39: 162–168. 6. Vernhet A, Pellerin P, Belleville MP, Planque J, Moutounet M. Relative impact of major wine polysaccharides on the performances of an organic microfiltration membrane. Am J Enol Vitic 1999;50:51–56.  7. Vernhet A, Cartalade D, Moutounet M. Contribution to the understanding of fouling build-up during microfiltration of wines. J Membr Sci 2003;211:357–370. 8. Ulbricht M, Ansorge W, Danielzik I, Konig M, Schuster O. Fouling in microfiltration of wine: The influence of the membrane polymer on adsorption of polyphenols and polysaccharides. Sep Purif Technol 2009;68:335–342. 9. El Rayess Y, Albasi C, Bacchin P, Taillandier P, Mietton-Peuchot M, Devatine A. Cross-flow microfiltration of wine: effect of colloids on critical fouling conditions. J Memb Sci 2011;385:177–186. 10. El Rayess Y, Albasi C, Bacchin P, Taillandier P, Mietton-Peuchot M, Devatine A. Analysis of membrane fouling during cross-flow microfiltration of wine. Innov Food Sci Emerg Technol 2012;16:398–408. 11. Boissier B, Lutin F, Moutounet M, Vernhet A. Particles deposition during the cross–flow microfiltration of red wines—incidence of the hydrodynamic conditions and of the yeast to fines ratio. Chem Eng Process 2008;47:276–286. 12. El Rayess Y. Microfiltration tangentielle appliquée à l’oenologie: compréhension et maîtrise des phénomènes de colmatage. Ph.D. Thesis. Université de Toulouse, Institut National Polytechnique de Toulouse, France, 2011;225:p. 13. Espinasse B, Bacchin P, Aimar P. On an experimental method to measure critical flux in ultrafiltration. Desalination 2002;146:91–96. 14. Jaffrin MY. Dynamic shear-enhanced membrane filtration: a review of rotating disks, rotating membranes and vibrating systems. J Membr Sci 2008;324:7–25. 15. TMCI PADOVAN S.P.A. Apparatus and method for filtering liquids, particularly organic liquids. Patent WO/2011/033537, 2011. 16. Santos CP, Gonçalves F, De Pinho MN. Optimisation of the method for the determination of the temperature of saturation in wines. Anal Chim Acta 2002;458:257–261. 17. Soares PAMH, Geraldes V, Fernandes C, dos Santos PC, de Pinho MN. Wine tartaric stabilization by electrodialysis: prediction of required deionization degree. Am J Enol Vitic 2009;60:183–188. 18. Benitez JG, Macias VMP, Gorostiaga PS, Lopez RV, Rodriguez LN. Comparison of electrodialysis and cold treatment on an industrial scale for tartrate stabilization of sherry wines. J Food Eng 2003;58:373–378. 19. Forsyth K. Comparison between electrodialysis and cold treatment as a method to produce potassium tartrate stable wine. AWRI Report 2010, Project Number: PCS 10004.

162 

 7 Integrated membrane processes in winemaking

20. Kiss I, Vatai G, Bekassy-Molnar E. Must concentrate using membrane technology. Desalination 2004;162:295–300. 21. Santana I, Gurak P, da Matta V, Freitas S, Cabral L. Concentration of grape juice (Vitis labrusca) by reverse osmosis process. Desalination and Water Treatment 2011;27:103–107. 22. Versari A, Ferrarini R, Parpinello GP, Galassi S. Concentration of grape must by nanofiltration membranes. Food Bioprod Process 2003;81:275–278. 23. Santos FR, Catarino I, Geraldes V, de Pinho MN. Concentration and rectification of grape must by nanofiltration. Am J Enol Vitic 2008;59:446–450. 24. Goncalves F, de Pinho MN. Integrated nanofiltration process to reduce the alcohol content of alcoholic beverages. Patent WO/2004/113489, 2003. 25. Diban N, Athes V, Bes M, Souchon I. Ethanol and aroma compounds transfer study for partial dealcoholization of wine using membrane contactor. J Membr Sci 2008;311:136–146. 26. Liguori L, Russo P, Albanese D, Di Matteo M. Effect of process parameters on partial dealcoholization of wine by osmotic distillation. Food Bioprocess Technol 2012, DOI 10.1007/s11947-012-0856-z. 27. Blank A, Vidal JC. Development of a membrane contactor for the exact management of dissolved gases in wine. 9ème édition du Symposium International d’Enologie Eno. 15–17 juin 2011, Bordeaux (France).

8 Membrane operations in the sugar and brewing industry Frank Lipnizki and René Ruby-Figueroa 8.1 Introduction The concept of membrane processes is not a recent invention. In fact, it has existed as long as life has existed. Since the invention of the phase “inversion membrane” by Sidney and Sourirajan in the 1960s [1], membrane operations have established themselves as efficient processes for industrial separations with high selectivity. In particular, in the food industry membranes demonstrated a rapid growth as low-energy processes providing a gentle treatment of the products at low-to-moderate temperatures. Membrane applications in the food industry are separation, fractionation, concentration, purification, and clarification of various streams, and membrane processes are often recognized as best available technology (BAT) in the industry. This chapter focuses on membrane applications in the sugar and brewing industries. Both industries are established globally, operating with large process streams and having a wide range of separation needs. Despite these needs, membrane processes in these two industries are still in their infancy. The key challenge of membrane processes here is membrane fouling caused by the presence of fouling components in the various feed streams. Membrane cleaning is one approach to reverse membrane fouling, but it requires regular cleaning intervals and cleaning chemicals. Improved pretreatment methods, such as gravity separation and pre-filtration plus new low fouling membranes, are under investigation but most of the applications in the sugar and brewing industries are still under development. Based on this, the chapter will not only focus on established applications of membrane processes in the sugar and brewing industries but also on potential membrane applications in these industries. The first section of this chapter will provide an overview on membrane applications in the sugar industry, covering both beet and cane sugar. The following section focusses on the brewing industry. In the final section of the chapter, a concluding outlook on membrane processes for both the sugar and the brewing industry will be given. The sections of this chapter are self-contained and the selective reader is therefore encouraged to move directly to the section of interest.

8.2 Beet and cane sugar production Due to the lack of historical data, it is difficult to determine when sugar became the principal sweetener in any given part of the world. The process currently known for creating sugar by pressing the juice and then boiling it into crystals, was developed

164 

 8 Membrane operations in the sugar and brewing industry

in India around 500 BC [2]. Since then, worldwide sugar production has grown constantly. In 2009–10, the world production was equal to 158,830 Mt and was expected to reach 170,375 Mt in the year 2010–11 according to the data reported by the International Sugar Organization (ISO) [3]. Sugar is an essential product for human life and as such the human body needs a daily sugar intake. When using the term “sugar” it is important to consider that sugar is a product obtained by an industrial process, in which sugar cane or sugar beets are the source required for refined sugar. Nowadays, sugar is produced in approximately 115 countries around the world: 67 countries produce sugar from sugar cane, 39 from sugar beets and nine countries from both sugar cane and sugar beets. Overall, 70% of the produced sugar is from sugar cane and 30% from sugar beets [4]. In the past, the sugar industry produced only sugar but nowadays the industry is also involved in the production of electricity, bioethanol and biochemicals and as such one of the key segments in food industry. Furthermore, sugar cane is also a valuable crop for bioproducts, which have a very high demand in the market [5]. Sugar production is energy-intensive and it requires large quantities of, for example steam, heating and electricity [6]. Hence, increasing fuel prices and stringent environmental regulations are forcing sugar producers to search for alternative energy-efficient and environmentally-friendly processes [7]. Membrane technology can play an important role in the future of the sugar industry by supporting the development of sustainable processes. The initial research effort in the development of membranes and membrane processes for the sugar industry can be dated back to the beginning of the 1970s, when R. F. Madsen, of DDS (De Danske Sukkerfabrikker, now Nordic Sugar part of Nordzucker) proposed the use of membrane processes in the beet sugar industry. This initial work led to the foundation of DDS Filtration, one of the leading European membrane producers, today Alfa Laval Nakskov (Denmark). Even though different applications of membrane technology in the sugar industry have been reported in the literature, the presence of membranes in the industry is very limited until now.

8.2.1 Membrane applications on beet sugar production An overview of the classical production process of beet sugar is shown in Figure 8.1. The major unitary operations will be described as follows: – Pretreatment: After the arrival at the sugar mill, the beets are cleaned with water to remove particles such as rocks, trash, and leaves. Subsequently, to improve the extraction of sugar, the beets are sliced in very thin V-shaped pieces named cossettes. – Extraction of the beet sugar by diffusion: In the diffusion tower, the sugar is extracted from the beet cossettes by hot water extraction, leaving a nutrient-rich pulp with very little sugar, which has uses such as cattle feed. In this stage, the optimum temperature for the extraction is 70–73°C.

8.2 Beet and cane sugar production 

Sugar beets

Beet washing and slicing

Sugar beet cossettes

Evaporation Juice concentration

Thick juice filtration

Raw juice

Diffusion tower Sugar extraction

Crystallization

Liming and carbonation

Ion exchange Demineralization

Thin juice filtration

Centrifugation

 165

Cooling and drying

White sugar

Figure 8.1: Traditional beet sugar production

–

–

–

Beet juice purification by liming: The purification stage allows the removal of proteins, pectins, inorganic salts and coloring substances from the raw juice by the addition of lime milk, followed by carbonation and demineralization. To carry out this operation, the juice should be heated to between 80–90°C. Beet juice concentration by evaporation: The cleaned juice, which in the sugar industry is called thin juice, has a sugar content of 14–16°Brix and is concentrated by multi-stage evaporation to obtain the “thick juice” with 60–75°Brix. Crystallization: The concentrated juice is separated in white sugar and sugar syrup (molasses) by several steps of boiling and crystallization.

In the following sections, the potential membrane applications for the different stages of the beet sugar production will be described in detail.

8.2.1.1 Sugar beet press water and pulp recycling In the conventional extraction process, the pulp obtained in the diffusion tower is separated by pressing it in a filter press into beet pulp and press water. The membrane operations can be coupled to this process as shown in Figure 8.2. Traditionally, the press water recovered in the pulp press operation is directly recycled to the diffusion tower. This press water contains typically around 1–3% of total solids, of which

Press water

Pulp press Fresh water

Beet pulp Diffusion tower Sugar extraction

Sugar beet cossettes

pH-Adjustment Pre-filtration Reverse osmosis

Pre-treatment Diafiltration water Ultrafiltration

To dryer

Pectin isolated

Raw juice

To low grade sugar crystalization Figure 8.2: Integration of sugar beet press water and pulp recycling by membrane technology

166 

 8 Membrane operations in the sugar and brewing industry

60–80% corresponds to sugars and the remaining 20–40% corresponds to dissolved species, such as salts, colloids and suspended impurities [8]. At an industrial scale, the amounts of water and impurities returned back to the extraction process are large, for example, the processing of 10,000 t/day of beets produces approximately 1,100 t/day of white sugar and 6,000 t/day of press water containing around 30 t/day of impurities [9]. These components are undesirable as their presence affects the sugar extraction efficiency as well as the juice concentration and purification and consequently lowers the overall productivity. The integration of nanofiltration (NF) and reverse osmosis (RO) has been proposed to improve the efficiency of the conventional sugar extraction. By using NF or RO, it is possible to concentrate most of the undesirable components, including the sugar from the press water, and send this stream to the low grade crystallization. The NF/RO permeate stream, which is almost pure water, can be recycled to the extraction unit. In 1993 and 1994, Bogliolo et al. [9] tested the use of RO by a pilot plant capable of treating up to 5 m3h of press water at two Italian sugar factories (see more details in Table 8.1). In these experiments, the press water was pretreated by liming, carbonation, sedimentation, alkalization, phosphatization and filtration. At the end of the RO trials (11 concentration runs and about 350 h of operations), a 50% decrease of pure water flux compared to the initial value was detected. In addition, it was reported that with an increasing press water concentration of up to 10.2°Brix, the permeate flux decreased from 20–30 l/m2h to 7–15 l/m2h and then remained constant during the following 80 hours of operation. Furthermore, the beet pulp obtained in the pulp press contains pectin, hemicelluloses, celluloses and small amounts of protein, lignin, phenolic, fats and ash that

Table 8.1: Sugar beet press water and pulp recycling. Adapted from [10] Membrane type

Module

Feed

Sugar beet press water OSMO 411T-MS10 4″ × 40″ Spiral Pretreated NaCl retention: 99% GE Osmonics, press GE Osmonics, USA USA water

Sugar beet pulp Polysulfone MWCO: 10, 30, 100 kDa (Millipore, USA)

Minitan (Millipore, USA)

Sugar beet pulp solution

TMP Temp. Flux (bar) (°C) (l/m2h)

30

60

0.5

20

7–60

Remarks

Ref.

High sucrose, glucose, fructose and raffinose retention Acid and alkaline cleaning every 2–80 h

[9]

5–25 Cleaning for 1 h with [11] (for 30 kDa) 0.5 w/v Ultrasil-11 (Ecolab, Germany) at 50°C for 1 h

8.2 Beet and cane sugar production 

 167

can be recovered by membrane operation. Hatziantoniou and Howell [11] studied the isolation of the sugar beet pectin, which can be used as thickener, fat replacer or fat mimic in the food industry by applying ultrafiltration (UF) with diafiltration. The laboratory scale experiments were carried out after initial pH adjustment and prefiltration, more details are included in Table 8.1. An initial concentration was conducted until a VCF of 4 was reached. After this, diafiltration with five times the feed volume followed by a final concentration step was carried out. Even though it was possible to isolate the pectin fraction from the sugar beet pulp by removing components with a molecular weight less than 6 kDa from the retentate stream, the fluxes achieved were low because of high fouling of the pectin.

8.2.1.2 Raw juice purification One of the main challenges in the sugar production process is the separation of nonsucrose components. The class of non-sucrose components covers both organic and inorganic components, such as the nitrogen-containing components (e.g., betaine, amino acids, amides and proteins, organic acids, vitamins), and mineral substances (e.g., carbonates, chlorides, sulfates and silicates). In particular, salts with monovalent cations are undesirble as they are melassigenic [12]. Furthermore, colored matters with a molecular weight of 500–20,000 Da have the tendency to form inclusions in the sugar crystals or to be adsorbed onto their surface [13]. The traditional process to remove the non-sucrose components is sedimentation with the addition of calcium-oxide and carbon dioxide. This method removes about 35% of the non-sucrose components but is also a challenge for the environment [12]. From an environmental point of view, membrane separation could therefore be an attractive alternative to treat and purify the raw juice. The first contribution in this area was the use of UF for the purification and clarification of beet juice patented by Rud F. Madsen in 1971 [14]. The idea was to replace the classical lime purification partially or fully by UF, separating raw juice into a retentate stream containing the high molecular components and collecting most of the sugars purified in the permeate. The first results obtained by Madsen [14] using cellulose acetate membrane resulted in very low fluxes. However, during 1979–1981, Nielsen et  al. [15] obtained the first promising results using a M35 plate-and-frame module and polymeric membranes (both from Alfa Laval Nakskov, Denmark), see Table 8.2. Furthermore, Hatissens et al. [16] stated in 1984 that fouling problems could be eliminated by a tangential velocity of 4 m/s and that no pre-filtration of the raw juice was needed before the clarification by UF but conversely it was still recommended to use additional further treatment steps, such as liming (0.05%), electrodialysis (ED) or ionexchange to remove remaining impurities. More recently, Attridge et al. [17] have reported the use of UF spiral wound (SW) module and tubular modules on a commercial scale. The SW system consisted of two parallel housings with each four hard-wrapped 8″ SWs and the tubular system

Polymeric UF membrane MWCO: 200 KDa (ITT-PCI, UK)

½″ tubular

Polymeric membranes in tubular module Polymeric UF membrane ½″ tubular FEG MWCO: 45–100 kDa (Koch Membrane Systems, USA)

Clarified raw juice

Unfiltered raw juice from clarifier overflow Spiral retentate (VCF: 5)

Clarified and150 μm prefiltered raw juice

Polymeric membranes in spiral-wound module Polymeric UF membrane 8″ Spiral MWCO: 45–100 kDa (Koch Membrane Systems, USA)

Feed

Clarified and limed raw juice

Module

Polymeric membranes in Plate&Frame module GR 61 PP (Polysulfone) M35 MWCO: 20 kDa Plate&Frame (DDS/DSS now Alfa Laval, Denmark)

MF/UF membrane type

85

80

–

–

80

80–88

80

110 (VCF 1.5–5)

20–100

50–175

100 (VCF: 6)

50–130

Temp. (°C) Flux (l/m2h)

–

0.17

2.8–4.5

TMP (bar)

Table 8.2: Raw beet juice purification by ultrafiltration and microfiltration. Adapted from [10]

Total operating time: 1,700 h Average cycle: 35 h VCF: 1.5 to 20 Reductions in permeate: 2–30% color 60% dextran 75% oxalates 100% suspended solids

Total operating time: 900 h Average cycle: 50 h VCF: 2–20 Total operating time: 300 h Average cycle: 30 h VCF: 20–50

Total operating time: 1,400 h Average cycle: 31 h Reductions in permeate: 17.7% color 98.6% turbidity 94.1% polysaccharides 100% suspended solids

28% color reduction

Remarks

[18]

[17]

[17]

[15]

Ref.

168   8 Membrane operations in the sugar and brewing industry

8.2 Beet and cane sugar production 

 169

consisted of 98.5″ FEG membranes arranged in two-pass configurations (both from Koch Membrane Systems, USA). Table 8.2 summarizes the results and it can be seen that by applying a SW module it is possible to purify and concentrate a clarified and prefiltered juice up to a volume concentration factor (VCF) of 6 and that the tubular modules can process unfiltered clarified juice and SW retentate up to a VCF of 20. Similar results were reported by Tyndall [18] for clarified juice up to a VCF of 5 using 0.5″ tubular membranes (ITT-PCI Membrane Systems, UK) on pilot scale with an 18-tube housing in single-stage operation (see Table 8.2 for details). An overall concept for the treatment of raw juice with polymeric hollow fiber (HF) microfiltration (MF) is the so-called “A.B.C. process” (Honiron Engineering, USA), which combines a continuous screening (“A”) with UF followed by an optional softening and alkaline adjustment before evaporation (“B”) and adsorption (“C”) [19]. In the screening step, the juice from the diffuser is prefiltered by a 150 μm screen, followed by heating to 80°C and further screening with a rotary filter. The clarification of the beet juice is carried out by MF. In the MF step, the raw juice is separated into a concentrated retentate and a purified permeate stream. The retentate stream can be either blended with factory molasses for Brix adjustment or evaporated for usage as animal feed and a purified permeate stream. The permeate stream is further processed to sugar by optional softening with weak cationic ion-exchange resin and alkaline adjustment followed by evaporation to 60–70°Brix and adsorption to remove colorants and viscosity precursor components. See Table 8.2 for details of the MF step. Apart from polymeric membranes, the use of inorganic – ceramic and metallic – membranes to achieve commercially interesting fluxes and permeate purities/rejections has been investigated. Hinkova et al. [8] observed at pilot scale that, by using a ceramic membrane, it was possible to obtain a juice suitable for direct crystallization as shown in Table 8.3. On industrial scale, stainless steel Scepter® membranes (Graver Technologies, USA) have been applied for diffusion juice [20]. In a three-stage membrane system consisting of 416 m2 MF membrane area, 50 m3h of diffusion juice with 13–16°Brix can be treated. The recovery rate of the system is 38% in the first stage, 72% in the second stage and 97% in the third stage, resulting in a retentate stream of 1.5 m3h with 1.6°Brix. One of the key factors in the successful implementation of UF and MF for clarification of beet juice is the pretreatment of the juice. Schrevel [21] compared results from different membranes – ceramic and polymeric – in different module configurations – tubular, HF and SW. From the results shown in Table 8.4 it can be seen that pre-limed juice results in significantly higher fluxes than raw juice. Based on this work it was concluded that supplementary treatments such as pre-filtration with a mesh screen and settling or pre-liming, settling, carbonization and pre-filtration with a mesh screen, support the subsequent UF of the beet juice to achieve the required sugar qualities. In line with this, Tebble et al. [22] proposed the integration of membranes in a side stream approach to reduce the need for lime and boost the overall capacity of the process.

170 

 8 Membrane operations in the sugar and brewing industry

Table 8.3: Raw beet juice purification by ultrafiltration and microfiltration. Adapted from [10] MF/UF memb- Module rane type

Feed

TMP (bar) Temp. (°C)

Polymeric membranes in hollow fiber module Hydrophilic Hollow 150 μm – polymer fiber prefiltered 0.1 to 0.2 μm diffuser A.B.C. juice passed process through a (Honiron rotary filter Engineering, USA) Inorganic membranes in tubular module Membralox Ceramic Diluted ceramic UF tubular raw juice Pore size: with 19 concentrate 20 μm channels (Pall Exekia, ø 0.4 mm France)

0.1–0.2

Septer® Stainless Diffusion juice 5 metallic MF steel with (Graver tubular 0.3% Technologies, suspended USA) solids

80

Flux (l/m2h) Remarks

Ref.

–

[19]

30, 50, – 60

75

117

–

Average [8] cycle: 3–10 h Reductions in permeate: 99% turbidity 30–50% color Average [20] cycle: 22 h VCF: 30

Overall, it appears that both UF and MF are generally suitable for the purification of raw beet juice as alternatives to liming or to minimize liming. However, fluxes and investment costs have to be balanced to offer a viable alternative to the conventional process. Further, the pretreatment of the beet juice seems to be the key to successful implementation of membrane technology. Figure 8.3 shows two potential concepts for integrating membrane technology in the clarification of the raw juice. The first concept uses MF/UF as additional purification process after partial/standard liming, the second concept is a two-stage MF/UF process and is used as a full replacement of the liming and minimize sugar losses in the retentate. Other works concentred their efforts on the removal of color impurities from raw sugar by UF. Mak [23] reported the use of an Alfa Laval (Sweden) filtration unit with PM10 HF modules that are 1.1 m long and contain a membrane area of 2.46 m2 per module. In this UF process, it was possible to remove proteins, starches, gums, colloids and color impurities. Anion-exchange resins are usually used to remove high molecular colorant like melanins, melanoidines, caramels and polyphenols from sugar liquor. The colorants are first adsorbed onto the resins and then the resins are regenerated with an

8.2 Beet and cane sugar production 

 171

Table 8.4: Influence of module configuration and membrane material on beet juice clarification. Adapted from [10] TMP (bar) Temp. Flux (l/m2h) (°C)

Remarks

Ref.

Ceramic pH adjusted, tubular 150 μm prefiltered and settled raw juice Spiralwound

0.4–0.7

–

50–190

VCF: 2.3 Average purity increase: 80%

[21]

Ceramic Pre-limed tubular juice

1.5

–

300

VCF: 5

Ceramic tubular

3–4

150

VCF: 4–5

Hollow fiber

0.5

300

VCF: 4–5

MF/UF membrane type

Module

Ceramic MF membrane Pore size: 0.2 μm

Polymeric UF membrane MWCO: 20 and 40 kDa Kerasep ceramic MF Pore size: 0.1 μm (NovaSep-Orelis, France) Kerasep ceramic UF MWCO: 15 kDa (NovaSep-Orelis, France) Polymeric MF membrane Pore size: 0.2 μm (X-Flow, The Netherlands)

Feed

MF/UF unit

CO2 Lime kiln

Carbonation

Purified raw juice

Lime milk Raw juice

Liming or partial liming

(a)

MF/UF unit Raw juice

(b)

Pre-treatment

Clarifier

Drum filter

Pre-treatment

Purified raw juice

Drum filter Clarifier

Figure 8.3: Concepts for the integration of membrane technology in purification of sugar beet raw juice

alkaline sodium chloride solution, resulting in a stream characterized by high salinity, high amount of colored organic matter and high chemical oxygen demand (COD), which represents a serious pollution problem [8]. Wadley et al. [24] proposed the use of NF membranes for the regeneration of this stream. By using a SelRO MPT–30 or

172 

 8 Membrane operations in the sugar and brewing industry

MPT–31 (Koch Membrane Systems, USA) with an NaCl retention of 10–50% and an organic compound retention in the range of 80–97%, it was possible to reduce the effluent volume by 30% and the salt consumption by 60%. Even better results were obtained by Cartier et al. [25], achieving a 89% water reduction and a 74% reduction in salt consumption using SW modules with Desal 5 (GE Osmonics, USA) and NF45 (Dow FilmTec, USA) membranes.

8.2.1.3 Demineralization of beet juice The thin juice obtained after carbonation and filtration has amounts of alkali cations (Na+, K+ and Ca2+) that should be removed. Various systems such as ion-exchange resins, synthetic adsorbents, coagulants, and membrane processes have been reported in the literature [26–31]. ED has been proposed as alternative or addition to the classical demineralization of the beet sugar juice by ion-exchange. By application of ED it is possible to demineralize beet sugar juice of 30°Brix by up to 80% [32 ]. The successful integration of ED in the European sugar industry started in 1996 and has been documented by Lutin et  al. [20 ]. The target was to reduce the load to an ion-exchange demineralization unit, thus doubling the plant capacity without increasing the load on the ionexchange and the need for regeneration. Therefore, ED was integrated as a pretreatment before the ion-exchange to demineralize the juice by 55%. To treat 40 m3h of 12 or 24°Brix beet sugar juice, 12 ED stacks were arranged in three parallel lines to minimize treatment time and achieve 50% conductivity reduction. The sugar loss in the process is less than 0.5% of the sugar production. The greatest technical challenges of applying ED in the sugar industry were the short membrane life, in particular for the anion-exchange membranes, the high viscosity of sugar syrup and the ED operating temperature of less than 40°C. In 2002, Elmidaoui et al. [33] published the work using a new generation of membranes manufactured for sugar applications by Tokuyama (Japan). The new membrane, Neosepta AXE 01 (Tokuyama, Japan), has a facilitated transport of organic anions and resistance to cleaning with NaOH at high temperatures. The membrane was tested in an ED pilot plant of 2.5 m2 at a production rate of 24 m3/d. The Neosepta AXE 01 membrane (Tokuyama, Japan) showed high alkali resistance with only 10% decrease of exchange capacity and 22% burst strength decrease, compared to 90% and 67% of the AFN Neosepta membrane (Tokuyama, Japan), the most suitable ED membrane for the sugar industry until then. In addition, the juice quality improved by demineralization reaching 9% in purity and 10% of the decoloration rate. At the end of the operation, 75% of Na+, 86% of K+ and 65% of Ca2+ were removed and no fouling was observed during 200 h of operation. The above results demonstrate that ED can be considered an alternative or addition to conventional demineralization by ion-exchange, offering control of organic fouling, reduction in waste effluents and pollution load, reduction in volume of molasses and reduce of capital cost in a continuous operation.

8.2 Beet and cane sugar production 

 173

8.2.1.4 Preconcentration of thin juice After the demineralization process, the thin juice enters a multi-effect evaporator system in which the sugar is concentrated from 15% up to 60% [34], see Figure 8.1. Evaporation is the most energy-intensive process in sugar factories and consumes about 50% of the total energy, which results in high operating cost and environmental challenges [34, 35]. Furthermore, heating sugar juice can decrease the product quality by changing the color and flavor. In this context, membrane processes, which are operating at ambient temperatures with no phase inversion, could be considered interesting candidates for this application. Several membrane processes such as membrane distillation (MD) [36], osmotic distillation (OD) [37–42], NF [43–46] and RO are suitable for the concentration of sugar syrup. The initial work on using RO for thin juice concentration was carried out at the start of the 1970s using cellulose acetate membranes, which limited the operating temperature to 25–30°C [7]. The low fluxes obtained around 10 l/m2h were not interesting from a commercial point of view and the low-operating temperatures lead to the risk of microbial growth. The development of new types of membrane, i.e., thin-film RO, made it possible to concentrate the thin juice at temperatures up to 80°C. The first experiments were carried out on pilot scale in the beginning of the 1980s using an M30 plate-and-frame module with HR95PP thin-film RO membranes (both Alfa Laval Nakskov, Denmark) and temperatures around 75–80°C. Even though higher fluxes have been obtained (Table 8.5), the osmotic pressure restricted this approach to a water reduction of 20% and further concentration by evaporation was required. Alternatively, Koekoek et  al. [47] tested NF SW modules (Toray, Japan and GE Osmonics, USA) and an NF tubular module (Pentair/X-Flow, The Netherlands) for the concentration of thin juice on pilot scale. The system was designed for 5 m2h of raw juice and consisted of 7 × 4″ SW modules (45 m2 membrane area) and a tubular module with seven tubes (13.3 m2 membrane area). The results obtained are summarized in Table 8.5 and show a 30% reduction in steam consumption by concentrating the thin juice up to 22%, but unsatisfactory long-term performance of the membranes caused by fouling was observed. In addition, the cleaning process could not restore the initial fluxes. Similar results were also observed in the preliminary tests conducted by Hinkova et al. [8]. In a test of six different NF membranes (Hydranautics – NittoDenko, USA/ Japan and GE Osmonics, USA) the fluxes of all membranes were reduced to close to or below 10 l/m2h after a period of 4 h. In 1981, Sandre [48] presented a technical feasibility study of RO for the preconcentration of thin juice. This work clearly demonstrated the benefits and advantages of membrane technology, principally a lower cost, using a PA300 membrane (Koch Membrane Systems, USA) to concentrate thin juice from 13% to 30% at high temperatures. Similar results were also obtained by Bichsel and Sandre [35] for the concentration of thin juice from 13–30% sugar content using PA300 (Koch Membrane Systems, USA) and FT30 RO membranes (Dow FilmTec, USA). Figure 8.4 provides an overview of potential membrane applications related to thin juice concentration.

174 

 8 Membrane operations in the sugar and brewing industry

Table 8.5: Clarified thin juice concentration by reverse osmosis and nanofiltration. Adapted from [10] RO/NF membrane type

Module

Feed

TMP (bar)

Polymeric membranes in Plate&Frame module HR95PP (ThinM30Plate Thin 40 film) Sugar &Frame juice with retention: 15–20% TS 99.975% (DDS/ DSS now Alfa Laval, Denmark)

Temp. (°C)

Flux Remarks

75

30–40 (l/m2h)

Polymeric membranes in spiral-wound and tubular modules Toray 610T 50 μm 30 70 41–72 4″ spiral (thin-film) prefiltered (kg/m2h) (Toray, Japan) thin juice with16– 22% TS and 30–35 mg CaO/l from ionexchange Desal DL (Thin-film) (GE Osmonics, USA)

STORK WFNX (X-Flow, The Netherlands)

4″ spiral

41–101 kg/m2h

Tubular module with seven tubes

10–20 (kg/m2h)

NF/RO unit

Pre-concentrated thin juice

Thin juice Permeate 99 % water

Ref.

20% water reduction in retentate Cleaning 0.5% HNO3 and Ultrasil-10 (Ecolab, Germany)

[15]

Permeate composition: 97% water 2% sugars < 1% non-sugars

[47]

Permeate composition: 99% water 1% sugars < 1% non-sugars Permeate composition: 97% water 1.5% sugars < 1% non-sugars

Concentrated thin juice

Evaporator Condensate RO unit

Permeate recovered water Retentate concentrated COD/BOD

Figure 8.4: Overview of potential membrane applications in the thin juice concentration

8.3 Membrane application in cane sugar production 

 175

Furthermore, Madaeni and Zereshki [34] presented the effect on energy consumption by using a two-stage RO system for preconcentration of thin juice up to 20°Brix. The related experiments were conducted using a BW30 RO membrane (Dow FilmTec, USA) at 22 bar transmembrane pressure and an operating temperature in the range of 30–45°C. The results were used in a subsequent simulation of energy and membrane area requirements using data obtained at the Bisotoun Sugar Factory-Kermanshah, Iran. Based on a thin juice flow rate of 80 m3h it was estimated that the energy consumption for the RO-evaporator combination would be 6,553 kW whereas the evaporator alone would require 9,740 kW. Hence, using RO for the preconcentration of the thin juice from 15 to 20°Brix prior to final concentration by evaporation reduced the energy costs by 33%. The required membrane areas were calculated at 2,801 m2 for stage one and 1,114 m2 for stage two. The recent work by Gul and Harasek [7] presents a new multi-stage pressure-driven membrane separation process for the preconcentration of thin juice. This new membrane concept can concentrate thin sugar juice from an initial feed concentration of 15% to 50% by use of membranes at a moderate transmembrane pressure of 32 bar at 80°C. Hence, 82% of the water is removed without any phase change. After this, the preconcentrated juice can be further concentrated up to 65–70%. It is stated that this process can save more than 80% energy in the concentration step and allow about 70% reduction in heat and heat-transfer area for pre-heating before evaporation.

8.3 Membrane application in cane sugar production Nowadays, a can sugar factory has often become a combination of sugar mill, power plant and distillery, which produce sugar, electricity and ethanol, respectively [6]. The production of cane sugar is quite similar to the beet sugar process and an overview of the production is given in Figure 8.5. The major steps in the cane sugar production process are as follows: – Pretreatment: The cut cane is delivered in bundles to a cane sugar mill, where it is chopped and shredded using rotating knives and hammer mills within 24 h of delivery.

Water Sugar cane

Cane milling

Lime Raw juice

Liming

Clarifier

Clarified raw juice

Juice concentration by evaporation

Raw sugar transportation Raw sugar crystallisation

Sugar mill Sugar refinery

Mud removal Drum filter Mud Refined sugar

Crystallization

Figure 8.5: Traditional cane sugar production

Decolorization by ion exchange

Raw sugar remelt

176 

–

–

– –

 8 Membrane operations in the sugar and brewing industry

Extraction of cane sugar juice by roller mills or diffusion: The extraction of the cane sugar is conducted as a countercurrent process using fresh hot water pumped through the chain of multiple roller mills or the continuous diffuser to extract raw juice and bagasse. Both products are separated in different places. Generally, bagasse moves to a co-generation plant for the production of electricity and raw juice is moved for juice clarification and juice treatment. Cane juice purification: The dark-green cane juice has some non-sugars, impurities that are removed in the clarification process. A mix of chemical reactants such as sulfur and lime are used as clarifying agents. Heating of the juice is required for the chemical reaction that finally separates the raw juice into two parts. The solid form is called mud or filter cake. Alternatively, sulphitation or/and carbonation is used to produce directly consumption sugar. Cane juice concentration by evaporation: The cane juice is concentrated in a multi-stage evaporator to 60–75°Brix and is now called syrup/raw syrup. Crystallization: The syrup is boiled and crystallized in several steps and separated into white sugar and remaining sugar syrup/molasses.

It is common that the initial extraction and concentration of cane sugar juice and also the crystallization of sugar are carried out in two different plants. In the first one, the raw syrup or raw sugar is produced by an initial crystallization of cane sugar by evaporation. In the second one, the raw sugar is refined by purification, decolorization and crystallization means. The following section will show the applications of membrane technology for both cane sugar mills and refineries.

8.3.1 Raw sugar cane juice purification The purification of sugarcane juice by membrane filtration promises a significant improvement in the sugar quality and yield [49]. In spite of extensive laboratory trials [15, 50–54], field tests aimed at full-scale operation have been conducted only in the last decade [19, 55–57]. Clarification of cane sugar is similar to the process described for beet sugar. Rud F. Madsen [14] patented in 1971 for the first time a membrane purification process. The major advantage of membranes over the conventional process is that it is possible to removal color and macromolecules, such as fat, starch and dextrans contemporaneously. In 1982, Nielsen et al. [15] reported the early results at pilot scale (see Table 8.6 for more details). An M35 plate-and-frame module with 42 m2 membrane area equipped with polymeric membranes (both Alfa Laval Nakskov, Germany) were used for the experiments. In general, the successful application of membranes in the clarification of cane juice is related to the use of inorganic membranes and the integration of membrane filtration in the overall process flow.

Module

Feed

Septer® metallic MF (Graver Technologies, USA)

–

–

–

95

–

TMP (bar) Temp. (°C)

Clarified – and 100 μm prefiltered juice (plus 60 μm safety filter)

Stainless steel Clarified cane tubular juice

Inorganic membranes in tubular module NAP Process Ceramic Kersep ceramic UF tubular with Pore size: 0.02 μm 19 channels ø (NovaSep-Orelis, France) 2.5 mm

Polymeric membranes in Plate&Frame module GR 61 PP (Polysulfone) M35 Clarified and MWCO: 20 kDa Plate&Frame limed cane (DDS/DSS now Alfa Laval, juice Denmark)

MF/UF membrane type

Results in improved final product

Remarks

200–300 Total operating time: four seasons Average cycle: 24–48 h Cleaning for 2 hours using sodium hypochlorite (300 ppm) and phosphoric acid (0.5 wt.%) VCF: 10 Reductions in permeate: 10–15%color > 99% turbidity 85% hardness 170 Average cycle: 24 h Reductions in permeate: Turbidity < 1 NTU Dextrane free based on Haze test.

50–130

Flux (l/m2h)

Table 8.6: Raw sugar cane juice purification by ultrafiltration and microfiltration. Adapted from [10]

[58]

[55]

[15]

Ref.

8.3 Membrane application in cane sugar production 

 177

178 

 8 Membrane operations in the sugar and brewing industry

The New Applexion Process (NAP) (Applexion, France) that involves a combination of two membrane stages was first installed in a cane sugar factory in 1994 [59]. In the first stage of NAP, using UF ceramic tubular membranes, high molecular weight components such as starch, dextran, wax, and gums are removed from prefiltered lime clarified cane juice. Later, in the second stage, ion-exchange is applied to remove calcium and magnesium salts from the previously ultrafiltered cane juice. Limed juice is used in this process; therefore, carbonation and/or sulphitation are not necessary in white sugar production. The UF process was designed to treat 380 l/min of prefiltered, lime clarified cane juice. The plant involved two parallel lines each with three loops of 10 modules, each with 99 ceramic tubular Kerasep ultrafiltration membranes (NovaSep-Orelis, France) that represented a total membrane area equal to 940 m2. Table 8.6 summarizes the operation data during the four seasons (1994–1998) after which a membrane exchange became necessary [60]. The syrup obtained in the UF plant was used for the production of very low color (VLC) sugar [59]. In 2002, Chou [61] developed a modification of the NAP process described previously, called the “SAT process”. The modification includes two processes [approved by the US Food and Drug Administration (FDA)] added before the UF step. In addition to the NAP and the SAT process the A.B.C. process – described in section 8.2.1.2 – has been adopted for cane juice [62]. However, the fluxes achieved with HF MF membranes were not sustainable, and then it is necessary to use the clarified cane juice instead of raw diffusion juice [58]. The use of tubular stainless steel Scepter® MF membranes (Graver Technologies, USA) has been reported for the clarification of cane juice [63]. The system was a threeloop unit with a total membrane area of 527.5 m2 to treat around 89 m3h during 24 h of production [58]. Even though inorganic membranes have shown successful results in the cane juice clarification, the application of polymeric membranes was necessary because of their lower cost, higher packing density and ease of exchange. In the season of 1997–98, a field test was carried out, where the successful application of SW modules (Koch Membrane Systems, USA) in the filtration of clarified cane was confirmed [64]. Similar results on a commercial scale were obtained by McArdle [65] during the season 1997–98. Afterwards, to boost a sugar recovery higher than 99%, the retentate obtained in the SW modules that achieved a VCF of 10, was further concentrated up to VCF 20 using tubular carbon membranes (Koch Membrane Systems). Steindl and Doyle [58] had confirmed the flux levels and operating conditions shown in Table 8.7. Table 8.8 shows details of the successful application of SW UF for the purification of prefiltered raw juice after diffusion, reported by Martoyo et  al. [66]. The results obtained in two sugar factories – fluxes in the range of 50 to 120 l/m2h and VCFs up to 30 – were interesting fluxes from an economical point of view.

8.3 Membrane application in cane sugar production 

 179

Table 8.7: Raw sugar cane juice purification by ultrafiltration and microfiltration. Adapted from [10]. MF/UF mem- Module brane type

Feed

TMP (bar)

Temp. (°C)

Flux (l/m2h)

Remarks

Ref.

Polymeric membranes in spiral-wound module combined with inorganic membranes in tubular module PVDF UF 8″ Spiral Clarified – 95 75–120 Average cycle: [58] membrane with and 48–72 h MWCO: 60 mm prefiltered Cleaning 45–200 kDa spacer cane juice using caustic, (Koch hypochlorite, and Membrane detergent Systems, VCF: 7–10 USA) Reductions in permeate: 10–22% color 100% suspended solids 97–99.8% turbidity 3% Brix Carbon MF Carbon Cane juice – – 80–175 Average cycle: membrane tubular concentrate 48–72 h Pore size: with from spiralCleaning using 0.1 and inner wound caustic, phosphoric 0.2 μm tube module acid, sodium (Koch ø 6 mm hypochchlorite and Membrane detergent Systems, VCF: 20–50 USA) Combined with spiral-wound and diafiltration 99% sugar recovery in permeate

Further evidence about the use of SW UF modules have been reported by Ghosh et al. [67, 68] and this is shown in Table 8.8. The results obtained in the experiments carried out with 2.5″ and 3.2″ modules (Permionics, India and Inchema, Switzerland) showed a permeate stream with an expected quality. However, the fluxes obtained were significantly lower; see Table 8.8 for more details. Similar results were also achieved in a large-scale pilot test with 809  m2 of membrane area using an 8″ SW module (Permionics, India). The low fluxes were related to strong fouling [68]. The application of tubular polymeric membranes has also been tested for the purification of raw juice. Tyndall [18] has reported the successful application of 0.5″ tubular membrane (ITT-PCI Membrane Systems, UK) at pilot scale (Table 8.9) for filtration of raw juice before clarifier and raw juice from clarifier underflow.

Module

Feed

3.2″ Spiral with 32 mil spacer

8″ Spiral

Cellpore® (modified polysulfone) MF membrane Pore size: 0.8 mm (Inchema, Switzerland)

Polyethersulfone UF membrane MWCO: 20 kDa (Permionics, India)

Prefiltered clarified raw juice

2.5″ Spiral with Prefiltered raw juice 40, 60 and 80 mm spacer

Polyethersulfone UF membrane MWCO: 20 kDa (Permionics, India)

Polymeric membranes in spiral-wound module PVDF UF membrane Spiral with ø Prefiltered raw juice after MWCO: 30–50 kDa 100 mm diffusion

MF/UF membrane type

85

Temp. (°C)

1.6–3.1 91–97

1.0–6.0 50–65

2.7–3.5

TMP (bar)

7

13–47

10–41

47–129

Flux (l/m2h)

Table 8.8: Raw sugar cane juice purification by ultrafiltration and microfiltration. Adapted from [10]

Total operating time: 980 h Average cycle: 24–72 h (depending on VCF) Chemical cleaning VCF: 5 to 30 Average cycle: up to 2 h Cleaning using 0.5% HCl and 0.1% caustic at 50°C Reductions in permeate: 8–11% Brix 18% Non-sugars 5–10% Sugars Average cycle: up to 2 h Cleaning using 0.5% HCl and 0.1% caustic at 50°C Reductions in permeate: 9% Brix 22% Non-sugars 5% sugars Average cycle: 8–14 h Cleaning using acid, alkali, and detergent Reductions in permeate: 2% Brix 7% Non-sugars

Remarks

[68]

[67]

[66]

Ref.

180   8 Membrane operations in the sugar and brewing industry

8.3 Membrane application in cane sugar production 

 181

Table 8.9: Raw sugar cane juice purification by ultrafiltration and microfiltration. Adapted from [10] MF/UF mem- Module brane type

Feed

TMP (bar) Temp. (°C)

Polymeric membranes in tubular module Polymeric UF ½″ Raw juice – membrane tubular before MWCO: clarifier 200 kDa (ITT-PCI, UK) Raw juice 2.8–4.1 from clarifier underflow

–

93

Flux (l/m2h)

Remarks

– Reductions in permeate: > 70% dextrans 100% suspended solids 110–160 Cycle: 21 h VCF: 5 Reductions in permeate: > 70% dextrans 100% suspended solids

Ref.

[18]

It seems that the use of membrane technology for the clarification of raw sugar cane juice results in VLC sugar after evaporation and crystallization and consequently in higher sugar quality. It should be noted that a clarification step as pretreatment before the UF unit and the operation at high temperatures ensure the high fluxes and therefore the success of membranes in this application. In addition, tubular modules with inorganic membranes have a proven track-record on industrial scale, while polymeric membranes in SW and tubular module configuration have the potential to establish themselves in the market based on the published pilot data.

8.3.2 Concentration of clarified cane juice Even though evaporation has been traditionally used for the concentration of clarified cane juice, some alternatives have been reported in the literature. Nene et al. [36] described the viability of concentrating clarified sugar cane juice by MD using flat membranes (Membrane, Germany). This application became one of the latest to be introduced in the cane sugar industry. MD is a membrane procedure in which two process streams with different temperatures are separated by a non-wetted microporous membrane. The DF of the mass transfer is thus the vapor pressure difference resulting from the gradient temperature across the membrane.

8.3.3 Molasses treatment Molasses is produced after the crystallization process in the sugar cane mill, and is a by-product of the refinery process. Cane molasses contains typically 55% of sucrose,

182 

 8 Membrane operations in the sugar and brewing industry

10% of ash and 5%–10% of organic non-sucrose components and 18%–25% of water, and for this reason it is generally sold at a low price as cattle feed and for alcohol products [69]. A great quantity of sucrose cannot be crystallized and remains in the cane molasses due to the inhibition imposed by the presence of ionic components. UF with polymeric membranes can be used to concentrate high molecular weight components and ash in the retentate and sucrose in the permeate according to the data reported by Nielsen et al. [15]. Cartier et al. [70] highlighted the use of ceramic membranes at this point to achieve a flux of 33 l/m2h and a final VCF of 7. Considering the overall cane sugar production, high molecular weight components should be removed at an early production stage because they tend to complicate the production in different ways. In the case of molasses, without high molecular weight components, ED can be used to obtain a concentrate stream containing potassium, which could be used as fertilizer, and an ion-free stream containing sucrose for crystallization [71].

8.3.4 Decolorization of remelted raw sugar Despite the raw cane juice or syrup obtained from the cane sugar mill, the purified process has not been completely satisfactory. In this context, MF and UF were initially proposed to purify and decolorize the remelted raw sugar in the refinery. Figure 8.6 shows the proposal of integration processes such as flocculation/coagulation followed by pre-decolorization using MF/UF process and finally a decolorization by means of ion-exchange. Conversely, ED has been proposed by Thampy et  al. [72] to remove the remaining inorganic matters from the raw sugar. In the 1980s, only polymeric membranes were considered for this purpose. However, with the development of inorganic membranes and the successful integration of these in the NAP process, the research on the decolorization of the raw sugar shifted towards these membranes. Furthermore, the research has been focused on the effect of different start-up modes, for example, abrupt or progressive modes increase crossflow velocity and transmembrane pressure. Dornier et al. [73] have studied the effect of start-up modes on the performance of tubular ceramic microfiltration membranes (Pall Exekia, France). The experiments were carried out at pilot level with a constant temperature of 80°C. It was found that a progressive mode gave initially 10–25% higher flux, but over longer operating times this difference diminished. Tubular MF/UF unit Remelted raw sugar

Flocculation / coagulation

Decolorization by ion exchange Pre-decolorization

Figure 8.6: Decolorization of raw sugar remelt with membranes technology

Crystallization

8.4 The brewing industry 

 183

Conversely, the influence of flocculation before MF/UF was studied by Cartier et  al. [70, 74]. A tubular ceramic Kerasep membrane (NovaSep-Orelis, France) was used for the pilot-scale experiments; see more details in Table 8.10. The results showed that combined with flocculation, the UF membrane with MWCO of 300 kDa showed the most promising results achieving fluxes in the range of 60–65 l/m2h at a decolorization rate of 50% and a turbidity removal equal to 90%. Hence, this process can support the subsequent decolorization by ion-exchange resin. Decloux et  al. [75] have determined that using a tubular ceramic Kerasep UF membrane with a MWCO of 15 kDa (Novasep-Orelis, France), the optimal operating conditions were 60°C and a transmembrane pressure of 3 bar to obtain the best decolorization; more details in Table 8.10. Hamachi et  al. [76] reported the limitations of UF for decoloration using three different tubular ceramic Membralox SCT UF membranes (Pall Exekia, France). Even when using the tight UF membrane of 1 kDa, color removal did not exceed 60% as some of the color components were too small to be retained by UF (Table 8.10). By using polymeric membranes (Pall Filtron, USA) with an MWCO of 30, 50 and 100 kDa in lab stirred cells (GE Osmonics, USA) and tubular carbon membranes (Novasep-Orelis, France) with an MWCO of 15, 30 and 50 kDa in single tube test unit, Karode et al. [77] have demonstrated that pretreatment alone did not always increase the decolorization rate. The fluxes obtained were stable around 50 l/m2h and the color reduction was 50% at a temperature of 80°C. In this case, pretreatment of the raw sugar with coagulation did not affect the overall decolorization; see Table 8.11 for more details. From the above it can be seen that MF and UF are interesting alternatives to reduce the load on the ion-exchange conventionally used for decolorization, thus increasing operating cycles and efficiency. The first commercial unit for this application was installed at a European factory in 1997 for a capacity of 100 t/day of raw sugar [78].

8.4 The brewing industry Beer is an ancient beverage and the brewing process can be traced back almost 5,000 years, according to the clay tablets found in Mesopotamia that described the beer brewing process. Over time, different types of starchy plants have been used for brewing, including maize (in South America), soy (in India and Persia), millet and sorghum (in Africa), and rice (in the Far East). Nowadays, beer production from barley malt is the most common brewing process worldwide [79]. Beer is the fifth most consumed beverage in the world after tea, carbonates, milk and coffee and it continues to be a popular drink with an average consumption of 9.6 liters per capita by population aged above 15 [80]. Western and European markets have been stagnant or declining for some years now, whereas the Americas, the Far East and South-east Asian markets are growing. Today, beer is the world’s favorite alcoholic beverage with annual production running at over 1.5 billion hl worldwide. In the last decade, the market for beer has

Module

bwith

flocculation. flocculation.

awithout

Kersap ceramic UF MWCO: 15 kDa (NovaSep-Orelis, France) Membralox ceramic MF MWCO: ~50 (20 nm), 5,1 kDa (Pall Exekia, France) Carbosep carbon UF MWCO: 15, 30 and 50 kDa (NovaSep-Orelis, France)

Kersap ceramic UF MWCO: 15 kDa NovaSep-Orelis, France

Ceramic tubular 19 channels ø 2.7 mm Ceramic tubular 19 channels ø 4 mm Carbon tubular 1 channel ø 6 mm

Inorganic membranes in tubular module Membralox ceramic MF Ceramic tubular with Pore size: 1.4 μm 19 channels ø 4 mm (Pall Exekia, France) Kersap ceramic MF Ceramic tubular Pore size: 0.1 and 0.2 μm 8 channels (NovaSep-Orelis, France) ø 4.5 mm

MF/UFMembrane Type

3–5

4.4

Raw cane sugar (28 and 46°Brix) Raw cane sugar (50°Brix)

VCF 9–10 Decolorization (%): 23a45–55b Turbidity removal (%): 89a–93/89–91b

40–60a 60–80b

11–217

Decolorization (%): 55–90a,b

Decolorization (%): 23–59

15–75a,b VCF: 5–10 Decolorization (%): 23–39a /45–58b Turbidity removal (%): 95%a /92%a 47–114 Decolorization (%): 53–65

Reductions in permeate: Turbidity: 90–150 NTU

Remarks

29–74

70/80 10–50

30– 90

60– 80

85

2.0

3

80

Temp. Flux (°C) (l/m2h)

2.25

TMP (bar)

Raw cane sugar remelt (30°Brix)

Raw cane sugar remelt partly prefiltered with 360 μm Raw cane sugar remelt (50°Brix and 3000 ICUMSA) pretreated with/ without flocculation

Feed

Table 8.10: Decolorization of raw sugar by microfiltration and ultrafiltration. Adapted from [10]

[77]

[76]

[75]

[70, 74]

[73]

Ref.

184   8 Membrane operations in the sugar and brewing industry

8.4 The brewing industry 

 185

Table 8.11: Decolorization of raw sugar by microfiltration and ultrafiltration. Adapted from [10] MF/UF membrane type

Module

Feed

TMP (bar)

Polymeric membranes in Plate&Frame module Polyethersulfone UF Sepa ST Raw cane 2.9 membrane (OMEGA) stirred sugar MWCO: 5, 30, 50 and cell (GE (50°Brix) 100 kDa Osmonics, (Pall Filtron, USA) USA)

Temp. Flux (°C) (l/m2h)

Remarks

Ref.

80

VCF: 2–3 Decolorization (%): 50–80a,b

[77]

7–50

awithout bwith

flocculation. flocculation.

changed, with China becoming the biggest brewing market and overtaking the traditional brewing markets – the USA and Germany – in the process. This change in market leadership has been accompanied by major investments in the brewing industry in Asia. These new breweries are using the latest technology to maximize their productivity and quality. The industry is heavily consolidated with four major players (ABI, SABMiller, Carlsberg and Heineken) accounting for more than 50% of the world production and revenue. However, there is a trend towards micro-breweries, small local breweries with limited annual output.

8.4.1 Membrane applications in the brewing process During production, beer alternately goes through four chemical and biochemical reactions (mashing, boiling, fermentation and maturation) and three solid–liquid separations (wort separation, wort clarification and rough beer clarification) [81]. An overview of the general production process of beer is shown in Figure 8.7. The major unitary operations are as follows: – Raw materials/ingredients: Four ingredients are required for beer production: malt, hops, water and yeast. Yeast is often described as a raw material but is really Hops

Malt and water

Malting

Fermentation and marturation

Worth boiler

Tank bottom

Figure 8.7: Traditional beer production

Recovered beer

Beer recovery from tank bottoms by MF

Concentrated yeast Botteling

Beer clarification by MF

Sterile filtration Pasteurization

Bright beer cellar

Dealcoholization by RO

186 

–

– –

–

–

 8 Membrane operations in the sugar and brewing industry

the actual “factory” that produces the beer from the brew house wort (described below). For most beers, the yeast is removed after fermentation and is not present in the final product. Malting: Malt is produced from malting of (barley) grains. The grains are steeped in water and thus brought to germination. During germination starch-degrading enzymes are formed, the walls of the starch granules are degraded and the starch is made accessible for the growth of the grain. After germination, the growth process is stopped by drying and heating the grains in a kiln. The produced dry malt is suitable for storage and still contains enzymes produced during malting. These enzymes are essential to the later conversion of starch to sugar during mashing. Apart from malted barley, other starch sources such as rice and maize grits can be used. Breweries in barley producing areas have their own malting plant(s) – others do not. Milling: The malt is crushed in mills making the starch content available to conversion by the enzymes. Mashing: The mixture of milled malt, gelatinized adjunct and water is called “mash”. The purpose of mashing is to obtain a high yield of extract (sweet wort) from the malt grits and to ensure product uniformity. Mashing consists of mixing and heating the mash in the mash tun, and takes place through infusion, decoction or a combination of the two. During this process, the starchy content of the mash is hydrolyzed, producing liquor called sweet wort. In the infusion mashing process, hot water between 71 and 82°C is used to increase the efficiency of wort extraction in the insulated mashing tuns. The mashing temperature is dictated by wort heating using steam coils or jackets. In decoction mashing, a portion of the mashing mixture is separated from the mash, heated to boiling and re-entered into the mash tun. This process can be carried out several times and the overall temperature of the wort increases with each steeping [82]. Lautering: The husk parts and other insoluble parts of the malt are strained off in a lauter tun or in the mash filter and the liquid, containing dissolved sugar, amino acids, etc. is now called “sweet wort” or just “wort”. Wort boiling: The next step, wort boiling, involves the boiling and evaporation of the wort (about a 4–12% evaporation rate) over a 1–1.5-h period. The strong rolling boil is the most fuel-intensive step of the beer production process. Hackensellner [83] estimated that 44–46 kBtu/barrel is used for conventional wort boiling systems in Germany. The boiling sterilizes the wort, coagulates grain protein, stops enzyme activity, drives off volatile compounds, causes metal ions, tannin substances and lipids to form insoluble complexes, extracts soluble substances from hops and cultivates color and flavor. During this stage, hops, which extract bitter resins and essential oils, can be added. Hops can be fully or partially replaced by hop extracts, which reduce boiling time and remove the need to extract hops from the boiled wort [82]. If hops are used, they can be removed after

8.4 The brewing industry 

–

–

–

–

–

 187

boiling with different filtering devices such as sedimentation, filtration, centrifugation or whirlpool. Wort clarification: The precipitated proteins, hops and malt are called hot trub and must be removed before fermentation. This is most often done in a whirlpool. However, high speed separators may alternatively be used and decanters may improve the efficiency of the whirlpool process. Fermentation (primary): The hot wort has a temperature around 96–99°C and should be cooled to get the correct pitching temperature. Pitching temperatures vary depending on the type of beer being produced, for example, for lagers the temperature is 6–15°C, while pitching temperatures for ales are higher at 12–25°C. In addition, the wort is aerated and yeast is added to start the fermentation. During fermentation, the yeast metabolizes the fermentable sugars (glucose, fructose, sucrose, maltose and maltotriose) in the wort to produce alcohol and carbon dioxide (CO2). The process consists of a series of complex biochemical reactions. These reactions, known as the “glycolytic pathway” or “Embden-Myerhof-Parnas pathway”, involve a number of enzymes and the reactions take place anaerobically inside the cells of the brewing yeast [82]. Maturation (secondary fermentation): Some brewing methods require a second fermentation, sometimes in maturation tanks (traditional) or in the same tank as the fermentation (Uni-tank process) where sugar or fresh, yeasted wort is added to start the second fermentation. In this stage, the beer will be saturated with carbon dioxide. Off-flavor compounds in the green beer will be reduced or converted during maturation. Filtration/stabilization/blending/carbonation: After fermentation/maturation the beer is filtered for clarity in a DE-filter or a membrane filter, stabilized (shelf-life) with stabilizing agent(s), blended with other beer streams and/or de-aerated water (“high gravity brewing”), and carbonated to a final CO2 level to form the final product. Various additives may be added at this point. Bright beer transfer to packaging: The finished product is called bright beer: It is stored in bright beer tanks (BBTs) awaiting packaging/canning/bottling. When transferred to packaging, the beer may be pasteurized in a flash pasteurizer.

Membrane operations have been proposed and introduced in different steps of the brewing process with successful results. The applications have been applied in stages such as lautering, clarification, dealcoholization, and recuperation from tank bottoms.

8.4.1.1 Filtration in the lautering process The lautering process consists basically of a process in which the mash is separated into clear liquid called wort and residual grain. In the conventional method, the mixture is filtered to remove the insoluble components. In 1989, Daoud patented a

188 

 8 Membrane operations in the sugar and brewing industry

system consisting of two stages of MF process preferably with an open, tubular membrane removing up to 95% of the total suspended solids in the first stage followed by second stage with tubular membranes to remove remaining solids [84]. In 1995 the use of mash filtration by membranes to obtain wort [85] was patented. According to information published, it was possible to separate the spent grain from the mash to form a clear wort with at least one membrane filtration unit, but preferably by using a multi-stage filter, for example, a multi-stage counter-flow filtering apparatus, such as a three-stage apparatus or a multi-stage crossflow filtering apparatus with a membrane pore size not exceeding 2.0 μm, preferably ranging from 0.1 to 1.5 μm to obtain a clarity of the wort of 0.25 to 5 EBC units at 65°C. According to Daufin et al. [86], the separation of hot trub and wort by MF is an application still under development. The aim is to produce sweet wort of high-quality by removing more than 90% spent grain and maximizing the solids content in the mash. Using a two-stage MF process with preferably open tubular membranes (40–60 μm) around 95% removal of total suspended solids in the first stage can be obtained. Alternatively, a decanter followed by tubular ceramic MF membranes with 1.3 μm pores [87] or rotating MF membranes have been regarded as an option [88].

8.4.1.2 Beer clarification At the final step in the traditional brewing process after fermentation and maturation, the clarification of rough beer and pasteurization of beer represent one of the most important operations in the brewing process when it comes to producing clear and bright beer and ensuring the microbiological stability of the final product [89]. Standard filtration consists of the retention of solid particles (yeast cells, macrocolloids, and suspended matter) and solutes responsible for haze. The variety of compounds (chemical diversity, wide size range) to be retained makes this operation one of the most difficult to control [89]. Clarified beer is currently obtained by conventional dead-end filtration with a filter press using filter aids, mainly diatomaceous earth (Kieselguhr) but also with perlites, cellulose or active carbon [89]. Diatomaceous earth has been the historic foundation of beer filtration and clarification, as a standard industrial practice for more than 100 years. In 2005, Pall (USA) reported that its use as a filter corresponds to 98% of the beer industry [90]. However, diatomaceous earth is associated with handling and disposal of the powder as well as large amounts of effluents. This type of problem and the ongoing challenges related to its disposal and warehousing, and its handling in large-scale operations, has led the companies to develop a cost-effective system that does not compromize beer quality or taste [90]. In this context, beer filtration with membrane technology has been proposed to overcome the problems associated with kieselguhr filtration. In 2001, X-Flow (now Pentair, The Netherlands) introduced a MF HF system for beer clarification followed in 2003 by Alfa Laval (Sweden) and Sartorius (Germany) combining a centrifuge with cassette membrane system for the beer clarification. The Alfa Laval/Sartorius (Sweden/ Germany) crossflow system (Figure 8.8) includes polyethersulfone membranes

8.4 The brewing industry 

 189

Retentate

Cooling

Bleed Storage tanks

Unfiltered beer

Separator Beer pre-clarification

Permeate Filtered beer Microfiltration for beer clarification

Figure 8.8: Beer clarification by a membrane crossflow system

(asymmetric pore structure) specially designed for filtering beer. This technology was created for full-scale breweries with a production from 100 hl/h to 500 hl/h or more. The overall advantages of the process include: – Beer quality identical to kieselguhr filtration, see Table 8.12. – Consistent beer quality in every batch. – Simple operation and maintenance. – Continuous and automated operation. Furthermore, the typical costs of the beer clarification process are in the same cost range as kieselguhr filtration and total up to around €0.45/hl based on the production of 100 hl/h, including the depreciation of investment, operating costs and membrane replacement costs [91]. In 2005, Pall (USA) [90] presented a system called PROFI that combines centrifuge and membrane technology by which yeast and other particles can be separated. It is designed to effectively operate in a continuous production system 24 h/day, 7 days/ week without the need for either filter aids, or for operator or workflow interruptions. Table 8.12: Beer quality after kieselguhr and after crossflow filtration

Original extract (%) Apparent extract (%) Real extract (%) Alcohol (%) Degree of fermentation app. (%) Color (EBC) V-max (filterability) (ml) Viscosity (MPas) Bitter units (BE) Total nitrogen 12% (mg/l) Total polyphenols (mg/l) Anthocyanogens (mg/l) β-Glucan (mg/l) Foam stability R&C (mg/l) Turbidity 0°C (EBC)

Unfiltered beer

After kieselguhr

After crossflow

11.40 2.221 3.93 3.84 80.60 7.20 – 1.62 27.80 870.20 113.40 36.70 242.30 – 32.00

11.37 2.21 3.91 3.83 80.60 6.70 1,830.00 1.57 27.20 868.90 110.70 36.20 228.00 112.00 0.53

11.39 2.21 3.92 3.84 80.60 7.00 6,940.00 1.56 27.60 868.10 112.80 36.40 203.80 111.00 0.41

190 

 8 Membrane operations in the sugar and brewing industry

Fillaudeau et al. [89] proposed the use of rotating and vibrating filtration for clarification of rough beer at pilot scale under severe and controlled operating conditions: 0–4°C, CO2 pressure of 1000 and 1500 mbar, in order to satisfy product quality and industrial process requirements. The system includes a flat disc of stainless membranes (coarse membranes) with a ceramic selective layer (TiO2, ZrO2 or SiO2) and a pore diameter from 0.60 to 4 μm. The results obtained constitute a significant improvement in comparison with conventional techniques reported in the literature. The selected membrane may satisfy the quality and flux required by the brewing industry validated for VCF higher than 10. It was observed that the major contribution of fouling was the yeast cells but it was not critical during the concentration phase. Even though different applications of membrane technology in the beer clarification have been reported in the literature, efforts have been very limited until now. At the moment, beer clarification via membrane processes has only 10% of the market volume [92].

8.4.1.3 Dealcoholization of beer The production of beers with low-alcohol content had different historical reasons in the past century. For example, during the World Wars (1914–1918 and 1939–1945) it was the shortage of raw materials that lead to the production of beers with low original extract (often with a high proportion of adjuncts) and thus of low-alcohol content. Conversely, in the years between 1919 and 1933 it was the prohibition to manufacture, sell and consume alcohol that increased the production of low-alcohol beer (LAB) in the USA [93, 94]. In the late 20th century, efforts of breweries to expand the assortment of products with beers with low-alcohol content was motivated mainly by the wish to increase the overall production by bringing out new products and also provide beer consumers with an alternative prior to or during different activities such as driving motor vehicles or operating machinery, sports, pregnancy and medication. In addition, there is the desire to penetrate beverage markets in countries where alcohol consumption is forbidden for religious reasons [95]. In the actuality, the markets are growing constantly. Spain is the largest consumer of low-alcohol beer in the EU at 9.5%, while Germany was the largest European beer market in 2010 [95]. In most EU countries, beers with low-alcohol content are divided into alcohol-free beers (AFBs) containing   ≤  0.5% alcohol by volume (ABV), LABs with no more than 1.2% ABV. In the USA, alcohol-free beer means that there is no alcohol present, while the upper limit of 0.5% ABV corresponds to so-called non-alcoholic beer or “near-beer” [96]. In countries that enforce religious prohibition, the alcohol content in beverages must not exceed 0.05% by volume [95]. The strategies to ethanol removal can be divided into two main groups: physical and biological processes. The physical methods are based on gentle removal of alcohol from regular beer by means of thermal process (rectification, evaporation) or membrane processes such as dialysis and RO [95]. At industrial scale, systems such

8.4 The brewing industry 

 191

as vacuum distillation (rectification) and vacuum evaporators (single or multi-stage) have been implemented successfully. Even so, a great loss of beer flavor and liveliness can occur during thermal processes [95]. In this context, membrane technology appear as a great alternative because it can be operated automatically at lower temperatures, which is essential when sensitive aroma compounds are intended to be separated [97–109]. The first successful application of crossflow membrane filtration in beer production was introduced by DDS Filtration (now Alfa Laval Nakskov, Denmark) in the 1970s. In this process, typically thin-film composite RO membranes in SW configuration are used for the alcohol reduction, achieving a low-alcohol beer with similar quality to standard beer, low-energy consumption compared to evaporation, no beer heating, no oxygen pick-up and long membrane lifetime. In practice, the RO is carried out in the so-called diafiltration mode. In the first phase the beer concentration is carried out at 7–8°C. During this step, a permeate stream containing mainly water and alcohol is separated from a concentrated beer stream, which contains most of the flavors. Subsequently during the diafiltration phase, the permeate removed from beer is quantitatively replaced by demineralized water that has to be sterile, completely demineralized (conductivity < 50 μS) and deaerated (oxygen content < 0.1 ppm). This continues until a desired alcohol concentration is reached in the beer. The last step in the dealcoholization is the taste adjustment by adding syrups and hops according to the brew master’s taste. Figure 8.9 shows the dealcoholization process for a batch of 690 hl using a three-loop plant with 4″ SW membranes with cellulose acetate membranes. The quality of beer dealcoholized by RO is summarized in Table 8.13. It should be noted that the alcohol-rich permeate is not necessarily a waste product but can be used in the production of other alcoholic drinks such as the currently popular “alcopops”. Another membrane alternative for reducing the alcohol content in beer is dialysis. By using standard dialysis membranes the alcohol content of the beer can be reduced by 8 to 10 times similar to RO. Dialysis is typically operated in counter current flow to maximize the DF between the beer on one side of the membrane and the alcoholfree dialysate on the other side of the membrane. In order to maintain a high DF and recycle the dialysate, a steam stripping column is often integrated into the process for the continuous alcohol removal. This increases the complexity of the process and makes it less favorable compared to the straightforward RO. Furthermore, in order to reduce CO2 losses from the beer, some CO2 is added to the dialysate and the operating pressure on the beer side is kept close to the saturation pressure of CO2 [95]. Generally, the currently available non-alcoholic beers present a poor flavor profile compared with the alcoholic beers. One of the most potential processes for aroma recovery is pervaporation (PV), where the recovered aromas can be incorporated into non-alcoholic beers. PV membranes are very selective for several chemical groups that constitute typical beverage aroma profiles [100–103]. PV has also been applied for ethanol removal and aroma recovery from alcoholic beverages [104–107]. In 2008, the

Number of loops: Membrane: Module: Number of elements: Membrane area:

Key plant data:

Permeate 2450 hl/day

Number of loops: Membrane: Module: Number of elements: Membrane area:

Key plant data:

Figure 8.9: Dealcoholization of beer by reverse osmosis/diafiltration process

Feed pressure: 35–40 bar

Batch tank 690 hl

Defiltration water 2450 hl/day

Feed pressure: 35–40 bar

Batch tank (start) 1040 hl/day

Permeate 350 hl/day

3 CA995-PE 4˝ spiral 42 1480 m2

3 CA995-PE 4˝ spiral 42 1480 m2

Batch tank 690 hl

Diafiltration Water + taste adjustment

192   8 Membrane operations in the sugar and brewing industry

8.4 The brewing industry 

 193

Table 8.13: Beer quality after dealcoholization by reverse osmosis (RO) Alcohol content (%) pH Turbidity (EBC) Acetaldehyde (mg/l) DMS (mg/l) Ethylacetate (mg/l) n-propanol (mg/l) i-butanol (mg/l) i-amylacetate (mg/l) i-amylalcohol (mg/l) Ethylcaproate (mg/l) Diacetly (mg/l) 2,3 pentanedione (mg/l)

0.56 4.4 0.57 1 0.013 1 2 5 0.1 19 0.00 0.009 0.009

integrated process was patented that considers the extraction of beverage aroma compounds before the dealcoholization and subsequent addition to the treated beverage [108]. Catarino et al. [109] reported the extraction of beer aroma compounds by PV using response surface methodology (RSM) to evaluate the effect of operating conditions in the response variable (flux and selectivity). Previously, Catarino and Mendes [110] demonstrated the extraction of aroma compounds from beer at industrial scale using plate-and-frame POMS/PEI composite membranes with a total effective area of 40 m2, operating between 1 mbar and 8 mbar of permeate pressure and between −75°C and −85°C of condensing temperature. It was observed that the operating conditions affect the permeate pressure and condensation temperature. The optimal operating conditions to achieve maximum permeate delivery and a good equilibrium of aroma profile were 25°C of feed temperature and 500 l/h of feed-flow rate. Most recently, del Olmo et al. [111] tested hydrophobic PDMS membranes for the initial recovery of the aromas, which were then added to an alcohol-free beer resulting in beer with good taste.

8.4.1.4 Beer from tank bottoms After fermentation, yeast settles at the bottom of the fermentation vessels. This precipitate represents 1.5–2% of the total beer volume and contains – in addition to the yeast – a high proportion of beer, which is lost if not recovered. Membrane technology appears to be a novel and attractive application in order to recover the beer and concentrate the yeast up to 20% DM. There are different possibilities for running the process for the recovery of beer from spent yeast, such as continuous process, semi-batch and full batch process. In 2002, Pall presented the Keraflux system, a full batch process using ceramic membranes from Membralox (Pall Exekia, France). During filtration the surplus yeast is taken from a batch tank, pumped through the membrane modules and is directly recirculated in the tank. The system was designed for breweries between 0.5 and 1.3 m hl/year [112].

194 

 8 Membrane operations in the sugar and brewing industry

Recovered beer tank

Yeast tank

Tank bottoms

Beer recovery from tank bottoms by MF Concentreated beer and yeast tank

Figure 8.10: New concept: beer recovery from tank bottoms

Table 8.14: Beer quality before and after recovered beer addition

Original gravity (°P) Alcohol % w/w Extract (°P) Color (EBC) pH Bitterness (EBU) Haze (EBC) Foam retention (s) FAN (free amino acids) (ppm) Anthocyanogen (ppm) Polyphenols (ppm)

Reference beer

Reference beer plus 10% recovered beer

10.85 3.58 3.87 6.00 3.98 20 0.50 76 13 26 45

10.54 3.58 3.54 5.90 4.21 22 0.60 72 25 26 60

Figure 8.10 shows a continuous membrane process, which separates the beer from the yeast using a PVDF MF membrane in a plate-and-frame module. The key advantages of this process include a recovered beer of high-quality (no oxygen pick-up, virtually sterile beer) as shown in Table 8.14. Other benefits are low membrane replacement costs and continuous operation that involves less yeast autolyzis and a reduction in the number of tanks. An important point to consider is the associated costs, including the initial investment and operating of the beer recovery plant. In this case the costs are balanced by the beer recovered from the yeast, for example, for a brewery with an annual production of 2 million hl, the recovered beer amounts to 24,000 hl or about 1% of the annual production.

8.5 Conclusions and outlook The membrane market in the sugar and brewing industries will grow with increasing acceptance of membrane processes in the food industry, which is estimated to be currently around €1 billion with average annual growth rates of 5–8%. Key drivers for the

8.6 References 

 195

growth of the market for the established membrane processes MF, UF, NF and RO will be procsss intensification environmental targets like reduction in CO2 emission and water consumption. In the area of PI, the development of integrated process solutions such as synergies and hybrid processes still offers many opportunities, such as the combination of evaporators and NF/RO for the concentration of thin juice in the sugar industry or of high speed separators and MF for beer clarification. While in the area of environmental targets there is great future potential for recycling of water and valuable products/by-products within processes as well as combining membrane bioreactors (MBRs) for wastewater treatment with RO to close the water loop in factories. Additionally, the development of the emerging membrane technologies such as membrane contactors and PV will stimulate further growth of membrane technology in the sugar and brewing industries. In the sugar industry, the use of MD and OD for the concentration of thin and clarified cane juice are interesting alternatives [36–42]. Furthermore, in the brewing industry the use of MCs for the CO2 removal from beer to obtain a dense foam head, for the oxygen removal to preserve the beer and for production of deoxygenized water for dilution of high gravity brewed beer are under development [113], as well as the use of PV for recovery of flavors from the original beer with alcohol for incorporation into non-alcohol beer [111]. Furthermore, the new concept of biorefineries for the simultaneous production of food, biofuels and biochemical to thus optimize the utilization of the feedstock are already partly implemented in the sugar industry but might have also potential for the brewing industry. Membrane processes as low-energy processes with a high selectivity can play an important part in the development of this new concept. Overall, crossflow membrane processes are established in the sugar and brewing industries and supported by new development and trends, and it can be predicted that the importance of membranes for these two industries will increase in the future.

8.5.1 Acknowledgements The authors would like to thank Hanne Jonck for proof reading the manuscript.

8.6 References 1. Loeb S, Sourirajan S. Seawater demineralization by means of a semipermeable membrane. In: Gould R., ed. Advances in Chemistry ACS Series No. 38. American Chemical Society: Washington D.C., USA; 1962:117–132. 2. Galloway JH. The Sugar Cane Industry: An Historical Geography from its Origins to 1914. Cambridge University Press: Cambridge, UK; 2005. 3. International Sugar Organization. World sugar balances 1997/98–2010/11. November 2010. 4. Contreras AM, Rosa E, Pérez M, van Langenhove H, Dewulf J. Comparative life cycle assessment of four alternatives for using by-products of cane sugar production. J Clean Prod 2009;17:772–779.

196 

 8 Membrane operations in the sugar and brewing industry

5. Renouf MA, Wegener MK, Nielsen LK. An environmental life cycle assessment comparing Australian sugarcane with US corn and UK sugar beet as producers of sugars for fermentation. Biomass and Bioenerg 2008;32:1144–1155. 6. Chauhan MK, Chaudhary VS, Samar SK. Life cycle assessment of sugar industry: A review. Renew Sust Energ Rev 2011;15:3445–3453. 7. Gul S, Harasek M. Energy saving in sugar manufacturing through the integration of environmental friendly new membrane processes for thin juice pre-concentration. Appl Ther Eng 2012;43:128–133. 8. Hinkova A, Budnik Z, Kadlec P, Pridal J. Potential of separation membranes in the sugar industry. Sep Purif Technol 2002;26:101–110. 9. Bogliolo M, Bottino A, Capannelli G, De Petro M, Servida A, Pezzi G, Vallini G. Clean water recycle in sugar extraction process: Performance analysis of reverse osmosis in the treatment of sugar beet press water. Desalination 1996;108:261–271. 10. Lipnizki F, Carter M, Trägårdh G. Applications of membrane processes in the beet and cane sugar industry. Sugar Industry/Zuckerindustrie 2006;131:29–38. 11. Hatziantoniou D, Howell JA. Influence of the properties and characteristics of sugar-beet pulp extract on its fouling and rejection behaviour during membrane filtration. Desalination 2002;148:67–72. 12. Djurid M, Gyura J, Zavargo Z. The analysis of process variables influencing some characteristics of permeate from ultra- and nanofiltration in sugar beet processing. Desalination 2004;169: 167–183. 13. Coca M, Garcia MT, Gonzalez G, Peña M, Garcia JA. Study of coloured components formed in sugar beet processing. Food Chem 2004;86:421–433. 14. Madsen RF. Process for the Purification and Clarification of Sugar Juice. British Patent No. 1.361;674:1971. 15. Nielsen WK, Kristensen S, Madsen RF. Prospects and possibilities in application of membrane filtration systems within beet and cane sugar industry. Sugar Technol Rev 1982;9:59–117. 16. Hatissens TR, van Nispen JGM, Koerts K, Nie LH. Ultrafiltration as an alternative for raw juice purification in the beet sugar industry. Int Sugar J 1984;86:227–229. 17. Attridge L, Eringis A, Jaferey I. Ultrafiltration of beet diffusion juice using spiral and tubular polymeric membranes. In: Biennial Meeting 2001 of the American Society of Sugar Beet Technologists (ASSBT). Vancouver, Canada, 28 February – 3 March, 2001. 18. Tyndall TJ. Recent developments in sugar clarification with tubular polymeric membranes. In: Symposium on advanced technology for raw sugar and cane and beet sugar refined sugar production. New Orleans, USA, 8–10 September, 1999. 19. Willet CC. The A.B.C. process for the production of refined sugar from beets. Int Sugar J 1997;99:7–8. 20. Lutin F, Bailly M, Barb D. Process improvements with innovative technologies in the starch and sugar industries. Desalination 2002;148:121–124. 21. Schrevel G. Separation of macromolecules in beet sugar processing. Zuckerindustrie 2002;127:768–769. 22. Tebble I, Nelson R, Sargent D. The use of membrane filtration to reduce lime usage in purification. Zuckerindustrie 2002;127:600–603. 23. Mak FK. Removal of colour impurities in raw sugar by ultrafiltration. Int Sugar J 1991;93:263–265. 24. Wadley S, Brouckaert CJ, Baddock LAD, Buckley CA. Modelling of nanofiltration applied to the recovery of salt from waste brine at a sugar decolorization plant. J Memb Sci 1985;102:163–175. 25. Cartier S, Theoleyre MA, Decloux M. Treatment of sugar decolorizing regeneration waste using nanofiltration. Desalination 1997;113:7–17.

8.6 References 

 197

26. Gula F, Lancrenon X. The use of ion-exchange resins for beet sugar juice processing. In: Biennial Meeting 2007 of the American Society of Sugar Beet Technologists (ASSBT). Salt Lake City, Utah, USA, 28 February 28–3 March, 2007. 27. L’Hermine GJA, Lundquist EG. Decolorization of Sugar Syrups using Functionalized Adsorbents. US Patent No. 5.972.121, 1998 28. McGravey FX. An evaluation of a multiple bed deionization process for beet sugar recovery. J Sugar Beet Res 1967;13:252–280. 29. Schoenrock KWR. The application and utility of magnesium exchange in the beet sugar process. J Am Soc Sugar Beet 1971;16:299–312. 30. Asadi M. Beet-sugar Handbook. John Wiley & Sons, Inc.: Hoboken, NJ; 2007, 99–466. 31. Schneider HG, Mikule J. Recovery of sugar from beet molasses by the P & L exclusion process. Int Sugar J 1975;77:259–264. 32. Lutin F. Electrodialysis in the sugar industry as a purification technology. In: SPRI Conference. Porto, Portugal, 9.–12. April, 2000. 33. Elmidaoui A, Lutin F, Chay L, Taky M, Tahaikt M, My Alaoui Hafidi R. Removal of melassigenic ions for beet sugar syrups by electrodialysis using a new anion-exchange membrane. Desalination 2002;148:143–148. 34. Madaeni SS, Zereshki S. Reverse osmosis alternative: Energy implication for sugar industry. Chem Eng Process 2008;47:1075–1080. 35. Bichsel SE, Sandre AM. Application of membrane technology to juice concentration. Int Sugar J 1982;84:266–268. 36. Nene S, Kaur S, Sumod K, Joshi B, Raghavarao KSMS. Membrane distillation for the concentration of raw cane-sugar syrup and membrane clarified sugarcane juice. Desalination 2002;147:157–160. 37. Bui VA, Nguyen MH, Muller J. A laboratory study on glucose concentration by osmotic distillation in hollow fibre module. J Food Eng 2004;63:237–245. 38. Karode SK, Kulkarni SS, Ghorpade MS. Osmotic dehydration coupled reverse osmosis concentration: steady-state model and assessment. J Memb Sci 2000;164:277–288. 39. Hogan PA, Canning RP, Peterson PA, Johnson RA, Michaels AS. A new option: osmotic distillation. Chem Eng Proc 1998;94:49–61. 40. Beaudry EG, Herron JR, Jochums CE, Medina LE. Apparatus for direct osmotic concentration of fruit juices. US Patent No. 5.281.430, 1994. 41. Versari A, Ferrarini R, Tornielli GB, Parpinello GP, Gostoli C, Celotti E. Treatment of grape juice by osmotic evaporation. J Food Sci 2004;69:E422–E427. 42. Courel M, Dornier M, Herry JM, Rios GM, Reynes M. Effect of operating conditions on water transport during the concentration of sucrose solutions by osmotic distillation. J Memb Sci 2000;170:281–289. 43. Donovan M, Jansen RP, Reisig RC, Hlavacek M, Walker G, Williams JC. Sugar-beet membrane filtration process. US Patent No. 6.440.222, 2002. 44. Seres Z, Gyura J, Eszterle M, Vatai G. Coloured matter removal from sugar-beet industry syrup by ultra- and nanofiltration. Acta Alimentaria 2004;33:119–127. 45. Saska M, de Lataillade J. Concentration and decolorization of dilute products from cane molasses desugarization with reverse osmosis and nanoflitration membranes. In: Sugar Industry Technologists Meeting. Honolulu, Hawaii, USA, 8–11, May, 1994. 46. Regiec P. Application of nanofiltration for condensing model sugar juices. Listy Cukrov 2004;120:132–134. 47. Koekoek PJW, van Nispen J, Vermeulen DP. Nanofiltration of thin juice for improvement of juice purification. Zuckerindustrie 1998;123:122–127.

198 

 8 Membrane operations in the sugar and brewing industry

48. Sandre AM. Technical feasibility of reverse osmosis in the concentration of thin juice. In: Biennial Meeting 1981 of the American Society of Sugar Beet Technologists (ASSBT). San Diego, CA, 22–26. February, 1981. 49. Balakrishnan M, Dua M, Bhagat JJ. Evaluation of ultrafiltration for juice purification in plantation white sugar manufacture. Int Sugar J 2000;102:21–24. 50. Madsen RF. Application of ultrafiltration and reverse osmosis to cane juice. Int Sugar J 1973; 75:163–167. 51. Kishihara S, Fujii S, Komoto M. Ultrafiltration of cane juice: influence of flux and quality of permeate. Int Sugar J 1981;83:35–39. 52. Kishihara S, Fujii S, Komoto M. Improvement of flux in ultrafiltration of cane juice. Int Sugar J 1983;85:99–102. 53. Kishihara S, Tamaki H, Fujii S, Komoto M. Clarification of technical sugar solutions through a dynamic membrane formed on a porous ceramic tube. J Memb Sci 1989;41:103–114. 54. Verma SK, Srikanth R, Das SK, Venkidachalam G. An efficient and novel approach for clarification of sugarcane juice by micro- and ultrafiltration methods. Indian J Chem Technol 1996;3:136–139. 55. Kwok RJ. Production of super VLC raw sugar in Hawaii. Experience with the New NAP Ultrafiltration/Softening Process. In: 55th Annual Sugar Industry Technologists Meeting. Durban, South Africa, 26–29 May, 1996. 56. Cartier S, Theoleyre M, Lancrenon X, Decloux M. Membrane technology in the sugar industry. In: Sugar Processing Research Institute Workshop on Separation Processes in the Sugar Industry. New Orleans, Louisiana, USA, 14–17 April, 1996. 5 7. Saska M, McArdle J, Eringis A. Filtration of clarified cane juice using spiral polymeric membrane configuration. In: International Society of Sugarcane Technologists, XXIII Congress. New Delhi, India, 22–26 February, 1999. 58. Steindl RJ, Doyle CD. Applications and benefits of membrane filtration for the Australian sugar industry. Proc Aust Soc Sugar Cane Technol 1999;21:406–411. 59. Kwok RJ. Production of super VLC raw sugar in Hawaii. Experience with the new NAP ultrafiltration/softening process. Int Sugar J 1996;98:490–496. 60. Lancrencon X. Membrane crossflow filtration for the production of refined sugar at a mill and in a refinery. Int Sugar J 2004;105:390–393. 61. Chou CC. White and refined sugar production from cane sugar factories. In: First Biennial World Conference on Recent Developments in Sugar Technologies. Delray Beach, Florida, USA, 16–17 May, 2002. 62. Monclin JP, Willett SC. The A.B.C. process for the production of refined sugar from cane mixed juice. In: Sugar Processing Research Institute Workshop on Separation Processes in the Sugar Industry. New Orleans, Louisiana, USA, 14–17 April, 1996. 63. Wittwer S. Scepter® Stainless Steel Membranes by Graver Technologies. In: Practical Membrane Technology Short Course. Las Vegas, Nevada, USA, 11–12 July, 2004. 64. Saska M, McArdle J, Eringis A. Filtration of clarified cane juice using spiral polymeric membrane configuration. Indian Sugar 1999, April, 7–13. 65. McArdle J, Eringis A, Eaton B. Filtration of clarified cane juice using spiral polymeric membrane configuration. In: Symposium on Advanced Technology for Raw Sugar and Cane and Beet Sugar Refined Sugar Production. New Orleans, Louisiana, USA, 8–10 September, 1999. 66. Martoyo T, Hino M, Nagase H, Bachtiar A. Pilot test on ultrafiltration of cane raw juice at the Indonesian plantation white Kedawoeng sugar factory. Zuckerindustrie 2000;125:787–792. 67. Ghosh AM, Balakrishnan M, Dua M, Bhagat JJ. Ultrafiltration of sugarcane juice with spiral-wound modules: On-site pilot trials. J Memb Sci 2000;174:205–216. 68. Ghosh AM, Balakrishnan M. Pilot demonstration of sugarcane juice ultrafiltration in an Indian sugar factory. J Food Eng 2003;58:143–150.

8.6 References 

 199

69. Trägårdh G, Gekas V. Membrane technology in the sugar industry. Desalination 1988;69:9–17. 70. Cartier S, Théoleyre MA, Decloux M. Combination of flocculation and crossflow filtration for removing colored impurities in raw cane sugar solutions. In: Conference on Sugar Processing Research. New Orleans, Louisiana, USA, April 14–17, 1996. 71 . Smith RN, Lacey RE. Electromembrane processes for the food industry. In: Scientific Conference, Corn Refiners Association. Lincolnshire, IL, 16–18 June, 1982. 72. Thampy SK, Narayanan PK, Trivedi GS, Gohil DK, Indusekhar VR. Demineralisation of sugar cane juice by electrodialysis. Int Sugar J 1999;101:355–356. 73 . Dornier M, Petermann R, Decloux M. Influence of start-up procedure on crossflow microfiltration of raw cane sugar. J Food Eng 1995;24:213–224. 74. Cartier S, Tatoud L, Théoleyre MA, Decloux M. Sugar refining process by coupling flocculation and crossflow filtration. J Food Eng 1997;32:155–166. 75. Decloux M, Tatoud L. Importance of the control mode in ultrafiltration: Case of raw cane sugar remelt. J Food Eng 2000;44:119–126. 76. Hamachi M, Gupta BB, Ben Aim R. Ultrafiltration: A means for decolorization of cane sugar solution. Sep Purif Technol 2003;30:229–239. 7 7. Karode SK, Gupta BB, Courtois T. Ultrafiltration of raw Indian sugar solution using polymeric and mineral membranes. Sep Sci Technol 2000;35:2473–2483. 78. Lancrenon X, Herve D, Rousset F. A new generation of cane sugar refineries? Int Sugar J 1998;100:490–498. 79. Food and Agriculture Organization of the United Nations. Agribusiness Handbook: Barley, Malt and Beer, 2009. 80. Organization for Economic Co-operation and Development. OECD Health Data 2005. 81. Moll M. Bières et Coolers – Définition, Fabrication, Composition. Tech & Doc: Paris, France, 1991, 15–263. 82. Olajire AA. The brewing industry and environmental challenges. J Cleaner Prod 2012, doi:10.1016/j.jclepro.2012.03.003. 83. Hackensellner T. Efficient Energy Use in the Brewhouse. The Huppman Group: Kitzingen, Germany, 2000. 84. Daoud IS. Separation of Wort from Mash. US Patent No. 4.844.932, 1989. 85. Versteegh CW. Process for Membrane Filtration of Mash to Produce Wort. US Patent No. 5.453.285, 1995. 86. Daufin G, Escudier JP, Carrère H, Bérot S, Fillaudeau L, Mecloux M. Recent and emerging applications of membrane processes in the food and dairy industry. Food Bioprod Process 2001;79:89–102. 87. Bühler TM, Burrell K, Eggars HU, Reed RJR. The application of membranes for new approaches to brewery operations. In: European Brewery Convention: Proceedings of the 24th Congress. Oslo, Norway, 1993; 691–700. 88. Lotz M, Randhahn H, Berndt R, Muller G. Mashing Process. US Patent No. 6.916.425, 2005. 89. Fillaudeau L, Boissier B, Moreau A, Blanpain-avet P, Ermolaev S, Jitariouk N, Gourdon A. Investigation of rotating and vibrating filtration for clarification of rough beer. J Food Eng 2007;80:206–217. 90. Atkinson S. Beer clarification. Membrane Technology 2005;7:8–9. 91. Borremans E, Modrok A. Membrane Filtration technology in the brewing industry. Drink Technol Marketing 2003, 7. 92. van der Sman RGM, Vollebreg HM, Mepschen A, Noordman TR. Review of hypotheses for fouling during beer clarification using membranes. J Memb Sci 2012;396:22–31. 93. Meussdoerffer FG. A comprehensive history of beer brewing. In: Esslinger HM, ed. Handbook of Brewing. Wiley-VCH Verlag: Weinheim, Germany; 2009; 1–42.

200 

 8 Membrane operations in the sugar and brewing industry

94. Silva DP, Branyik T, Teixeira JA, Almeida E, Silva JB. Alcohol-free beer. In: Venturini Filho WG, ed. Bebidas Alcoolicas: Ciencia e Tecnologia, Vol.1. Edgard Blucher: Sao Paulo, Brazil; 2010; 69–83. 95. Branyik T, Silva DP, Baszczynski M, Lehnert R, Almeida e Silva JB. A review of methods of low alcohol and alcohol-free beer production. J Food Eng 2012;108:493–506. 96. Montanari L, Marconi O, Mayer H, Fantozzi P. Production of alcohol-free beer. In: Preedy VR, ed. Beer in Health and Disease Prevention. Elsevier Inc.: Burlington, MA, 2009; 61–75. 97. Pereira CC, Rufino JRM, Habert AC, Nobrega R, Cabral LMC, Borges CP. Aroma compounds recovery of tropical fruit juice by pervaporation: Membrane material selection and process evaluation. J Food Eng 2005;66:77–87. 98. Bluemke W, Schrader J. Integrated bioprocess for enhanced production of natural flavors and fragrances by Ceratocystis moniliformis. Biomol Eng 2001;17:137–142. 99. Raisi A, Aroujalian A, Kaghazchi T. Multicomponent pervaporation process for volatile aroma compounds recovery from pomegranate juice. J Memb Sci 2008;322:339–348. 100. Shepherd A, Habert AC, Borges CP. Hollow fibre modules for orange juice aroma recovery using pervaporation. Desalination 2002;148:111–114. 101. Sampranpiboon P, Jiraratananon R, Uttapap D, Feng X, Huang RYM. Separation of aroma compounds from aqueous solutions by pervaporation using polyoctylmethyl siloxane (POMS) and polydimethyl siloxane (PDMS) membranes. J Memb Sci 2000;174:55–65. 102. Baudot A, Souchon I, Marin M. Total permeate pressure influence on the selectivity of the pervaporation of aroma compounds. J Memb Sci 1999;158:167–185. 103. Dobrak A, Figoli A, Chovau S, Galiano F, Simone S, Vankelecom IFJ, Drioli E, van der Bruggen B. Performance of PDMS membranes in pervaporation: Effect of silicalite fillers and comparison with SBS membranes. J Colloid Interface Sci 2010;346:254–264. 104. Verhoef A, Figoli A, Leen B, Bettens B, Drioli E, van der Bruggen B. Performance of a nanofiltration membrane for removal of ethanol from aqueous solutions by pervaporation. Sep Purif Technol 2008;60:54–63. 105. Takacs L, Vatai G, Korany K. Production of alcohol free wine by pervaporation. J Food Eng 2007;78:118–125. 106. Karlsson HOE, Trägårdh G. Applications of pervaporation in food processing. Trends Food Sci Technol 1996;7:78–83. 107. Brazinha C, Crespo JG. Aroma recovery from hydro alcoholic solutions by organophilic pervaporation: modelling of fractionation by condensation. J Memb Sci 2009;341:109–121. 108. Mendes A, Madeira LM, Catarino M. Process for enriching the aroma profile of a dealcoholized beverage. WO Patent No. WO2008099325, 2008. 109. Catarino M, Ferreira A, Mendes A. Study and optimization of aroma recovery from beer by pervaporation. J Memb Sci 2009;341:51–59. 110. Catarino M, Mendes A. Non-alcoholic beer – A new industrial process. Sep Purif Technol 2011;79:342–351. 111. del Olmo A, Blanco CA, Palacio L, Prádanos P, Hernández A. Setting Up of a Method of Pervaporation for Improving Alcohol-Free Beer. Euromembrane: London, UK; 23–27 September 2012. 112. Bock M, Rögener F. Beer recovery from spent yeast. Brewing and Beverage Industry International 2002;3:8–10. 113. Gableman A, Hwang ST. Hollow fibre membrane contactors. J Memb Sci 1999;159:61–106.

9 Processing of stevioside using membrane-based separation processes Sourav Mondal and Sirshendu De 9.1 Introduction Stevia rebaudiana Bertoni, also known as “sweet herb” is a herbaceous plant that originated in the tropical grasslands and mountain terrains of Paraguay, and has a sweet little secret. The leaves of this plant contain diterpene glycosides, mostly stevioside and rebaudioside, which is one of the most demanding high potency natural sweetener and dietary supplements available at present. Stevioside has been known to be 300 times sweeter than sugar and contains added therapeutic benefits to human health. Some of the unique properties of stevia sweeteners are its zero glycemic index, zero carbohydrate and zero calories [1], compared to other conventional sweeteners. Being the world’s only natural sweetener in this category, it is very popular. The sweet part of the herb is extracted and then blended with other all-natural ingredients to create the delicious and healthy sweetener. It does not induce any carcinogenic or toxicological effect and has been approved for human consumption, worldwide. However, the major challenge to the scientific community is the extraction of steviol glycosides from the leaves in an efficient and economic manner, keeping its nutritional qualities intact, for commercial usage as sugar substitute. Stevia is a new world genus distributed from the South American Andes to southern USA, through Argentina, the Brazilian highlands and Central Mexico [2]. Stevia rebaudiana is a 30–60 cm tall herbaceous plant (shown in Figure 9.1) with perennial rhiozomes, simple, opposite and narrowly elliptic to oblanceolate leaves, trinerved venation, paniculate-corymbose inflorescences with white flowers, and achenes that bear numerous, equally long pappus awns [3]. Stevia rebaudiana Bertoni is one of 154 members of the genus Stevia and one of only two that produce sweet steviol glycosides [3–5]. Stevia was first brought to the attention of Europeans in 1887 [6] when M. S. Bertoni learned of its unique properties from the Paraguayan Indians and Mestizos [7]. Stevioside, the most abundant sweet constituent present in the leaves of S. rebaudiana, was first isolated in impure form in the first decade of 20th century [8, 9], but the final chemical structure was determined 60 years later by Mosettig et al. [10]. The second major sweet diterpene glycoside from S. rebaudiana was identified in 1970 [11]. Furthermore, six less abundant sweet component glycosides were isolated from the species: rebaudioside B–E, dulcoside A and steviolbioside [12–14]. In June 2008, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) concluded that steviol glycosides are safe for use in foods and beverages and

202 

 9 Processing of stevioside using membrane-based separation processes

Figure 9.1: Stevia rebaudiana Bertoni plant

established an acceptable daily intake (ADI) of 4 mg/kg body weight [15, 16]. JECFA established specifications for the identity and purity for steviol glycosides, requesting a minimum content of 95% of the sum of the seven steviol glycosides: stevioside, rebaudioside A, rebaudioside C, dulcoside A, rebausoside, steviolbioside, rebaudioside B [17, 18]. In December 2008, the US FDA stated that it had no objection regarding the conclusion of expert panels that stevia containing a minimum of 95% rebaudioside A is generally recognized as safe (GRAS) for use as a general purpose sweetener in foods and beverages. The Food Standards Australia New Zealand (FSANZ) organization has completed evaluation of use of steviol glycosides in foods in 2008 and recommended the Australia and New Zealand Food Regulation Ministerial Council (Ministerial Council) to allow the use of steviol glycosides in food [19]. In Europe, steviol glycosides have been recently approved for use as a sweetener by the European Food Safety Authority [20]. In Japan, China, Korea, Brazil, Paraguay and several other countries worldwide, steviol glycosides are considered natural food constituents and, as such, are implicitly accepted for food use. Steviol glycosides were first commercialized as a sweetener in 1971 by the Japanese firm Morita Kagaku Kogyo Co., Ltd., a leading stevia extract producer in Japan. Since then, it has been cultivated and manufactured by several companies in different parts of the globe. Currently, China is the largest exporter of stevioside in the world. In Japan, the demand foe stevia for sweetening and flavoring purposes has increased enormously [21]. Cultivation of S. rebaudiana for the Japanese market mainly occurs in the Republic of China, Taiwan, Thailand and some parts of Malaysia [22 ]. Stevioside has also been consumed in Korea since 1995,

9.2 Physical and biological properties of steviol glycosides 

 203

with the majority of its use in the sweetening of the beverage, soju and is supplied by China [21]. Stevioside extract, containing 60% stevioside, and free from steviol and isosteviol is approved for use in ditectic foods, beverages, medicines, soft drinks, etc. in Brazil, where production occurs in the southern province of the country for the local market. Significant proportions of cultivation and stevioside processing is also done in Canada [23], the Czech Republic [24], India [25] and Russia [26].

9.2 Physical and biological properties of steviol glycosides The sweetness potency of stevioside has been rated to be 300 times the relative sweetness intensity of 0.4% sucrose solution. The compound exhibits a slightly menthollike bitter aftertaste [27]. The sweetness intensities (i.e., sweetening power relative to sucrose, which is taken as 1) of the other S. rebaudiana sweet components have been determined as [28]: – dulcoside A: 50–120 – rebaudioside A: 250–450 – rebaudioside B: 300–350 – rebaudioside C (previously known as Dulcoside B): 50–120 – rebaudioside D: 250–450 – rebaudioside E: 150–300 – rteviolbioside: 100–125 Also, stevioside has been found to be synergistic with aspartame, acesulfame-K, and cyclamate, but not with saccharin [27]. The solubility of stevioside in aqueous systems is fairly low. However, the second most abundant component in S. rebaudiana leaves, rebaudioside A, which has a more pleasant taste than stevioside, is six to seven times more soluble in water, as its molecule contains an additional glucose unit [11, 22]. The high water holding capacity of the stevia leaf powder, due to high protein content, enhances the swelling ability, an important function of protein in preparation of viscous foods such as soups, gravies, dough and baked products. Formation and stabilization of the emulsion is aided by the protein content and is critical in many food applications, such as cake, batters, coffee whiteners, milks, frozen desserts and others. This property depends heavily on composition and the stress under which the product is subjected during processing [29]. Fat absorption capacity has been attributed to the physical entrapment of oil. Stevioside also possesses a reasonable fat absorption capacity, which plays an important role in food processing, as fat acts on flavor retainers and increases the mouth-feel of foods. Stevioside is a thermally stable molecule up to 100°C, in the pH range of 3–9. However, it decomposes rapidly at higher alkaline pH levels [30]. Both stevioside and

204 

 9 Processing of stevioside using membrane-based separation processes

rebaudioside A have been found to be stable when formulated in acidulated beverages at room temperature for a minimum of 3 months [31]. Solid stevioside is stable for 1 h at 120°C, and does not undergo browning or caramelization [32], but decomposes when temperatures exceed 140°C [33]. The chemical composition of the stevia leaves were listed in Table 9.1 [29, 34, 35]. Stevia leaf extract exhibits a high degree of antioxidant activity and has been reported to inhibit hydroperoxide formation in sardine oil with potency greater than that of either DL-α-tocopherol or green tea extract. The antioxidant activity of stevia leaf extract has been attributed to the scavenging of free radical electrons and superoxides [36]. A recent study assessing in vitro potential of ethanolic leaf extract of S. rebaudiana indicates that it has a significant potential for use as a natural antioxidant [37]. Stevia is thought to inhibit the growth of certain bacteria and other infectious organisms [38, 39]. It has also been reported in the literature that the bactericidal activity of Escherichia coli (and other food-borne pathogenic bacteria), other microorganisms like Salmonella typhimurium, Bacillus subtilis, and Staphylococcus aureus has also been found to be inhibited by the fermented stevia leaf extract [40–42]. Stevia glycosides possess valuable biological properties. Regular consumption of these compounds decreases the content of sugar, radionuclides, and cholesterol in the blood [43], improves cell regeneration and blood coagulation, suppresses neoplastic growth and strengthens blood vessels [44–47]. It also exhibits choleretic [48], anti-inflammatory [49, 50] and diuretic properties; prevents sulceration in the gastrointestinal tract [48], including antihypertensive [51–53], antihyperglycemic [52, 54, 55]; anti-human rotavirus activities [56, 57], glucose metabolism [56, 58] and renal function [59]. It presents potential applications as antidiarrhoeal therapeutics [60]. In addition, the stevia plant and stevioside are beneficial in the treatment of cancer and as alternatives for saccharose in the treatment of diabetes [54, 55], obesity and hypertension [51, 53, 61–63]. It can also act as an anti-carcinogenic product [56, 64, 65], and as antigingivitis [64].

Table 9.1: Proximate analysis of dry Stevia rebaudiana leaves Component Moisture Protein Fat Crude fiber Ash Carbohydrates Reducing sugar Non-reducing sugar

% (w/w) of dry weight basis 4.2–6.5 6.2–20.42 2.5–5.6 13.6–18.5 8.5–13.1 35.2–52.8 5.6–6.1 9.6–9.9

9.3 Extraction methods of steviol glycosides 

 205

9.3 Extraction methods of steviol glycosides Popular extraction methods of stevioside include ion-exchange, solvent extraction, adsorption and chromatographic separation.

9.3.1 Ion-exchange In ion-exchange, positively and negatively charged impurities are removed from an aqueous stream by oppositely charged resin [66] using Amberlite IR-120 ion-exchange resin followed by crystallization to obtain stevioside crystals. Adduci et  al. [67] reported isolation of stevioside using three main steps consisting of hot water extraction, decoloration by electrolysis followed by ion-exchange. The purity of stevioside attained was 70–80% and the yieldwas apprximately 10%. A mixture of Amberlite IR-120 and Amberlite IRA-401 was used to remove color and to get a clear solution. Amberlite XAD-2 was also used for this purpose [68]. The use of consequent positive and negatively charged resins was also reported [69]. However, ion-exchange needs to be coupled with other unit operations. Payzant et al. [70] proposed a two-step ion-exchange resin for the purification of glycoside from stevia leaves: 95% purity of rebaudioside A was reported. Kumar et al. [71] used a strong cation-exchange resin followed by a weak anion-exchange resin as a final treatment method: the final purity of stevioside was 65%. High purity (beyond 90%) stevioside and rebaudioside A was obtained using ion-exchange resin as a final step [72–74].

9.3.2 Solvent extraction Organic solvents are very common for extraction of phytochemicals from the plants because of their selectivity towards organic components. Water immiscible solvents, methanol, chloroform, n-butanol, ethanol and fatty alcohols have been used for extraction of stevioside from stevia leaves [75–81]. Stevioside and rebaudioside A were further purified by crystallization and recrystallization from the solvent. The use of water and methanol as the extraction medium under pressurized conditions was studied by Pol et al. [82 ]. In a recent study, Chayya et al. [83] used a systematic approach to evaluate the optimum conditions for water extraction of stevioside from stevia leaves using RSM. The optimum conditions of extraction were: temperature 78°C; time of heating 56 min, and a leaf:water ratio 1:14 (w/v). At these conditions, extracted stevioside was 10.5 g per 100 g of dry stevia leaves. It can be noted that the above figures correspond to the Indian variety of the plant.

206 

 9 Processing of stevioside using membrane-based separation processes

9.3.3 Extraction by chelating agents In order to overcome the use of toxic chemicals and slow, specialized and expensive equipment such as ion exchangers or chromatographs, chelating agents are sometimes used so that undesired/desired components form complexes with these chemicals, leading to easier separation. Chelating agents, such as carboxylic acid, citric acid, calcium hydroxide, bentonite, celite, aluminium hydroxide, iron hydroxide, zeolite, pectinase enzyme were used by several researchers to remove organic and inorganic materials as primary clarification [84–91].

9.3.4 Adsorption and chromatographic separation Selective separation of a species from a liquid stream can generally be attained by adsorption. Highly specific components can be selectively separated by chromatographic separation processes. Extremely high purity glycosides were obtained using adsorption in a chromatographic column [92–101]. Clarification of the extract by adsorption on zeolite and activated charcoal was also reported [102, 103].

9.3.5 Ultrasonic extraction A Chinese patent [104] described ultrasonic assistance during extraction of stevia leaves. Stevioside powder 85–98% was produced by adopting various unit operations (flocculation, filtration, absorption, decoloration, concentrating and spray-drying) after ultrasonic assisted extraction.

9.3.6 Microwave-assisted extraction Several advantages associated with microwave-assisted extraction (MAE) make it competitive compared to other conventional processes. These are faster extraction, reduced solvent use and higher recovery. MAE for extraction of stevioside and rebaudioside A was first reported by Jaitak et al. [105]. It was concluded that yield of stevioside (8.64%) and rebaudioside A (2.34%) was higher than with the conventional method (6.54% and 1.20%, respectively). Microwave power of 80 W, 50°C and 1 min duration gave the optimum results. Methanol-water (80:20 v/v) was the medium that responded to microwave assistance to the highest level.

9.3.7 Super critical fluid extraction (SCFE) Using super critical carbon di oxide, extraction of steviol glycosides was also attempted [106]. Bitter components were removed more easily. Pasquel et al. [107] proposed

9.4 State-of-the-art membrane-based processes 

 207

a two-step process: (i) pretreatment of stevia leaves by SCFE with CO2, and (ii) extraction of stevia glycosides by CO2 + water, CO2 + ethanol and CO2 + water + ethanol. The best result was obtained at 16°C and 120 bar with water as co-solvent. Large amount of rebaudioside A was obtained. The quality of the final powder was as good as any conventional method but in terms of yield of rebaudioside A, SCFE fared better.

9.4 State-of-the-art membrane-based processes Membrane-based processes can provide solution to clarification and purification, concentration, even fractionation by a judicious selection of the appropriate membrane. A pretreatment of aqueous extract by microfiltration (MF) can replace centrifugation and primary clarification steps using chelating agents. During microfiltration, chlorophylls, cell debris, most of the high molecular weight proteins, organic matter, etc., are removed. Fine filtration can be replaced by appropriately selected ultrafiltration (UF). In this process, the glycosides permeate through the membrane and the remaining high molecular weight undesired substances are removed. The permeate stream can be concentrated using an appropriately selected nanofiltration (NF) membrane. During this process, water from the glycoside rich solution is extracted, leading to the concentrated solution that can easily be spray-dried or vacuum dried to produce powder. Additionally, the membrane-based systems are rate-governing, have high throughput and offer commercially viable solutions. Having higher throughput, these systems are less time-consuming, and can do away with the complicated operations and expensive equipment like column chromatographs, etc. Removal of various unwanted non-glycosidic materials from the extract leads to clarification and purification of the glycoside containing streams. Fuh and Chiang [108] investigated the possibility of ultrafiltration followed by diafiltration as clarification and purification steps. They also compared the efficiency of UF with that of chelating agents (i.e., the use of inorganic salts). Three types of chelating agents were used. These were: (i) 3% AlCl3 solution adjusted to pH 7.0–7.5 by Ca(OH)2; (ii) saturated Ca(OH)2 solution adjusted to pH to 8–8.5 by bubbling carbon dioxide gas; (iii) 3% Ferric sulphate solution adjusted to pH 7.0–7.5 with NaOH. It was observed that a 100 kDa membrane recovered more stevioside and caused more depigmentation compared to a 25 kDa membrane. Over and above, the permeate flux of 100 kDa membrane was much greater compared to that of the 25 kDa membrane. Recovery of stevioside increased upto 90% for the combined UF and DF method. A fully fledged membrane-based process for extraction of stevioside was reported by Kutowy et al. [109]. The invention had the following steps: 1. Aqueous extraction in a column. 2. Pretreatment by MF. 3. Purification by 2–3 kDa UF membrane. 4. Concentration by NF at an elevated temperature.

208 

 9 Processing of stevioside using membrane-based separation processes

About 78% recovery in stevioside and 80% recovery for rebaudioside A were obtained during MF. In case of UF in diafiltration mode, 1.51 g/l concentration of stevioside was obtained at a permeate flux 35 l/m2h. With an increase in diafiltration volume, recovery of stevioside decreases and at the same time permeate flux increases considerably. Microfiltration of stevia extract using ceramic membrane of various pore sizes were performed by Silva et al. [110]. The best results obtained were 90% yield of stevioside, 95% yield of sebaudioside A, and almost 100% clarification was achieved by using 0.05 μm membrane at 2 bar. Vanneste et al. [111] proposed tailor-made membranes with high selectivity for clarification of stevia extract. They concluded that tailor-made 27% polyethersulfone (PES) and 24% (PES) membrane were better performers than commercial membrane. The molecular weight of stevioside and rebaudioside A was 804 Da and 967 Da, respectively. Thus, the selection of an appropriate NF membrane with cut-off in the range 200–400 Da would be able to retain the glycosidic compounds and allow water and smaller molecular weight impurities to permeate through the membrane, thereby, concentrating the glycoside rich stream. This concentration would reduce load on drying equipment to produce powder. RO can also be used to concentrate the clarified stevia extract but at the expense of energy. Fuh and Chiang [108] used RO for concentration of ultrafiltered clarified extract. They concentrated the extract 10 times and stored it at 5°C for 12 h. It was observed that crystals of stevioside were precipitated and the crystal purity was about 56%. The supernatant of RO concentrate was passed through two mixed bed ion-exchange resins (Amberlite 458 and IRC 50). The purity of stevioside was 66% after first ion-exchange column and it was increased to 90% at the exit of second column and overall recovery was about 80%. The solution was dried and powder was obtained. NF of UF clarified stevia extract was performed using 400 Da membrane at 80°C and 517 kPa in a diafiltration mode by Kutowy et al. [109]. The impurities were reduced by 55% in the retentate and no trace of glycosides was found in the permeate stream. Rao et al. [112] also used NF to concentrate their clarified extract by UF. They used a NF membrane of MWCO between 200–250 Da operated at 1,500 kPa transmembrane pressure. This process removed 80–90% of water as permeate, thus concentrating the product in the retentate.

9.5 Detailed membrane-based clarification processes Powdered dry stevia leaves were used as raw material for preparing stevia extract. Dry stevia leaves powder was obtained from M/s, RAS Agro Associates (Maharashtra, India). The steps of different stages of clarification of stevia leaf extract are outlined in Figure 9.2.

9.5.1 Hot water extraction A hot water extraction method was used for preparing aqueous stevia extract. A specific ratio of leaf:water (weight:volume) was measured and the dry stevia leaves were

9.5 Detailed membrane-based clarification processes 

 209

Crushing and drying of S. rebaudiana leaves

Hot water extraction of dried leaves

Primary clarification by microfiltration

Primary clarification by centrifugation

Ultrafiltration of stevia extract

Concentration by nanofiltration

Enhancement of purity by diafiltration

Figure 9.2: Steps of different stages of extraction, clarification and purification of stevia extract using membrane technology

mixed with hot water. This sample was exposed to a particular temperature for a fixed duration. After the termination of the heating process, the stevia extract was allowed to cool and was then filtered using Whatman filter papers. This extract was analyzed for its stevioside concentration and color. Thermostatic water bath was used to control the temperature ( ± 1°C) of the process. RSM was used in this process to obtain the optimum conditions for maximum stevioside extraction from S. rebaudiana leaves. The independent variables selected for this optimization process were heating temperature, time and S. rebaudiana leaf:water ratio. These variables were represented as T, t and R, respectively, in terms of their actual values. The experimental ranges selected for independent variables were: heating temperature (30–90°C), time (10–120 min) and S. rebaudiana leaf:water ratio in g:ml (1:5–1:20). After selection of independent variables and their ranges, experimental design was applied to generate the experimental combinations for conducting the experiments using a commercial statistical package, Stat-Ease Design Expert 7.0.0 software. A three-variable (five level of each variable) second-order rotatable central composite experimental design [113, 114] and RSM were employed to understand the linear, quadratic, and interaction effects of temperature of water, time of heating and leaf:water ratio on two responses: stevioside concentration and color of the extract. Stevioside extraction was highly significant at probability level p < 0.001 for first order terms of heating temperature, heating time and leaf:water ratio. Among the quadratic terms, only two – heating temperature and ratio – had a significant effect on stevioside extraction at p < 0.05 and p < 0.01, respectively. The contribution of heating temperature and leaf:water ratio interaction term was not significant (p < 0.05) but the other two interaction terms were significant at p < 0.05 for stevioside extraction (Table 9.2). The regression equation of the model showing the net effect of independent parameters on stevioside extraction, in coded level of the parameters, is given in eqn. 9.1: Stevioside (%) = 9.61 + 116*X1 + 0.75*X2 + 0.54*X3 – 0.24*X12 – 0.19*X22 –0.32*X32 – 0.30*X1*X2 – 0.20*X1*X3 + 0.28*X2*X3

(9.1)

210 

 9 Processing of stevioside using membrane-based separation processes

Table 9.2: The regression coefficients of the second-order polynomial model for the response functions (stevioside and color) in coded level Coefficients of the regression model b0 (intercept) b1 (temperature) b2 (time) b3 (ratio) b12 b22 b32 b12 b13 b23 Lack of fit R2 Adjusted R2 CV (%)

F-values

Coefficients values (coded)

Stevioside (%)

Color (A420)

Stevioside (%)

Color (A420)

36.09*** 175.02*** 72.70*** 38.25*** 7.82* 4.73 14.36** 6.85* 3.12 6.07* 1.64 – –

81.53*** 56.21*** 174.99*** 338.0*** 6.76* 0.37 107.69*** 0.79 3.69 39.10*** 3.24 – –

9.61 1.16 0.75 0.54 –0.24 –0.19 –0.32 –0.30 –0.20 0.28

12.53 0.93 1.65 –2.29 –0.32 0.074 1.26 0.14 –0.31 –1.02

0.97 0.94 3.56

0.98 0.97 3.48

*Significant at p < 0.05; **significant at p < 0.01; ***significant at p < 0.001.

The coefficient of determination value of above equation is 0.97. It indicated that the model explained 97% of the variability of the stevioside extracted in the liquid during the extraction process. The F-value for lack of fit of this model was 1.64 with probability level of p < 0.05. The model coefficient of the regression equation explained that the model was highly significant (p < 0.001). Positive linear terms of independent parameters indicated that while increasing these paremeters the stevioside extraction also increases. The quadratic coefficients of both the significant terms of heating temperature and ratio showed negative response. Among interaction terms, a temperature-time combination term had a negative effect and a timeratio combination term had a positive effect on the stevioside extraction process. Figure 9.3 represents a three dimensional plot, demonstrating the variation of stevioside concentration with respect to any two parameters when other parameter is kept constant at center point. From Figure 9.3(A), it is clear that stevioside extraction increases significantly with increase in heating time or temperature while keeping the leaf:water ratio constant at center point. Figure 9.3(B) shows that when time is kept constant at center point, stevioside extraction efficiency increases with increase in leaf:water ratio or temperature with respect to a particular temperature or ratio, respectively. The same trend of stevioside extraction can be observed in Figure 9.3(C), where the operating temperature is fixed at the center point. Stevioside extraction increases with an increase in leaf:water ratio at a particular time. Similarly, concentration of extracted stevioside increases significantly with an increase in time at a particular ratio. At low leaf:water ratio the percentage of stevioside extracted

9.5 Detailed membrane-based clarification processes 

 211

11.5 10

Stevioside (%)

8.5 7 5.5 4

90.00 120.00 101.67

75.00 83.33

60.00 65.00 46.67

45.00 28.33

Time (min)

(a)

10.00 30.00

Temperature (°C)

11.5

10

Stevioside (%)

8.5

7

5.5

4

90.00

(b)

20.00 18.33 16.67 15.00 13.33 11.67 10.00 8.33 6.67 5.00

Ratio (g/ml)

75.00 60.00 45.00 30.00

Temperature (°C)

11.5 10

Stevioside (%)

8.5 7 5.5 4

120.00 101.67 83.33

20.00 18.33 16.67 15.00 13.33 11.67 10.00

65.00 46.67 28.33

8.33

(c)

Ratio (g/ml)

6.67 5.00

10.00

Time (min)

Figure 9.3: Response surface plot for stevioside as a function of (a) time and temperature (leaf to water ratio kept constant at center point), (b) leaf:water ratio and temperature (time kept constant at center point) (c) leaf:water ratio and time (temperature kept constant at center point)

212 

 9 Processing of stevioside using membrane-based separation processes

gets less and as the ratio increases, i.e., as the volume of water increases the percentage of stevioside extraction also increases. The reason for is that when water volume is less it is not able to extract all the stevioside from the leaves but when the volume of water is increased the extraction efficiency increases, but again after a certain limit of leaf:water ratio the further increase in volume of water does not add any significant extraction efficiency. Thus, the recommendations for practical applications are: 1. Water should be used as extracting medium of stevioside for human consumption. 2. The optimum operating conditions for maximum extraction of stevioside from S. rebaudiana leaves (Indian variety) are, temperature 78°C, time of heating 56 min and leaf:water ratio 1:14 (g:ml). At these conditions, the amount of stevioside extracted is 10.45 g per 100 g of dry S. rebaudiana leaves. 3. RSM coupled with numerical optimization undertaken can be useful for maximum extraction of stevioside from any variety of S. rebaudiana leaves in aqueous medium.

9.5.2 Selection of operating conditions and membrane Clarification is a vital unit operation for removal of these suspended particles from the stevia extract. Centrifugation and microfiltration are two popular methods used for clarification of liquid extracts or fruit juices. The advantages of using these two methods are that they are chemical free, and thus can be used safely for clarification process in food industries. Microfiltration was generally used as pretreatment method prior to ultrafiltration in diafiltration mode. Chhaya et  al. [115] analyzed and reported a comparison of the effectiveness of centrifugation and microfiltration as a primary clarification step of stevia extract, and the effects of transmembrane pressure drop as well as the stirring speed in the filtration cell on the permeate flux and permeate concentration of the stevia extract during microfiltration. The centrifugation of stevia extract, after hot water extraction, was performed using a laboratory scale under a batch mode of operation. The operating parameters taken into account were rotation speed (g) and time (min) of operation. Microfiltration membranes with average pore size of 0.2 μm and water permeability of 5.14 × 10–9 m/Pa s were used. Response surface analysis was carried out to determine the optimum centrifugal speed and the relative effect of these primary clarification methods on the overall quality of the clarified extract. The various combinations of experimental conditions and values of all the responses are presented in Table 9.3. From Table 9.3, it is obvious that the linear terms of centrifugation speed and time are highly significant (p < 0.001) and both showed negative response on the concentration of stevioside in the clarified extract. Among quadratic terms, the centrifugation speed is significant at p < 0.05 and centrifugation time is significant at p < 0.01. Both have a positive effect on stevioside concentration during the clarification process. The interaction term is less significant (p < 0.05) compared to terms at linear level and has

9.5 Detailed membrane-based clarification processes 

 213

Table 9.3: Experimental conditions and responses for two variables (at coded level) for centrifugation of stevioside extract Sl. No.

1 2 3 4 5 6 7 8 9 10 11 12 13

Speed (g) R (Z1)

2000 (–1) 9000 (+1) 2000 (–1) 9000 (+1) 550.25 (–1.682) 10449.75 (+1.682) 5500 (0) 5500 (0) 5500 (0) 5500 (0) 5500 (0) 5500 (0) 5500 (0)

Time (min) T (Z2)

15 (–1) 15 (–1) 45 (+1) 45 (+1) 30 (0) 30 (0) 8.79 (–1.682) 51.21 (+1.682) 30 (0) 30 (0) 30 (0) 30 (0) 30 (0)

Responses Color A420

Clarity %T

Total solid g/100 ml

Stevioside %

7.5 7.5 7.5 7.3 7.5 7.3 7.6 7.4 7.3 7.3 7.3 7.3 7.3

3.3 5.0 4.0 7.0 2.7 6.9 3.7 6.6 6.5 6.4 6.0 6.4 6.6

3.1 2.8 2.8 2.6 3.0 2.6 3.0 2.6 2.6 2.7 2.7 2.6 2.6

95.8 90.0 88.5 87.5 93.6 87.1 94.7 87.1 88.7 89.2 88.3 89.4 87.1

R and T represent the actual values of independent parameters, centrifugation speed and time, respectively.

a positive effect. The regression model representing the effect of centrifugation speed and time on the stevioside concentration retained in the clarified stevia extract, in terms of their coded level, is given in eqn. 9.2: Stevioside (%) = 88.54 – 2.01Z1 – 2.5Z2 + 0.89Z 12  + 1.14Z 22  + 1.22Z1Z2

(9.2)

The R2 value is 0.96 for above equation. The stevioside retained in the clarified extract (as compared to feed) varies in 87.1–95.8% of that present in crude extract throughout the experiments. Figure 9.4 represents the percentage variation of stevioside (retained in the clarified stevia extract) as a function of centrifugation speed and time. The graph indicates that the stevioside content decreases with the increase in centrifugation speed and time. The optimization revealed that centrifugation speed (g) 5334 and time (min) 25.62, yield maximum enhancement of the overall extract quality, color (A) 7.4, clarity (%T) 6.0, total solid (g/100 ml) 2.7 and stevioside (% of that in crude extract) 89.5. Primary clarification of stevia extract is also performed using microfiltration membrane in a stirred batch cell. The range of operating conditions varied from 100 to 300 kPa and stirring speed from 500 to 2,500 rpm. Table 9.4 represents various properties of the clarified stevia extract and shows that as the operating pressure increases, the color of the extract decreases and the clarity increases. But, at the same time, recovery of stevioside decreases. At higher operating pressures, although the color decreases by 0.8% and clarity increases by 15%, recovery of stevioside decreases by

214 

 9 Processing of stevioside using membrane-based separation processes

98 96 94

Stevioside (%)

92 90 88 86

2000.00 2875.00 3750.00 4625.00 5500.00 6375.00 7250.00 8125.00 9000.00

Speed (g)

45.00 39.00 33.00 27.00 21.00 15.00

Time (min)

Figure 9.4: Variation of stevioside as a function of centrifugal speed and time

Table 9.4: Property table of stevia extract clarified by microfiltration process

Transmembrane Pressure (kPa) 138

207

276

Stirring speed (rpm) 500 1,500 2,500 500 1,500 2,500 500 1,500 2,500

Color (A420)

Clarity (%T)

7.5 ± 1.2 7.3 ± 1.5 7.8 ± 1.3 7.4 ± 1.4 7.5 ± 1.2 7.2 ± 1.5 7.2 ± 1.2 7.2 ± 1.4 7.5 ± 1.5

5.6 ± 1.2 5.9 ± 1.4 5.5 ± 1.6 6.2 ± 1.5 6.0 ± 1.3 6.5 ± 1.5 6.8 ± 1.3 6.5 ± 1.2 6.6 ± 1.4

Total solid (g/100 ml)

Stevioside (%)

Steady-state permeate flux (l/m2h)

2.7 ± 0.2 2.7 ± 0.5 2.7 ± 0.4 2.7 ± 0.3 2.7 ± 0.5 2.7 ± 0.3 2.7 ± 0.2 2.7 ± 0.4 2.7 ± 0.2

87.4 ± 2.2 89.1 ± 2.5 88.5 ± 2.7 86.7 ± 2.4 85.1 ± 2.3 87.7 ± 2.5 84.8 ± 2.2 84.4 ± 2.6 81.3 ± 2.23

5.2 ± 0.3 8.1 ± 0.1 12.5 ± 0.5 9.0 ± 0.5 11.7 ± 0.6 18.1 ± 0.2 11.3 ± 0.8 20.5 ± 0.1 27.9 ± 0.7

5%. This occurs because, at higher transmembrane pressure drop, more solutes are deposited on the membrane surface, forming a cake-type layer. This layer of solutes acts as a dynamic membrane to retain some more stevioside, thereby reducing the recovery of stevioside in the permeate at higher operating pressures. As it is required to maximize the recovery of stevioside, a lower operating pressure may be selected, for example, 138 kPa. Recovery of stevioside is almost the same at 138 kPa and both 1,500 and 2,500 rpm. But, the permeate flux at 2,500 rpm is around 50% more than that at 1,500 rpm. Thus, 138 kPa pressure and 2,500 rpm can be selected as the preferable operating condition of microfiltration of stevia extract. Two methods, centrifugation and microfiltration are carried out for the clarification of stevia extract. After clarification, both the clarified extracts along with the feed are compared on the basis of their properties. The most suitable operating

9.5 Detailed membrane-based clarification processes 

 215

Table 9.5: Comparison of the properties of the extract clarified by two different clarification methods (centrifugation and microfiltration) Clarified extract

Color (A420)

Clarity (%T)

TS (g/100 ml)

Stevioside (%)

7.4 7.8 11.0

6.0 5.5 0.12

2.7 2.7 3.2

89.5 88.5 13.5 g/l

Centrifugation Microfiltration Crude extract

conditions and the product quality of the clarified extract using both of these methods are presented in Table 9.5, which shows that the color, clarity, total solid and stevioside content in the properties of the clarified extract in both the cases were similar. In fact, performance of centrifugation is marginally better. The choice of a suitable membrane for ultrafiltration of primary clarified stevia extract is key to the entire process. For identification of a suitable membrane, four polymeric ultrafiltration membranes of molecular cut-off 5, 10, 30 and 100 kDa were used. Figure 9.5 shows tevioside in the permeate for different membranes at 414 kPa pressure. As the pore size of the membrane increases (as the MWCO increases), pores get blocked first. There are various mechanisms of pore blocking: partial, standard and intermediate [116]. This leads to an initial sharp decline in permeate flux. Once pores get blocked, the solute particles start depositing over the membrane surface, leading to a gradual flux decline. Therefore, pore blocking is severe for a 100  kDa membrane (having highest pore size) and lower for lower cut-off membranes. Thus, higher cut-off membranes do not necessarily produce higher permeate flux. This phenomenon was also observed by other researchers [117, 118]. The recovery of stevioside in the permeate was the lowest for the 100 kDa membrane, followed by 10, 5 and 30 kDa membranes. For the 100 kDa membrane, due to 60

3.0

Stevioside (%) Permeate flux (l/m2 . h)

30  2 °C 2.5

50

Stevioside yield (%)

2.0 40 1.5 30 1.0 20 0.5

0

10

0

5 kDa

10 kDa

30 kDa

100 kDa

Figure 9.5: Yield of stevioside and steady-state permeate flux at 414 kPa, comparison of the different membranes

216 

 9 Processing of stevioside using membrane-based separation processes

larger pore size, most of the pores are blocked faster and cake formation was severe. Because of this blockage of pores, even smaller sized particles are rejected by the membrane and the rejection of stevioside is more in this case. However, for 5 and 10  kDa membranes, this effect results in almost similar stevioside rejection by the cake layer that behaves like another dynamic membrane. For example, at 414 kPa pressure the yield of stevioside in the permeate was 37% and 20% for 5 and 10 kDa membranes, respectively. Conversely, for the 30 kDa membrane, the yield of stevioside in the permeate was about 50% at this pressure. For lower cut-off membranes such as 5 and 10 kDa, pore blocking effects are less and most of the higher molecular weight solutes are rejected because of the smaller pore size of the membranes and that the dynamic cake layer formed over it rejects some more solutes, resulting in low recovery of stevioside in permeate. Pore blocking and consequent cake layer (dynamic membrane) formation have optimal occurrence in the case of the 30 kDa membrane resulting in higher recovery of stevioside along with high throughput in the studied range of the membranes. Thus, the permeate flux and stevioside in the permeate are maximum for a 30 kDa membrane. Hence, this membrane was selected for clarification of centrifuged stevia extract. Using a 30  kDa MWCO PES membrane, a well planned set of experiments were designed using pretreated hot water extracted centrifuged feed, in order to observe the effects of operating conditions (transmembrane pressure drop and the stirring speed) on the permeate flux and permeate quality. In order to get higher permeate flux it is important to select higher operating pressure and stirring speed. However, the selection of the operating conditions also depends on the permeate quality. The results of the experiments on permeate quality at steady-state are reported in Table 9.6. At lower operating pressures (276 kPa and 414 kPa), recovery of stevioside in the permeate varies between 46 to 50%. The values of color were 0.74–0.87 in this range of operating pressure. There is a marked decrease in recovery of stevioside in permeate accompanied by lower values of color at higher operating pressure (552 and 690 kPa). Stirring does not have significant effect on the properties of the permeate. The purity and selectivity of stevioside in permeate can be estimated by eqns 9.3 and 9.4: purity =

stevioside concentration in permeate concentration of total solids in permeate

(9.3)

stevioside concentration in permeate   concentration of LMW in permeate

(9.4)

selectivity =

Table 9.6 shows that both purity and selectivity of stevioside in the permeate are higher at lower operating pressures. Hence, the operating conditions should be selected such that maximum stevioside in the permeate with a reasonable flux would be obtained. At lower pressures, the stevioside yield is more but the permeate flux is less. Conversely, at higher pressures the flux is more but the stevioside yield is less in the permeate. In the lower pressure range, at both 276 and 414 kPa the stevioside recovered in the permeate (average over three stirrer speeds) is maximum among the

9.5 Detailed membrane-based clarification processes 

 217

Table 9.6: Properties of the permeate with the operating conditions under stirred continuous mode of ultrafiltration using 30 kDa membrane Operating pressure (kPa)

276

Stirring Color speed (A420) (rpm)

600 1,200 1,800 414 600 1,200 1,800 552 600 1,200 1,800 690 600 1,200 1,800 Centrifuged extract (feed) Actual stevia extract

0.87 ± .03 0.77 ± 0.02 0.79 ± 0.02 0.76 ± 0.02 0.74 ± 0.02 0.77 ± 0.02 0.48 ± 0.01 0.44 ± 0.01 0.43 ± 0.01 0.41 ± 0.01 0.37 ± 0.01 0.40 ± 0.01 10.4 ± 0.3 12.3 ± 0.4

Clarity (%T) 86.9 ± 2.6 86.9 ± 2.6 84.5 ± 2.5 85.0 ± 2.6 86.3 ± 2.6 86.6 ± 2.6 91.8 ± 2.8 91.8 ± 2.8 92.6 ± 2.8 89.5 ± 2.7 88.7 ± 2.7 95.0 ± 2.8 1.62 ± 0.05 0.02 ± 0.006

Purity of Selectivity of stevioside stevioside

0.61 0.67 0.61 0.61 0.65 0.60 0.64 0.50 0.52 0.64 0.56 0.65

1.77 1.77 1.75 1.77 1.87 1.71 1.42 1.11 1.04 1.42 1.12 1.30

Stevioside yield (%) 47.6 ± 1.4 47.8 ± 1.4 47.1 ± 1.4 47.8 ± 1.4 50.5 ± 1.8 46.0 ± 1.4 38.3 ± 1.2 30.0 ± 0.9 28.5 ± 8.0 38.2 ± 1.2 30.3 ± 0.9 35.0 ± 1.0

pressure values studied herein: about 44.5%. Fuh and Chiang (1990) [108] reported 45% recovery of stevioside using a 25 kDa membrane. Between 276 and 414 kPa, the permeate flux is higher at higher pressures and stirring speeds. Therefore, 414 kPa pressure at 1,800 rpm (with a flux of 36 l/m2/h and about 44.5% recovery of stevioside) can be selected as a suitable operating condition for continuous stirred ultrafiltration experiments under total recycle mode for clarification of stevia extract.

9.5.3 Crossflow ultrafiltration In crossflow ultrafiltration, the growth of a gel layer is arrested by the convection of the feed, as it flows over the membrane surface, and thus prevents decrease in throughput. Crossflow ultrafiltration experiments of centrifuged stevia extract in both total recycle and batch mode is performed. The steady-state flux values at different operating pressure drop and crossflow rates are presented in Figure 9.6. The properties of the permeate at the steady-state with different operating conditions are presented in Table 9.7. As the operating pressure drop increases, the stevioside recovery in the permeate decreases. The selectivity and purity of stevioside in the permeate are almost independent of flow rate. Both selectivity and purity decrease with transmembrane pressure drop. At higher pressure drop, the cake layer becomes compact (associated with increasing porosity) and it acts as a dynamic membrane. Therefore, this layer retains some of the stevioside and recovery of stevioside in the permeate becomes lower. For example, average (over various crossflow rates)

218 

 9 Processing of stevioside using membrane-based separation processes

30 60 l/h 80 l/h 100 l/h 120 l/h Error bar:  3%

Steady state permeate flux (l/m2. h)

27 24 21 18 15 12 9 6 3 0

0

100

200

300

400

500

600

700

Operating pressure (kPa) Figure 9.6: Variation of steady-state permeate flux with transmembrane pressure drop and crossflow rate

Table 9.7: Various properties of the crossflow ultrafiltered liquor at different operating conditions under total recycle mode of operation Operating Flow rate (l/h) pressure (kPa) 276

414

552

690

Centrifuged extract (feed) Crude extract

Permeate color (A420)

Permeate clarity (% – T)

Purity of stevioside

Selectivity of stevioside

60 80 100 120 60 80 100 120 60 80 100 120 60 80 100 120 _

1.5 ± 13 1.4 ± 0.12 1.1 ± 0.13 1.1 ± 0.15 0.9 ± 0.14 0.9 ± 0.13 0.9 ± 0.12 0.8 ± 0.14 0.8 ± 0.15 0.7 ± 0.16 0.7 ± 0.12 0.6 ± 0.13 0.6 ± 0.14 0.6 ± 012 0.6 ± 0.15 0.4 ± 0.12 10.31 ± 1.13

53.1 ± 1.1 62.5 ± 1.3 69.2 ± 2.3 69.5 ± 1.3 80.9 ± 1.5 81.3 ± 2.1 81.8 ± 2.3 81.5 ± 2.4 83.4 ± 1.5 84.9 ± 2.6 85.5 ± 2.4 87.3 ± 1.8 85.3 ± 1.7 86.1 ± 1.6 86.9 ± 2.2 87.5 ± 2.3 1.3 ± 0.5

0.67 0.61 0.64 0.70 0.55 0.51 0.52 0.58 0.55 0.58 0.63 0.59 0.53 0.43 0.49 0.43

2.0 1.6 1.8 2.4 1.2 1.0 1.1 1.4 1.2 1.4 1.7 1.4 1.1 0.7 1.0 0.8

_

_

_

9.5 Detailed membrane-based clarification processes 

 219

recovery of stevioside at 276 kPa is 56% but 44% at 414 kPa, 40% at 552 kPa and 31% at 690 kPa. This dynamic cake-type layer retains other solids at higher pressure drop, thereby dramatically increasing the clarity of the permeate at higher operating pressures. Clarity is about 87% at 690 kPa and 120 l/h crossflow rate whereas, that in the feed of ultrafiltration is only 1.26%. Thus, the total solids in the permeate also decreases at higher operating pressures. It is also noted from this table that the stevioside recovery decreases marginally with crossflow rates. Excepting the first two experiments, at 276 kPa, 60 and 80 l/h the variation of permeate recovery for different cross rates at a fixed pressure value is insignificant. This is because the membrane for the first experiment is fresh and after one experiment some irreversible fouling occurs that reduces the stevioside recovery drastically. However, this fouling is present for subsequent experiments, but it is marginal and the stevioside recovery shows a declining trend (although extremely small) with the crossflow rates at a fixed transmembrane pressure drop. In the batch mode of operation, the permeate is not recycled back. In fact, for clarification of the stevia extract, the permeate is the product and this is the most favorable mode of operation. Three experiments were conducted in this case, at 100 l/h crossflow rate, the operating pressure difference were varied as 276, 414 and 552 kPa. The permeate flux profile along with the volume concentration ratio is presented in Figure 9.7, as a function of time. The permeate flux decline is more in this case compared to the total recycle mode, however, no steady-state exists. In this mode of operation, the permeate is not recycled to the feed tank; as a result, the volume of the feed tank reduces, leading to an increase in feed concentration. As the feed concentration

1.40

Flow rate  100l/h 276 kPa 414 kPa 552 kPa

Permeate flux (l/m2.h)

10

1.35 1.30

8

1.25

6

1.20 1.15

4 1.10 2 1.05 0

0

1

2

3

4

5

6

7

8

9

10

1.00 11

Volume concentration factor (VCF)

12

Time (h) Figure 9.7: Flux decline profile and variation of volume concentration factor with transmembrane pressure drop in batch concentration mode of crossflow ultrafiltration

220 

 9 Processing of stevioside using membrane-based separation processes

increases, the concentration polarization becomes more severe. More solutes are convected towards the membrane surface, resulting in a thicker cake layer. This increases the resistance against the solvent flux and the permeate flux declines. At higher operating pressures, solute deposition on the membrane surface is augmented by forced convection, leading to a further decline in permeate flux. As the above phenomena increases as time of filtration increases, a steady-state is never attained. The properties of the permeate were also monitored over the filtration period. Profiles of color, clarity, total solids, stevioside recovery and purity in permeate are presented in Figure 9.8 (A–E). It is observed from these figures that color, total solids and the stevioside recovery in the permeate decrease with time, and clarity increases with time. As discussed earlier, with progress in filtration time, the cake-type layer that grows on membrane surfaces acts as dynamic membrane and retains the solutes. Conversely, the crossflow rate should be at maximum to obtain a higher permeate flux (see Figure 9.6). Thus, total solids and stevioside recovery decreases. Although color decreases with time, its variation is marginal. Thus, among the operating conditions studied herein, 276 kPa pressure and 120 l/h crossflow rate are suitable operating conditions for ultrafiltration of stevia extract with a 30 kDa membrane. An interesting observation is made from Figure 9.8(D). Stevioside recovery at the end of 10 h for all three operating pressures ranges from 30% (at higher pressures such as 552 kPa) to 38% (at lower pressures such as 276 kPa). This is due to the enhanced retention of the dynamic membrane at higher pressures, making it more compact. Another interesting feature that can be observed in Figure 9.8 (E) is that the purity for lower pressures decreases on increasing time of operation, which is not the case for higher pressures. The compactness of the dynamic cake layer over the membrane is not significant enough to screen other solids at lower pressures. This results in an increase in total solids in the permeate (Figure 9.7C), thereby, decreasing purity. It must be noted here that purity does not exclusively depend on the amount of stevioside present in the permeate, rather the relative ratio of the amount of stevioside to total solids in the permeate. So, even for the same amount of stevioside content, purity can be increased if the total solid content is decreased. Ideally, the batch operation at lower pressures and high flowrates should be limited up to 5 h (as observed from the present study) for maintaining high purity of stevioside in the permeate.

9.5.4 Nanofiltration The nanofiltration experiments were conducted in batch cells using a 400 Da MWCO membrane. Profiles of permeate flux and VCF with transmembrane pressure drop are shown in Figure 9.9 (A–D) at various stirring speeds: we can see the permeate flux declines with time and the flux is higher at higher operating pressure, as expected. At 1241 kPa pressure and 1,500 rpm, flux is the highest, and hence VCF is the maximum at about 2 in the test cell.

9.5 Detailed membrane-based clarification processes 

1.8

 221

120

1.5 100

Clarity (%T)

Color (A)

1.2 0.9 0.6

80

0.3

60

0

0

2

(a)

4

6

8

10

Time (h)

0

2

(b)

3.0

4

6

8

10

6

8

10

Time (h)

100

2.7 80

Stevioside Recovery(%)

Total solid (g/100 ml)

2.4 2.1 1.8 1.5 1.2 0.9 0.6

0 (c)

2

4

6

8

60 40 20 0

10

Time (h)

0

2

(d)

4 Time (h)

0.70

Flow rate  100 l/h 276 kPa 414 kPa 552 kPa

0.65 0.60

Purity

0.55 0.50 0.45 0.40

0 (e)

2

4

6

8

10

Time (h)

Figure 9.8: Profiles of permeate properties for various operating conditions in batch concentration mode of ultrafiltration. (a) color, (b) clarity, (c) total solids, (d) recovery of stevioside, (e) purity of stevioside

Various properties of the permeate are reported in Table 9.8. We can see that the clarity of permeate is more than 99% in most of the cases. Color and total solids in the permeate are quite low.

 9 Processing of stevioside using membrane-based separation processes

1.3

25

1.2 20 1.1 15

Permeate flux (l/m2.h)

(a)

0.4

0.6

0.8

1.0

Time (h) 50 45 40 35 30 25 20 15

Operating pressure  1103 kPa

1.4 1.2 1.0

0

0.2

0.4

0.6

0.8

1.0

Time (h) 500 rpm 1000 rpm

40

1.5 1.4

30 1.3 25

1.2

20

1.1

15

1.0

0.2

0.4

0.6

0.8

1.0

Time (h) 60 55 50 45 40 35 30 25 20

Operating pressure  1241 kPa

2.0 1.8 1.6 1.4 1.2 1.0

0 (d)

1.6

35

(b)

1.8 1.6

Operating pressure  965 kPa

0

Permeate flux (l/m2.h)

0.2

1.0 1.2

Permeate flux (l/m2.h)

1.4

0

(c)

1.5

Volume concentration factor

30

1.7

45

0.2

0.4

0.6

0.8

Volume concentration factor

1.6

Operating pressure  827 kPa

Volume concentration factor

Permeate flux (l/m2.h)

35

1.0

Time (h)

Volume concentration factor

222 

1500 rpm Error bars:  3%

Figure 9.9: Flux decline profiles and variation of volume concentration factor with operating conditions during stirred batch nanofiltration. (a) 827 kPa, (b) 965 kPa, (c) 1103 kPa, (d) 1241 kPa

It can be concluded that the purity of overall process (UF + NF) is constant around 60%. However, the overall recovery of stevioside increases with stirring and transmembrane pressure drop. Maximum recovery is obtained at 1,241 kPa and 1,500 rpm.

9.5.5 Diafiltration To extract maximum amounts of stevioside (or any other product) in this case, multiple filtration of the retentate is often productive [119]. Multi-stage diafiltration aims at recovering the desired component, present in the original feed, by concentrating the retentate. It has been observed that single-stage ultrafiltration for extraction of sweeteners from stevia leaf extract can be achieved upto a maximum of 47% at a particular set of operating conditions and a specified membrane [120]. A multi-stage membrane diafiltration is carried out in crossflow mode to increase the recovery of stevioside beyond 80%, by maintaining a VRR of 1.5 at every stage. Additional permeate quality parameters, including color and clarity, were also determined to observe the change in properties of the product during filtration. The initial total solid concentration after hot water extraction was 3.2 % w/w and after centrifugation it was 1.6 % w/w. The feed concentration of stevioside and

500 1000 1,500 500 1,000 1,500 500 1,000 1,500 500 1,000 1,500 552 kPa, 100 l/h Optimum operating condition Optimum operating condition

827

Crude extract

UF extract (feed for NF) Centrifuge extract

1241

1103

965

Stirring speed (rpm)

Operating pressure (kPa)

Permeate Clarity (%T) 99.5 ± 0.4 99.9 ± 0.3 99.1 ± 0.4 99.0 ± 0.5 99.5 ± 0.4 99.6 ± 0.3 99.3 ± 0.2 98.6 ± 0.3 98.1 ± 0.2 99.3 ± 0.4 99.3 ± 0.4 99.1 ± 0.3 92.8 ± 0.3 _

0.006 ± 0.003

Permeate Color (A420)

0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.003 0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.004 0.01 ± 0.005 0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.004 0.02 ± 0.002 0.02 ± 0.005 0.62 ± 0.003

10.6 ± 1.4

12.3 ± 1.6

2.9 ± 0.4

_

0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.02 0.2 ± 0.06 0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.03 0.1 ± 0.04 0.1 ± 0.03 0.2 ± 0.04 0.9 ± 0.05

Permeate Total solid (g/100 ml)

15,699 ± 4.4

14,129 ± 6.2

127.9 ± 1.4 143.1 ± 1.4 196.8 ± 2.4 173.0 ± 3.4 107.7 ± 2.4 208.5 ± 1.4 348.0 ± 3.4 179.2 ± 2.4 349.5 ± 3.4 246.5 ± 2.4 276.7 ± 1.4 201.3 ± 2.4 4,945 ± 5.4

Permeate Stevioside (mg/l) 1.1 ± 0.05 1.2 ± 0.04 1.3 ± 0.02 1.2 ± 0.06 1.3 ± 0.05 1.3 ± 0.04 1.3 ± 0.05 1.4 ± 0.04 1.4 ± 0.03 1.4 ± 0.04 1.5 ± 0.03 1.6 ± 0.04

Product (retentate) Total solid (g/100 ml) 6,486 ± 5 7,250 ± 3 7,462 ± 6 7,283 ± 4 7,606 ± 8 7,976 ± 3 7,565 ± 5 8,233 ± 2 8,529 ± 7 7,952 ± 6 8,726 ± 8 9,594 ± 10

Product (retentate) Stevioside (mg/l) 41.3 46.2 47.5 46.4 48.4 50.8 48.2 52.4 54.3 50.6 55.6 61.1

59.0 60.4 57.4 60.7 58.5 61.4 58.2 58.8 60.9 56.8 58.2 60.0

Overall Overall recovery (%) purity (%) (UF + NF) (UF + NF)

Table 9.8: Various properties of permeate of nanofiltration at the end of the experiment (Feed is ultrafiltration permeate at 552 kPa and 100 l/h)

9.5 Detailed membrane-based clarification processes 

 223

224 

 9 Processing of stevioside using membrane-based separation processes

rebaudioside to crossflow ultrafiltration was 1215 mg/l and 268 mg/l, respectively. Three-stage diafiltration were performed at 690 kPa and 50 l/h. Figure 9.10(A) represents the variation of stevioside and rebaudioside concentration during ultrafiltration. We can see from the figure that the rebaudioside concentration reaches a maximum in 3–5 h, and decreases thereafter. In the subsequent stages rebaudioside was present in trace amounts. As rebaudioside was present in the stevia leaf extract in small amount compared to stevioside, most of it was permeated in the initial stage. The flux decline profile is illustrated in Figure 9.10. As the filtration stage increases, the feed was diluted to maintain constant feed volume and the resultant permeate flux increases. As the VRF was fixed for a particular stage, the time required for the second and third stage was much less than for the first stage to obtain an equivalent quantity of permeate volume. It can be inferred from the figure, that the permeate flux gets doubled from the first to the second stage of filtration. The increase in permeate flux was also supported from the total solids value in Table 9.9. The total solids content in the

100

Stevioside stage 1 Stevioside stage 2 Stevioside stage 3 Rebaudioside stage 1

Initial feed (%)

80

60

40

20

0

0

1

2

(a)

3

4

5

6

7

Time (h) 20

Stage 1 Stage 2 Stage 3

18

Permeate flux (l/m2h)

16 14 12 10 8 6 4

0 (b)

1

2

3

4

5

6

Time (h)

Figure 9.10: (a) Stevioside and rebaudioside profile. (b) Permeate flux variation with time of operation at 690 kPa and 50 l/h crossflow rate

0.44 0.44 0.5 0.34 0.66 0.46 0.39

89.12 90.16 86.50 92.47 66.44 72.44 71.61

Centrifuged product (feed to first stage)

1.0 2.0 3.0 4.0 5.0 6.0 7.0

0.75 0.76 0.77 0.77 0.82 0.65 0.73

Total solids (%w/w) 0.5 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Time (h)

Clarity (%T)

Time (h)

Color (A)

Stage 2

Stage 1

0.62 0.21 0.24 0.76 0.54 0.63 0.88 0.77

Color (A)

74.82 93.32 85.50 56.88 82.03 71.12 60.25 62.81

Clarity (%T)

0.36 0.28 0.40 0.22 0.15 0.09 0.10 0.13

Total solids (%w/w) 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time (h)

Stage 3

3.55

0.33 0.35 0.29 0.36 0.36 0.26 0.32 0.33

Color (A)

1.53

87.01 87.01 89.95 87.50 86.5 92.26 87.90 87.30

Clarity (%T)

Table 9.9: Properties of permeate (color, clarity and total solids) for the experimental conditions: TMP – 690 kPa and crossflow rate 50 l/h

1.6

0.03 0.03 0.01 0.03 0.04 0.03 0.01 0.01

Total solids (%w/w)

9.5 Detailed membrane-based clarification processes 

 225

226 

 9 Processing of stevioside using membrane-based separation processes

permeate decreases many-fold on increasing the stages of diafiltration, for example, the total solid content decreases by 50-fold from the feed to the first stage (centrifuged product) to the permeate from the third stage. The color and clarity data presented in Table 9.9 indicates that as time progresses the clarity decreases, while color is fixed. As the permeate was not recycled back to the feed, the feed gets concentrated and hence the color and clarity in the permeate decreases. But, in the third stage, the color and clarity was almost constant. This is because, the feed was diluted significantly and the total solids in the permeate was also constant. The stevioside and rebaudioside A (also other glycosides) present in the solution are completely soluble and do not impart any change in color or clarity of the system. However, the presence of other insoluble (undesired) solids comprised of cell debris, lignin, cellulose, starch, chlorophyll, etc., are responsible for color and clarity. Thus, the presence of stevioside and rebaudioside in a clear solution is favorable. The cumulative recovery of stevioside was greater than 80% after stage 3 as evident from Figure 9.10. However, the stevioside and rebaudioside recovery was more in the permeate during ultrafiltration at a lower transmembrane pressure. Thus, increasing the number of stages and ultrafiltration at lower pressures, the recovery of the steviol glycosides can be enhanced.

9.6 References 1. Donnell K, Kearsley M. Sweeteners and Sugar Alternatives in Food Technology. Wiley Blackwell: London; 2012. 2. Grashoff JE. A Systematic Study of the North and Central American Species of Stevia. Ph.D Dissertation. University of Texas, University of Michigan Microfilm, 1972;1–608. 3. Robinson BL. Observations on the genus Stevia. Contrib Gray Herb Harvard Univ 1930, 36–58. 4. Soejarto DD, Kinghorn AD, Farnsworth NR. Potential sweetening agents of plant origin. III. Organoleptic evaluation of Stevia leaf herbarium samples for sweetness. J Nat Prod 1982;45: 590–599. 5. Soejarto DD, Compadre CM, Medon PJ, Kamath SK, Kinghorn AD. Potential sweetening agents of plant origin II. Field search for sweet-tasting Stevia species. Econ Bot 1983;37:71–78. 6. Bertoni MS. Revista de Agronomia de l’Assomption 1899, 1, 35. 7. Lewis WH. Early uses of Stevia rebaudiana (Asteraceae) leaves as a sweetener in Paraguay. Econ Bot 1992;46:336–337. 8. Bertoni MS. La Stevia rebaudiana Bertoni. La estevina y la rebaudina, neuvas substancias edulcorantes. Anales Cientificos Paraguayos, Series II 1918;6:29–134. 9. Bertoni MS. Le Kaá hê-é: sa nature et ses propriétés. Anales Cientificos Paraguayos, Serie I 1905;5:1–14. 10. Mosettig E, Beglinger U, Dolder F, Lichti H, Quitt P, Waters JA. The absolute configuration of steviol and isosteviol. J Am Chem Soc 1963;85:2305–2309. 11. Kohda H, Kasai R, Yamasaki K, Murakami K, Tanaka O. New sweet diterpene glycosides from Stevia rebaudiana. Phytochem 1976;15:981–983. 12. Kobayashi M, Horikawa S, Degrandi IH, Ueno J, Mitsuhashi H. Dulcosides A and B, new diterpene glucosides from Stevia rebaudiana. Phytochem 1977;16:1405–1407. 13. Tanaka O. Steviol-glycosides: new natural sweetener. Trends Analyt Chem 1982;1:246–248.

9.6 References 

 227

14. Yamasaki K, Kohda H, Kobayashi T, Kasai R, Tanaka O. Structures of Stevia diterpeneglucosides: Applications of 13C NMR. Tetrahedron Lett 1976, 1005–1008. 15. Food and Agriculture Organization of the World Health Organization. Steviol glycosides. FAO JECFA Monographs 5, 2008. 16. Food and Agriculture Organization of the World Health Organization. List of Substances Scheduled for Evaluation and Request for Data. Food and Agriculture Organization of the United Nations/ World Health Organization, Joint FAO/WHO Expert Committee on Food Additives. Seventy-third meeting Food additives and Contaminants. Geneva, 8–17 June, 2010. http://www.who.int/ipcs/food/jecfa/jecfa73.pdf. (Accessed October 14, 2013) 17. World Health Organization. Joint FAO/WHO Expert Committee on Food Additives. Sixty-ninth meeting, Summary and Conclusions, Steviol Glycosides. Issued July 4, 2008. 18. World Health Organization. Joint FAO/WHO Expert Committee on Food Additives. WHO Food Additive Series: 60, 2009. Safety evaluation of certain food additives. Steviol Glycosides (addendum). 19. Food Standards Australia New Zealand. Final Assessment Report, Application A540, Steviol Glycosides as Intense Sweeteners, 2008. 20. European Food Safety Authority. Scientific opinion on the safety of steviol glycosides for the proposed uses as a food additive. EFSA Panel on Food Additives and Nutrient Sources added to Food (ANS). EFSA J 2008, 8, 1537. 21. Kinghorn AD, Wu CD, Soejarto DD. Stevioside. In: Nabors L, ed. Alternative Sweeteners. 3rd ed. Marcel Dekker: NY; 2001, 167–183. 22. Kinghorn AD, Soejarto DD. Stevioside. In: Nabors L, Gelardi RC, ed. Alternative Sweeteners, 2nd ed. Marcel Dekker: NY; 1991, 157–171. 23. Brandle JE, Starratt AN, Gizjen M. Stevia rebaudiana: its agricultural, biological, and chemical properties. Can J Plant Sci 1998;78:527–536. 24. Nepovim A, Drahosova H, Valicek P, Vanek, T. The effect of cultivation conditions on the content of Stevioside in Stevia rebaudiana plants cultivated in the Czech Republic. Pharm Pharmacol Lett 1998;8:19–21. 25. Chalapathi MV, Thimmegowda S, Sridhara S, Parama VRR, Prasad TG. Natural noncalorie sweetener Stevia (Stevia rebaudiana Bertoni)—a future crop of India. Crop Research 1977;14:347–350. 26. Dzyuba O, Vseross O. Stevia rebaudiana (Bertoni) Hemsley—a new source of natural sweetener for Russia. Rastitel’nye Resursy (Plant Resources) 1998;34:86–95. 27. Bakal AI, Nabors L. Stevioside. In: Nabors L, Gelardi RC, ed. Alternative Sweeteners, 2nd ed. Marcel Dekker: NY; 1986, 295–307. 28. Crammer B, Ikan R. Progress in the chemistry and properties of the Rebaudiosides. In: Grenby TH, ed. Developments in Sweeteners – 3. Elsevier Applied Science: London; 1987, 45–64. 29. Savita S, Sheela K, Sunanda S, Shankar A, Ramakrishna P. Stevia rebaudiana – A functional component for food industry. J Human Eco 2004;15:261–264. 30. Kinghorn AD, Soejarto DD. Current status of stevioside as a sweetening agent for human use. In: Wagner H, Hikino H, Farnsworth R, ed. Economic and Medicinal plant research. Academic Press: London; 1985, 1–52. 31. Chang SS, Cook JM. Stability studies of stevioside and rebaudioside A in carbonated beverages. J Agri Food Chem 1983;31:409–412. 32. Abou-Arab A, Abou-Arab A, Abu-Salem MF. Physico-chemical assessment of natural sweeteners Steviosides produced from Stevia rebaudiana Bertoni plant. Africa J Food Sci 2010;4:269–281. 33. Kroyer GT. The low calorie sweetener stevioside: stability and interaction with food ingredients. LWT-Food Sci Technol 1999;32:509–512. 34. Anish T, Rema S. Preliminary studies on Stevia rebaudianci leaves proximal composition, mineral analysis and phytochemical screening. J Med Sci 2006;6:321–326.

228 

 9 Processing of stevioside using membrane-based separation processes

35. Kinghorn AD, ed. Stevia: The Genus Stevia. CRC Press: NY; 2002. 36. Thomas J, Glade M. Stevia: It’s not just about calories. The Open Obesity J 2010;2:101–109. 37. Shukla S, Mehta A, Bajpai V, Shukla, S. In vitro antioxidant activity and total phenolic content of ethanolic leaf extract of Stevia rebaudiana Bert. Food Chem Toxic 2009;47:2338–2343. 38. Patil V, Ashwini K, Reddy P, Purushotham M, Prasad T, Udaykumar M. In vitro multiplication of Stevia rebaudiana. Curr Sci 1996;70:960. 39. Sivaram L, Mukundam U. In vitro culture studies on Stevia rebaudiana. In Vitro Cell Dev-Pl 2003;39:520–523. 40. Tomita T, Sato N, Arai T, Shiraishi H, Sato M, Takeuchi M, et al. Bactericidal activity of a fermented hot-water extracts from Stevia rebaudiana Bertoni and other food-borne pathogenic bacteria. Microbiol Immunol 1997;41:1005–1009. 41. Debnath M. Clonal propagation and antimicrobial activity of an endemic medicinal plant Stevia rebaudiana. J Med Plant Res 2008;2:45–51. 42. Ghosh S, Subudhi E, Nayak S. Antimicrobial assay of Stevia rebaudiana Bertoni leaf extracts against 10 pathogens. Int J Integrat Bio 2008;2:27–31. 43. Atteh J, Onagbesan O, Tona K, Decuypere E, Geuns J, Buyse J. Evaluation of supplementary stevia (Stevia rebaudiana, bertoni) leaves and stevioside in broiler diets: effects on feed intake, nutrient metabolism, blood parameters and growth performance. J Anim Physiol Anim Nutr (Berl) 2008;92:640–649. 44. Barriocanal L, Palacios M, Benitez G, Benitez S, Jimenez J, Jimenez N, Rojas V, et al. Apparent lack of pharmacological effect of steviol glycosides used as sweeteners in humans, a pilot study of repeated exposures in some normatensive and hypotensive individuals and in type 1 and type 2 diabetics. Regul Toxicol Pharma 2008;51:37–41. 45. Jeppesen PB, Gregersen S, Rolfsen SE, Jepsen M, Colombo M, Agger A, Xiao J, Kruhøffer M, Orntoft T, Hermansen K. Antihyperglycemic and blood pressure-reducing effects of stevioside in the diabetic Goto-Kakizaki rat. Metabolism 2003;52:372–378. 46. Maki K, Curry L, Reeves M, Toth P, Mckenney J, Farmer M, et al. Chronic consumption of Rebaudioside A, a steviol glycoside, in men and women with type 2 diabetes mellitus. Food Chem Toxicol 2008;46:47–53. 47. Wingard R, Brown J, Enderlin F, Dale J, Hale R, Seitz C. Intestinal degradation and absorption of the glycosidic sweeteners stevioside and rebaudioside A. Cell Mol Life Sci 1980;36:519–520. 48. Kochikyan V, Markosyan A, Abelyan L, Balayan A, Abelyan, V. Combined enzymatic modification of stevioside and rebaudioside A. Appl Biochem Microbio 2006;42:31–37. 49. Jayaraman S, Manoharan M, Illanchezian S. In-vitro antimicrobial and antitumor activities of Stevia rebaudiana (Asteraceae) leaf extracts. Trop J Pharm Res 2008;7:1143–1149. 50. Sehar I, Kaul A, Bani S, Pal H, Saxena A. Immune up regulatory response of a non-caloric natural sweetener, stevioside. Chem-Bio Interac 2008;173:115–121. 51. Chan P, Linson B, Chen Y, Liu J, Hsieh M, Cheng J. A double blind placebo-controlled study of the effectiveness and tolerability of oral stevioside in human hypertension. Br J Clinic Pharm 2000;50:215–220. 52. Jeppesen P, Gregersen S, Alstrupp K, Hermansen K. Stevioside induces antihyperglycaemic, insulinotropic and glucagonostaticeffects in vivo: Studies in the diabetic Goto-Kakizaki (GK) rats. Phytomed 2002;9:9–14. 53. Lee CN, Wong K, Liu J, Chen Y, Chan P. Inhibitory effect of stevioside on calcium influx to produce antihypertension. Planta Med 2001;67:796–799. 54. Chen J, Jeppesen P, Abudula R, Dyrskog S, Colombo M, Hermansen K. Stevioside does not cause increased basal insulin secretion or b-cell desensitization as does the sulphonylurea, glibenclamide: Studies in vitro. Life Sci 2006;78:1748–1753.

9.6 References 

 229

55. Jeppesen P, Gregersen S, Poulsen C, Hermansen K. Stevioside acts directly on pancreatic a cells to secrete insulin: Actions independent of cyclic adenosine monophosphate and adenosine triphosphate-sensitive K+ channel activity. Metabolism 2000;49:208–214. 56. Suanarunsawat T, Chaiyabutr N. The effect of steviosides on glucose metabolism in rat. Canad J Physiol Pharm 1997;75:976–982. 57. Takahashi K, Matsuda M, Ohashi K, Taniguchi K, Nakagomi O, Abe Y, Mori S, Sato N, Okutani K, Shigeta S. Analysis of anti-rotavirus activity of extract from Stevia reabudiana. Antiviral Res 2001;49:15–24. 58. Toskulkao C, Sutheerawatananon M, Wanichanon C, Saitongdee P, Suttagit M. Effect of stevioside and steviol on intestinal glucose absorption hamsters. J Nutr Sci Vitamin 1995;41:105–113. 59. Jutabha P, Toskulkao C, Chatsudthipong V. Effect of stevioside on PAH transport by isolated perfused rabbit renal proximal tubule. Canad J Phys Pharm 2000;78:737–744. 60. Chatsudthipong V, Muanprasat C. Stevioside and related compounds: Therapeutic benefits beyond sweetness. Pharma Therapeutic 2009;121:41–54. 61. Goyal S, Samsher, Goyal R. Stevia (Stevia rebaudiana) a bio-sweetener: A review. Int J Food Sci Nutr 2010;61:1–10. 62. Hsieh MH, Chan P, Sue YM, Liu JC, Liang TH, Huang TY, Tomlinson B, Chow MS, Kao PF, Chen YJ. Efficacy and tolerability of oral stevioside in patients with mild essential hypertension: A two-year, randomized, placebo-controlled study. Clin Ther 2003;25:2797–2808. 63. Pól J, Hohnová B, Hyötyläinen T. Characterization of Stevia rebaudiana by comprehensive two-dimensional liquid chromatography time-of-flight mass spectrometry. J Chromatogr A 2007;1150:85–92. 64. Blauth de Slavutzky S. Stevia and sucrose effect on plaque formation. J für Verbraucherschutz Lebensmittelsicherheit 2010;5:213–216. 65. Persinos GJ. Method of Producing Stevioside. US Patent 1973, 3723410. 66. Adduci J, Buddhasukh D, Ternai B. Improved isolation and purification of stevioside. J Sci Soc Thailand 1987;13:179–183. 67. Yutaka K, Hiroaki I. Sweetener. Japanese Patent 1987, 62–025949. 68. Korean Patent KR9007421, Purification process of Stevioside, Pacific Chemical Co., 1990. 69. Payzant JD, Laidler JK, Ippolito RM. Method of Extracting Selected Sweet Glycosides from Stevia rebaudiana Plant. US Patent 1999, 5962678. 70. Kumar JK, Babu GK, Kaul VK, Ahuja PS. Process for Production of Stevioside from Stevia rebaudiana Bertoni. International Patent 2006, WO 2006/038221 A1. 71. Purakayastha S, Markosyan A, Malsagov M. Highly Purified Stevioside, Rebaudioside A and A Purified Sweet Steviol Glycoside Mixture. US Patent 2010, 0227034. 72. Malasagov M, Tomov T, Somann T, Abelyan VH. Process for Manufacturing a Sweetener and Use Thereof. US Patent 2011, 7862845B2. 73. Morita T, Morita K, Kanzaki S. Novel Stevia Variety and Method of Producing Sweeteners, Japanese Patent 2011, AA01H500F. 74. Kageyama I. Purifcation of Stevioside. Japanese Patent 1980, 55–092400. 75. Fumio M. Stevioside Extracted from Stevia Containing Sweetener. Japanese Patent 1980, 55–0007039. 76. Tadaaki H, Ryoichi I, Teruo K. A Method for Purifying Stevioside. Japanese Patent 1976, 51–131900. 77. Shigeji S. Preparation of Stevioside. Japanese Patent 1980, 55–162953. 78. Tadashi K, Masato K. Production of Stevia Sweetener. Japanese Patent 1995, 07–143860. 79. Toyoshige M, Usei B. Sweetener Obtained From Plant Body of Variety of Stevia Rebaudiana Cultivatable From Seed. Japanese Patent 2002, 2002–262822. 80. Jackson MC, Francis GJ, Chase RG. High Yield Method of Producing Pure Rebaudioside A. US Patent 2006, 0083838.

230 

 9 Processing of stevioside using membrane-based separation processes

 81. Pol J, Ostra EV, Karasek P, Roth M, Karolinka B, Kaslavsky J. Comparison of two different solvents employed for pressurized fluid extraction of stevioside from Stevia rebaudiana: methanol versus water. Anal Bioanal Chem 2007;388:1847–1857. 82. Chhaya, Majumdar GC, De S. Optimization of process parameters for water extraction of Stevioside using response surface methodology. Sep Sci Technol 2012;47:1–9. 83. Kumar S. Method for Recovery of Stevioside. US Patent 1986, 4599403. 84. Giovanetto RH. Method for the Recovery of Steviosides From Plant Raw Material. US Patent 1990, 4892938.  85. Deji W. High-efficiency Method for Continuously Extracting Stevioside From Stevia Leaf. China Patent 2009, 200810216065. 86. Weiping H, Zhou JH. Process for Extracting Sweet Diterpene Glycosides. US Patent 1999, 6228996.  87. Abelyan VH, Ghochikyan VT, Markosyan AA, Adamyan MO, Abelyan LA. Extraction, sepration and modification of sweet glycosides from the Stevia rebaudiana plant. US Patent 2010, 7838044B2. 88. Kotaro K, Tokuo O. Extraction and Purification of Sweetener Component From Dry Leaf of Stevia. Japanese Patent 1987, 62–166861. 89. Yukio O, Hajime I, Taku T. Purifying Method of Stevioside Solution. Japanese Patent 1983, 58–028247. 90. Taku T, Yukio O. Preparation of Stevioside. Japanese Patent 1983, 58–028246.  91. Moriata T, Fujita I, Iwamura J. Sweetening Compound, Method of Recovery, and Use Thereof. US Patent 1978, 4082858. 92. Matsushita S, Ikushigo T. Separation of Sweet Component From Natural Sweet Extracts. US Patent 1979, 4171430. 93. Hideaki U, Ryoichi I, Teruo K. Purification of Stevia Sweetening Agent. Japanese Patent 1979, 54–030199. 94. Susumu O. Isolation of principal sweetening component of stevia. Japanese Patent 1982, 57–086264. 95. Masashi O, Tadashi Y. Purification of Stevioside. Japanese Patent 1982, 57–075992. 96. Koji I, Takeshi I. Purification of Stevioside. Japanese Patent 1979, 54–041898. 97. Ryoichi I, Isamu H. Separation and Purification of Stevioside Sweetening. Japanese Patent 1979, 54–132599. 98. Dobberstein RH, Ahmed MS. Extraction, Separation and Recovery of Diterpene Glycosides From Stevia rebaudiana Plants. US Patent 1982, 4361697. 99. Chiang C, Evans JC, Hahn JJ, Heylen AAJ, Ohmes AK, Patist A, Rhinemus TA, Stangler J, Tyler CA, Vercauteren RLM. Separation of Rebaudioside A From Stevia Glycosides Using Chromatography. US Patent 2011, 20110087011. 100. Liu J, Zhang K, Guo S. Separation and Purification of Stevioside and Rebaudioside A. US Patent 2012, 0083593 A1. 101. Moraes EP, Machado NRCF. Clarification of Stevia rebaudiana (Bert.) Bertoni extract by adsorption in modified zeolite. Acta Scientiarum 2011;23:1375–1380. 102. Rajab R, Mohankumar C, Murugan K, Harish M, Mohanam PV. Purification and toxicity studied of Stevioside from Stevia rebaudiana Bertoni. Toxicol Int 2009;16:49–54. 103. Sun J, Gu R. Method for ultrasonic extraction of stevioside. WIPO Patent Application WO/ 2011/113373. 104. Jaitak V, Bandna, Singh B, Kaul VK. An efficient microwave-assisted extraction process of Stevioside and Rebaudioside A from Stevia rebaudiana (Bertoni). Phytochem Anal 2009;20:240–245.

9.6 References 

 231

105. Kienle U. Method of Making A Natural Sweetener Based on Stevia rebaudiana and Use There of. US Patent 1992, 5112610. 106. Pasquel A, Meireles MAA, Marques MOM, Petenate AJ. Extraction of Stevia glycosides with CO2 + water, CO2 + ethanol and CO2 + water + ethanol. Brazil J Chem Eng 2000;17:271–282. 107. Das S, Das A, Murphy R, Punwani I, Nasution M, Kinghorn A. Evaluation of the cariogenic potential of the intense natural sweeteners stevioside and rebaudioside A. Caries Res 1992;26:363–366. 108. Fuh WS, Chiang BH. Purification of steviosides by membrane and ion-exchange process. J Food Sci 1990;55:1454–1457. 109. Kutowy O, Zhang SQ, Kumar A. Extraction of Sweet Compounds from Stevia rebaudiana Bertoni. US Patent 1999, 5972120. 110. Silva FV, Bergamasco R, Andrade CMG, Pinheiro N, Machado NRCF, Reis MHM, de Araujo AA, Rezende SL. Purification process of Stevioside using zeolites and membranes. Intl J Chem React Eng 2007;5:A40. 111. Vanneste J, Sotto A, Coutin CM, van Craeyveld V, Bernaerts K, van Impe J, Vandeur J, Taes S, van der Bruggen B. Application of tailor-made membranes in a multi-stage process for the purification of sweeteners from Stevia rebaudiana. J Food Eng 2011;103:285–293. 112. Rao AB, Reddy GR, Ernala P, Sridhar S, Ravikumar YVL. An improvised process of isolation, purification of Steviosides from Stevia rebaudiana Bertoni leaves and its biological activity. Int J Food Sci Technol 2012;47:2554–2560. 113. Khuri AI, Cornell JA. Response Surfaces: Designs and Analyses. Marcel Dekker: NY; 1989. 114. Myers RH. Response Surface Methodology. Allyn and Bacon: Boston, MA; 1971. 115. Chhaya, Majumdar GC, De S. Primary clarification of stevia extract: A comparison between centrifugation and microfiltration. Sep Sci Technol 2013;48:113–121. 116. Jonsson G, Pradanos P, Hernandez A. Fouling phenomena in microporous membranes: Flux decline kinetics and structural modifications. J Membr Sci 1996;112:171–183. 117. Chamchong M, Noomhorm A. Effect of pH and enzymatic treatment on microfiltration and ultrafiltration of tangerine juice. J Food Proc Eng 1991;14:21–34. 118. Fukumoto LR, Delaquis P, Girard B. Microfiltration and ultrafiltration ceramic membranes for apple juice clarification. J Food Sci 1998;63:845–850. 119. Zhang SQ, Kumar A, Kutowy O. Membrane based separation scheme for processing sweeteners from Stevia leaves. Food Res Int 2000;33:617–620. 120. Chhaya, Sharma C, Mondal S, Majumdar GC, De S. Clarification of Stevia extract by ultrafiltration: Selection criteria of the membrane and effects of operating conditions. Food Bioprod Proc 2012;90:525–532.

10 Production of value-added soy protein products by membrane-based operations Martin Mondor 10.1 Introduction 10.1.1 Soy as the most important source of plant protein ingredients Soybeans and soybean products have long been used by millions of people in the East as their chief source of protein and as medicine. First cultivated in South East Asia, soybean is now present all around the world. Soy is an abundant source of dietary protein, containing 40% protein on average, and is available at relatively low cost. In 2011, soy accounted for approximately 68% of world protein meal consumption, over other plant protein sources such as rapeseed (13%), cottonseed (6%) and sunflower seed (5%) [1]. From a nutritional point of view, purified soy proteins in the form of concentrates and isolates can be considered similar to animal proteins [2, 3]. Nutritional and health benefits for humans have been attributed for a long time to the consumption of soy foods, especially soy proteins. These benefits include hypocholesterolemic effects [4] and the prevention of heart disease [5] and breast cancer [6]. In 1999, the US FDA approved a health claim for soy proteins stating that 25 g of soy proteins each day, as part of a diet low in saturated fat and cholesterol, may reduce the risk of heart disease [7, 8]. This recognition is a good indication of the additional benefits, beyond basic nutrition, associated with the consumption of soy proteins. Soy protein products offer more than just nutritional and health benefits for humans. Advances in soy ingredient technology have resulted in products that are industrially produced for a variety of purposes. Soy products can be found as emulsifiers, texture enhancers, and ingredients to increase or replace protein content in food products such as bread, pastries, beverages and meat [9].

10.1.2 Production of soy protein isolates by isoelectric precipitation Isolates are the most highly refined soy protein products that are commercially available, containing at least 90% protein on a dry basis. Commercial soy protein isolates are usually prepared from dehulled and defatted soybeans by isoelectric precipitation [10, 11]. The proteins are extracted from soybean flakes or flours with water adjusted to pH 8–11 using a base, at a solids/solvent ratio of 1:10 to 1:20 and at a temperature of 30–50°C. The insoluble fibrous residue is then removed by a centrifugation step, and the pH of the resulting soy protein extract is adjusted to pH 4.2–4.5 (i.e., the isoelectric point of the proteins) using a mineral acid, such as HCl, to precipitate the proteins. The proteins are then recuperated by a second centrifugation

234 

 10 Production of value-added soy protein products by membrane-based operations

step, which is followed by multiple washings to remove minerals and sugars, in order to increase the protein content, and neutralization of the proteins to pH 7 with a dilute base, such as NaOH. The resulting soy protein dispersion is fed to a spray dryer to produce an isolate. Although soy protein isolates produced by isoelectric precipitation usually have superior functional properties compared to protein isolates from other plant sources, soy protein isolates may still have limited functional properties because of protein denaturation [12, 13]. Soy protein isolates produced by isoelectric precipitation also have a high phytic acid content (1–3% w/w), which alters the solubility of the isolates, especially at low pH [14, 15]. In addition, from an environmental point of view, the isoelectric precipitation process requires a large amount of water (for the extraction, precipitate washing and neutralization steps) and generates a large volume of effluents (in the isoelectric precipitation and washing steps). With a high biochemical oxygen demand that occurs because of whey-like proteins that remain soluble in the pH range of 4.2 to 4.5, the effluent generated following the isoelectric precipitation step is especially problematic and constitutes a serious water pollution threat unless properly processed. The whey-like proteins found in this effluent are difficult to recover because the low solids concentration, varying from 1% to 3% [16], makes isolation of those proteins not economically viable.

10.1.3 Soy bioactive peptides It has been well-known for several decades that bioactive peptides can be derived from dietary proteins. Bioactive peptides may be present as independent entities or encrypted in the parent protein. During food processing or gastrointestinal digestion, these peptides are released from the parent protein and act as compounds with hormone-like activities [17]. In general, bioactive peptides derived from food contain two to nine amino acids [18], although this range may be extended to 20 or more amino acid units [17]. As an important protein source, soybean is also a potential source of bioactive peptides. As of 2005, ExPASy databases listed a total of 1411 protein entries (266 Swiss-Prot entries and 1145 TrEMBL entries) for soybean [19]. Acidic hydrolysis and enzymatic hydrolysis are the two main methods to generate soybean peptides. The acidic hydrolysis method is simple and less expensive but may result in amino acid damage. Enzymatic methods, in contrast, are easier to control, use mild conditions and do not cause amino acid damage. Therefore, enzymatic hydrolysis is the most commonly used method to produce food-grade protein hydrolysate and release bioactive peptides from their parent protein. The type of enzyme used for hydrolysis is very important, given that it will impact the biological activities of the generated peptides. Proteinases (endopeptidases) such as trypsin, subtilisin, chymotrypsin, thermolysin, pepsin, proteinase K, papain and plasmin are commonly used for the proteolysis of food proteins [20]. Enzymes are also often combined to produce bioactive peptides. Heat treatment can be applied to proteins prior to

10.2 Membrane technologies in the processing of soy protein products 

 235

enzymatic hydrolysis, because such treatment will impact the generation of the bioactive peptides. For example, it was observed that soybean meals that were heattreated at high humidities had higher levels of aggregated peptides [21]. Fermentation has also been considered for producing bioactive peptides. Bioactive peptides can be released by the microbial activity of fermented foods or through enzymes derived from microorganisms [17]. Interest in fermented soybean products, such as natto, tempeh, soy sauce and soy paste, has grown in recent years. However, fermentation is not enough to fully hydrolyse soybean proteins, and further enzymatic degradations are needed to produce peptides with high activities [22]. The bioactive peptides are then isolated using different techniques, such as ionexchange resins, different chromatography techniques and ultrafiltration. Salting-out and solvent extraction are often used before further purification steps. After centrifugation, the supernatant is filtered and lyophilized for liquid chromatography analysis. Chromatography is the most powerful technique available for isolating and purifying bioactive peptides. Different chromatography techniques can be used for peptide recovery, with high performance liquid chromatography (HPLC) the most common separation method. Commercially available reversed-phase columns allow for rapid separation and detection of the peptides in a mixture, whereas normal-phase liquid chromatography is preferred for the separation of hydrophilic peptides. Ion-exchange chromatography, capillary electrophoresis, and capillary isoelectric focus on separate peptides based on their charge properties, whereas size-exclusion chromatography is a separation method based solely on molecular size. Chromatography/ion-exchange resins are the techniques used for large-scale separation of bioactive peptides. The main limitation of those techniques is their high cost. Depending on the initial protein source, the enzyme that is used and the processing conditions, the biological activities of the peptides differ. When different enzymes hydrolyse soy protein, the protein yields antioxidant peptides [23], peptides with anti-cancer properties [24], or peptides with hypotensive activity [25]. Also, several bioactive peptides that inhibit angiotensin-converting enzyme (ACE) have been found in enzyme hydrolysates of soy proteins.

10.2 Membrane technologies in the processing of soy protein products To overcome the aforementioned disadvantages of the traditional isoelectric precipitation process, membrane technologies such as ultrafiltration (UF) and electrodialysis (ED) with bipolar membranes, used individually or in combination, have been considered for the production of soy protein isolates. In the following sections, the basic principles of various membrane technologies used in soy processing will be reviewed, and a few selected applications of membrane technologies for the production of soy protein isolates and the isolation of soy bioactive peptides will be briefly presented.

236 

 10 Production of value-added soy protein products by membrane-based operations

10.2.1 Ultrafiltration UF is a membrane separation process with important industrial applications for the concentration, purification and separation of colloidal or macromolecular species in solution. UF consists of the application of a transmembrane pressure to force a solution against a semipermeable membrane with a pore size ranging from 1 to 100 nm. Water and solutes in the feed stream that are below the NMWCO of the membrane are allowed to permeate through the membrane pores into a permeate stream, while the larger feed components are retained by the membrane (Figure 10.1). This step is presented in terms of volume concentration ratio (VCR = V0/VUF) where V0 is the starting volume and VUF the final volume after UF. UF systems can also be operated in continuous or in discontinuous diafiltration modes. Continuous diafiltration is achieved by adding distilled water to the feed solution at a rate equivalent to the permeate flux. This step is presented in terms of the volume permeated (VD = Vp/VDF), where Vp is the volume of permeate generated during the continuous diafiltration step and VDF is the volume prior to diafiltration. Diafiltration can also be carried out in discontinuous mode. In that case, after the UF step between V0 and VUF, the solution is rediluted to V0 and the concentration step is repeated. Each of these steps is represented by the VCR. The main advantages of UF in food applications are: mild operating conditions; product purification with concentration; separation of dissolved molecules as long as the appropriate membrane is used; the absence of a change in phase or state of the solvent during the process; and low equipment cost and energy requirements. UF is particularly attractive in protein systems, which are quite sensitive to extreme changes in the environment, such as the changes that would occur during thermal evaporation and concentration [26, 27].

Retentate Soy proteins and polysaccharides

UF Membrane P Water

Salts and sugars Permeate

Figure 10.1: Separation characteristics of the ultrafiltration membrane process

10.2 Membrane technologies in the processing of soy protein products  

 237

10.2.1.1 Membranes The membrane is the key component of an ultrafiltration process and is defined as a thin barrier through which solutes are selectively transported. Commercial UF membranes are made from inorganic materials, such as metals and ceramic materials, or prepared from organic polymers, such as cellulose acetate, polyamide, polyethersulfone, polysulfone and polyvinylidene difluoride [27, 28]. However, membranes made from inorganic materials, although widely used in the water industry, have very limited application in protein processing [29]. From a structural point of view, polymeric membranes can be classified into two types: symmetric (membranes that have similar structural morphology at all positions within them) and asymmetric. In practice, most membranes used in UF are asymmetric and are characterized by a thin skin on the surface of the membrane. The layers underneath the skin consist of a microporous layer that serves to support the skin layer. Rejection of the components occurs at the skin layer but, because of that layer’s unique ultrastructure, the retained components that are above the NMWCO do not enter the main body of the membrane, and consequently the membrane rarely gets plugged. However, asymmetric membranes are still susceptible to flux-lowering phenomena [27]. By definition, the sieving properties of asymmetric ultrafiltration membranes are expressed in terms of nominal rating. For a given membrane, the nominal rating refers to the molecular size or molecular weight of a solute above which a certain percentage of this solute in the feed stream will be retained by the membrane under controlled conditions [27]. UF devices can be classified into two configurations, dead-end or crossflow, based on the feed mass transfer characteristics of the membrane. In dead-end mode, the feed stream is pumped perpendicular to the membrane surface, resulting in the continuous build-up of the solutes on the membrane surface. In crossflow mode, the feed stream is pumped tangentially to the membrane surface, resulting in high shear rate and/or turbulence in the immediate vicinity of the membrane. The main advantage of crossflow UF is the fact that the solutes that tend to accumulate on the membrane surface are partly back-transported away from the membrane, making the process more efficient than the dead-end configuration. Consequently, industrial UF processes are usually carried out in crossflow mode. Different crossflow UF membrane modules are available on the market, including flat-sheet tangential flow, SW, tubular, and HF modules [29].

10.2.1.2 Membrane fouling Membrane fouling refers to the loss of membrane performance caused by the adsorption and deposition of components present in the feed stream on the membrane surface and within the pores. In addition to lowering the average permeate flux, membrane fouling may also result in an alteration of the membrane selectivity, which means that some components that would generally permeate the membrane may be

238 

 10 Production of value-added soy protein products by membrane-based operations

retained by the cake formed on its surface. Also, depending on the nature and extent of fouling, restoring the flux will eventually require extensive cleaning or replacement of the membrane [30]. In membrane systems, the cost of fouling is significant and can account for roughly 10–20% of operating costs during the first year, increasing by an additional 10% of operating costs annually under heavy fouling conditions [31]. In practice, the amount of permeate that passes through a unit area of membrane per unit of time (the flux) is directly related to the different hydraulic resistances. Evolution of the different hydraulic resistances is often described using Darcy’s law: J = ____  ΔP   μRG

(10.1)

where J is the permeate flux (m/s), ΔP is the pressure DF (Pa), μ is the permeate viscosity (Pa.s), and RG is the global resistance (/m). The global resistance represents the sum of the membrane resistance, the resistance caused by concentration polarization, the resistance caused by irreversible fouling, and the resistance caused by cake formation. The relative importance of the different resistances changes during UF, and their values are experimentally approximated from measurements of flux and transmembrane pressure with pure water.

10.2.1.3 Operating variables Among the various operating parameters that may affect the permeate flux, the most important ones are transmembrane pressure, temperature, feed concentration, and tangential flow velocity [32]. Theoretically, an increase in the applied transmembrane pressure should result in a proportional increase in the permeate flux. In practice, however, an increase in the applied transmembrane pressure results in a greater convective rate for the transport of solute to the membrane surface, which will in turn result in an increase in the global resistance and will limit the permeate flux increase. Assuming that there are no effects of temperature on membrane fouling, an increase in temperature will usually result in an increase in diffusivity and a decrease in feed stream viscosity and thus in higher permeate flux. The impact of an increase in feed concentration on the permeate flux is complex, given that an increase in feed concentration will alter the viscosity, density and diffusivity of the feed solution. Thus, in practice, permeate flux may increase with feed concentration, decrease, or remain the same. However, it is more common to observe a decrease in the permeate flux with an increase in feed concentration, because an increase in feed concentration will result in an increase in feed stream viscosity and density. Finally, an increase in tangential velocity will generally increase the permeate flux by lowering the concentration polarization and the probability of fouling owing to the greater turbulence observed near the membrane surface [27, 33]. The module configuration (flat sheets, tubular modules, HFs, plate units, SW modules) also affects membrane performance.

10.2 Membrane technologies in the processing of soy protein products  

 239

10.2.2 Electrodialysis ED is a membrane process in which an electrical potential difference is applied to an ion-exchange membrane, resulting in the migration of ions through the membrane and thus enabling the modification of the ionic composition of adjacent liquids. Conventional ED can be performed with two types of ion-exchange membranes: anion-exchange membranes, which have a fixed positive charge, and cationexchange membranes, which have a fixed negative charge. The fixed charge carried by the membranes allows them to facilitate passage of ions of the opposite charge (counter-ions) and repel ions of the same charge (co-ions). This exclusion, which is a result of electrostatic repulsion, is called “Donnan exclusion”. A special field of ED is bipolar-membrane ED, which uses special types of membranes known as bipolar membranes. Upon the application of a direct electrical potential, these membranes allow the electro-dissociation of water molecules into protons and hydroxyl ions and can be used to adjust the pH value of solutions. Advantages of the use of ED include low-energy consumption, product purification with no dilution, rapid and controlled salt removal from a product stream, ion substitution from the adjacent solution, and pH variation and adjustment with no addition of external solutions.

10.2.2.1 Conventional electrodialysis 10.2.2.1.1 Monopolar membranes Monopolar electrodialysis membranes are thin sheets or films of anion- or cationexchange resins reinforced with a thermoplastic polymer such as polyethylene, polypropylene, polystyrene, or another engineered polymers. The objective is always to produce a polymer that is chemically and thermally stable, has the appropriate mechanical properties (high stability but some flexibility), and contains sufficient ionic groups to ensure that the membrane has acceptable conductivity and selectivity [34]. The concentration and type of fixed charges on the polymer determines membrane permselectivity and electrical resistance. There are two types of ion-exchange membranes: cation-exchange membranes, which contain negatively charged groups fixed to the polymer matrix, and anion-exchange membranes, which contain positively charged groups [35]. For cation-exchange membranes, the fixed ionic groups are usually sulfonic acids, which completely dissociated over nearly the entire pH range, and carboxylic acids, which are virtually undissociated at pH values below 3. For anion-exchange membranes, the typical fixed ionic groups are quaternary ammoniums, which completely dissociated over the entire pH range, and secondary ammoniums, which are only weakly dissociated. The properties of ion-exchange membranes, such as electrical resistance, permselectivity, ion-exchange capacity, solvent transfer, and stability, determine to a large extent the technical feasibility and economic success of an industrial process.

240 

 10 Production of value-added soy protein products by membrane-based operations

10.2.2.1.2 Basic principles The principles of conventional ED are illustrated in Figure 10.2 for a dilutionconcentration configuration with alternating cation- and anion-exchange membranes placed between an anode and a cathode to form individual cells. A cell consists of a volume with two adjacent membranes. An industrial ED stack may have up to 200 cells stacked between the electrodes [35]. When an electrical potential is established between both electrodes, anions (Y−) move in the direction of the anode and permeate through the adjacent anionic membranes but are retained by the negatively charged cation-exchange membranes, while cations (X+) move in the direction of the cathode and permeate through the adjacent cationic membranes but are retained by the anion-exchange membranes. When several membranes are stacked in a dilutionconcentration configuration with alternating cation- and anion-exchange membranes, ion permeation through the membranes results in an ion concentration increase in compartments known as the concentrate compartments and an ion concentration decrease in the adjacent compartments known as the diluate compartments.

10.2.2.1.3 Operating variables The limiting current density is one of the first operating variables that must be determined. When an electrical current is passed through an ion-exchange membrane, the ion concentration on the surface of the membrane facing the diluate stream is decreased and eventually reduced to zero at the limiting current density. If the ion concentration of the diluting stream at the membrane surface is reduced to zero, there will be no more ions available to carry the electric current through the membrane.

C

A

C

X+

Y–

A

X+

Y–

Concen- Diluate Concentrate trate Cell pair Figure 10.2: Conventional electrodialysis cationic-anionic configuration A, anionic membrane; C, cationic membrane

10.2 Membrane technologies in the processing of soy protein products  

 241

The net result is a large increase in voltage drop, higher energy consumption, and the generation of water dissociation [36]. Water dissociation has consequences for the ED process, because it results in a loss of current utilization and drastic shifts in pH at membrane surfaces. For this reason, it is generally desirable to conduct ED at a current density below the limiting current density, at the concentration of maximum conductivity of the diluate stream, and at a maximum flow rate. ED is usually carried out at 80% of the limiting current density to prevent any undesirable consequences resulting from operating above it. Solutions used in ED must be free of suspended particles, and the feed solution can therefore be filtered to remove particulate material that may otherwise foul the surface of the ion-exchange membranes [37]. In terms of operating temperature, because a high process temperature will result in a low electrical resistance and lower viscosities of the fluids, the operating temperature should be as high as possible without impairing membrane and product integrity. Flow pressure during operation of the stack must be equal throughout and equilibrated with the pressure of the diluting stream to prevent transfer through pressure gradient. On an industrial scale, the ED stack can be operated on a once-through continuous basis, on a feed-and-bleed basis, or on a batch basis. In a continuous mode of operation, the feed stream is circulated through the stack only once, and the degree of ion removal rarely exceeds 50%. To obtain the desired degree of removal, some type of staging is required. This means that the treated solution has to go through a series of stacks until the required degree of depletion or concentration is obtained. In the feed-and-bleed mode of operation, some of the solution that has already been processed is pumped into the recirculation loop of the membrane system, so that a steady-state condition results. Recirculation of the processed solution is less efficient than once-through flow in terms of stack utilization and energy consumption, because the same solution must be pumped and electrodialysed repeatedly. Batch recirculation of the diluate stream avoids some of the inefficiencies of the feedand-bleed operation mode. In batch operation, the solution to be electrodialysed is placed in a reservoir, circulated through the stack, and recirculated to the reservoir until the desired degree of ion removal or enrichment is attained. One advantage of the batch operation mode is the fact that the process cycle can be adjusted toward the desired end product and is independent of feed concentration [34].

10.2.2.2 Bipolar membrane electrodialysis 10.2.2.2.1 Bipolar membranes and basic principles Bipolar membranes are composed of two layers of ion-exchange membranes (one anionic layer and one cationic layer) joined by a hydrophilic junction also known as the transition layer. When a direct electrical potential is applied, the diffusion of water from both sides of the bipolar-membrane allows the electro-dissociation

242 

 10 Production of value-added soy protein products by membrane-based operations

of water molecules into protons and hydroxyl ions, which further migrate from the transition layer through the cationic exchange layer toward the cathode for the H+ and through the anionic exchange layer toward the anode for the OH−. Under an applied electrical potential field, the water dissociation in the transition layer of the bipolar membranes is accelerated 50 million times in comparison to the rate of water dissociation in aqueous solutions due to the Wien effect [35]. ED with bipolar membranes can therefore be advantageously used to adjust stream pH without the external addition of acids or bases, which are often sources of impurities [34].

10.2.2.2.2 Operating variables A typical bipolar-membrane ED configuration is the three-compartment cell illustrated in Figure 10.3. In a three-compartment cell, the basic repeating unit consists of a cation-exchange membrane, a bipolar-membrane, and an anion-exchange membrane [38]. In this configuration, when an electrical potential is applied across the ED stack, the cations (X+) present in the diluate compartment (between the anion-exchange membrane and the cation-exchange membrane) move in the direction of the cathode, permeate through the cationic membranes and combine with the OH− ions, generated by the bipolar membranes, to form the corresponding base (i.e., XOH). At the same time, the anions (Y−) move in the direction of the anode, permeate through the anionic membranes and combine with the H+ ions, generated by the bipolar membranes, to form the corresponding acid (i.e., HY). The energy required in bipolar-membrane ED is the sum of (i) the energy required for water splitting in the transition layer of the

BP

A

C

X+

BP

X+ OH–

H+ Y–

Y–

Acid

Salt solution

Base

Repeating cell unit Figure 10.3: Bipolar membranes three-compartment electrodialysis configuration A, anionic membrane; BP, bipolar-membrane; C, cationic membrane

10.2 Membrane technologies in the processing of soy protein products  

 243

bipolar-membrane, and (ii) the electrical energy required to transfer the salt ions from the feed solution as well as the protons and hydroxyl ions from the transition layer of the bipolar-membrane into the acid and base compartments [35]. In the food industry, a significant portion of these applications are directed to the regeneration of acids and bases that have been used in a chemical process either for neutralization or regeneration of ion-exchange resins [39]. Two-compartment cells can also be used where it is not practical to use threecompartment cells, such as cases where high purity of both the acid and base is not possible to obtain or may even generate problems during the process. ED with two-compartment cells may use a stack configuration with alternating cationic and bipolar membranes or alternating anionic and bipolar membranes. ED with bipolar membranes is operated at a higher current density than conventional ED, and the voltage drop across a cell unit is also higher [35]. Consequently, a limited number of cell units can be used in a stack because of the heat that is generated and must be dissipated.

10.2.3 Integrated electrodialysis-ultrafiltration process Conventional ED has some serious limitations related to the size of the ionic species that can migrate through the ion-exchange membranes. In order to migrate through the ion-exchange membranes, the ionic species must have a radius or molecular weight that does not exceed a certain limit allowed by the membrane porosity [40, 41]. The idea to replace one or more ion-exchange membranes with a porous membrane acting as a molecular barrier has proven to be profitable, because it allows the separation of charged molecules with a molecular weight higher than the weight allowed by the traditional ion-exchange membranes [42, 43]. Separation that appeared impossible to carry out with ED using conventional ion-exchange membranes was achieved when some ion-exchange membranes were replaced by porous membranes [44]. The use of a conventional ED cell, in which some ion-exchange membranes were replaced by ultrafiltration membranes (ED-UF), was reported for the fractionation of bioactive peptides from different sources [45–49]. In the present section, only the configuration used for the separation of soy peptides will be presented. Part of the cell stack used for the fractionation of soy peptides is shown in Figure 10.4. An ED cell consisting of an anionic membrane located at one extremity of the stack to separate the electrode rinsing solution from the KCl1 receiving solution, a first UF membrane to separate the KCl1 receiving solution from the hydrolyzed protein solution, a second UF membrane to separate the hydrolyzed protein solution from the KCl2 receiving solution, and a cationic membrane to separate the KCl1 and the KCl2 receiving solutions. The sequence consisting of: KCl1 receiving solution compartment → UF membrane → hydrolyzed protein compartment → UF membrane → KCl2 receiving solution

244 

 10 Production of value-added soy protein products by membrane-based operations

A

UF

UF

C

UF

UF

X+

E

KCI1

F.H.

KCI2

Y–

Y–

KCI1

F.H.

C

X+

KCI2

E

Repeating cell unit Figure 10.4: Integrated electrodialysis-ultrafiltration configuration A, anionic membrane; C, cationic membrane; UF, ultrafiltration membrane

compartment → cationic membrane, can be repeated in order to increase the membrane surface area available for the separation. A cationic membrane is located at the other extremity of the stack to separate the KCl2 solution from the electrode rinsing solution [49]. Theoretically, negatively charged peptides should migrate toward the anode into the KCl1 compartment, and positively charged peptides should migrate toward the cathode into the KCl2 compartment. Compared to conventional pressure-driven processes (UF, nanofiltration), the integrated ED-UF process would have better selectivity and a lower tendency toward membrane fouling [48, 50]. However, one possible limitation is the transport of water through the semipermeable UF membranes separating the hydrolysate solution and the receiving solutions, from the solution that is dilute in solute to the solution that is concentrated, a phenomenon known as osmosis. It can therefore be expected that the concentration of peptides in the receiving solutions will be limited by the gradual transfer of water from the hydrolysate solution.

10.3 Production of soy protein isolates by membrane technologies 10.3.1 Ultrafiltration UF has been suggested as a possible means of overcoming the problems encountered with the traditional isoelectric precipitation process. Porter and Michaels [51] first suggested using UF to process soy protein extracts, and a number of researchers have

10.3 Production of soy protein isolates by membrane technologies 

 245

since investigated different aspects of the use of ultrafiltration to purify soy extracts [7, 14, 26, 52–76]. Different processes using UF to produce soy protein isolates or concentrates have been considered. Generally, however, ultrafiltration is applied in place of the isoelectric protein precipitation step and the washing step. The protein extraction step, using soy flour or soy flakes, is the same as in the conventional process, and the spray-drying step is still required to obtain the final isolate in dried form. An exception is the work of Vishwanathan et al. [76], who used UF membranes to eliminate non-protein substances from okara protein extract (okara is a by-product of the processing of soybean for soy milk). UF membranes with various MWCO values have been considered for the concentration or purification of soy protein extracts, but a 50 kDa MWCO seems to represent a good trade-off between a high permeate flux and high protein rejection. One of the advantages of UF over the isoelectric precipitation step is the fact that UF recovers essentially all of the solubilized protein and avoids the generation of whey-like products, resulting in increased protein recovery [68]. Another advantage is the fact that the undesirable components of soy protein extracts, such as oligosaccharides and phytic acid, can be selectively separated from the proteins as long as the proper membrane and operating parameters have been chosen [7, 14, 54, 55, 63]. In addition, isoflavones can be recovered in high amounts in the isolate [74, 75]. It has also generally been reported that soy protein isolates or concentrates produced by UF have better functional properties than the corresponding products of isoelectric precipitation [67, 70, 71, 73]. Nevertheless, as previously discussed, the main disadvantage of using the UF process to produce soy protein isolates remains membrane fouling.

10.3.1.1 Removal of undesirable components of soy protein extracts 10.3.1.1.1 Phytic acid Phytic acid occurs at fairly high levels in soy, as a soluble salt at concentrations of 1% to 2.3%, dry basis [77, 78], and represents roughly 70% of the total phosphorus in soy [54, 79, 80]. Phytic acid is a moderately strong acid with six phosphate groups and 12 exchangeable protons: six protons that are strongly dissociated and have a pKa of 1.1–2.6; two protons that are dissociated and have a pKa of 4–6.5; and four protons that are very weakly dissociated and have a pKa higher than 8 [79, 80, 81]. Dissociated phosphate groups confer a negative charge over the entire pH range, which allows phytic acid to bind with positively charged molecules, including mineral cations and proteins at pH values below their isoelectric points. For pH values above the protein isoelectric point, the formation of a ternary complex involving protein, divalent cations (Ca2+ and Mg2+) and phytic acid is supported by the results of several studies [14, 54, 82]. The relative importance of this ternary complex is difficult to establish, however, because it is dependent on the relative amount of divalent cations and on the pH. Phytic acid can also interact with monovalent or divalent cations to form a salt.

246 

 10 Production of value-added soy protein products by membrane-based operations

Although some research suggests that phytic acid may have some benefits in human nutrition, including anti-carcinogenic and antioxidant effects [83, 84], the adverse effect of the formation of the aforementioned complexes on the digestibility of proteins [80] and the bioavailability of minerals [85, 86] in the intestine has been recognized. The interactions of soy proteins with phytic acid also modify the functional properties of the soy isolate, including its solubility [79]. In general, the solubility of soy proteins for pH values below the isoelectric point of the proteins is reduced when a high amount of phytic acid is present in the isolate. The interactions between phytic acid and the proteins also explain why isolates prepared by isoelectric precipitation at pH 4.5 contain 60–70% of the phytic acid present in soybeans [79]. For the above reasons, significant efforts have been made to reduce the level of phytic acid in soy products. Different techniques have been considered to achieve this goal, including the use of ion-exchange resins [87, 88], the addition of the metallic cations Ca2+ and Ba2+ [88], and enzymatic hydrolysis followed by UF-diafiltration [54, 89]. UF-diafiltration alone has demonstrated good phytic acid removal effectiveness and good protein purification [14, 54, 90, 91, 92]. Those authors found that phytic acid removal depends both on the pH of the soy extract and on the UF-diafiltration conditions. Phytic acid removal appears to be optimal within a pH range of about 5–6.7.

10.3.1.1.2 Oligosaccharides The main objective in producing soy protein isolate is to remove the oligosaccharides from the soy protein extract, thereby increasing the protein content of the extract to at least 90%, dry basis. Because oligosaccharides are smaller in molecular size than proteins, UF can selectively remove these undesirable components. The major oligosaccharides in soy that are to be removed from the extract are sucrose, raffinose and stachyose. In their work, Omosaiye et al. [63] used a 50 kDa HF membrane to purify a soy protein extract. Their results indicated that the rates of oligosaccharide removal during UF closely followed theoretical behavior for a non-rejected solute. The concentrations of sucrose, raffinose and stachyose in the retentate remained practically constant or increased only slightly during UF. Skorepova and Moresoli [91] also reported similar observations. They used a high shear tangential flow HF UF module with a 100 kDa membrane to purify a soy protein extract that was electroacidified (pH 6) or non-electroacidified (pH 9). They reported that the removal of carbohydrates during filtration was always consistent with the theoretical predictions (based on free permeability assumption) for both the electroacidified and the non-electroacidified feeds.

10.3.1.2 Production of soy protein isolate with a high amount of isoflavones It is well-known that soy is an excellent source of isoflavones, which have been suggested to possess anti-carcinogenic [93–98] and antiosteoporotic [95, 99, 100] properties,

10.3 Production of soy protein isolates by membrane technologies 

 247

to help in reducing cardiovascular risk factors [95, 99, 101] and to aid in treating menopausal symptoms [95, 102, 103]. It would therefore seem valuable to produce soy protein isolate with high levels of isoflavones. Unfortunately, the conventional isoelectric precipitation process used to produce soy protein isolate causes much of the isoflavones to remain solubilized following the acidic protein precipitation step, and they are thus discarded [75]. However, Singh [75] reported that the use of ultrafiltration with membranes that have an MWCO between 5 and 30 kDa, to produce soy protein isolate, helped retain a significant fraction of the isoflavones present in the extract, even though isoflavones typically have a molecular weight less than 1500 Da. It is believed that isoflavones may be retained because of their complexation with proteins. Batt et al. [74] also reported high retention of isoflavones (at least 82% recovery) when soy protein concentrates were produced by UF-diafiltration using a membrane system (300 kDa) consisting of two TiO2-coated stainless steel tubular parallel-pass ultrafiltration modules.

10.3.1.3 Functionality of soy protein isolate produced by ultrafiltration Functionality is defined as any property of a food or food ingredient, with the exception of its nutritional properties, that affects its use. It is well-known that the development of new food products depends on knowledge of the functional properties of individual proteins as an important step towards their evaluation and use. Consequently, the functional characteristics of soy proteins in either concentrate or isolate form play a major role in determining their acceptability as ingredients in prepared food products. The main functional properties that are evaluated for new product development are solubility, ability to emulsify, ability to bind water or fat, and ability to form foams or gels. These properties are intrinsic physicochemical characteristics that affect the behavior of proteins in food systems during processing, manufacturing, storage and preparation. Among the aforementioned functional properties, solubility is an excellent indication of protein functionality. In general, other functional properties are positively correlated to the aqueous solubility of proteins. The positive correlation between solubility and a protein’s ability to function as an emulsifier, gelling agent and viscosity builder has been reported in many studies. Factors such as pH, ionic strength and the presence of antinutritional factors, including phytic acid, will affect the functional properties of proteins, although the protein structure (native vs. denatured) is the most important factor. In this context, membrane processing such as UF seems to be a valuable process for keeping the proteins in their native state and producing soy protein isolates with good functional properties, in comparison with the traditional isoelectric precipitation process [67, 71, 73]. Manak et  al. [67] compared the functional properties of soy protein isolates processed using UF-diafiltration with the functional properties of a commercial soy protein isolate. For UF, a membrane with a 10 kDa MWCO (PM10; Romicon Inc.,

248 

 10 Production of value-added soy protein products by membrane-based operations

Woburn, MA, USA) was used, and the extracts, which were prefiltered to 20 μm, were concentrated by applying a VCR of 4.5 prior to a discontinuous diafiltration step consisting of adding either one or two volumes of filtered tap water to the concentrated feed and then reconcentrating it. Another approach consisted of a continuous diafiltration step in which filtered tap water was added to the extract during concentration, at the same rate as the UF permeate was being removed. Water addition began after the initial extract volume had been reduced by 60% and continued until the total volume of the ultrafiltration permeate to be recovered at the end of the run was equal to 1.5 times that of the initial extract volume. The results indicated that all the isolates had protein contents equal to or greater than the commercial isolate (92.3–94.6%, dry basis, as compared with 91.8%). The soy protein isolates produced by membrane technologies exhibited higher nitrogen solubilities than the commercial isolate did (87.7–100.0 as compared with 67.3). The authors attributed this enhancement in solubility to the inclusion of highly soluble whey-like proteins in the isolates produced by UF-diafiltration. In terms of foaming properties, all the membrane-isolated soy proteins yielded foam viscosities in excess of 200,000 cps as compared with 33,000 cps for the commercial isolate, as well as volume increases of approximately twice that of the commercial isolate. The emulsifying capacity was also superior for the soy protein isolates produced by UF-diafiltration as compared with the commercial isolate. The membrane-isolated proteins emulsified from 13.45 to 17.32 ml of oil per ml of solution containing 0.625% protein as compared with 9.22 ml of oil for the commercial soy isolate. The membrane-isolated proteins also demonstrated a higher fat adsorption capacity than their commercial counterpart (1.71–2.52 ml of oil/g of isolate as compared with 1.57). In terms of gel strength, the commercial isolate demonstrated similar or greater strength as compared with the soy protein isolates produced by UF-diafiltration. A comparative study on the solubility, water hydration capacity and emulsifying properties of a commercial acid-precipitated soy protein isolate and a soy protein concentrate produced by UF was carried out by Rao et al. [71]. The membrane-isolated concentrate was produced using the method of Shallo et al. [89] with 300 kDa tubular UF membranes (Graver Separations, Inc., Glasgow, DE, USA). For the pH range 3–10, statistical analysis of the nitrogen solubility means for both soy protein ingredients indicated that the acid-precipitated isolate had lower solubility, regardless of the pH. This lower solubility in the commercial isolate was attributed to the more severe denaturation that may have occurred during processing. In contrast, the water hydration capacity of the commercial acid-precipitated isolate was higher than that of the soy protein concentrate produced by UF (5.64 ± 0.02 vs. 2.61 ± 0.09 g of water/g of dry product). In terms of emulsifying properties, the researchers reported that, as compared with the concentrate, the commercial isolate exhibited a slightly higher emulsifying activity index (7.27 ± 0.09 vs. 6.0 ± 0.05 m2/g) but a significantly lower emulsion stability index (23.12 ± 0.35 vs. 41.91 ± 0.88 m2/g). The higher emulsifying activity observed for the commercial

10.3 Production of soy protein isolates by membrane technologies 

 249

isolate was attributed to the fact that the isolate was more hydrophobic than its membrane-purified counterpart. Hojilla-Evangelista et  al. [73] also reported superior functional properties for soy proteins isolated by UF-diafiltration as compared with those isolated by isoelectric precipitation. For UF, a membrane with a 5 kDa MWCO was used, and the extracts, which were prepared by adapting the method of Sessa [104], were concentrated by applying a VCR of 6.67 prior to a discontinuous diafiltration step consisting of adding water (50% of the initial volume prior to UF) to the concentrated feed and then reconcentrating it. The procedure for recovering proteins by isoelectric precipitation was adapted from the method of Thanh and Shibasaki [105]. The researchers found that UF-diafiltration produced only protein concentrates (73% protein, dry basis), whereas isoelectric precipitation produced protein isolates (about 90% protein, dry basis). The soy proteins isolated by UF-diafiltration showed markedly higher solubility values up to pH 7.0 than the isoelectrically precipitated soy protein isolate (65% vs. 45%). Surface hydrophobicity index and emulsifying activity index values were also superior for the membrane-isolated proteins (844.3 ± 24.6 vs. 529.5 ± 21.7 and 98.7 ± 4.2 vs. 56.0 ± 3.3 m2/g, respectively). Foam capacity and emulsion stability index values were similar for both ingredients (144 ± 3 vs. 131 ± 7 and 15.0 ± 0.2 vs. 15.0 ± 1.5, respectively), whereas the isoelectrically precipitated proteins showed better foam stability than the membrane-processed product (95.0 ± 1.8% vs. 77.4 ± 3.0%). From the aforementioned results, it is possible to conclude that UF-diafiltration generally had no adverse effects on, and in most cases even improved, the functional properties of soy protein concentrates or isolates, as compared with soy proteins isolated by the traditional isoelectric precipitation process.

10.3.2 Electrodialysis with bipolar membranes Bazinet et al. [106] were the first researchers to consider the use of ED with bipolar membranes to isoelectrically precipitate soy proteins. Their approach is similar to the traditional isoelectric precipitation process with the exception that the pH of the soy extract is adjusted to the protein isoelectric point (pH 4.2–4.5) by ED with bipolar membranes instead of by means of a mineral acid, such as HCl. Specifically, the pH adjustment is achieved using a bipolar-cationic configuration. The soy protein extract is pumped into the compartments receiving the H+ ions generated by the bipolar membranes. The protons come into contact with the protein stream, bringing the proteins to their isoelectric point and causing them to precipitate [107]. In order to counterbalance the H+ ions generated at the bipolar membranes, cations present in the protein stream permeate through the cationic membranes, therefore keeping the protein solution electrically neutral [108]. The permeating cations form the corresponding bases with the OH− generated by the bipolar membranes. The precipitated proteins can then

250 

 10 Production of value-added soy protein products by membrane-based operations

be recuperated by centrifugation, washed, neutralized and spray-dried in the same manner as in the conventional process. Bazinet et  al. [109] studied the effects of soy protein concentration (15, 30 and 60 g/l) in combination with KCl solution concentration (0.06, 0.12 and 0.24 M) on the performance of bipolar-membrane electroacidification. In their study, they used a bipolar-cationic configuration with the soy protein solution fed into the compartments receiving the protons and a KCl solution fed into the compartments receiving the hydroxyl ions. A 20 g/l Na2SO4 solution was circulated at the electrodes. The temperature of the electrolytes was maintained at 20°C, and the current density was 25 mA/cm2. That study found that increasing the soy protein concentration from 15 to 60 g/l and the KCl concentration from 0.06 to 0.24 M decreased the energy consumption (excluding the pumping energy) from 0.693 kWh/kg of isolate for a soy protein concentration of 15 g/l with 0.06 M KCl to 0.453 kWh/kg of isolate for a soy protein concentration of 60 g/l with 0.24 M KCl. In another study, the same authors [110] examined the effects of temperature on the performance of bipolar-membrane electroacidification. The ED cell used in that study was the same as the one in the authors’ earlier study [109]. The concentration of the soy protein solution was 30 g/L, and the KCl concentration was 0.06 M. Three different temperatures, namely 10°C, 20°C and 35°C, were considered. A drop in energy consumption occurred as the temperature of the soy extract was increased (0.728 kWh/kg for a temperature of 10°C, 0.455 kWh/kg for a temperature of 20°C, and 0.371 kWh/kg for a temperature of 35°C). The results of these studies indicate that soy protein concentration and temperature are the two primary factors influencing the energy required for the generation of the H+ and OH− and for the transport of ions through the cationic membranes. There are several advantages to using ED with bipolar membranes rather than the conventional acidification step with a mineral acid for the isoelectric precipitation of soy proteins. Because ED with bipolar membranes does not use any chemical acids during the protein precipitation step, this technology generates less effluent than the traditional approach. In addition, since the protein extract is partially demineralised during the acidification step, the final isolate from ED with bipolar membranes has a superior chemical composition, including a lower salt content, compared to soy protein isolates produced by the traditional isoelectric process. However, the industrial-scale application of ED with bipolar membranes for soy protein isoelectric precipitation is limited because of gradual protein precipitation in the cell [111]. That precipitation results in increased cell resistance (decreased system efficiency), causes protein losses (decreased yield) and complicates the passage of the soy protein extract through the cell. An on-line centrifugation step has been proposed by Bazinet et  al. [107] to allow the recovery of precipitated proteins and decrease cell fouling. Pourcelly and Bazinet [111] have also suggested that there is a need to optimize the hydrodynamic design of the ED cell in order to minimize fouling. More recently, Mondor et al. [90, 112, 113], Alibhai et al. [114], Skorepova and Moresoli [91] and Ali et al. [92, 115] have proposed a combination of ED with bipolar membranes and UF-diafiltration to decrease this fouling and produce soy protein isolates with low phytic acid content.

10.3 Production of soy protein isolates by membrane technologies 

 251

10.3.3 Electrodialysis with bipolar membranes in combination with ultrafiltration-diafiltration Mondor et al. [90, 112] were the first to consider the combination of ED with bipolar membranes and UF for the laboratory-scale production of soy protein isolates with low phytic acid content. When compared to the conventional isoelectric precipitation process, the new process would use the same protein extraction step, but the protein precipitation and washing steps would be replaced by a combination of ED with bipolar membranes (to adjust the pH of the soy extract to 6–7) and UF-diafiltration (to concentrate and remove some minerals and sugars). The neutralization to pH 7 (if required) and the spray-drying step would not be modified. One advantage of adjusting the pH range to 6 to 7 instead of 4.5 (the isoelectric point of the proteins) is the fact that fouling of the ED cell is greatly minimized, given that protein precipitation at these pH levels is minimal. Also, as previously mentioned, it is in this pH range that phytic acid, an antinutritional factor, is found mainly in its free form, making its removal by UF optimal. ED experiments were carried out using a bipolar-cationic configuration consisting of four cationic membranes (Neosepta CMX, Tokuyama Soda Ltd., Tokyo, Japan), including two at the extremities of the stack, and three bipolar membranes (Neosepta BP-1, Tokuyama Soda Ltd.). This arrangement created three closed loops containing the soy protein extract (3.5 l; 70 g/l), a 0.1 M KCl solution (8 l) and a 20 g/l Na2SO4 solution (8 l) used as a rinsing solution for the electrodes. Each closed loop was connected to a separate external reservoir, allowing for continuous recycling. The electrolyte temperature and current were controlled at 30–35°C and 1–2 A, respectively. Using a dead-end laboratory-scale UF cell equipped with a 100 kDa regenerated cellulose membrane and operated at 30 psi, Mondor et al. [112] demonstrated that a VCR of 2.5 for the ultrafiltration step followed by a discontinuous diafiltration step with a VCR of 2.8 enabled the production of a concentrate with 84.7% protein (dry basis) and only 3.8% ash when the pH of the extract was adjusted to 6 by ED with bipolar membranes prior to the UF-diafiltration steps. In comparison, the same UFdiafiltration sequence resulted in a concentrate with a protein content of 87.6% (dry basis) and 7.5% ash when the pH of the extract was 9. In addition, the phosphorus content of the concentrates was 5.98 and 11.57 mg/g of dry concentrate for extract pH levels of 6 and 9, respectively. Given that phytic acid is the main source of phosphorus in soy, these results suggest that phytic acid removal was better with the purification sequence carried out at pH 6 compared to the sequence carried out at pH 9. It was also shown that the solubility profile of the concentrate was improved (from 25% to 70%) for the pH range below the protein isoelectric point for the concentrate produced at pH 6 vs. the concentrate produced at pH 9. In another study, Alibhai et al. [114] used a low-shear tangential flow UF system equipped with flat-sheet polyethersulfone membranes with 100 or 200  kDa as the MWCO. The feed consisted of a soy protein extract that was non-electroacidified (pH 9) or electroacidified (pH 6) [90] and contained approximately 60% protein, 10% ash and 30% sugars, at 2 or 4 wt%. Those researchers applied a VCR of 2 during the UF

252 

 10 Production of value-added soy protein products by membrane-based operations

step and used a continuous diafiltration step with a dilution volume (VD) of 0.67, which corresponds to the addition of a volume of water equivalent to 67% of the initial feed volume. Their results indicate that the electroacidified extract (pH 6) caused more fouling than the extract at pH 9, with a significant contribution associated with reversible fouling (cake formation). This observation suggests that filtration improvement, i.e., fouling reduction, could be achieved by means of a high shear tangential flow UF system. In this context, Skorepova and Moresoli [91] used a high shear tangential flow HF UF module equipped with a 100 kDa membrane to process a soy protein extract that was non-electroacidified (pH 9) or electroacidified (pH 6) [90]. Filtration performance was evaluated by comparing the filtration time and the final product composition for a UF-diafiltration sequence with a VCR of 4.5 for the UF step and a VCR of 4 for the discontinuous diafiltration step. The removal of carbohydrates during filtration was always consistent with the theoretical predictions (based on free permeability assumption) for both feeds. However, higher removal of calcium, magnesium and phytic acid (estimated as total phosphorus) was achieved during the filtration of the electroacidified feed compared to the non-electroacidified feed. For the extract at pH 6, the phosphorus content was reduced by 57.2%, whereas a reduction of only 13.7% was observed at pH 9. However, the removal of phosphorus at pH 6 was still lower than the theoretical expectations based on free permeability (i.e., 88%), indicating that the phosphorus is somewhat retained by the membrane but to a much lower extent than at pH 9. These results suggest that, for the extract at pH 9, phytic acid permeation through the membrane is limited as compared to permeation observed for the extract at pH 6. At pH 9, most of the phytic acid molecules would be associated with the protein via the formation of a ternary complex [82], leading to the inability of the phytic acid to permeate through the membrane. In contrast, for the extract at pH 6, the conditions for the formation of a ternary complex are limited (the proteins are negatively charged but less so than at pH 9), and a significant fraction of the phytic acid is in free form, which facilitates its removal. From a negative point of view, the electroacidification pretreatment had a negative impact on the permeate flux and resulted in more significant membrane fouling with correspondingly longer filtration times. Discontinuous diafiltration enhanced the removal of carbohydrates and minerals, thus yielding a product with higher protein content, but was unable to improve the permeate flux for the electroacidified feed. Recently, Mondor et al. [113] studied the impact of four different UF-diafiltration sequences with a total permeate volume of 1.5–1.6 times the initial volume on membrane fouling and permeate flux as well as on isolate composition. The sequences that they investigated were as follows: a VCR of 2 followed by two-time discontinuous diafiltration with a VCR of 2, i.e., (re-VCR 2) × 2; a VCR of 2 followed by continuous diafiltration with a VD of 2; a VCR of 5 followed by discontinuous diafiltration with a VCR of 5, i.e., re-VCR 5; and a VCR of 5 followed by continuous diafiltration with a VD of 4. Experiments were carried out using a high shear tangential flow HF UF module equipped with a 100 kDa membrane to process an electroacidified (pH 6) extract. The

10.3 Production of soy protein isolates by membrane technologies 

 253

pH 6 electroacidified extract was produced as described by Mondor et al. [90] with the exception that the starting soy protein extract consisted of 4 l of a 30 g/l protein solution, and the electrolyte temperature and current density were controlled at 30°C and 1.43 mA/cm2, respectively. The results indicate that the VCR 5–VD 4 sequence showed the most severe fouling and consequently the greatest permeate flux decline. At the same time, this sequence achieved the most efficient purification. When compared to the phosphorus/protein ratio of the starting extract (i.e., 12.38 ± 0.07 mg P/g protein), the phosphorus/protein ratio of the isolates was decreased by 76% for the VCR 5–VD 4 sequence, 72% for the VCR 5–re-VCR 5 sequence, 61% for the VCR 2–VD 2 sequence, and 45% for the VCR 2–(re-VCR 2) × 2 sequence. These observations suggest better phytic acid removal with the VCR 5–VD 4 sequence than with the other three sequences. In agreement with the above observations, the isolate produced from the VCR 5–VD 4 sequence showed the highest solubility, especially for the pH range of 2 to 4. The results of the aforementioned works demonstrate the advantages of combining ED with bipolar membranes and UF-diafiltration. When compared to isoelectric precipitation of the proteins by electroacidification, the combined method results in limited fouling of the ED cell when the pH of the extract is adjusted to 6, but that fouling is still not negligible. Therefore, Ali et al. [115] studied the impact of the addition of KCl (0.12 or 0.24 M) to the starting extract on the efficiency of the electroacidification process as well as on the subsequent purification by UF-diafiltration. The solubility of the resulting soy protein isolates was also studied. ED experiments were carried out as described by Mondor et al. [113]. A high shear tangential flow HF ultrafiltration module equipped with a 100 kDa membrane was used to process the electroacidified (pH 6) extract with or without added KCl, and a VCR 5-re-VCR 5 UFdiafiltration sequence was applied. The results indicate that the addition of KCl to the initial soy protein extract at pH 9 improved protein solubility, making it possible to adjust the pH of the extracts from 9 to 6 while also preventing fouling of the spacers by precipitated proteins. Furthermore, the addition of KCl to the starting extracts had the effect of doubling ED productivity, from 0.094 kg to 0.205 kg of pH 6 extract per m2 of membrane. This increase in productivity was attributed to the increase in conductivity of the starting extract as a result of KCl addition. Another benefit of KCl addition was the fact that the amount of energy required to lower the pH of the extract from 9 to 6 was decreased from 0.331 to 0.139 kWh/kg of pH 6 extract per m2 of membrane after the addition of KCl. The reduction in the amount of energy required was due mainly to the increase in the speed of electroacidification with the addition of KCl, which resulted in a significant decrease in pumping energy. For the UF-diafiltration process, the pH 6 extracts with added KCl showed a higher permeate flux for both steps. The permeate flux during ultrafiltration-diafiltration was improved by as much as 20% for the extracts with added KCl as compared to the extract without added KCl. These differences in permeate flux were attributable in part to the differences in protein size distribution, with larger aggregates forming at pH 6 with added KCl than at pH 6 without added KCl. In terms of the composition of the isolates, the protein

254 

 10 Production of value-added soy protein products by membrane-based operations

content was the same for all three pH 6 isolates and was not affected by the addition of KCl to the extract prior to the pH adjustment step by ED with bipolar membranes. However, the amount of ash in the pH 6 isolate produced without the addition of KCl to the extract was two times lower than the amount in the isolates produced with the addition of KCl to the extract (2.7 ± 0.0 for the pH 6 isolate, 5.6 ± 0.1 for the pH 6 isolate with 0.12 M KCl, and 5.9 ± 0.1 for the pH 6 isolate with 0.24 M KCl). The phosphorus/ protein ratio was also significantly lower for the pH 6 isolate compared to the pH 6 isolate with 0.12 M KCl and the pH 6 isolate with 0.24 M KCl (3.5 ± 0.1, 5.6 ± 0.1 and 5.1 ± 0.3, respectively). Finally, the solubility profiles of the different pH 6 isolates indicated that the solubility of the pH 6 isolates produced with added KCl was lower than the solubility of the pH 6 isolate produced without added KCl for the pH range 2–3.5. As explained by Ali et al. (2010), this difference is probably attributable to the small reduction in phytic acid content due to the greater protein precipitation in the electroacidification cell during preparation of the pH 6 soy protein isolate as compared to the pH 6 isolate with 0.12 M KCl and the pH 6 isolate with 0.24 M KCl. From an environmental point of view, the main advantage of combining ED with bipolar membranes and UF-diafiltration is the fact that the volume of water required in the process is reduced compared to both of the isoelectric precipitation processes (conventional and with bipolar membranes). Also, the amount of proteins in the UF permeate is in the order of only 0.016% w/w (adapted from Skorepova and Moresoli [91]) when a 100 kDa UF membrane is used to process, to a VCR of 4.5, a soy extract with 1.8% total solids and for which the pH was adjusted to 6 using ED with bipolar membranes. This value is significantly lower than the 1–3% reported for the supernatant generated during the conventional isoelectric precipitation step [16] and the 1.5–1.7% reported for the supernatant after isoelectric precipitation by ED with bipolar membranes [116]. Another advantage to using ED with bipolar membranes rather than isoelectric precipitation is the fact that cell fouling is minimized, given that most of the soy proteins remain soluble at the pH range 6–7. This was found to be especially true when KCl was added to the pH 9 extract (0.12 or 0.24 M KCl) prior to the pH adjustment to 6 [115]. In terms of product functionality, the combination of ED with bipolar membranes to adjust the pH of the extract to 6 and UF resulted in a soy protein isolate with improved solubility characteristics compared to a soy protein isolate produced by the traditional approach. The main limitation of the combined process is the declining permeate flux observed during UF.

10.4 Separation of soy peptides by membrane technologies Not only are soy protein isolates of interest for their high protein content and their functional properties, but they can also be used as a source of nutraceutical peptides by the food and pharmaceutical industries [19, 117]. Fractionation and purification processes can be used to select specific peptides based on their physico-chemical

10.4 Separation of soy peptides by membrane technologies 

 255

properties. Such technologies may include chromatography, ion-exchange resins, barometric membrane processes such as UF and nanofiltration (NF) and, more recently, the integrated ED-UF approach [49].

10.4.1 Ultrafiltration UF seems to be a commonly used technology for separating peptides according to their molecular weights, and it allows a large amount of purified peptide fractions to be produced [118]. A typical approach involves the sequential UF of a soy protein hydrolysate, in which the hydrolysate is processed using a membrane with a given MWCO in order to produce an initial concentrate, concentrate 1, and an initial permeate, permeate 1. Then, permeate 1 is processed using a second membrane with a MWCO lower than that of the first membrane to obtain concentrate 2 and permeate 2. The procedure can be repeated any number of times using a membrane with a lower MWCO for each successive sequence. It is also possible to process the initial concentrate 1 using a second membrane with a MWCO higher than that of the first membrane to obtain concentrate 2 and permeate 2. In that case, the procedure can be repeated any number of times using a membrane with a higher MWCO for each successive sequence. Sequential UF has been applied successfully by different research teams to produce bioactive soy peptide fractions from soy protein hydrolysates [119–124]. Deeslie and Cheryan [119] hydrolyzed a soy protein isolate (Promine-D, 93.3% protein, dry basis; Central Soya Co. Inc., Fort Wayne, IN, USA) at two levels of substrate conversion (a low level of 26% or a high level of 80%) using a protease (Pronase) obtained from Calbiochem-Behring Corp. (La Jolla, CA, USA). The continuous membrane reactors that were used have been described elsewhere [125, 126]. The soy protein hydrolysates were then processed through several UF membranes in series in order of increasing pore size (5, 10, 50 and 100 kDa) to obtain different soy peptide fractions. Molecular weight distribution of the hydrolysates was determined by gel permeation chromatography using Sephadex G 15, G 50 and G 75 gel media with 0.067M phosphate buffer (pH 7.4) as the eluant. The solubility and foaming properties of the different fractions obtained after sequential ultrafiltration of the low-conversion hydrolysate were determined at pH 4.5 as described earlier by Deeslie and Cheryan [127]. The results indicated that the 5 and 10  kDa membranes produced high conversion hydrolysate fractions (permeate) showing two large peaks with molecular weights of about 2300 and 1000 Da. The 50 kDa permeate also contained a large proportion of peptides below 2300 Da, whereas the 100  kDa permeate consisted predominantly of three fractions: 25,000 Da, 13,000 Da and a small fraction below 2300 Da. Regardless of the membrane used for fractionation, the low-conversion samples had a greater proportion of higher molecular weight components compared to the high conversion samples. The solubility results indicated that the solubility of the fractions decreased as the MWCO increased, probably because of the corresponding shift in molecular

256 

 10 Production of value-added soy protein products by membrane-based operations

weight distribution. Concerning the foaming properties, no major differences were observed between the fractions for the initial foam volume. However, foam stability (measured as foam volume remaining after 30 min) increased as the size of the peptides in the fractions increased. Small peptides found in extensively hydrolyzed proteins, which were in the 5 and 10 kDa permeates in this study, had poor foam stability. Wu et al. [120] reported the fractionation of a soy protein hydrolysate using 100, 50 and 20 kDa ultrafiltration disc membranes. The hydrolysate was produced from a soy protein isolate predenatured by mild alkali at pH 10 and heated at 50°C for 1 h prior to partial hydrolysis by papain at pH 7 and 38°C for 1 h. The starting soy protein isolate, the hydrolysate and four UF fractions (i.e., the 100 kDa retentate and the 100, 50 and 20 kDa permeates) were analyzed for molecular weight distribution, surface hydrophobicity, protein solubility, emulsifying activity index and emulsion stability index. The SDS-PAGE patterns of the isolate contained three bands (80, 76 and 50 kDa) identified as 7S globulin and two bands (35 and 25 kDa) identified as 11S globulin. Minor bands identified as subunits of 11S globulin were found at 38 and 33 kDa, as was a band identified as lipoxygenase at 94 kDa. The SDS-PAGE patterns of the hydrolysate indicated that the subunits of 11S globulin were more susceptible to papain hydrolysis than those of 7S globulin, as evidenced by the diffused SDS-PAGE pattern. In addition, the band of lipoxygenase disappeared. The SDS-PAGE pattern of the 100 kDa retentate was similar to the pattern of the original hydrolysate with the exception of the band at 28 kDa, which was substantially reduced in the retentate. Three major bands (32, 25 and 15 kDa) were observed in the SDS-PAGE pattern of the peptides in the 100 kDa permeate, two major bands (25 and 14 kDa) were observed in the 50 kDa permeate, and one major band of 14 kDa was observed in the 20 kDa permeate. Two weak bands (45 and 35 kDa) were also observed in all permeates. When compared to the soy protein isolate, the hydrolysate demonstrated higher surface hydrophobicity, protein solubility, emulsifying activity index and emulsion stability index. The surface hydrophobicity of the hydrolysate was 26.2 as compared to 12.2 for the original isolate. After hydrolysis, protein solubility at pH 7 increased from 56% for the original isolate to 94% for the hydrolysate. Hydrolysis improved the emulsifying activity index of the isolate from 102 to 207 m2/g and the emulsion stability index from 37.2 to 48.9 min. The different permeates had higher protein solubility and emulsifying activity index values but lower surface hydrophobicity values compared to the retentate and hydrolysate. The solubility of peptides in the permeates (100, 50 and 20 kDa) was almost 100% at all pH ranges studied with the exception of pH 5, where the solubility of the 100 kDa permeate was 89%. The surface hydrophobicities of the peptides in the permeates were 1.35, 1.40 and 1.42 for the 100, 50 and 20 kDa permeates, respectively, which were significantly lower than the surface hydrophobicity of the isolate. Lastly, the emulsifying activity indices of the peptides in the permeates (287, 309 and 269 m2/g for the 100, 50 and 20 kDa permeates, respectively) were significantly higher than the emulsifying activity indices of the peptides in the hydrolysate and the original isolate.

10.4 Separation of soy peptides by membrane technologies 

 257

Nass et al. [121] reported the use of UF to fractionate a soy protein hydrolysate produced from a soy protein isolate using a method mimicking gastrointestinal conditions. Proteolytic hydrolysates were prepared from soy protein with various combinations of pepsin, trypsin, chymotrypsin and/or elastase. In order to remove low-digested protein and inactivate residual protease activity immediately after digestion, hydrolysis was carried out in a stirred UF cell (Amicon, Millipore GmbH, Schwalbach, Germany) equipped with a 5 kDa membrane disc filter. To inactivate residual trypsin activity, hydrolysates were heated for 15 min in a boiling water bath. UF was carried out to separate the hydrolysate into the following fractions: > 5, 3–5, 1–3 and < 1 kDa. The different fractions were then examined to determine whether peptides derived from food proteins might influence bile acid synthesis. A reporter gene cell line that carries a cholesterol 7R-hydroxylase promoter fragment fused to firefly luciferase (cyp7a-luc) was used to screen for nutritive peptides affecting cyp7a expression, the enzyme catalyzing the rate-limiting step in bile acid synthesis. Different fractions that lowered the activity of the cyp7a-luc reporter gene at a fixed concentration of 5 mg/ml were isolated, but activity depended on the protease combinations used for digestion and the molecular mass range of the product. The < 1 kDa fraction produced from trypsin digestion alone showed the lowest relative specific luciferase activity among the different fractions, with 0.35 ± 0.10 compared to 1.00 ± 0.07 for the control. In the work of Park et  al. [122], soy protein hydrolysate was obtained by hydrolysing soy protein with alcalase (EC 3.4.21.62, from Bacillus licheniformis). A 10% (w/v) defatted soy protein solution was prepared and hydrolyzed with alcalase for 6 h at 50°C and pH 8 with an enzyme/substrate ratio of 0.4 AU/g of protein. Hydrolysis was stopped by heat treatment at 90°C for 10 min. The antioxidant activity of the hydrolysate was determined on the basis of the lipid peroxidation inhibitory activity [128–131], and lipid oxidation was evaluated by measuring thiobarbituric acid (TBA) and the peroxide value (PV). The activity of the hydrolysate was compared with that of α-tocopherol. The antioxidant activity results indicated that the soy protein hydrolysate exhibited concentration-dependent antioxidant activity, and its specific activities were 1.60 and 0.81%/mg for the TBA and PV methods, respectively. The potent antioxidant peptides were then isolated using 30, 10 and 3 kDa ultrafiltration membranes to obtain four different fractions ( < 3, 3–10, 10–30 and > 30 kDa), and the fraction with the most significant antioxidant activity was further purified using consecutive chromatographic methods, including fast protein liquid chromatography (FPLC) and reverse-phase HPLC. The antioxidant activities of the different ultrafiltration fractions were significantly higher compared to those of the initial hydrolysate, especially for the < 3  kDa fraction, with specific activities of 6.34 and 5.50%/mg for the TBA and PV methods, respectively. These activities were equivalent to 60 and 37% of those of α-tocopherol at the same concentration for the TBA and PV methods, respectively. After purification by chromatography, the < 3 kDa fraction had a specific activity of 108.13%/mg for the TBA method, which represents a 67.6-fold enhancement compared with the initial hydrolysate. In the amino acid composition of the final potent

258 

 10 Production of value-added soy protein products by membrane-based operations

antioxidant peptide, the most abundant amino acids were hydrophobic amino acids, of which phenylalanine was especially abundant. Tsou et  al. [123, 124] applied sequential filtration to a soy protein hydrolysate produced through limited hydrolysis of a soy protein isolate (Gemfont Corp., Taipei, Taiwan) by the enzyme complex Flavourzyme 1000 MG (1000 unit/g; Novo Nordisk A/S, Copenhagen, Denmark) or Neutrase 0.5 L (0.5 unit/g; Novo Nordisk). A 2.5% (w/v) soy protein isolate solution and 1% (w/w of isolate) enzyme were used to produce each hydrolysate. The optimum temperature and pH conditions were 50°C and 7 for Flavourzyme and 45°C and 6 for Neutrase. The samples were immediately heated in a boiling water bath for 10 min to inactivate the enzyme. A SW membrane module with MWCO of 30, 10 and 1 kDa was used sequentially. For the hydrolysate obtained with Flavourzyme, fresh water was added to each concentrate twice to replace 10% of the permeate volume in order to remove permeate components from the concentrate. The molecular weight distribution of the hydrolysate was analyzed using a high performance size-exclusion chromatography (HPSEC) system equipped with a Superdex HR 10/30 column (Amersham Biosciences Ltd., Pittsburgh, PA, USA) connected to a UV detector (Shimadzu Co., Kyoto, Japan) set at 214 nm. The mobile phase was 0.02M phosphate buffer (pH 7.2) containing 0.25M NaCl, and the flow rate was set at 0.5 ml/min. The different hydrolysates and fractions were evaluated for their suppression of glycerol-3-phosphate dehydrogenase (GPDH) activity and relative lipid accumulation (RLA) in 3T3-L1 preadipocytes during differentiation [123, 124]. The hydrolysates revealed much higher suppression of GPDH activity and RLA compared to intact soy protein isolate. Lower GPDH activity or RLA indicates higher anti-adipogenic activity. Sequentially fractionating the hydrolysates using membranes with MWCO from 30 to 1 kDa to obtain the 1 kDa permeate resulted in further reduction of GPDH activity. A comparison of the HPSEC profiles shows that the most active peptide fraction for antiadipogenic activity was composed primarily of small peptides with molecular weights less than 1300 Da for the Flavourzyme experiment and between 1300 and 2200 Da for the Neutrase experiment.

10.4.2 Integrated electrodialysis – ultrafiltration approach Langevin et  al. [49] studied the separation of soy peptides using the ED-UF setup previously described in the theory section 10.2.3 with 10 kDa UF membranes. Soy peptides were generated by enzymatic hydrolysis as described by Vilela et al. [132] using Profam 974 soy protein isolate, pepsin from porcine gastric mucosa (P7000–100G) and pancreatin from porcine pancreas (P1625–100G) obtained from Sigma-Aldrich Canada Ltd. (Oakville, ON, Canada). The applied voltage was 8 V, which corresponded to a current in the range 0.16–0.20 A at the start of the process. Separation was carried out for 180 min at pH 3, 6 and 9. In parallel, the researchers separated soy peptides using a NF system composed of an HL thin-film composite membrane from

10.4 Separation of soy peptides by membrane technologies 

 259

GE Osmonics TF, with a MWCO of 300–500 Da (Sterlitech Corp., Kent, WA, USA) and a negative zeta potential over the pH range 3–9. The feed volume was 2 l at 0.1% (w/v) (1 g/l; prepared from UF permeate). The system was operated at a flow rate of 1.8 l/min, and a VCR of 2 was applied under a constant transmembrane pressure of 50 bars. The temperature was kept constant at 24°C. The effective area of the membrane was 140 cm2. Separation was carried out for 15 min (the duration required to reach the VCR of 2) at pH 3, 6 and 9. A different NF membrane was used for each pH condition. Each ED-UF and NF experiment was carried out in triplicate. The samples were analyzed for their protein content using the bicinchoninic acid (BCA) method. A standard curve of bovine serum albumin (BSA) was used to determine the protein content of the different fractions. The absorbance was read at 562 nm on a microplate reader after incubation of the microplate at 37°C for 30 min and cooling at room temperature for 15  min [133]. The BCA analysis was used to calculate the mass flux and mass balance. The samples were analyzed in duplicate by liquid chromatography/mass spectrometry (LC/MS; 1100 series, Agilent Technologies Canada Inc., Mississauga, ON, Canada) to determine their molecular weight profiles as described by Langevin et al. [49]. The total peptide content was measured by the O-phthalaldehyde (OPA) assay as described by Church et al. [134]. A standard curve of Phe-Gly solution was used on a range 0–1000 μM. The absorbance was read at a wavelength of 340 nm. Amino acid analyses of each fraction were performed using the method of Beaulieu et al. [135] with the exception that an Alliance Waters (e2695) separation module with a fluorescence detector (Waters Corp., Milford, MA, USA) was used. The antioxidant capacity of the fractions was analyzed by performing the oxygen radical absorbance capacity (ORAC) assay and measuring hydrogen peroxide (H2O2) degradation [136]. The results indicated that, for KCl1, the mass flux increased from 109 ± 6 mg/m2h at pH 3 to 165 ± 65 mg/m2h at pH 9. For KCl2, however, the mass flux decreased from 238 ± 147 mg/m2h at pH 3 to 97 ± 31 mg/m2h at pH 9. The difference in migration of the charged peptides appears to indicate that the hydrolysate contains more cationic peptides than anionic peptides. However, as mentioned by the authors, these results must be used with caution owing to the large standard deviations. For the NF experiments, no significant difference was observed for the mass flux as a function of the pH of the feed hydrolysate. The mass flux obtained for the NF process averaged 10,200 ± 900  mg/m2h for the three pH values tested, which is significantly higher than the fluxes observed for the ED-UF process. A mass balance indicated that the losses averaged 26 ± 16% for ED-UF and 23 ± 10% for NF. For both processes, the loss of peptides was attributed to the system hold-up volume and to membrane fouling [49]. The LC/MS analysis indicated that, for the experiments carried out at pH 3, the three ED-UF fractions (hydrolysate after electrodialysis-ultrafiltration, KCl1 fraction and KCl2 fraction) presented similar peptide profiles to the feed hydrolysate with the exception of a decrease in the 700 to 800 Da range for the KCl2 fraction. At pH 6, however, the concentrations of peptides in the molecular weight range 400–500 Da

260 

 10 Production of value-added soy protein products by membrane-based operations

were almost two- and four-times higher than the concentration in the feed hydrolysate for the KCl1 and KCl2 fractions, respectively. For the same pH, the proportion of peptides with molecular weights over 700 Da was very low for both the KCl1 and KCl2 fractions, indicating that no large peptides had migrated even though the MWCO of the UF membrane was 10 kDa. For the pH 9 condition, peptides that were 600–700 Da in size were found at a higher percentage in the KCl2 fraction than in the feed hydrolysate by a factor of two. However, the abundance of this fraction in the 500–600 Da range was 2% compared to 30% in the feed. In general, the change in pH seemed to influence the specific molecular weight recovery in the three final products for ED-UF. For the NF fractions, no significant difference, as a function of pH, was observed for the peptide molecular weight profile. The majority of the peptides with molecular weights less than 400 Da migrated in the permeate, as did some peptides with molecular weights of 600 and 900 Da. This may have occurred because a certain amount of high molecular weight peptides had permeated through the membrane with the help of the pressure and/or because some small peptides found in the permeate had recombined by hydrophobic peptide-peptide interactions to form larger peptides. The total peptide content of the different fractions was expressed as a ratio over the OPA value of the feed hydrolysate. For the three ED-UF fractions, the ratio was always less than 0.1, indicating that the total peptide content in the fractions was very low compared to the content in the feed hydrolysate (2508.24 ± 9 μM Eq Phe-Gly/l) [49]. For the KCl1 solution, this ratio increased with the pH, indicating that the pH of the solution significantly influenced the migration of peptides by changing their charge. In contrast, the amount of peptides in KCl2 was stable for the three pH values. For NF, the ratio was not a function of the pH values and was in the order of 1.6 in the retentate. The high peptide concentration for the NF retentate is a result of the application of a VCR of 2. In terms of amino acid profiles, significant differences were observed for some specific amino acids among the different samples. For ED-UF at pH 3, an increase in glutamic acid was observed in KCl1. The pKa of 4.1 attributed to the R group results in a negative charge at pH 3, which would explain the migration in the KCl1 fraction. Significant increases in valine, leucine, lysine, isoleucine and alanine were observed for the KCl2 fraction. Given that the isoelectric point of valine and leucine is 6, at pH 3 they would be positively charged, which would explain their migration in the KCl2 compartment. At pH 6, significant increases in lysine (almost three-fold) and arginine and a decrease in tyrosine were observed in KCl1. These amino acids have in common a side chain charge that could improve the mobility of the global peptide that they are part of. At pH 9, increases in valine, leucine and isoleucine were observed in KCl2. If those amino acids are located at the N-terminus, they could still have their charge on the NH3+ group (9.7 < pKa < 9.8), which could explain this observation. For NF, a significant increase in methionine in the permeate at pH 6 and significant increases in tyrosine, methionine, leucine, arginine and phenylalanine in the permeate at pH 9 were observed compared to the hydrolysate and to the other fractions.

10.5 Concluding remarks and perspectives 

 261

The ORAC results were expressed as a ratio of the peptide fraction value over the feed hydrolysate value and showed significant increases in antioxidant capacity for KCl1 at pH 3 and KCl1 at pH 6 compared to the feed hydrolysate. No significant increases were observed for the NF permeate and retentate regardless of the pH. The results obtained by H2O2 degradation confirmed some of the results obtained by ORAC for the KCl1 fractions at pH 3 and pH 6, showing that these fractions had higher antioxidant capacities than the feed hydrolysate. Furthermore, the KCl1 fraction at pH 9 and the NF permeate also showed higher antioxidant capacities than the feed hydrolysate. Compared to traditional techniques, the integrated ED-UF process appears to have lower costs and better productivity than chromatographic/ion-exchange resin techniques, as well as better selectivity than conventional pressure-driven processes (UF and NF). However, the productivity of conventional pressure-driven processes remains significantly higher than that of the integrated ED-UF process, and some selective separation is possible assuming that sequential UF/NF is carried out.

10.5 Concluding remarks and perspectives Since the early 1970s, when UF was first used to process soy protein extracts to obtain soy protein concentrates or isolates, there have been a number of potential applications developed in the soy protein sector for UF, bipolar-membrane ED, the combination in series of UF and bipolar-membrane ED, and the integration of conventional ED with UF. Ultrafiltration-diafiltration has been shown to be an efficient process for reducing the levels of phytic acid and oligosaccharides in soy protein extracts while at the same time enabling the retention of whey-like proteins and isoflavones, thus producing soy protein isolates with improved nutritional and functional properties as compared with the isolates produced by the traditional isoelectric precipitation process. The retention of whey-like proteins also results in increased protein recovery and reduces the pollution problems associated with effluent generation. When used in place of the conventional acidification step with a mineral acid for the isoelectric precipitation of soy proteins, ED with bipolar membranes has been shown to generate less effluent and produce soy protein isolates with superior chemical composition, including lower salt content. However, gradual protein precipitation in the ED cell remains a limitation. The combination of ED with bipolar membranes and UFdiafiltration to decrease this fouling has yielded some promising results in addition to producing soy protein isolates with low phytic acid content. For the isolation of soy bioactive peptides, sequential UF and, more recently, the integration of conventional ED with UF have shown some interesting results. Nevertheless, the use of membrane technologies in the soy protein industry is still in its infancy, given that only a limited number of the novel applications studied so far on the laboratory and pilot scales have been applied on the industrial scale. This is certainly because of the high cost of membranes and equipment as well as the often limited lifetime of membranes for applications with high fouling potential. As illustrated above, however, some membrane technology applications have

262 

 10 Production of value-added soy protein products by membrane-based operations

great potential for use in isolating soy bioactive peptides or producing soy protein isolates with improved nutritional and functional properties. Membrane technologies could also be beneficial in terms of environmental protection. For those reasons, the increasing demand for healthy food products and the strengthening of environmental policies should promote the application of membrane technologies in the soy protein industry in coming years.

10.5.1 Acknowledegments The author would like to thank François Lamarche for the preparation of the figures.

10.6 References 1. World protein meal consumption 2011. American Soybean Association: Saint Louis, Missouri; 2012. http://soystats.com/2012/Default-frames.htm (Accessed July 23rd, 2012). 2. Friedman M. Nutritional value of proteins from different food sources: A review. J Agric Food Chem 1996;44:9−26. 3. Endres GJ. Protein quality and human nutrition. In: Endres GJ., ed. Soy Protein Products: Characteristics, Nutritional Aspects and Utilization. Revised and Expanded Edition. AOCS Press: Champaign, IL; 2001;10–19. 4. Sugano M, Yamada Y, Yoshida K, Hashimoto Y, Matsuo T, Kimoto M. The hypocholesterolemic action of the undigested fraction of soybean protein in rats. Atherosclerosis 1988;72:115–122. 5. Castiglioni S, Manzoni C, D’Uva A et al. Soy proteins reduce progression of a focal lesion and lipoprotein oxidiability in rabbits fed a cholesterol-rich diet. Atherosclerosis 2003;171: 163–170. 6. Barnes S, Grubbs C, Setchell KD, Carlson J. Soybeans inhibit mammary tumors in models of breast cancer. Prog Clin Biol Res 1990;347:239–253. 7. Kumar NSK, Yea MK, Cheryan M. Soy protein concentrates by ultrafiltration. J Food Sci 2003;68: 2278–2283. 8. Messina M. Insights gained from 20 years of soy research. J Nutr 2010;140:2289S–2295S. 9. Lusas EW, Riaz MN. Soy Protein Products: Processing and Use. J Nutr 1995;125:573S–580S. 10. Mounts TL, Wolf WJ, Martinez WH. Processing and utilization. In: Wilcox JR., ed. Soybean: Improvement, Production and Uses. American Society of Agronomy: Madison, WI; 1987;819–866. 11. Lusas EW, Rhee KC. Soy protein processing and utilization. In: Erickson DR., ed. Practical Handbook of Soybean Processing and Utilization. AOCS Press: Champaign, IL, and United Soybean Board: St. Louis, MO; 1995;117–160. 12. Petruccelli S, Anon MC. Relationship between the method of obtention and structural and functional properties of soy protein isolates. 1. Structural and hydration properties. J Agric Food Chem 1994;42:2161–2169. 13. Wagner JR, Sorgentini DA, Anon MC. Relation between solubility and surface hydrophobicity as an indicator of modifications during preparation processes of commercial and laboratory-prepared soy protein isolates. J Agric Food Chem 2000;48:3159–3165. 14. Omosaiye O, Cheryan M. Low-phytate, full-fat soy protein product by ultrafiltration of aqueous extracts of whole soybeans. Cereal Chem 1979a;56:58–62. 15. Brooks JR, Morr CV. Effect of phytate removal treatments upon the molecular weight and subunit composition of major soy protein fractions. J Agric Food Chem 1985;33:1128–1132.

10.6 References 

 263

16. Meena G, Usha B. Soy protein isolate: production technology, functional properties, nutritive value and applications – a review. Indian Food Packer 1990;90:5–22. 17. Korhonen H, Pihlanto A. Food-derived bioactive peptides: opportunities for designing future foods. Curr Pharm Des 2003;9:1297–1308. 18. Kitts DD, Weiler K. Bioactive proteins and peptides from food sources. Applications of bioprocesses used in isolation and recovery. Curr Pharm Des 2003;9:1309–1323. 19. Wang W, Gonzalez de Mejia E. A new frontier in soy bioactive peptides that may prevent age–related chronic diseases. Compr Rev Food Sci Food Safety 2005;4:63–78. 20. Yamamoto N, Ejiri M, Mizuno S. Biogenic peptides and their potential use. Curr Pharm Des 2003;9:1345–1355. 21. Fischer M, Gruppen H, Piersma SR, Kofod LV, Schols HA, Voragen AGJ. Aggregation of peptides during hydrolysis as a cause of reduced enzymatic extractability of soybean meal proteins. J Agric Food Chem 2002;50:4512–4519. 22. Gibbs BF, Zougman A, Masse R, Mulligan C. Production and characterization of bioactive peptides from soy hydrolysate and soy-fermented foods. Food Res Int 2004;37:123–131. 23. Pena-Ramos EA, Xiong YL. Antioxidant activity of soy protein hydrolysates in a liposomal system. J Food Sci 2002;67:2952–2956. 24. Kim SE, Kim HH, Kim JY, Kang YI, Woo HJ, Lee HJ. Anticancer activity of hydrophobic peptides from soy proteins. BioFactors 2000;12:151–155. 25. Wu J, Ding X. Hypotensive and physiological effect of angiotensin converting enzyme inhibitory peptides derived from soy protein on spontaneously hypertensive rats. J Agric Food Chem 2001;49:501–506. 26. Cheryan M. Mass transfer characteristics of hollow fiber ultrafiltration of soy protein systems. J Food Process Eng 1977;269–287. 27. Cheryan M. Ultrafiltration and Microfiltration Handbook. Technomic Publishing Company Inc.: Chicago, IL; 1998. 28. Cuperus FP, Nijhuis HH. Applications of membrane technology to food processing. Trends Food Sci Tech 1993;7:277–282. 29. Ghosh R. Ultrafiltration-based protein bioseparation. In: Pabby AK, Rizvi SSH and Sastre AM., eds. Handbook of Membrane Separations: Chemical, Pharmaceutical, Food, and Biotechnological applications. CRC Press Taylor and Francis Group: Boca Raton, FL; 2009;497–511. 30. Zeman LJ, Zydney AL. Microfiltration and Ultrafiltration: Principles and Applications. Marcel Dekker: NY, USA; 1996. 31. Sheikholeslami R. General description. In: Sheikholeslami R., ed. Fouling in Membranes and Thermal Units. Desalinations Publications: L’Aquila, Italy; 2007;5–18. 32. Scott K. Handbook of Industrial Membranes. Elsevier: Oxford, UK; 2003. 33. Strathmann H. Synthetic membranes and their preparation. In: Porter MC, ed. Handbook of Industrial Membrane Technology. Noyes Publications: Noyes, Park Ridge, NJ; 1990;1–60. 34. Davis TA, Genders JD, Pletcher D. A First Course in Ion Permeable Membranes. The Electrochemical Consultancy. Alresford Press: Alresford, UK; 1997. 35. Strathmann H. Ion-exchange Membrane Separation Processes. Elsevier: Amsterdam, The Netherlands; 2004. 36. Tanaka Y. Ion-exchange Membranes: Fundamentals and Applications. Elsevier: Amsterdam, The Netherlands; 2007. 37. Fidaleo M, Moresi M. Electrodialysis Applications in the Food Industry. In: Taylor S., ed. Advances in Food and Nutrition Research. Academic Press: San Diego, CA; 2006;51:265–360. 38. Thate S, Eigenberger G, Rapp H-J. Introduction. In: Kemperman AJB., ed. Handbook on Bipolar-Membrane Technology. Twente University press: Enschende, The Netherlands; 2000;7–16.

264 

 10 Production of value-added soy protein products by membrane-based operations

39. Davis TA, Laterra A. Onsite generation of acid and base with bipolar membranes. Paper Presented at the 48th Annual Meeting of the International Water Conference. Pittsburgh, PA; 1987. 40. Van der Bruggen B, Koninckx A, Vandecasteele C. Separation of monovalent and divalent ions from aqueous solution by electrodialysis and nanofiltration. Water Res 2004;38:1347–1353. 41. Aider M, de Halleux D, Bazinet L. Potential of continuous electrophoresis without and with porous membranes (CEPM) in the bio-food industry: review. J Membr Sci 2008;19:351–362. 42. Ruckenstein E, Zeng X. Albumin separation with Cibacron Blue carrying macroporous chitosan and chitin affinity membranes. J Membr Sci 1998;142:13–26. 43. Galier S, Roux-de Balmann H. Study of biomolecules separation in an electrophoretic membrane contactor. J Membr Sci 2004;241:79–87. 44. Causserand C, Lafaille J-P, Aimar P. Transmission of bio-molecules through porous membranes triggered by an external electric field. J Control Rel 1994;29:113–123. 45. Poulin J-F, Amiot J, Bazinet L. Simultaneous separation of acid and basic bioactive peptides by electrodialysis with ultrafiltration membrane. J Biotechnol 2006;123:314–328. 46. Poulin J-F, Amiot J, Bazinet L. Improved peptide fractionation by electrodialysis with ultrafiltration membrane: Influence of ultrafiltration membrane stacking and electrical field strength. J Membr Sci 2007;299:83–90. 47. Poulin J-F, Amiot J, Bazinet L. Impact of feed solution flow rate on peptide fractionation by electrodialysis with ultrafiltration membrane. J Agric Food Chem 2008;56:2007–2011. 48. Firdaous L, Dhulster P, Amiot J, Gaudreau A, Lecouturier D, Kapel R, Lutin F, Vézina L-P, Bazinet L. Concentration and selective separation of bioactive peptides from an alfalfa white protein hydrolysate by electrodialysis with ultrafiltration membranes. J Membr Sci 2009;329:60–67. 49. Langevin M-E, Roblet C, Moresoli C, Ramassamy C, Bazinet L. Comparative application of pressure- and electrically-driven membrane processes for isolation of bioactive peptides from soy protein hydrolysate. J Membr Sci 2012;403–404:15–24. 50. Lapointe J-F, Gauthier SF, Pouliot Y, Bouchard C. Fouling of a nanofiltration membrane by a β-lactoglobulin tryptic hydrolysate: impact on the membrane sieving and electrostatic properties. J Membr Sci 2005;253:89–102. 51. Porter MC, Michaels AS. Applications of membrane ultrafiltration to food processing. Membrane ultrafiltration. Chem Tech 1971;1–633. 52. Iacobucci GA, Myers DV, Okubo K. Process for Preparing Protein Products U.S. Patent 3,736,147 (1973). 53. Pompei C, Lucisano M, Maletto S. Production d’isolats de soja par ultrafiltration et diafiltration. Proc IV Int Congress Food Sci and Technol 1974;V:125–134. 54. Okubo K, Waldrop AB, Iacobucci GA, Myers DV. Preparation of low-phytate soybean protein isolate by ultrafiltration. Cereal Chem 1975;52:263–271. 55. Goodnight KC, Hartman GH, Marquardt RF. Aqueous Purified Soy Protein and Beverage U.S. Patent 3,995,071 (1976). 56. Lawhon JT, Mulsow D, Cater CM, Mattil KF. Production of protein isolates and concentrates from oilseed flour extracts using industrial ultrafiltration and reverse osmosis systems. J Food Sci 1977;42:389–394. 57. Lawhon JT, Hensley DW, Mulsow D, Mattil KF. Optimization of protein isolate production from soy flour using industrial membrane systems. J Food Sci 1978;43:361–364. 58. Lawhon JT, Hensley DW, Mizuhoshi M, Mulsow D. Alternate processes for use in soy protein isolation by industrial ultrafiltration membranes. J Food Sci 1979;44:213–219. 59. Lawhon JT, Golightly NH, Lusas EW. Utilization of membrane-produced oilseed isolates in soft-serve frozen desserts. J Am Oil Chem Soc 1980, September; 302–306. 60. Lawhon JT, Manak LJ, Rhee KC, Rhee KS, Lusas EW. Combining aqueous extraction and membrane isolation techniques to recover protein and oil from soybeans. J Food Sci 1981a;46:912–916.

10.6 References 

 265

61. Lawhon JT, Rhee KC, Lusas EW. Soy protein ingredients prepared by new processes – aqueous processing and industrial membrane isolation. J Am Oil Chem Soc 1981b, March; 377–383. 62. Cheryan M, Schlesser JE. Performance of a hollow fiber system for ultrafiltration of aqueous extracts of soybeans. Lebensm Wiss u Technol 1978;11:65–69. 63. Omosaiye O, Cheryan M, Matthews ME. Removal of oligosaccharides from soybean water extracts by ultrafiltration. J Food Sci 1978;43:354–360. 64. Omosaiye O, Cheryan M. Ultrafiltration of soybean water extracts: processing characteristics and yields. J Food Sci 1979b;44:1027–1031. 65. Khan MN, Lawhon JT. Baking properties of oilseed protein and isolates produced with industrial membrane systems. Cereal Chem 1980;57:433–436. 66. Lah, CL, Cheryan M. Emulsifying properties of a full-fat soy protein product produced by ultrafiltration. Lebensm Wiss u Technol 1980;13:259–263. 67. Manak LJ, Lawhon JT, Lusas EW. Functioning potential of soy, cottonseed, and peanut protein isolates produced by industrial membrane systems. J Food Sci 1980;45:236–245. 68. Nichols DJ, Cheryan M. Production of soy isolates by ultrafiltration: factors affecting yield and composition. J Food Sci 1981a;46:367–372. 69. Nichols DJ, Cheryan M. Production of soy isolates by ultrafiltration: process engineering characteristics of the hollow fiber system. J Food Process Pres 1981b;5:103–118. 70. Thomas RL, Ndife LI, Shallo H, Nelles LP. Soy Proteins and Method for Their Production U.S. Patent 6, 313, 273 B1 (2001). 71. Rao A, Shallo, HE, Ericson AP, Thomas RL. Characterization of soy protein concentrate produced by membrane ultrafiltration. J Food Sci 2002;67:1412–1418. 72. Noordman TR, Kooiker K, Bel W, Dekker M, Wesselingh JA. Concentration of aqueous extracts of defatted soy flour by ultrafiltration: Effect of suspended particles on the filtration flux. J Food Eng 2003;58:135–141. 73. Hojilla-Evangelista MP, Sessa DJ, Mohamed A. Functional properties of soybean and lupin protein concentrates produced by ultrafiltration-diafiltration. J Am Oil Chem Soc 2004;81:1153–1157. 74. Batt HP, Thomas RL, Rao A. Characterization of isoflavones in membrane processed soy protein concentrate. J Food Sci 2003;68:401–404. 75. Singh N. Process for Producing a High Solubility, Low Viscosity, Isoflavone-enriched Soy Protein Isolate and the Products Thereof U.S. Patent 7,306,821 B2 (2007). 76. Vishwanathan KH, Govindaraju K, Singh V, Subramanian R. Production of okara and soy protein concentrates using membrane technology. J Food Sci 2011;76:E158–E164. 77. De Boland AR, Garner GB, O’Dell BL. Identification and properties of phytate in cereal grains and oilseed products. J Agric Food Chem 1975;23:1186−1189. 78. Lolas GM, Markakis P. Phytic acid and other phosphorus compounds of beans. J Agric Food Chem 1975;23:13−15. 79. Cheryan M. Phytic acid interactions in food systems. Crit Rev Food Sci Nutr 1980;13:297−335. 80. Selle PH, Ravindran V, Caldwell RA, Bryden WL. Phytate and phytase: consequences for protein utilisation. Nutr Res Rev 2000;113:255−278. 81. Martin C J, Evans W J. Phytic acid-metal ion interactions. II. The effect of pH on Ca(II) binding. J Inorg Biochem 1986;27:17−30. 82. Grynspan F, Cheryan M. Phytate-calcium interactions with soy protein. J Am Oil Chem Soc 1989; 66:93−97. 83. Graf E, Eaton JW. Suppression of colonic cancer by dietary phytic acid. Nutr Cancer 1993;19:11−19. 84. Vucenik I, Shamsuddin AM. Protection against cancer by dietary IP6 and inositol. Nutr Cancer 2006;55:109−125. 85. Dendougui F, Schwedt G. In vitro analysis of binding capacities of calcium to phytic acid in different food samples. Eur Food Res Technol 2004;219:409−415.

266 

 10 Production of value-added soy protein products by membrane-based operations

86. Kamchan A, Puwastien P, Sirichakwal PP, Kongkachuichai R. In vitro calcium bioavailability of vegetables, legumes and seeds. J Food Compos Anal 2004;17:311−320. 87. Brooks JR, Morr CV. Phytate removal from soy protein isolates using ion-exchange processing treatments. J Food Sci 1982;47:1280−1282. 88. Han YW. Removal of phytic acid from soybean and cottonseed meals. J Agric Food Chem 1988;36:1181−1183. 89. Shallo HE, Rao A, Ericson AP, Thomas RL. Preparation of soy protein concentrate by ultrafiltration. J Food Sci 2001;66:242−246. 90. Mondor M, Ippersiel D, Lamarche F, Boye JI. Effect of electro-acidification treatment and ionic environment on soy protein extract particle size distribution and ultrafiltration permeate flux. J Membr Sci 2004b;231:169–179. 91. Skorepova J, Moresoli C. Carbohydrate and mineral removal during the production of low-phytate soy protein isolate by combined electroacidification and high shear tangential flow ultrafiltration. J Agric Food Chem 2007;55:5645–5652. 92. Ali F, Ippersiel D, Lamarche F, Mondor M. Characterization of low-phytate soy protein isolates produced by membrane technologies. Innov Food Sci Emerg Tech 2010;11:162–168. 93. Coward L, Barnes NC, Setchell KDR, Barnes S. Genistein, daidzein and their β – Glycoside conjugates: Antitumor isoflavones in soybean foods from American and Asian diets. J Agric Food Chem 1993;41:1961–1967. 94. Peterson TG, Coward L, Kirk M, Falany CN, Barnes S. The role of metabolism in mammary epithelial cell growth inhibition by the isoflavones genistein and biochanin A. Carcinogenisis Sep 1996;17:1861–1869.  95. Setchell KDR. Phytoestrogens: the biochemistry, physiology, and implications for human health of soy isoflavones. Am J Clin Nutr 1998;68:1333S–1346S. 96. Messina M. Soy, soy phytoestrogens (isoflavones), and breast cancer. Am J Clin Nutr 1999;70: 574–575. 97. Zhou J, Gugger ET, Tanaka T, Guo Y, Blackburn GL, Clinton SK. Soybean phytochemicals inhibit the growth of transplantable human prostate carcinoma and tumor angiogenesis in mice. J Nutr 1999;129:1628–1635. 98. Cohen LA, Zhou Z, Pittman B, Scimeca JA. Effect of intact and isoflavone-depleted soy protein on NMU–induced rat mammary tumorigenesis. Carcinogenesis 2000;21:929–935.  99. Potter SM, Baum JA, Teng H, Stillman RJ, Shay NF, Erdman JW. Soy protein and isoflavones: their effects on blood lipids and bone density in postmenopausal women. Am J Clin Nutr 1998, 68(suppl), 1375S–1379S. 100. Messina M, Messina V. Soyfoods, soybean isoflavones, and bone health: A brief overview. J Renal Nutr 2000;10:63–68. 101. Anthony MS, Clarkson BC, Hughes CL, Morgan TM, Burke GL. Soybean isoflavones improve cardiovascular risk factors without affecting the reproductive system of peripubertal Rhesus monkeys. J Nutr 1996;126:43–50. 102. Kurzer MS. Hormonal effects of soy isoflavones: Studies in premenopausal and postmenopausal women. J Nutr 2000;130:660S–661S. 103. Scambia G, Mango D, Signorile G, Angeli RA. Clinical effects of a standardized soy extract in postmenopausal women: A pilot study. Menopause 2000;7:105–111. 104. Sessa D. Processing of soybean hulls to enhance the distribution and extraction of value-added proteins. J Sci Food Agric 2003;84:75−82. 105. Thanh VH, Shibasaki K. Major proteins of soybean seeds: A straightforward fractionation and their characterization. J Agric Food Chem 1976;24:1117–1121. 106. Bazinet L, Lamarche F, Labrecque R, Toupin R, Boulet M, Ippersiel D. Systematic study on the preparation of a food grade soyabean protein. Report for the Canadian Electricity Association 9326 U 987, 1996, Research and Development, Montreal.

10.6 References 

 267

107. Bazinet L, Lamarche F, Ippersiel D. Bipolar-membrane electrodialysis: Applications of electrodialysis in the food industry. Trends Food Sci Tech 1998a;9:107–113. 108. Bazinet L, Lamarche F, Ippersiel D. Ionic balance: a closer look at the K+ migrated and H+ generated during bipolar-membrane electro-acidification of soybean proteins. J Membr Sci 1999;154:61–71. 109. Bazinet L, Lamarche F, Labrecque R, Ippersiel D. Effect of KCl and soy protein concentrations on the performance of bipolar-membrane electro-acidification. J Agric Food Chem 1997a;45: 2419–2425. 110. Bazinet L, Lamarche F, Labrecque R, Ippersiel D. Effect of number of bipolar membranes and temperature on the performance of bipolar-membrane electroacidification. J Agric Food Chem 1997b;45:3788–3794. 111. Pourcelly G, Bazinet L. Developments of bipolar-membrane technology in food and bio-industries. In: Pabby AK, Rizvi SSH and Sastre AM., eds. Handbook of Membrane Separations: Chemical, Pharmaceutical, Food, and Biotechnological Applications. CRC Press Taylor and Francis Group: Boca Raton, FL; 2009;581–633. 112. Mondor M, Ippersiel D, Lamarche F, Boye JI. Production of soy protein concentrates using a combination of electroacidification and ultrafiltration. J Agric Food Chem 2004a;52:6991–6996. 113. Mondor M, Ali F, Ippersiel D, Lamarche F. Impact of ultrafiltration/diafiltration sequence on the production of soy protein isolate by membrane technologies. Innov Food Sci Emerg Tech 2010;11:491–497. 114. Alibhai Z, Mondor M, Moresoli C, Ippersiel D, Lamarche F. Production of soy protein concentrates/ isolates: Traditional and membrane technologies. Desalination 2006;191:351–358. 115. Ali F, Mondor M, Ippersiel D, Lamarche F. Characterization of low-phytate soy protein isolates produced by membrane technologies. Innov Food Sci Emerg Tech 2011;12:171–177. 116. Bazinet L, Lamarche F, Ippersiel D. Comparison of chemical and bipolar-membrane electrochemical acidification for precipitation of soybean proteins. J Agric Food Chem 1998b;46:2013–2019. 117. Hartmann R, Meisel H. Food-derived peptides with biological activity: from research to food applications. Curr Opin Biotech 2007;18:163–169. 118. Drioli E. Membrane processes in the separation, purification, and concentrationof bioactive compounds from fermentation broths. In: Asenjo JA. and Hong J., eds. Separation, Recovery, and Purification in Biotechnology. American Chemical Society: Washington, DC; 1986;52–66. 119. Deeslie WD, Cheryan M. Fractionation of soy protein hydrolysates using ultrafiltration membranes. J Food Sci 1991;57:411–413. 120. Wu WU, Hettiarachchy NS, Qi M. Hydrophobicity, solubility, and emulsifying properties of soy protein peptides prepared by papain modification and ultrafiltration. J Am Oil Chem Soc 1998;75:845–850. 121. Nass N, Schoeps R, Ulbrich-Hofmann R et al. Screening for nutritive peptides that modify cholesterol 7α-Hydroxylase expression. J Agric Food Chem 2008;56:4987–4994. 122. Park SY, Lee J-S, Baek H-H, Lee HG. Purification and characterization of antioxidant peptides from soy protein hydrolysate. J Food Biochem 2010;34:120–132. 123. Tsou M-J, Kao F-J, Tseng C-K, Chiang W-D. Enhancing the anti-adipogenic activity of soy protein by limited hydrolysis with Flavourzyme and ultrafiltration. Food Chem 2010a;122:243–248. 124. Tsou M-J, Lin W-T, Lu H-C, Tsui Y-L, Chiang W-D. The effect of limited hydrolysis with Neutrase and ultrafiltration on the anti-adipogenic activity of soy protein. Process Biochem 2010b;45: 217–222. 125. Deeslie WD, Cheryan M. Continuous enzymatic modification of proteins in an ultrafiltration reactor. J Food Sci 1981;46:1035–1042. 126. Cheryan M, Deeslie WD. Soy protein hydrolysis in membrane reactors. J Am Oil Chem Soc 1988;60:1112–1115.

268 

 10 Production of value-added soy protein products by membrane-based operations

127. Deeslie WD, Cheryan M. Functional properties of soy protein hydrolysates from a continuous ultrafiltration reactor. J Agric Food Chem 1988;36:26–31. 128. Chen H-M, Muramoto K, Yamauchi F. Structural analysis of antioxidative peptides from soybean β-conglycinin. J Agric Food Chem 1995;43:574–578. 129. Park PJ, Jung WK, Nam KS, Shahidi F, Kim S-K. Purification and characterization of antioxidative peptides from protein hydrolysate of lecithin-free egg yolk. J Am Oil Chem Soc 2001;78:651–656. 130. Saiga A, Tanabe S, Nishimura T. Antioxidant activity of peptides obtained from porcine myofibrillar proteins by protease treatment. J Agric Food Chem 2003;51:3661–3667. 131. Wu H-C, Chen H-M, Shiau C-Y. Free amino acids and peptides as related to antioxidant properties in protein hydrolysates of mackerel (Scomber austriasicus). Food Res Int 2003;36:949–957. 132. Vilela RM, Lands LC, Chan HM, Azadi B, Kubow S. High hydrostatic pressure enhances whey protein digestibility to generate whey peptides that improve glutathione status in CFTR-deficient lung epithelial cells. Mol Nutr Food Res 2006;50:1013–1029. 133. Wiechelman KJ, Braun RD, Fitzpatrick JD. Investigation of the bicinchoninic acid protein assay: identification of the groups responsible for color formation. Anal Biochem 1988;175:231–237. 134. Church FC, Swaisgood HE, Porter DH, Catignani GL. Spectrophotometric assay using ortho-phthaldialdehyde for determination of proteolysis in milk and isolated milk-proteins. J Dairy Sci 1983;66:1219–1227. 135. Beaulieu L, Thibodeau J, Bryl P, Carbonneau ME. Proteolytic processing of Atlantic mackerel (Scomber scombrus) and biochemical characterisation of hydrolysates. Int J Food Sci Tech 2009;44:1609–1618. 136. Singh M, Murthy V, Ramassamy C. Modulation of hydrogen peroxide and acrolein-induced oxidative stress, mitochondrial dysfunctions and redox regulated pathways by the Bacopa Monniera extract: potential implication in Alzheimer’s disease. J Alzheimers Dis 2010;21: 229–247.

11 Concentration of polyphenols by integrated membrane operations Iren Tsibranska and Bartosz Tylkowski 11.1 Introduction Phenolic compounds, known also as polyphenols, constitute one of the largest and recently very popular groups of phytochemicals, widely distributed in the plant kingdom, with more than 8000 phenolic structures currently found. Polyphenols are widely common secondary metabolites of plants, the content of which varies greatly between different species, and cultivars, and with maturity, season, region and yield. They are found in various amounts in large numbers of natural products, especially plant material such as fruits, vegetables as well as cereals, and beverages (coffee, tea, wine and beer) [1]. A particular rich source are: grapes [2], apples [3], olives [4], teas [5], honey [6] (particularly propolis [7]), potatoes [8], and many more. They arise biogenically from two main synthetic pathways: the shikimate pathway and the acetate pathway. Polyphenols are compounds comprising more than one phenolic group. The structure of natural polyphenols varies from simple molecules, such as phenolic acids, to highly polymerized compounds [9, 10]. Polyphenols can be divided into at least four different classes depending on their basic chemical structure: 1. Flavonoids: They comprise the most studied group of polyphenols. “Flavonoid” is a term often used to denote polyphenols in general, but more commonly in Europe to denote only the flavones. They are ubiquitous in plants and include at least 2,000 naturally occurring compounds [11]. Typical examples of flavonols are: quercetin, kaempferol and myricetin. They can exist naturally as aglycone or as O-glycosides (e.g., D-glucose, galactose, arabinose, rhamnose, etc.) Other forms of substitution such as methylation, sulphation and malonylation are also found. 2. Flavanols: The two most common flavanols are catechin and its stereoisomer epicatechin. 3. Proanthocyanidins: They are polymers of catechin and/or epicatechin and can contain up to eight units or more. These compounds are often called “proanthocyanidins”, “procyanidins” or “tannins”. Moreover the tannins greatly affect the taste, astringency and keeping qualities of wines, beers, fruit juices and especially coffee and tea. 4. Anthocyanins: They are colored substances, sometimes called anthocyanidins. Typical examples are: cyanidin, delphinidin and pelargonidin. They are important in wine and fruit juice colors, and their combination with metal salts may discolor these products.

270 

 11 Concentration of polyphenols by integrated membrane operations

Moreover, polyphenols are further sub-divided on the basis of the hydroxylation of the phenolic rings, glycosylation, acylation with phenolic acids and the existence of stereoisomers.

11.1.1 Beneficial effects of polyphenols Polyphenols have a large and diverse array of beneficial effects on both plants and humans. Plant-derived polyphenols have been shown to be strong antioxidants with potential health benefits as they are able to protect the heart, protect from premature skin ageing, and bind viruses and bacteria. The effect of polyphenols on human cancer cell lines is most often protective and induces a reduction of the number of tumors or of their growth [12 ]. These effects have been observed at various sites, including the mouth, stomach, duodenum, colon, liver, lung, mammary gland or skin. Epidemiological studies have repeatedly shown an inverse association between the risk of chronic human diseases and the consumption of polyphenolic rich diet [13]. Based on recent studies, polyphenolic compounds are also believed to have anti-inflammatory and immune boosting properties [14], and all these can prove immensely helpful for improving general wellbeing. Polyphenols are also beneficial in ameliorating the adverse effects of the ageing on the nervous system or brain [15]. Polyphenols can have many other health benefits, but the actual health-promoting actions of polyphenols are still being studied. The properties mentioned above give to natural phenols a great potential as active principles in the cosmetic and pharmaceutical industry and as antioxidant compounds in the food industry. In this context the recovery of polyphenols from plants, herbs, fruits, vegetables, etc. could facilitate the production of valuable natural products, which would guarantee both sustainability and satisfaction of consumer demands.

11.1.2 Separation/concentration of polyphenols by traditional methods Essential to the study of polyphenols and their industrial application are the means available for their separation. This section aims to present a brief unified summary of general techniques. In order to use the polyphenols they have first to be extracted from the natural products and then separated or concentrated. It is well-known that extraction of polyphenols from vegetable materials by organic solvents is a common operation applied in many industrial processes, particularly in the pharmaceutical industry. Different extraction techniques, such as: solid phase extraction [16], MAE [17], pressured liquid extraction [18] or SFE [19] have been reported.

11.1 Introduction 

 271

11.1.2.1 Separation of polyphenols at laboratory scale A number of techniques have been used for the separation/concentration of polyphenols. At the laboratory scale, the most used methods are: thin layer chromatography (TLC), paper chromatography (PC), gas chromatography (GC), HPLC, high speed countercurrent chromatography (HSCCC), supercritical fluid chromatography (SFC) and capillary electrophoresis (CE). Some of these are described below.

11.1.2.1.1 Thin layer chromatography (TLC) and paper chromatography (PC) Since the early 1960s, TLC has been used in polyphenols separation and analysis. PC was the first preferred method in the past, however over the years TLC slowly replaced it, as new stationary phases (such as silica, microcrystalline cellulose, and polyaminde) were developed. Today, the TLC is still being used as a separation tool for many antioxidant phytochemicals because of the convenience, low cost, simultaneous separation and detection of considerable amount of samples and the availability of new stationary phases [20].

11.1.2.1.2 Gas chromatography (GC) Despite the high resolution and sensitivity of GC, because of the lack of volatility of the majority of plant-derived antioxidants, its use in the separation has not been as popular as HPLC. Application of GC is also limited because of the difficulty of largescale separation and purification. Separation of antioxidant phytochemicals by GC has mostly been attempted for compounds in the essential oils of herbs. GC has been also used for identification of polyphenols extracted from wastewater olive oil samples [21], from modern and archeological vine derivatives [22] and grapes [23].

11.1.2.1.3 HPLC Amongst the different methods available, HPLC is preferred for the separation and quantification of polyphenolics in fruits, such as mango, apple, etc. [24–26]. Nevertheless, because of the disadvantages in detection limits and sensitivity, HPLC methods present limitations especially in complex matrix, such as crude plant extracts and environmental samples.

11.1.2.1.4 Capillary electrophoresis (CE) Although HPLC remains as the dominating separation technique for antioxidant phytochemicals, CE is gaining popularity [27]. CE has several unique advantages compared to HPLC: (i) it requires a very small sample size; (ii) it has high-efficiency due to nonparabolic fronting; (iii) it has a shorter analytical time; (iv) it is low cost, particularly when using capillary zone electrophoresis (CZE) and fused-silica capillary; and (v) it uses

272 

 11 Concentration of polyphenols by integrated membrane operations

no or only small amount of organic solvent, and therefore limits solvent waste. However, conversely, one of the major limitations of CE, compared to other techniques like GC or HPLC, is its low sensitivity in terms of solute concentration, and worse reproducibility compared to chromatographic techniques, which is caused by the short optical pathlength of the capillary used as detection cell and also by the small volumes that can be introduced into the capillary (normally, a few nanoliters) [28–30].

11.1.2.2 Concentration of polyphenols at industrial scale As it was demonstrated briefly above, the choice of methods and strategies for separation/concentration of biologically active compounds at laboratory scale varies from one to another research group and depends on the class of polyphenols studied. However polyphenolic class separations as may be achieved in trace amounts analytically cannot yet be applied on the scale required in industry. According to patents description, the traditional approaches used for concentrating of biologically active compounds (BAC), extracted from natural products, at industrial scale, include simple steam distillation and vacuum distillation, which generally require increased temperature and high energy consumption. The first is inappropriate for heat-sensitive products. These methods may also result in a loss of compounds of low molecular weight, which can be removed together with the solvent during evaporation. Another industrial technique, still widely used for polyphenols separation because of its simplicity and its value as an initial separation step, is a conventional open-column chromatography. Unfortunately, one of the major problems with this method is polyphenols sparing solubility in solvents employed in chromatography. Moreover, the polyphenols become less soluble as their purification proceeds. Poor solubility in the mobile phase used for a chromatographic separation can induce precipitation at the head of the column, leading to poor resolution, decrease in solvent flow, or even blockage of the column. Another method described in patents, particularly for polyphenols concentration from propolis extract, is the lyophilization process. However, this method shows some of the disadvantages of the previously mentioned processes, e.g., it involves a large amount of energy, comprises incubation at about −70°C, etc.

11.2 Concentration of polyphenols by integrated membrane operations Membrane technologies are successfully used to concentrate and/or selectively fractionate BAC from aqueous and organic solvent solutions, particularly soluble phenolic compounds. Membrane processes offer an improved efficiency and reduced operating cost in comparison with the traditional ones used in the food and pharmaceutical industries for concentration of juices, plant extracts and solvent recovery, as well as for treatment of polyphenols containing industrial waste liquids. For a number of industrially important applications, integrated membrane processes

11.2 Concentration of polyphenols by integrated membrane operations 

 273

have been proposed in recent years. Despite this increasing popularity of membrane technologies, the main problem for their large-scale application remains the membrane fouling, which heightens the investment costs and leads to oversized design of the membrane plants [31].

11.2.1 Membrane processes for concentration of plant extracts Polyphenols from plant materials are extracted by appropriate solvents (usually hydroalcoholic mixtures) and further treated by membrane operations in order to obtain high biologically active concentrates and to recover the solvent. Among the non-aqueous solvents, ethanol is preferred, especially when the concentrates are intended for direct food or medicinal applications. The multicomponent composition of the natural extracts poses several problems for their membrane separation: interactions among the compounds [32], insufficient selectivity [33] of the process, the membrane resistance to the organic solvent, as well as membrane fouling. In particular, nanofiltration (NF) has been intensively investigated [33–48] for concentration and fractionation of complex solutions by selecting a sequence of membranes with suitable MWCO in the range of 150 –1,000 Da. In recent years, a large number of potential applications of NF have been proposed. Many of them are focused on organic solvent nanofiltration (OSN) for concentration of polyphenols from different plant extracts [39–48]. UF [34, 49–52 ] and UF-NF processes [40, 48], coupled with solid-liquid extraction, are tested for a variety of materials with application in the pharmaceutical industry. Possibilities for extraction intensification for liberating polyphenols from the solid material are also discussed [45, 47, 53, 54]. Regarding the solvent recovery (in permeates), as well as the preservation of high biological activity in the obtained retentates, the results are fully encouraging [42–46]. Fractionation and purification are rather hindered by the similar molecular weights of various compounds in the feed. Considering possible solute-membrane interactions, model systems of phenolic compounds are used to investigate the influence of the type and position of the functional groups on NF/RO performance [55]. Generally, batch operation mode is preferred, which limits the required membrane area and investment costs, but involves stronger concentration changes and pronounced flux decline. The latter remains the major difficulty for larger-scale operations [34, 36, 44], and the methodology for its control is the focus of a great number of publications concerning polyphenols [31, 56, 57]. Most of the investigations are laboratory scale UF, NF or RO [47], but few examples on an industrial scale are also available, especially for concentrating aqueous extracts [34–36]. FO [58] and OD [59] have also been considered for large-scale concentrations (roselle extracts, anthocyanins from kokum extracts). Table 11.1 reports examples of investigated membrane processes for the concentration of antioxidant compounds from aqueous (Table 11.1a) and non-aqueous (Table 11.1b) plant extracts.

UF, NF

OD

SLE SLR 1:15; TE 30 min, T not reported[34]

SLR 1:5, TE 3 h[58]

Roselle extract[34, 58]

Polypropylene

NF: Polyamide (PA); cross link. PA; PA polysulfone; Polyethersulfone; (Dow, FILMTEC, Toray, Koch, GE Osmonics, Microdyn-Nadir)

UF: Composite polyamide; Polyethersulfone; (GE Osmonics, Microdyn Nadir)

Extraction Membrane Membrane method and process conditions Material (manufacturer)

Aqueous extracts

10.2

0.0155 + semiindustrial validation 2.5

Area (m2)

semiindustrial scale hollow fiber

Flat-sheet, crossflow

Operation mode

Table 11.1 (a): Concentration of aqueous plant extracts by membrane operations

2644 mg/kg total phenolics by Folin-Ciocalteu method;

29.1 g/kg TSS total phenols; 7.1 g/kg TSS anthocyanins content assessed by the pH differential method

Feed concentration, analysis Anthocyanins: delphinidin 3xylosyl-glucoside (delphinidin 3-sambubioside or hibiscin); cyanidin 3-xylosylglucoside (cyanidin 3-sambubioside or ossypicyanin)

Composition

Extract characterization

424 μmol Trolox/g by DM ORAC

5.96

405 μmol 6.25-fold Trolox/g by oxygen radical absorbance capacity (ORAC).

Antioxidant activity

Degree of concentration

274   11 Concentration of polyphenols by integrated membrane operations

Spiral module

0.9[36] 0.6[37]

Bark residues SLE NF from mate SLR 1:100 tree[36, 37] or 3:100. TE 6.5 or 3 min T 82,100°C HL2521TF (Osmonics, USA)

UF: hollow fiber

0.16

UF: polysulfone (China Blue Star Mem. Techn.)

Anthocyanins: cyanidin-3-glucoside and cyanidin3-sambubioside

Composition

1.6 chlorogenic acid equiv (mg CAE/ml) by Folin-Ciocalteu method, component analysis by HPLC

ABTS: 32.28 mM trolox in the permeate

Not reported

Not reported

Antioxidant activity

2.5

54

Degree of concentration

(Continued)

Gallic, chlorogenic 120.7 (EC50– 6.8[36] (5-caffeoyl quinic), mg sample/g 3.3[37] 3,4-dihydroxyben- DPPH) g DPPH zoic, 4,5-dicaffeoyl quinic acid; epigallocatechin gallate

1974.1 mg/l Flavonoids/ total flavonoids, anthocyanins 194.1 mg/l anthocyanins by colorimetric UV-VIS analysis

NF: Spiralwound

NF: polyamide, 1.6/2.1/ polypipe razine 2.6 amide, polyethersulfone (Mycrodin Nadir, Filmtec/Dow)

Blood orange PE peels[35, 52]

NF UF

49 mg/l total antocyanins, by colorimetric UV-VIS analysis

Cellulose triacetate 1.14 × 10–2 flat (Osmotek, Inc., membrane Corvallis, OR, USA) module

PESLR 1:2TE FO Garcinia indica Choisy and T not specified (kokum)[59]

Area (m2)

Extract characterization Feed concentration, analysis

Extraction Membrane Membrane method and process conditions Material (manufacturer) Operation mode

Aqueous extracts

11.2 Concentration of polyphenols by integrated membrane operations 

 275

NF

UF

SLE SLR 1:8; TE 15 h T 50°C

SLE SLR 1:25; TE 90 min T 25°C

Soybeans[38]

Castanea sativa leaves[33]

Area (m2)

0.07

Flat-sheet, crossflow incl. batch redilution of the retentate

tangential flow spiral module

Operation mode

33.8% of the freeze died solid, det. by FolinCiocalteau assay using gallic acid as standard

Total isoflavones: 1300 (μg/g dry mass) by HPLC analysis

Feed concentration, analysis

Antioxidant activity

Benzoic and cinnamic acids, flavonoids, and ellagic acid and gallic acid structures

1.73 for total isoflavones (β-glucosides and malonyl glucosides)

Degree of concentration

DPPH: 1.36 0.33 g/l; ABTS:0.75 g Tr/g; FRAP:1.24 mmol/g; Reducpower: 3.6 μmol ferric sulfate/g

Malonyl and Bioactivity not β-glucosides reported (genistin, daidzin, glycitin), aglycones (daidzein genistein, glycitein)

Composition

Extract characterization

Superscript numbers refer to studies listed in the reference list for this chapter. SLE, solid-liquid extraction; PE, press extraction; SLR, Solid/liquid ratio; TE, time of extraction; T, temperature; UF, ultrafiltration; NF, nanofiltration, RO, reverse osmosis; FO, forward osmosis; OD, osmotic distillation.

Modified polyethersulfone (Omega membranes, Minisette, Pall Filtron)

Polyvinylidene 0.9 difluoride (PVDF) (GE Osmonics, USA)

Extraction Membrane Membrane method and process conditions Material (manufacturer)

Aqueous extracts

Table 11.1 (a): Concentration of aqueous plant extracts by membrane operations (Continued)

276   11 Concentration of polyphenols by integrated membrane operations

Acetone/ UF (as H2O, 50:50; preparation SLR≈1:15; technique) TE 30 min; T room

NF

UF

50% EtOH T 35–40°C

Ginko biloba[39]

Persimmon[50] MeOH SLR 1:12 TE 30 min T 90°C

0.004

Polysulfone Not (Tianjin, Tianfang, reported China)

Not specified

Hollow fiber

Flat-sheet crossflow

Centrifugal UF membrane devices, diafiltration

Area (m2) Operation mode

Semipermeable, Not not specified, reported (Millipore, Bedford, USA)

Membrane Membrane process Material (manufacturer)

Almond skins[49]

Non-aqueous Solvent, and mixed SLE solvents conditions

Table 11.1 (b): Concentration of non-aqueous plant extracts by membrane operations

Tot. polyph. 91.1% of the extract, Folin Denis, GC and HPLC comp. anal.

Total flavones, analysis not specified

Comp. anal. by RP-LC-PAD/ ESI-MS, NP-LC-PAD, MALDI-TOF, FIA-ESIMS/MS

Feed concentration, analysis

Antioxidant activity

Flavone glycosides, derivatives of quercetin, kaempferol and isorhamnetin. Polyphenols, including condensed tannin

Hydroxyl radical scaveng. activities

Not reported

11 low MW phenolic Not reported comp.: benzoic acids, flavan-3-ols mono-, oligomers, flavonol and flavanone glycosides; high MW proanthocyanidins: di-decamers

Composition

Extract characterization

(Continued)

93.4% cond. tannin in ret. 87.4% tot. phen. in perm.

Not reported

Fractionation only

Degree of concentration 11.2 Concentration of polyphenols by integrated membrane operations 

 277

EtOH-H2O NF 8:2[41], H20; SLR 1:4; TE 7 days; EtOH-H2O 7:3[7, 42] SLR 1:10;20; TE 15 min; 24 h

Propolis

[7, 41, 42]

[40, 51]

EtOH-H2O 1:1(1st) 95% EtOH (2nd), SLR 0.2 g/ml; TE 1 h; [40] Me OH-H2O 80:20

UF[51] NF ⇒ MF ⇒ UF[40] NF ⇒ UF ⇒ MF ⇒ UF (incl. diafil.)[40] Millipore stirred UF cell[51] Tubular[40], tangential filtration

Tangential filtration[41] Dead-end filtration[7, 42]

Polyamide/ 0.6[41] polysulfone 0.0054 (NF90 Osmonics, [7, 42] USA)[41]; Cross-linked polyimide (Duramem, Evonik, UK)[7, 42]

Area (m2) Operation mode

Millipore type Not GS and HA[51]; reported[51] [40]NF: Polyamide 0.044[40] AFC40 UF: Polyethersulfone ES404; UF: Polysulfone, PU608, PU120; MF: Polyvinilidene fluoride FP200 (PCI Membrane Systems)

Membrane Membrane process Material (manufacturer)

Grape seeds

Non-aqueous Solvent, and mixed SLE solvents conditions

in EtOH/H2O: 98.74 mg GAE/g total phen., FolinCiocalteau[41]; in H2O: 36.57 mg/g; HPLC comp. anal.; 19.9 mg GAE/ml[7, 42]

Total polyphenols by Folin-Ciocalteu method[40, 51] and HPLC[40]; 142 mg/l[40] total phenols

Feed concentration, analysis

Antioxidant activity

flavones, flavonols, 93% DPPH[42] flavanones, dihydroflavonols (ex. pinocembrin, pinobanksin, caffeic acid, quercetin, pinobanksinmethilether, p-coumaric acid, crysin

Oligomeric Not reported proanthocyanidins; polyphenolic acids: cinnamic (coumaric, caffeic, ferulic, chlorogenic, neochlo rogenic); flavonoids: flavan-3-ol (catechin, epicatechin, their polymers or esters with galactic acid or glucose)

Composition

Extract characterization

Table 11.1 (b): Concentration of non-aqueous plant extracts by membrane operations (Continued)

1.1-fold in EtOH, 2.86-fold in H2O[41] 2.1–3.0-fold[7, 42]

11.4% of seed weight[51] Five fractions[40] (acids, aldehydes, monomers, proanthocyanidins

Degree of concentration

278   11 Concentration of polyphenols by integrated membrane operations

Cocoa seeds[47]

Rosemary[46]

Lemon balm[45]

EtOH-H2O NF incl. 50:50, diafilt. 80:20 incl scr.CO2 SLR 1:10; TE 110 min; T 40°C[45] EtOH, SLR 1:10 TE 9 h T 25°C[46] EtOH (incl. NF, RO scr CO2); TE 1 h; T 40°C,

NF

Sideritis [43, 44] EtOH SLR 1:15 TE 2 h T room

NF: DL, HL (GEOsmonics), NF-90 (FilmtecDow) RO: SG (GE Osmonics) BW-30 (FilmtecDow)

Cross-linked polyimide Duramem (Evonik, UK)

Cross-linked polyimide Duramem (Evonik, UK)

Membrane Membrane process Material (manufacturer)

Non-aqueous Solvent, and mixed SLE solvents conditions

3.14 × 10–4

0.0054

0.0054

Dead-end

Flat-sheet, tangential[45]; Dead-end[46]

Dead-end[43] flat-sheet, tangential[44]

Area (m2) Operation mode

Chlorogrnic acid, lavandulifolioside, verbascoside, leucoseptiside A flavonoid gluosides

Composition

HPLC analysis

DPPH, EC50 0.025 g/l[46]

84% DPPH antiox. activity[43]

Antioxidant activity

Polyphenols (mono to Not decamers) measured

0.118 kg/m3[46], Rosemarinic acid 5.2–6.7 g/l[45] Rosemarinic acid, by HPLC

1.17 mg GAE/ml (17.55 mg/g solid) tot. phen. by FolinCiocalteau; 0.38 mg/ml tot. flavonoids by AlCl3 assay

Feed concentration, analysis

Extract characterization

(Continued)

Not reported, retentions > 90%

2.9–3-fold[45, 46]

3-4-fold

Degree of concentration

11.2 Concentration of polyphenols by integrated membrane operations 

 279

Hexane, acetonitril, methanol SLR 1:2.25 TE 7 days

NF UF

Dead-end

Area (m2) Operation mode

polysulfone 38.4 × and polyamide 10–4 (NF-DK; Desal 5 DK, Osmonics); diacetate of cellulose, formamide, and acetone (Osmonis INOX)

Membrane Membrane process Material (manufacturer) Composition

168 mg/g (in carnosic and MeOH) rosmarinic acids 175 mg/g (in H2O) by FolinCiocalteu as Gallic Acid Equiv.

Feed concentration, analysis

Extract characterization

DPPH: 78% in H2O 74% in MeOH

Antioxidant activity

Up to 1.6-fold

Degree of concentration

Superscript numbers refer to studies listed in the reference list for this chapter. SLE, solid-liquid extraction; PE, press extraction; SLR, Solid/liquid ratio; TE, time of extraction; T, temperature; UF, ultrafiltration; NF, nanofiltration, RO, reverse osmosis; FO, forward osmosis; OD, osmotic distillation.

Thymus capitatis[48]

Non-aqueous Solvent, and mixed SLE solvents conditions

Table 11.1 (b): Concentration of non-aqueous plant extracts by membrane operations (Continued)

280   11 Concentration of polyphenols by integrated membrane operations

11.2 Concentration of polyphenols by integrated membrane operations 

 281

11.2.2 Membrane processes for concentration of juices Membrane processes are increasingly used for clarification and concentration of thermo-sensitive fruit juices as alternative for thermal evaporation, preserving the quality of the product (odor, color, sensory properties, nutritional value and biological activity). Table 11.2 illustrates several application examples. Besides microfiltration (MF), UF, NF [18, 60–62], RO [63] and MD [57, 64, 65] have been widely applied to many types of fruit juices. An important factor in the choice of the membrane process is the achievable final concentration. For the pressure-driven membrane processes it reaches about 25–35°Brix (e.g., RO), but it is essentially higher for the MD processes (55–65°Brix). For this reason a number of integrated membrane operations, including OD [59, 66–69], MD, the coupled process MD-OD or the coupled process RO-OD [68] have been proposed, thus achieving final soluble solid content of 63–72°Brix [68]. A comparison of the different membrane methods for concentration of fruit juices, their potential for full-scale plant application and economical analysis was reported recently [70].

11.2.3 Membrane processes for recovery/concentration of polyphenols from industrial waste waters (WW) This area of application is related to the major interest for natural products, containing compounds with biological activities as polyphenols, for which membrane technologies have proven their potential for recovery and concentration [71–73], thus transforming the waste effluents or by-products to source material for high-value compounds. Membrane operations in sequential design are particularly suitable alternative for polyphenol recovery; depending on the initial concentration in the waste effluents, the resulting concentrates are more or less valuable for industrial application [74]. A typical example with potential economic impact is the olive vegetation waste water [4, 31, 56, 75–80]. The obtained concentrates are rich in polyphenols and can be further used to improve the bioactive phenol content of the virgin olive oil [81]. Other examples are industrial waste liquors from wine making industry [73, 82, 83] and wastewater from cork processing industry [84]. A number of membrane operations have been investigated such as MF, UF, NF, RO, OD and VMD (Table 11.3). MF and UF are usually applied for preliminary treatment. The polyphenols retention by UF can be enlarged by adding surfactants to the waste stream, the formed large organic compounds-surfactant structures being further subjected to UF [80]. The concentration unit is usually represented by the NF and/or RO module. Its selectivity can be improved by combination with other processes such as adsorption, preceeding or succeeding the membrane separation step [82–85].

Blood orange[66] UF ⇒ RO ⇒ OD UF ⇒ OD UF: 0.23

RO: 1.12

OD: 1.4

UF: PVDF (teries-Cor HFM-251)

RO: composite polyamide (SWC2-2521)

OD: microporous polypropylene (fiber potting material polyethylene)

0.212 gGAE/l total phenolics, Folin-Ciocalteau; Flavonoids: HPLC anal.

Flavonoids: narirutin; 2, naringin; 3, hesperidin; 4, neohesperidin

Feed concentration, Antioxidants/ analysis composition

Polyphenols characterization

9–10 Brix

ABTS assay; total antioxidant activity (TAA): 7.33 mM Trolox final (UF-RO-OD) and 7.66 mM (UF-OD)

ABTS > 22 mM Trolox (in retentate from 450 Da NF)

Final Analysis of concentration antioxidant (TSS) activity

56–60 g/l total Flavonoids and phenyl- 60 Brix antocyanins, propanoids: 46.7 g /l hesperdin, flavanones 32.4 g/l narirutin (hesperidin, narirutin); RO: spiralHPLC anal. of flavonols: quercetin wound flavanones and as rutine (quercetin(Hydranautics) anthocyanins 7-rutino-side); OD: hollowanthocyanins: fibers Celgard cyanidin-3-glucoside, (Hoechstcyanidin-3-glucosideCelanese 6’’-malonyl Corp.)

UF: tubular membrane (Koch)

NF: monotubular

NF: 0.0048

NF: TiO2 (Inopor) 750 Da and 450 Da

Module

UF: 0.162. UF: hollow fiber 38.5 × 2. flat-sheet 10–4

UF ⇒ NF

Bergamot[18]

Area

(m2)

UF: 1. polysulfone (China Blue StarMem Techn.); 2. fluoropolymer (AlfaLaval)

Integrated Membrane membrane Material process proposed

System

Table 11.2: Concentration of fruit juices by membrane operations

282   11 Concentration of polyphenols by integrated membrane operations

Apple[64, 68, 58]

Semi-ind.scale hollow fibre

Polypropylene

OD[58] 10.2

MF: polymeric (Koch, MF: 0.05 Flat-sheet USA); RO: composite RO: 0.36 HR98PP (DSS, OD: 0.036 Silkeborg, Dk); OE: polytetrafuoroethylene (Pall-Gelman TF200 Fr)

Module MD 020 CP 2N, (Microdyn) 40 polypropylene capillaries

MF ⇒ RO ⇒ OD[68]

0.1

Polypropylene64

UF ⇒ OD UF ⇒ MD UF ⇒ ODMD[64] (coupled)

OD: 1.4

OD: microporous polypropylene

116 mg/l gallic acid (GAE)[64], 643 mg GAE/kg[68] 368 mg/kg[58]; Tot. phenolics by Folin-Ciocalteu method[58, 64, 68]; comp. anal. by HPLC[64]; GC[68]

chlorogenic acid, epicatechin, phloridzin

Phenolic fraction: hydrolysable tannins, anthocyanins (delphinidin, cyanidin, pelargonidin 3-glucosides, 3,5-diglucosides)

Feed concentration, Antioxidants/ analysis composition

Polyphenols characterization

g catechin/l.total phenolics; Prussian blue OD: hollowfibers Celgard spectroph. method (HoechstCelanese Corp.)

Module

UF: 0.0046 UF: hollow fiber

Area (m2)

UF: poly(ether ether ketone)

Integrated Membrane membrane Material process proposed

Pomegranate[67] UF ⇒ OD

System

65 Brix[64], 29 g/100g (RO)[68], 53 g/ 100 g (OE)[68], 570 g/kg[58]

520.0 g/kg

(Continued)

Not measured[64]; ABTS[68]: 10.5 μmol Trolox/g; ORAC; 43 μmol Trolox/g after OE[58]

ABTS assay: 10.2 mM Trolox in OD retentate

Final Analysis of concentration antioxidant (TSS) activity 11.2 Concentration of polyphenols by integrated membrane operations 

 283

Polypropylene

Polyvinylidene fluoride

OD[58]

UF[60]

Grape[58, 60]

0.0051

MOD: polypropylene

10.2

SGMD and VMD: 0.0159 polytetrafluoroethylene (K150, Osmonics, USA)

0.015

UF ⇒ MO D[57] SGMD[65] VMD[65]

Chokeberry (CB)[57] Red currant (RC)[57, 65] Cherry (CH)[57, 65]

Area

(m2)

UF: polyethersulfone

Integrated Membrane membrane Material process proposed

System

Flat-sheet

Semi-ind. scale hollow fiber

Flat-sheet membrane Tubular microdyn, ger Flat-sheet

Module

Table 11.2: Concentration of fruit juices by membrane operations (Continued)

Anthocyanins: cyanidin 3-rutinoside, delphinidin 3-rutinoside, delphinidin 3-glucoside, cyanidin 3-glucoside; flavonols; flavan-3-ols;

1061 mg/kg total flavonoids phenolics by Folin- (catechin, epicatechin, Ciocalteu method; quercetin, anthocyanins, procyanidins), and resveratrol (3, 5, 40trihydroxystilbene),

Not measured[57]; Relative phenol levels reported[65], Folin-Ciocalteu[65] in mg GAE/l[65]; HPLC comp. anal. of anthocyanins.

Feed concentration, Antioxidants/ analysis composition

Polyphenols characterization

660 g/kg

[57]in g/100 g: CB – 63.9; RC – 65; CH – 62.4[62]: 13.7 VRF (vol. red. factor)

ORAC assay: 55 μmol/g Trolox equiv after OE

[57]ABTS assay in mg Trolox/ ml: CB:104; RC: 37.69 CH: 21.27[65] not measured

Final Analysis of concentration antioxidant (TSS) activity

284   11 Concentration of polyphenols by integrated membrane operations

Tubular, PCI (Paterson Candy Int.)

42.9 mg/l total (the sum of the HPLC anal.) flavonols; 835.8 mg/l antocianins; HPLC anal.

944 mg/l gallic acid total phenolics by Folin-Ciocalteau method

Flavonols:myricetin, 25°Brix after quercetin, kaempferol; RO Anthocyanins: delphinidin 3-glucoside, delphinidin 3-rutinoside, cyanidin 3-glucoside, cyanidin 3-rutinoside.

Not measured

ABTS assay: 13–16 mM Trolox (permeate/ retentate) after UF

Final Analysis of concentration antioxidant (TSS) activity

62–65° MW < 1000 Da: Brix coumaric, caffeic after OD[69] acid and deriv.: chlorogenic, protocatechuic acids, deriv. of 3,4-dihydroxybenzoic acid; epicatechin, catechin, and procyanidins; flavonols as glycosides of quercetin and kaempferol

Feed concentration, Antioxidants/ analysis composition

Polyphenols characterization

Superscript numbers refer to studies listed in the reference list for this chapter. MF, microfiltration; UF, ultrafiltration; NF, nanofiltration; RO, reverse osmosis; OD, osmotic distillation; MD, membrane distillation; MOD, membrane osmotic distillation; VMD, vacuum membrane distillation; SGMD, sweeping gas membrane distillation.

0.9

AFC-99 thin-film composite

Module

Blackcurrant[63] RO

Area (m2)

cellulose acetate 0.00384[61] Flat sheet[61] (Nadir Filtration 0.23[62] Tubular[62, 69] [61] GmbH) polyvinylidenefluoride (Koch Series-CorTM)

Integrated Membrane membrane Material process proposed

Kiwifruit[61, 62, 69] UF[61, 62] UF ⇒ OD[69]

System

11.2 Concentration of polyphenols by integrated membrane operations 

 285

NF: 2.5 RO: 2.5

NF: polymeric, not spec.

RO: polymeric, not spec.

RO: spiral-wound

NF: spiral-wound

RO: spiral-wound UF: tubular

VMD: flat-sheet

VMD: 0.0055

RO: 7 UF: 0.24

UF ⇒ NF ⇒ RO

Olive mill WW[76, 77]

OD: hollow fibers

OD: 1.4

MF: tubular (ceramic) spiralwound (PES) UF: spiral-wound (PS/PES) tubular (ceramic)

MF ⇒ UF ⇒ RO Incl. diafiltr. produced RO water added to MF and UF concentrates

Olive mill WW[75]

MF: tubular NF: spiral-wound

MF: 0.0048 NF: 1.6

(Module)

MF: 0.35 (ceramic) 3.8 (PES) UF: 5 (PS) 8.36 (PES) 0.35

MF: Al2O3 NF: hydrofobic polyethersulfone OD: microporous polypropylene VMD: polypropylene (PP) polyvinylidenefluoride (PVDF)

MF ⇒ NF ⇒ OD/VMD

Olive mill WW[4]

Area (m2)

MF: ceramic Tami: zirconium oxide polymeric Nadir: polyethersulfone (PES) UF: polymeric osmonics: polysulfone (PS), polyethersulfone (PES) ceramic Tami: zirconium oxide RO: composite polyamide UF: ceramic (zirkonia)

Integrated membrane Membrane process proposed Material

System

Table 11.3: Recovery/concentration of polyphenols from waste waters (WW) by membrane operations

Low molecular weight polyphenols: Hydroxytyrosol, procatechic acid, tyrosol, caffeic acid, p-cumaric acid, oleuropein

Composition

0.5–0.7, total Total phenols polyphenols by UV-Vis, Price and Butler method

0.35 (after MF), HPLC component analysis

0.212, HPLC component analysis

Feed concentration in g/l, method of analysis

Polyphenols characterization

286   11 Concentration of polyphenols by integrated membrane operations

MF ⇒ UF ⇒ NF

MF ⇒ UF ⇒ NF ⇒ ⇒ RO

MF ⇒ DMCD

MEUF (micellar enhanced UF)

Olive mill WW[78]

Olive mill WW[31, 56]

Olive mill WW[79]

Olive mill WW[80]

MF: Osmonics JX UF: Osmonics GM NF: Osmonics DK RO: Osmonics SC TF200 (Gelman Science) polytetrafluoroethylene (PTFE) polymer and supported by a polypropylene DMCD: poly(vinyldene fluoride) (PVDF)

NF: Dow filmtec and selfmade, polyamide

MF: polypropylene UF: polyethersulfone (PES) (self-made); CSM polysulfone (PS)

Integrated membrane Membrane process proposed Material

System

0.00287

MF, DMCD: 0.0028

2.51

Not specified

Area (m2)

Flat-sheet

Flat-sheet

Pilot plant spiral-wound

Flat-sheet and spiral-wound

(Module)

8.3 g tyrosol equiv./l, total phenols by UV-Vis spectrophotom. Folin-Ciocalteu method and HPLC component analysis 4.1 g of tyrosol equivalents/L), total phenols by UV-Vis, Folin-Ciocalteu method

> 0.3, only Chemical oxygen demand measured (COD)

5.9–6.6, total phen. by UV-Vis, Folin-Ciocalteu method

Feed concentration in g/l, method of analysis

(Continued)

3,4-dihydroxyphenylglycol; gallic acid; hydroxytyrosol; p-dihydroxyphenyl acetic acid; tyrosol; oleuropein; ferulic acid

Composition

Polyphenols characterization

11.2 Concentration of polyphenols by integrated membrane operations 

 287

UF, NF

UF: cellulose acetate (CA0-CA5), NF: DS5-DK polysulfone-polyamide, (GE water technology)

Tubular, lengths: 0.305 m (Millipore) or 0.604 m (Inside Céram)

(Module)

UF: 2x 0.147 NF: spiral-wound NF: 2.09

Not specified

Area (m2)

0.36–0.41 (g tannic acid/L), total phenols, Folin-Ciocalteu method

0.17–4.4 g GAE/l, Folin-Ciocalteu colorimetric method, and expressed as gallic acid equivalents (GAE).

Feed concentration in g/l, method of analysis

MW 200–1200 Da: procyanidins, monomeric flavan-3-ols (catechin, epicatechin, epigallocatechin, quercetin, quercetin-3-glucoside), gallic acids esters (epigallocatechin gallate), benzoic acids (gallic and ellagic) polypehnols tannins MW 125 Da–91 kDa

Composition

Polyphenols characterization

Superscript numbers refer to studies listed in the reference list for this chapter. MF, microfiltration; UF, ultrafiltration; NF, nanofiltration; RO, reverse osmosis; OD, osmotic distillation; MD, membrane distillation; DCMD, direct contact membrane distillation; VMD, vacuum membrane distillation; MEUF, micellar enhanced ultrafiltration.

Cork industry WW[84]

[82, 83]

Millipore membranes: Nanomax 95 and 50, PA/ PS; Osmonics: DL2540, GE2540, TF Tami: Inside Céram, titania

Integrated membrane Membrane process proposed Material

Industrial UF ⇒ NF liquids from grape pomace

System

Table 11.3: Recovery/concentration of polyphenols from waste waters (WW) by membrane operations (Continued)

288   11 Concentration of polyphenols by integrated membrane operations

11.3 References 

 289

OD [4] has been proposed for treatment of olive vegetation waters as athermal membrane separation process, operated at room temperature and atmospheric pressure, suitable for concentration of solutions containing thermo-sensitive compounds. MD is a non-isothermal membrane separation process in which water vapor transport occurs through a non-wetted porous hydrophobic membrane; in the most used configuration, DCMD [79], where the liquid in both sides of the membrane is in direct contact with the hydrophobic microporous membrane and water at a lower temperature than the liquid in the feed side is used as condensing fluid in permeate side. Integrated membrane systems including several membrane operations, have been proposed and tested on pilot and semi-industrial scales (see Table 11.3), including long-term operation and proper fouling control [31] and feasibility analysis [76]. Related patents [86–89] have been converted to commercial applications [71].

11.3 References 1. Hervert-Hernández D, García OP, Rosado JL, Goñi I. The contribution of fruits and vegetables to dietary intake of polyphenols and antioxidant capacity in a Mexican rural diet: Importance of fruit and vegetable variety. Food Res Int 2011;44:1182–1189. 2. Rondeau P, Gambier F, Jolibert F, Brosse, N. Compositions and chemical variability of grape pomaces from French vineyard. Ind Crop Prod 2013;43:251–254. 3. Ceymann M, Arrigoni E, Schärer H, Bozzi Nising A, Hurrell RF. Identification of apples rich in health-promoting flavan-3-ols and phenolic acids by measuring the polyphenol profile. J Food Compos Anal 2012;26:128–135. 4. Garcia-Castello E, Cassano A, Criscuoli A, Conidi C, Drioli E. Recovery and concentration of polyphenols from olive mill wastewaters by integrated membrane system. Water Res 2010;44:3883–3892. 5. Qin XY, Cheng Y, Yu LC. Potential protection of green tea polyphenols against intracellular amyloid beta-induced toxicity on primary cultured prefrontal cortical neurons of rats. Neurosci Lett 2012;513:170–173. 6. Biesaga M, Pyrzyńska K. Stability of bioactive polyphenols from honey during different extraction methods. Food Chem 2013;136:46–54. 7. Tylkowski B, Trusheva B, Bankova V, Giamberini M, Peev G, Nikolova A. Extraction of biologically active compounds from propolis and concentration of extract by nanofiltration. J Membrane Sci 2010;348:124–130. 8. Anastácio A, Carvalho IS. Phenolics extraction from sweet potato peels: Key factors screening through a Placket-Burman design. Ind Crop Prod 2013;43:99–105. 9. Shoji T, Akazome Y, Kanda T, Ikeda M. The toxicology and safety of apple polyphenol extract. Food Chem Toxicol 2004;42:959–967. 10. Soto-Vaca A, Gutierrez A, Losso JN, Xu Z, Finley JW. Evolution of phenolic compounds from color and flavor problems to health benefits. J Agr Food Chem 2012;60:6658–6677. 11. Ho CT, Shahidi F. Phenolics in food and natural health products: An Overview. In: Phenolic Compounds in Foods and Natural Health Products. Vol. 909. American Chemical Society: Washington, DC; 2005, 1–8. 12. Khan N, Mukhtar H. Multitargeted therapy of cancer by green tea polyphenols. Cancer Lett 2008;269:269–280.

290 

 11 Concentration of polyphenols by integrated membrane operations

13. Rizvi KBPSI. Plant polyphenols as dietary antioxidants in human health and disease. Oxid Med Cell Longev 2009;5:270–278. 14. Leiro JM, Varela M, Piazzon MC, Arranz JA, Noya M, Lamas J. The anti-inflammatory activity of the polyphenol resveratrol may be partially related to inhibition of tumour necrosis factor-α (TNF-α) pre-mRNA splicing. Mol Immunol 2010;47:1114–1120. 15. Kennedy DO, Wightman EL. Herbal extracts and phytochemicals: plant secondary metabolites and the enhancement of human brain function. Adv Nutr 2011;2:32–50. 16. Bertin L, Ferri F, Scoma A, Marchetti L, Fava F. Recovery of high added value natural polyphenols from actual olive mill wastewater through solid phase extraction. Chem Eng J 2011;171:1287–1293. 17. Upadhyay R, Ramalakshmi K, Jagan Mohan Rao L. Microwave-assisted extraction of chlorogenic acids from green coffee beans. Food Chem 2012;130:184–188. 18. Conidi C, Cassano A, Drioli E. A membrane-based study for the recovery of polyphenols from bergamot juice. J Membrane Sci 2011;375:182–190. 19. Wu JJ, Lin JC, Wang CH, Jong TT, Yang HL, Hsu SL, Chang CMJ. Extraction of antioxidative compounds from wine lees using supercritical fluids and associated anti-tyrosinase activity. J Supercrit Fluid 2009;50:33–41. 20. Tsao R, Deng Z. Separation procedures for naturally occurring antioxidant phytochemicals. J Chromatogr B 2004;812:85–99. 21. Zafra A, Juárez MJB, Blanc R, Navalón A, González J, Vílchez JL. Determination of polyphenolic compounds in wastewater olive oil by gas chromatography-mass spectrometry. Talanta 2006;70:213–218. 22. Garnier N, Richardin P, Cheynier V, Regert M. Characterization of thermally assisted hydrolysis and methylation products of polyphenols from modern and archeological vine derivatives using gas chromatography-mass spectrometry. Anal Chim Acta 2003;493:137–157. 23. Viñas P, Campillo N, Martínez-Castillo N, Hernández-Córdoba M. Solid-phase microextraction on fiber derivatization for the analysis of some polyphenols in wine and grapes using gas chromatography-mass spectrometry. J Chromatogr A 2009;1216:1279–1284. 24. Schieber A, Ullrich W, Carle R. Characterization of polyphenols in mango puree concentrate by HPLC with diode array and mass spectrometric detection. Innov Food Sci Emerg 2000;1:161–166. 25. Amico V, Chillemi R, Mangiafico S, Spatafora C, Tringali C. Polyphenol-enriched fractions from Sicilian grape pomace: HPLC-DAD analysis and antioxidant activity. Bioresource Technol 2008;99:5960–5966. 26. Alonso-Salces RM, Barranco A, Corta E, Berrueta LA, Gallo B, Vicente F. A validated solid-liquid extraction method for the HPLC determination of polyphenols in apple tissues: Comparison with pressurised liquid extraction. Talanta 2005;65:654–662. 27. Caridi A. New perspectives in safety and quality enhancement of wine through selection of yeasts based on the parietal adsorption activity. Int J Food Microbiol 2007;120:167–172. 28. Caridi D, Trenerry VC, Rochfort S, Duong S, Laugher D, Jones R. Profiling and quantifying quercetin glucosides in onion (Allium cepa L.) varieties using capillary zone electrophoresis and high performance liquid chromatography. Food Chem 2007;105:691–699. 29. Liu Q, Cai W, Shao X. Determination of seven polyphenols in water by high performance liquid chromatography combined with preconcentration. Talanta 2008;77:679–683. 30. Ignat I, Volf I, Popa VI. A critical review of methods for characterisation of polyphenolic compounds in fruits and vegetables. Food Chem 2011;126:1821–1835. 31. Stolle M. Effective fouling inhibition by critical flux based optimization methods on a NF membrane module for olive mill wastewater treatment. Chem Eng J 2011;168:1140–1148. 32. El Rayess Y, Albasi C, Bacchin P, Taillandier P, Raynal J, Mietton-Peuchot M, Devatine A. Cross-flow microfiltration applied to oenology: A review. J Membrane Sci 2011;382:1–19. 33. Díaz-Reinoso B, Moure A, Domínguez H, Parajó JC. Membrane concentration of antioxidants from Castanea sativa leaves aqueous extracts. Chem Eng J 2011;175:95–102.

11.3 References 

 291

34. Cissé M, Vaillant F, Pallet D, Dornier M. Selecting ultrafiltration and nanofiltration membranes to concentrate anthocyanins from roselle extract (Hibiscus sabdariffa L.). Food Res Int 2011;44: 2607–2614. 35. Conidi C, Cassano A, Drioli E. Recovery of phenolic compounds from orange press liquor by nanofiltration. Food Bioprod Process 2012;90:867–874. 36. Prudêncio APA, Prudêncio ES, Amboni RDMC, Murakami ANN, Maraschin M, Petrus JCC, Ogliari PJ, Leite RS. Phenolic composition and antioxidant activity of the aqueous extract of bark from residues from mate tree (Ilex paraguariensis St. Hil.) bark harvesting concentrated by nanofiltration. Food Bioprod Process 2012;90:399–405. 37. Murakami ANN, Amboni RDMC, Prudêncio ES, Amante ER, Zanotta LM, Maraschin M, Petrus JCC, Teófilo RF. Concentration of phenolic compounds in aqueous mate (Ilex paraguariensis A. St. Hil) extract through nanofiltration. LWT – Food Sci Technol 2011;44:2211–2216. 38. Benedetti S, Prudêncio ES, Mandarino JMG, Rezzadori K, Petrus JCC. Concentration of soybean isoflavones by nanofiltration and the effects of thermal treatments on the concentrate. Food Res Int 2013;50:625–632. 39. Xu L, Wang S. The Ginkgo biloba extract concentrated by nanofiltration. Desalination 2013;323: 31–38. 40. Santamaría B, Salazar G, Beltrán S, Cabezas JL. Membrane sequences for fractionation of polyphenolic extracts from defatted milled grape seeds. Desalination 2002;148:103–109. 41. Mello BCBS, Petrus JCC, Hubinger MD. Concentration of flavonoids and phenolic compounds in aqueous and ethanolic propolis extracts through nanofiltration. J Food Eng 2010;96:533–539. 42. Tsibranska IH, Peev GA, Tylkowski B. Fractionation of biologically active compounds extracted from propolis by nanofiltration. J Membrane Sci Technol 2011;1:1–5. 43. Tylkowski B, Tsibranska I, Kochanov R, Peev G, Giamberini M. Concentration of biologically active compounds extracted from Sideritis ssp. L. by nanofiltration. Food Bioprod Process 2011;89:307–314. 44. Tsibranska I, Tylkowski B. Concentration of ethanolic extracts from Sideritis ssp. L. by nanofiltration: comparison of dead-end and cross-flow modes. Food Bioprod Process 2013; 91:169–174. 45. Peev G, Penchev P, Peshev D, Angelov G. Solvent extraction of rosmarinic acid from lemon balm and concentration of extracts by nanofiltration: Effect of plant pre-treatment by supercritical carbon dioxide. Chem Eng Res Des 2011;89:2236–2243. 46. Peshev D, Peeva, LG, Peev G, Baptista IIR, Boam AT. Application of organic solvent nanofiltration for concentration of antioxidant extracts of rosemary (Rosmarinus officiallis L.). Chem Eng Res Des 2011;89:318–327. 47. Sarmento LAV, Machado RAF, Petrus JCC, Tamanini TR, Bolzan A. Extraction of polyphenols from cocoa seeds and concentration through polymeric membranes. J Supercrit Fluid 2008;45:64–69. 48. Achour S, Khelifi E, Attia Y, Ferjani E, Hellal AN. Concentration of antioxidant polyphenols from Thymus capitatus extracts by membrane process technology. J Food Sci 2012;77:C703–C709. 49. Prodanov M, Garrido I, Vacas V, Lebrón-Aguilar R, Dueñas M, Gómez-Cordovés C, Bartolomé B. Ultrafiltration as alternative purification procedure for the characterization of low and high molecular-mass phenolics from almond skins. Anal Chim Acta 2008;609:241–251. 50. Gua HF, Li CM, Xu YJ, Hu WF, Chen MH, Wan QH. Structural features and antioxidant activity of tannin from persimmon pulp. Food Res Int 2008;41:208–217. 51. Nawaz H, Shi J, Mittal GS, Kakuda Y. Extraction of polyphenols from grape seeds and concentration by ultrafiltration. Sep Purif Technol 2006;48:176–181. 52. Ruby-Figueroa R, Cassano A, Drioli E. Ultrafiltration of orange press liquor: optimization of operating conditions for the recovery of antioxidant compounds by response surface methodology. Sep Purif Technol 2012;98:255–261.

292 

 11 Concentration of polyphenols by integrated membrane operations

53. Liu D, Vorobiev E, Savoire R, Lanoisellé JL. Intensification of polyphenols extraction from grape seeds by high voltage electrical discharges and extract concentration by dead-end ultrafiltration. Sep Purif Technol 2011;81:134–140. 54. Li Y, Skouroumounis GK, Elsey GE, Taylor DK. Microwave-assistance provides very rapid and efficient extraction of grape seed polyphenols. Food Chem 2011;129:570–576. 55. Arsuaga JM, Sotto A, López-Muñoz MJ, Braeken L. Influence of type and position of functional groups of phenolic compounds on NF/RO performance. J Membrane Sci 2011;372:380–386. 56. Stoller M, Bravi M. Critical flux analyses on differently pretreated olive vegetation waste water streams: some case studies. Desalination 2010;250:578–582. 57. Koroknai B, Csanádi Z, Gubicza L, Bélafi-Bakó K. Preservation of antioxidant capacity and flux enhancement in concentration of red fruit juices by membrane processes. Desalination 2008;228:295–301. 58. Nayak CA, Rastogie NK. Forward osmosis for the concentration of anthocyanin from Garcinia indica Choisy. Sep Purif Technol 2010;71:144–151. 59. Cissé M, Vaillant F, Bouquet S, Pallet D, Lutin F, Reynes M, Dornier M. Athermal concentration by osmotic evaporation of roselle extract, apple and grape juices and impact on quality. Innov Food Sci Emerg 2011;12:352–360. 60. Kalbasi A, Cisneros-Zevallos L. Fractionation of monomeric and polymeric anthocyanins from concord grape (Vitis labrusca L.) juice by membrane ultrafiltration. J Agr Food Chem 2007;55: 7036–7042. 61. Cassano A, Donato L, Conidi C, Drioli E. Recovery of bioactive compounds in kiwifruit juice by ultrafiltration. Innov Food Sci Emerg 2008;9:556–562. 62. Cassano A, Donato L, Drioli E. Ultrafiltration of kiwifruit juice: Operating parameters, juice quality and membrane fouling. J Food Eng 2007;79:613−621. 63. Pap N, Pongrácz E, Jaakkola M, Tolonen T, Virtanen V, Turkki A, Horváth-Hovorka Z, Vatai G, Keiski RL. The effect of pre-treatment on the anthocyanin and flavonol content of black currant juice (Ribes nigrum L.) in concentration by reverse osmosis. J Food Eng 2010;98:429–436. 64. Onsekizoglu P, Bahceci KS, Acar MJ. Clarification and concentration of apple juice using membrane processes: A comparative quality assessment. J Membrane Sci 2010;352:160–165. 65. Bagger-Jørgensen R, Meyer AS, Pinelo M, Varming C, Jonsson G. Recovery of volatile fruit juice aroma compounds by membrane technology: Sweeping gas versus vacuum membrane distillation. Innov Food Sci Emerg 2011;12:388–397. 66. Galaverna G, Di Silvestro G, Cassano A, Sforza S, Dossena A, Drioli E, Marchelli R. A new integrated membrane process for the production of concentrated blood orange juice: Effect on bioactive compounds and antioxidant activity. Food Chem 2008;106:1021–1030. 67. Cassano A, Conidi C, Drioli E. Clarification and concentration of pomegranate juice (Punica granatum L.) using membrane processes. J Food Eng 2011;107:366–373. 68. Aguiar IB, Miranda NGM, Gomes FS, Santos MCS, Freitas DGC, Tonon RV, Cabral L MC. Physicochemical and sensory properties of apple juice concentrated by reverse osmosis and osmotic evaporation. Innovat Food Sci Emerg Tech 2012;16:137–142. 69. Cassano A, Jiao B, Drioli E. Production of concentrated kiwifruit juice by integrated membrane process. Food Res Int 2004;37:139−148. 70. Sotoft LF, Christensen KV, Andrésen R, Norddahl B. Full scale plant with membrane based concentration of blackcurrant juice on the basis of laboratory and pilot scale tests. Chem Eng Process 2012;54:12–21. 71. Galanakis CM. Recovery of high added-value components from food wastes: Conventional, emerging technologies and commercialized applications. Trends Food Sci Tech 2012;26:68–87. 72. Li J, Chase HA. Applications of membrane techniques for purification of natural products. Biotechnol Lett 2010;32:601–608.

11.3 References 

 293

73. Crespo JG, Brazinha C. Membrane processing: Natural antioxidants from winemaking by-products. Filtr Separat 2010;47:32–35. 74. Mudimu OA, Peters M, Brauner F, Braun G. Overview of membrane processes for the recovery of polyphenols from olive mill wastewater. Am J Environ Sci 2012;8:195–201. 75. Russo C. A new membrane process for the selective fractionation and total recovery of poly-phenols, water and organic substances from vegetation waters (VW). J Membrane Sci 2007;288:239–246. 76. Arvaniti EC, Zagklis DP, Papadakis VG, Paraskeva CA. High-Added Value Materials Production from OMW: A Technical and Economical Optimization. Int J Chem Eng 2012; Article ID 607219. 77. Paraskeva CA, Papadakis VG, Tsarouchi E, Kanellopoulou DG, Koutsoukos PG. Membrane processing for olive mill wastewater fractionation. Desalination 2007;213:218–229. 78. Zirehpour A, Jahanshahi M, Rahimpour A. Unique membrane process integration for olive oil mill wastewater purification. Sep Purif Technol 2012;96:124–131. 79. El-Abbassi A, Hafidi A, Khayet M, García-Payo MC. Integrated direct contact membrane distillation for olive mill wastewater treatment. Desalination 2012;323:31–38. 80. El-Abbassi A, Khayet M, Hafidi A. Micellar enhanced ultrafiltration process for the treatment of olive mill wastewater. Water Res 2011;45:4522–4530. 81. Servili M, Esposto S, Veneziani G, Urbani S, Taticchi A, Di Maio I, Selvaggini R, Sordini B, Montedoro GF. Improvement of bioactive phenol content in virgin olive oil with an olive-vegetation water concentrate produced by membrane treatment. Food Chem 2011;124:1308–1315. 82. Díaz-Reinoso B, González-López N, Moure A, Domínguez H, Parajó HC. Recovery of antioxidants from industrial waste liquors using membranes and polymeric resins. J Food Eng 2010;96: 127–133. 83. Díaz-Reinoso B, Moure A, Domínguez H, Parajó HC. Ultra- and nanofiltration of aqueous extracts from distilled fermented grape pomace. J Food Eng 2009;91:587–593. 84. Bernardo M, Santos A, Cantinho P, Minhalma M. Cork industry wastewater partition by ultra/ nanofiltration: A biodegradation and valorisation study. Water Res 2011;45:904–912. 85. Soto ML, Moure A, Domínguez H, Parajó JC. Recovery, concentration and purification of phenolic compounds by adsorption: A review. J Food Eng 2011;105:1–27. 86. Tornberg E, Galanakis CM. Olive waste recovery. World Intellectual Property Organization 2008, WO/2008/082343. 87. Fernandez-Bolanos J, Heredia A, Rodrıguez G, Rodrıguez R, Guillen R, Jimenez A. Method for obtaining purified hydroxytyrosol from products and by-products derived from the olive tree. World Intellectual Property Organization 2002, WO/2002/064537. 88. Crea R. Method of obtaining a hydroxytyrosol-richcomposition from vegetation water. World Intellectual Property Organization 2002, WO/2002/0218310. 89. Ibarra A, Sniderman Zagiary N. Olive Polyphenols Concentrate. US Patent 2008/0014322 A1.

12 Valorization of food processing streams for obtaining extracts enriched in biologically active compounds Carla Brazinha and Joao G. Crespo 12.1 Introduction Recently, the recovery and purification of small bioactive molecules from food processing streams has gained interest. Small bioactive molecules comprise a large variety of compounds with a molecular weight typically below 1 kDa, which includes compounds that are valuable because of their use as flavors and fragrances, as building blocks or precursors in the fine-chemistry industry [1], and compounds with antimicrobial, antioxidant or anti-carcinogenic activity (among other types of desirable biological activity) [2–6]. The recovery of these compounds is usually difficult because of their low concentration, often vestigial, and the complexity of the original matrix where they have to be recovered from, as happens in the case of agro-industrial wastes containing lignocellulosic biomass (e.g., grape and wine pomaces). The reason for the increasing demand for bioactive compounds stems from the growing consumers concern with their quality of life. Additionally, with the increasing population and ageing and weight problems over the past decades, concerns about a careful and dedicated diet has grown noticeably. Moreover, the growing awareness of the link between diet and health has converted natural extracts into a particularly attractive market. This explains why the functional foods market, as well as the nutraceutical and cosmeceutical, have grown so rapidly. The production of extracts, where different biologically active compounds (BACs) recovered are present, has been claimed as the most attractive approach. Actually, several studies have shown that extracts may exhibit higher biological activity than purified compounds [7, 8], because of positive synergetic interactions between the different compounds present. The challenge remains on the production of these extracts with a desirable balance of target constituents, free of compounds with a detrimental activity (pesticides, heavy metals), using a recovery process that allows the use of the label “natural”.

12.2 Market of the natural extracts ingredients The world market for flavors and fragrances ingredients was estimated to be €13 billion in 2006 [9] and the US market was projected to be €5.5 billion in 2014 with the markets segments of food 36%, cosmetics and toiletries 27%, veverages 15%, and a forecast to rise 3.0% annually. In food applications, the market tends to grow in order

296 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

to compensate the present reformulation of food products towards reduced sodium, sugar and fat products. Additionally, there is a market trend for more complex, exotic and authentic (natural) flavors and fragrances [10]. The EU Flavor Directive 88/388/EEC of 1988 clearly differentiated between natural (botanical origin), natural identical (biotech origin) and artificial flavoring (synthetic origin) substances. On the contrary, the currently valid EU Flavor Directive (EC) 1334/2008 states that the difference between the terms nature identical or artificial flavoring substances no longer exists and they are both defined as flavoring substances. Nevertheless the price of the distinct flavor origins is radically different. Botanical flavoring ingredients have a price value two or three times higher than those of synthetic origin, as in the case of vanillin (respectively, €15,000/kg and below €10/kg) [11] or γ-decalactone, peach aroma (respectively, $1400/kg and $75/kg) [12]. The biotech origin flavors have an intermediate market value. Botanical and biotech natural flavors, typically abundant in low-cost residues, are particularly interesting from an economic point of view. An example is the bioproduction of vanillin using natural extracts enriched in ferulic acid, which is a precursor for its biosynthesis. This natural extract may be obtained from the processing of plant by-products, particularly maize bran, rice bran, wheat bran, wheat straw and brewer’s spent grain, with high contents in ferulic acid (the general crosslinking agent of plant cell-wall materials) [13–16]. Vanillin is used in food (mainly), cosmetic and pharmaceutical applications, exhibiting also antimicrobial and antioxidant properties [11]. Another example is botanical limonene, which may be recovered from citrus by-products (orange peels) [17, 18] and also presents antimicrobial properties [11]. Natural flavors may also be obtained from food streams, e.g., from the headspace of beer fermenters [19]. Natural extract additives with recognized health benefits (antioxidant, antiinflammatory, anti-cancer properties) are used in functional foods, and as nutraceutical and cosmeceuticals ingredients. The world market for nutraceutical ingredients was projected to be €23.7 billion by 2015, spread across Asia/Pacific 37%, North America 24%, Western Europe 23% and other regions 16%. An annual increase of 7.2% is predicted because of the raise in economic prosperity of countries such as Brazil, Russia, India and China. In particular, China is expected to become the largest producer and consumer of nutraceutical ingredients by 2020, larger than the USA and Western Europe, the current key players in this market. The segment of the nutraceutical ingredient market estimated to have the fastest grow consists of herbal and botanical extracts and animal- and marine-based derivatives with an annual increase of 8.9% through 2015 [20]. The US market of cosmeceutical products was projected to be $8.5 billion in 2015, with an annual increase of 5.8%, mainly on skincare products (63%). The US market of cosmeceutical ingredients was projected to be $15 billion in 2015, with an annual increase of 6.1%, with the antioxidant category as the most important category [21]. The increase of the markets of bioactive ingredients is generally explained by the growing concerns about health and wellness issues by a population that is getting

12.3 Production of natural extracts – process and final product requirements 

 297

older, wealthier and more demanding. And hence, the label “natural” is particularly appealing. Natural extracts enriched in BAC are usually obtained from the valorization of food by-products (from fruit, cereals, vegetables, animal or marine products). Natural extracts enriched in chlorogenic acid and caffeine (phenolic compounds) may be obtained from spent coffee to be used as cosmeceutical ingredients. Xylooligosaccharides (XO) may be recovered from several plant material biomasses, such as sugarcane bagasse, brewery spent grains for nutraceutical ingredients [22]. Isolated phenolic pigments may be recovered from by-product of cocoa, cocoa hulls, as colourant ingredients with bioactive antioxidant and antiradical properties [23], such as the anthocyanins recovered from grape pomace [24]. Hyaluronic acid (HA) may be purified from fish eyeball, a marine by-product for cosmeceutical ingredients and on medical applications [25]. Small biopeptides, free amino acids (taurine, creatine, etc.) may be recovered from fish by-products for food and pharmaceutical industries [26]. Phloroglucinol, mannitol, oleic, arachidonic and eicosapentaenoic acids, and fucosterol may be recovered from macro algae for nutraceutical ingredients [27]. Because of the positive synergetic effects of the different bioactive compounds present in natural extracts [28], they are claimed to possess higher biological activities than purified compounds, which opens new opportunities in the food, nutraceutical, cosmeceutical and pharma industries.

12.3 Production of natural extracts – process and final product requirements A production process for obtaining natural extracts should take into consideration the characteristics of the complex raw material, the sensitivity of the target compounds to the processing conditions and the safety and specifications of the intended final product. The general process for obtaining natural extracts follows the scheme shown in Figure 12.1. The raw materials pretreatment steps depend on the nature of the original raw material. They may consist of milling when dealing with solid and slurry type of streams, in order to increase the interfacial area of the material, turning the target compounds more accessible, but they may also involve a coarse filtration step or centrifugation when liquid streams containing particulates are processed. Indeed, because of the sensitivity of the BAC, mild operating conditions should be used throughout the global production process. Particularly, because of the general sensitivity of bioactive compounds to oxidation, one should operate with oxygendepleted headspaces (usually nitrogen-rich headspaces), in the absence of light and/ or adding non-expensive antioxidants such as sulphur dioxide in the extraction step [29]. Owing to the sensitivity to heat, mild temperature conditions should be used, and anthocyanins are an extreme example with recommended temperatures not higher than 30–35°C [24].

298 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

Raw material pre-treatment

Solvent extraction

Purification/fractionation and concentration using membrane processes

Pervaporation/ vapour permeation

Nanofiltration

extract

Ultrafiltration Nanofiltration

(Fractionated) condensation Flavor concentrate

concentrate

Reverse osmosis

Extracting solvent

Reverse osmosis

extract extract

concentrate Extracting solvent

Drying

Figure 12.1: Production scheme of natural extracts enriched in target biologically active compounds. In the case of the production of bioflavors from food by-products, fermentation may take place between the extraction and the membrane processes

The solvent extraction aims at releasing the non-bonded target compounds from the natural matrix raw material while maintaining their intrinsic bioactivity properties. The chemical (hydrophilicity/organophilicity) nature of the solvent is a key factor that determines the efficiency of extraction. It must exhibit a high affinity towards the target compounds and, preferably, the solvent should not be viscous. Biocompatible solvents should be used, such as aqueous and aqueous-ethanolic solvents, or other biocompatible solvents such as polypropylene glycol, enabling the use of the label “natural” in the final product. Moreover, the use of organic solvents is becoming stricter with the new REACH and costly because of the disposal costs required [30]. In order to increase the recovery of free target compounds, several treatments may be performed to break down bonds between those compounds and the natural matrix raw material, namely through enzymatic, alkaline or acidic hydrolyzis. For example, ferulic acid is linked to lignin and polysaccharides in the plant cell walls through ester bonds that enzymatic (with xylanase and ferulic acid esterase [14]) or alkaline [31, 32] treatments break. An increase of the efficiency of the extraction of anthocyanins from red grape skin was observed by using pectinases, which break down pectin from the cell walls of plants, and cellulases, which hydrolyze plant polysaccharides [33], or under acidic conditions that break glycoside bonds (yet, in this

12.3 Production of natural extracts – process and final product requirements  

 299

case, some acids may cause degradation during storage) [34, 35]. Bioactive peptides may be also obtained by enzymatic hydrolyzis from fish proteins of fish by-products. Supercritical CO2 may also be considered a biocompatible solvent, with the advantage that the solvent CO2 is removed at the end of the operation by decompression, as in the recovery of phenolic pigment from cocoa hulls [23] and β-carotene from Botryococcus braunii or Dunaliella microalgae [36, 37]. The supercritical CO2 technology is selective for target organophilic compounds and the solvent is regarded as nontoxic, noninflammable and noncorrosive, in line with the label “natural” [30]. Microwave-assisted and ultrasound-assisted extractions have been also proposed, combining the typical solvent extraction with, respectively, microwave and ultrasound radiations in order to increase the efficiency and rate of extraction while maintaining the bioactivity properties of sensitive target compounds, such as anthocyanins [30, 38, 39]. The step of fractionation and concentration of the natural extracts containing the target bioactive compounds is at heart of the process. The challenge here is the development of suitable downstream processing techniques, allowing for the recovery of these compounds from complex streams without affecting their structure and function, which ultimately translates into their bioactivity. Membrane processes offer potential sustainable solutions for this problem because they can operate under mild conditions of temperature, pressure and stress, without involving the addition of any mass agents such as solvents, avoiding product contamination and preserving the biological activity of the compounds recovered. The large variety of membrane materials available, as well as the diversity of membrane processes developed, underlines one of the strengths of membrane separations: the possibility of designing and fine tuning the membrane and the membrane process for a specific task. Purification of the natural extracts produced may be also required for the removal of undesirable compounds that may interact with target bioactive compounds, decreasing the overall bioactivity of the extract [40] (e.g., catechin and epicatechin in grape pomace may react with sugars and proteins to give glycosides and polypenolic proteins, respectively [30]). For assuring the quality of the final product, the removal of pathogenic microorganisms (mycotoxins) and chemical contaminants (heavy metals, pesticides, ethyl carbamate) is also required. Contaminants restricted by the Commission Regulation (EC) No 1881/2006 of 19 December 2006, setting maximum levels for certain contaminants in foodstuffs in the EU and by the Federal Food, Drug, and Cosmetic Act (FFDCA) of the US FDA [41], should be removed below legal limits. During the final step of drying, the solvent may be removed and the final product (ingredient) is obtained either as a solid powder or in a concentrated liquid phase. When aqueous-based extract fractions are obtained they may be concentrated in a first step by reverse osmosis (RO). For media with relatively low ionic strengths this process has clearly revealed as the most economic option. After a preconcentration by RO the extracts may be dried (and encapsulated) in a spray dryer. One of the key issues that requires an optimization study is the extent of preconcentration that

300 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

should be achieved by RO because, above a specific threshold value, solvent removal by RO becomes uneconomical. Drying is often used in order to stabilize the bioactive compounds present in the extracts. The most common drying technique is spray-drying, which may employ, when appropriate, stabilizing agents compatible with food, cosmetic and pharmaceutical applications, such as maltodextrin, hydroxypropyl methylcellulose (HPMC), and polyvinylpyrrolidone (PVP). The bioactive compounds of the extracts may also be incorporated in a formulated product, through emulsion solubilization/ encapsulation into carrier materials for protection, controlled release or easier incorporation into liquid dispersions.

12.4 Fractionation, concentration and purification of BAC with membrane-processing techniques 12.4.1 Fractionation with pervaporation/vapor permeation Aroma compounds are small molecules (typically below 170 Da) with a nonnegligible vapor pressure. Natural flavors/aroma compounds are present in very complex natural matrices, together with hundreds of other flavors at trace concentrations. As an example, the aroma profile of wine may contain 600 –800 volatile aroma compounds [42 ]. Additionally, natural aromas are highly diluted and are thermo-sensitive compounds. Organophilic pervaporation is a particularly suitable process for aroma recovery because it does not require the use of any additional extracting agent, as happens in other recovery techniques such as absorption and liquid-liquid extraction processes, including membrane-based solvent extraction [43]. Organophilic pervaporation/vapor permeation (VP) is also particularly suitable for concentrating dilute compounds, because the separation process is based on selective solute-membrane molecular interactions, it operates under mild conditions (pH and temperature) and complex media containing particles and colloids may be processed directly because intrapore fouling does not occur. Compared to traditional evaporative techniques, organophilic pervaporation/VP is: (i) generally more selective to aroma compounds, and (ii) particularly more selective to organophilic aroma compounds with valuable organoleptic properties (such as long chain esters, ketones), obtaining a final product with high-quality. An additional advantage is the flexible possibilities of obtaining different permeate compositions as a result of the selection of the operating parameters: operating feed parameters, parameters related to membrane transport (e.g., membrane chemistry and morphology) and downstream pressure. Optimization of such parameters is essential to assure a target permeate composition, directly linked to the optimization of the overall aroma recovery process [44]. Organophilic pervaporation/VP may be applied in post-reaction recovery processes but it may also be integrated in ongoing reaction/fermentation processes,

12.4 Fractionation, concentration and purification of BAC 

 301

not compromising the viability of the cells when operated properly [11, 45–47]. The integrated bioconversion and recovery processes may relieve most common product inhibition effects and, when applicable, the degradation of the product of interest in the fermentation media. Nevertheless, organophilic pervaporation/VP has several challenges to overcome. A major disadvantage are the low fluxes of target compounds obtained when compared with evaporative techniques, for the same feed temperature and feed composition, because non-porous membranes represent an important additional barrier for mass transport [44]. Increasing feed temperature would increase the DF and increase the partial fluxes of target compounds, but this is not a solution for thermosensitive aroma compounds. Moreover, high diffusivities of water in the commonly used pervaporation membranes – such as polydimethylsiloxane (PDMS), polyoctylmethylsiloxane (POMS), and poly(1-trimethylsilyl-1-propyne) (PTMSP) – decrease the values of selectivity towards the target aroma compounds present in dilute aqueous media [44, 48, 49]. In order to reduce the importance of the diffusion step in the mass transport, composite membranes with very thin non-porous layers have been used for increasing fluxes to a certain extent. Limitations of external mass transfer in the feed compartment are frequently observed when recovering compounds with a high affinity for the pervaporation membrane. Feed side concentration polarization of solutes is particularly severe when solutes have high affinity towards the membrane. This phenomenon occurs when the transport of a solute in the feed boundary layer towards the membrane is not fast enough to compensate the high sorption of the solute occurring at the feed side of the membrane, decreasing its concentration in the boundary layer near the membrane surface, and hence, its DF and flux. This phenomenon is commonly the bottleneck in organophilic pervaporation processes for the recovery of aroma compounds with high sorption coefficients [50]. In fact, the measured separation factors of hydrophobic organics can be 10–20% of their intrinsic separation factors in the absence of concentration polarization [51]. When processing streams sensitive to shear stress, as may happen in the hybrid process of pervaporation integrated with ongoing fermentation, operation under gentle fluid dynamic conditions favors the development of concentration polarization of target solutes [50]. In order to minimize feed side polarization of concentration, appropriate fluid dynamic conditions have to be used notably through the development of membrane modules with an improved design. Organophilic pervaporation/VP enables significant energy savings as compared to more classic evaporative processes, as only a small fraction of the feed is transported through the membrane to the vacuum downstream compartment where it is condensed. Nevertheless, organophilic pervaporation may become costly because of the energy involved in maintaining a controlled permeate pressure in the downstream compartment (usually relatively low) and the cooling down of the permeate stream in the condenser(s), particularly when non-condensable gases are produced in fermentation processes permeating the membrane. In order to improve the economical

302 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

viability of pervaporation/VP processes, there is a need for alternative ways to capture the permeating vapors, using techniques that do not require the energy for phase transition. Microencapsulation, possibly using supercritical CO2 fluids, has been recently proposed but not yet proved to be efficient on a large scale. Actually, sweeping gas pervaporation has the major advantage of being easier to integrate with these type of processes, when compared with vacuum pervaporation. Examples of aroma recovery by organophilic pervaporation, for the valorization of food product streams and food by-products, were mentioned above, namely the recovery of the beer aroma profile and limonene. Vanillin may be sustainably produced by biosynthesis via bioconversion of a natural extract enriched in ferulic acid, using an integrated fermentation-pervaporation process (see Figure 12.2). In order to increase the efficiency of the microbial production of vanillin, the integration of the bioconversion with a downstream technique has been proposed in the literature [11, 47]. This integrated process may solve the vanillin inhibition effect and the vanillin high degradation into vanillin alcohol or vanillic acid during the fermentation process. Vanillin recovery by organophilic pervaporation is rather attractive because, in addition to the general advantages of using pervaporation for aroma recovery from complex media, vanillin has the unique possibility of being recovered in one single step, as solid and free of contaminants, because of its high melting point, as shown in [52].

12.4.2 Extract fractionation and purification by nanofiltration Nanofiltration is a membrane process particularly adequate for the recovery of low molecular weight bioactive compounds. Nanofiltration membranes have MWCO values with a range 200–1000 Da, making possible the separation of small bioactive compounds from larger molecules and, also, the elimination of small contaminants from target bioactive species.

Bioreactor

Pervaporation module

Recirculation pump

P T

Condenser

Solid vanillin

Vacuum pump

Figure 12.2: Schematic representation of the integrated fermentation-pervaporation process for the bio-production of vanillin. Pure vanillin recovered from fermentation media in a single step pervaporation

12.4 Fractionation, concentration and purification of BAC 

Recovery yield of hydroxytyrosol (extract/feed) = 70 %

 303

Water 0.18 m3/h

2 m3/h

Water

2.4 kg/h

Spray dryer

Olive cake residue

Final product 0.2 m3/h

0.18 m3/h

Olive cake 2 m3/h

500 kg/h

1.8 m3/h

1.62 m3/h

Water Mixing tank

Centrifuge

Nanofiltration

Reverse Osmosis

Figure 12.3: Diagram of the process for the production of extracts rich in hydroxytyrosol, starting from the olive cake byproduct available at olive mills

The mechanism of transport/rejection of solutes through these membranes is not solely supported on size-exclusion phenomena. Actually, there is a large variety of solute molecular characteristics (size, geometry, dipole moment, potential for establishing Coulombic or Van der Waals interactions) that determine their interaction with the membrane. These interactions will depend also on the membrane surface chemistry and structure and on the environmental conditions used. Relevant parameters may be optimized: membrane material, temperature, pH, ionic strenght and transmembrane pressure. A particular attention shall be given to the fluid dynamic conditions employed, which will determine conditions for mass transfer near the membrane interface and, consequently, the local concentration of the various chemical species present in the media. An interesting example of the use of nanofiltration for the recovery of bioactive natural extracts is illustrated in Figure 12.3. In this process, the solid/slurry by-product stream resulting from the production of olive oil is used as a source of valuable bioactive compounds, such as hydroxytyrosol and tyrosol. These compounds, together with other bioactive molecules extracted by aqueous leaching, are recovered in the supernatant stream of this process. The problem results from the fact that, beside the extraction of compounds with desirable bioactivity, there is a number of other compounds, with a higher molecular weight, that were identified as detrimental from a bioactivity perspective. Figure 12.4 shows a chromatogram that illustrates the complexity of the raw extract and how this extract was cleared from higher (undesirable) compounds by using a nanofiltration process. In this particular case, a tight nanofiltration membrane (Desal DK) with a cut-off of approximately 250 Da assured the permeation of hydroxytyrosol, tyrosol and other desirable small phenolic compounds, and the retention of higher molecular weight compounds, including pesticides (if present in the original extract) and heavy metals. The permeate of the nanofiltration process, rich in small phenolic compounds, can be additionally concentrated by RO and dried and encapsulated in a spray dryer.

304 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

1400

???

???

1200

mA U

1000 800 600 400

Feed

200 0

Extract 0

10

20

30

40

50

60

70

80

90

100

Minutes

100 90 80 70 60 50 40 30 20 10 0

Tests performed with animal cell cultures infected with Ehrlichia ruminantum

Cell visbility (%)

Bacterial cell visbility (%)

Figure 12.4: Chromatogram of the original raw extract (in red) and of the extract after processing by nanofiltration (in blue). It can be clear seen that compounds with higher molecular weight are excluded from the final extract

100 90 80 70 60 50 40 30 20 10 0

Tests performed with cancer cells from the colon (cell line HT 29)

0 10 20 30 40 50 60 70 80 90 100

0 10 20 30 40 50 60 70 80 90 100

mg HT/l

mg HT/l

Figure 12.5: Left: In-vitro anti-inflammatory impact of the extract produced expressed in terms of its hydroxytyrosol concentration; Right: In-vitro anti-carcinogenic impact of the extract produced expressed in terms of its hydroxytyrosol concentration

The final product, a highly concentrated aqueous extract or a nano/micro particle powder obtained in the spray dryer, may be used in functional foods, nutraceuticals and cosmeceuticals. The bioactivity of these products (see Figure 12.5), which illustrates their anti-inflammatory and anti-carcinogenic potential, makes them extremely attractive for the pharmaceutical and cosmetic industries, as patients and clients are more and more interested in using compounds with a “natural” origin and label. Another interesting process that demonstrates the potential of nanofiltration for the recovery of valuable bioactive molecules is represented in Figure 12.6. In this process, a by-product stream obtained during the production of vegetable oils, the “deodistillate”, although extremely rich in fito-sterols is usually heavily contaminated with pesticides. Unfortunately, the separation of sterols and pesticides is

12.4 Fractionation, concentration and purification of BAC 

 305

Enzyme for reuse Esterification reaction UF

Objective: Recovery of enzyme for reuse in consecutive batch reactions

Objective: Enzymatic conversion of sterols in steryl esters NF Permeate waste (with pesticides)

Retentate rich in steryl esters (free of pesticides) Objective: Obtain an extract rich is steryl esters and free of pesticides

Figure 12.6: Diagram of the process for the production of extracts rich in steryl esters, free of pesticides, starting from a deodestillate byproduct stream from the vegetable oil production

not possible by nanofiltration because these types of compounds have very similar molecular weights and their resolution by direct nanofiltration is not efficient. The approach followed in this process involved, in the first place, the conversion of the fito-sterols to steryl esters, by enzymatic esterification with fatty acids present in the deodestillate, using an appropriate enzyme (an esterase). This reaction has two goals: (i) the conversion of sterols to steryl esters, which are known to be better absorbed by the human organism because of its easier transport through biological membranes; (ii) the increased molecular weight of steryl esters (higher than the original sterols) make their separation from pesticides possible if an adequate membrane process is specifically designed for this purpose. As can be seen in Figure 12.6, after the enzymatic reaction (80% of conversion yield for the sterols present in the reaction media), the enzyme is removed from the media by ultrafiltration, for reuse, while the stream containing the steryl esters is further processed for the removal of pesticides. As these compounds have now a molecular weight lower than the steryl esters, and as they are present in relatively low concentrations (ppm or ppb range), the best way to remove them is by dia-nanofiltration. In this process, a nanofiltration membrane is used to retain the compounds of interest (the steryl esters) while permeating the contaminants (the pesticides). In order to remove these contaminants to extremely low concentrations, compatible with EU legislation, a fresh biocompatible solvent is permanently added to the feed phase. By using this procedure the pesticides are washed out by the solvent from the stream containing the steryl esters, which are retained by the nanofiltration membrane. It is important to stress that this example involves a more complex situation because the nanofiltration membrane has to be stable, for long periods of operation, in solvents that are not aqueous-based. Solvent resistant nanofiltration membranes are therefore required to assure the success of this type of processes.

306 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

The literature describes other processes where solvent resistant nanofiltration membranes are required. In [53], ethanolic rosemary extracts containing caffeic and rosmarinic acids, were produced. Dia-nanofiltration using a DuramemTM membrane with MWCO of 200 Da enabled a partial separation of caffeic from rosmarinic acid, and hence, a fractionated extract purified in rosmarinic acid was obtained.

12.5 Concluding remarks The use of membranes for the recovery of small bioactive molecules is expected to grow significantly within the next years, in particular in what refers to the recovery of high added value compounds with impact on human health. The discovery of small molecules with desirable bioactive properties, present in natural matrices such as plants and marine sources, will boost the need for recovery processes regarded as clean and sustainable and which allow the use of the label “natural” in the final product. Membrane processes perfectly meet this demand because of the mild conditions under which they operate. Several membrane processes are expected to play a role in this field. Pervaporation, VP, nanofiltration, dia-nanofiltration and RO were discussed in this chapter in order to illustrate their potential use, but other membrane technologies, such as forward osmosis (FO), membrane distillation (MD) and electrodialysis, are expected to become more widely used when targeting electrically charged compounds (electrodialysis) or when aiming to produce highly concentrated extracts under mild temperature conditions (FO, MD).

12.6 References 1. Swift KAD. Catalytic transformations of the major terpene feedstocks. Top Catal 2004;27: 143–155. 2. Karakaya S. Bioavailability of phenolic compounds. Crit Rev Food Sci 2004;44:453–464. 3. Unno T, Sugimoto A, Kakuda T. Scavenging effect of tea catechins and their epimers on superoxide anion radicals generated by a hypoxanthine oxidase system. J Sci Food Agr 2000;80:601–606. 4. Goya L, Mateos R, Bravo L. Effect of the olive oil phenol hydroxytyrosol on human hepatoma HepG2 cells – Protection against oxidative stress induced by tert-butylhydroperoxide. Eur J Nutr 2007;46:70–78. 5. Vismara R, Vestri S, Kusmic C, Barsanti L, Gualtieri P. Natural vitamin E enrichment of Artemia salina fed freshwater and marine microalgae. J Appl Phycol 2003;15:75–80. 6. Andrich G, Nesti U, Venturi F, Zinnai A, Fiorentini R. Supercritical fluid extraction of bioactive lipids from the microalga Nannochloropsis sp. Eur J Lipid Sci Tech 2005;107:381–386. 7. Saucier CT, Waterhouse AL. Synergetic activity of catechin and other antioxidants. J Agr Food Chem 1999;47:4491–4494. 8. Chen CY, Milbury PE, Kwak HK, Collins FW, Samuel P, Blumberg JB. Avenanthramides and phenolic acids from oats are bioavailable and act synergistically with vitamin C to enhance hamster and human LDL resistance to oxidation. J Nutr 2004;134:1459–1466.

12.6 References 

 307

9. Guentert M. The flavour and fragrance industry – past, present and future. In: Berger RG, ed. Flavours and Fragrances, Chemistry, Bioprocessing and Sustainability. Springer-Verlag: Berlin, Germany; 2007;1–14. 10. Brochure of Flavors & Fragrances, US Industry Study with Forecasts for 2014 & 2019, Study #2732, http://www.freedoniagroup.com/brochure/27xx/2732smwe.pdf (Accessed 21 May 2013). 11. Berger RG. Biotechnology of flavours – the next generation. Biotechnol Lett 2009;31:1651–1659. 12. Lipnizki F, Olsson J, Trägårdh G. Scale-up of pervaporation for the recovery of natural aroma compounds in the food industry. Part 1: simulation and performance. J Food Eng 2002;54: 183–195. 13. Tilay A, Bule M, Kishenkumar J, Annapure U. Preparation of ferulic acid from agricultural wastes: its improved extraction and purification. J Agr Food Chem 2008;56:7644–7648. 14. Shin HD, McClendon S, Le T, Taylor F, Chen RR. A complete enzymatic recovery of ferulic acid from corn residues with extracellular enzymes from Neosartorya spinosa NRRL185. Biotechnol Bioeng 2006;95:1108–1115. 15. Salgado JM, Maxa B, Rodríguez-Solana R, Domínguez JM. Purification of ferulic acid solubilized from agroindustrial wastes and further conversion into 4-vinyl guaiacol by Streptomyces setonii using solid state fermentation. Ind Crop Prod 2012;39:52–61. 16. Moreira MM, Morais S, Barros AA, Delerue-Matos C, Guido LF. A novel application of microwaveassisted extraction of polyphenols from brewer’s spent grain with HPLC-DAD-MS analysis. Anal Bioanal Chem 2012;403:1019–1029. 17. Kulkarni PS, Brazinha C, Afonso CAM, Crespo JG. Selective extraction of natural products with benign solvents and recovery by organophilic pervaporation: fractionation of D-limonene from orange peels. Green Chem 2010;12:1990–1994. 18. Sahraoui N, Vian MA, El Maataoui M, Boutekedjiret C, Chemat F. Valorization of citrus by-products using Microwave Steam Distillation (MSD). Innov Food Sci Emerg 2011;12:163–170. 19. Scott JA, Cooke DE. Continuous gas (CO2) stripping to remove volatiles from an alcoholic beverage. J Am Soc Brew Chem 1995;53:63–67. 20. World Nutraceutical Ingredients. Industry Study with Forecasts for 2015 & 2020, Study #2799, November 2011. http://www.freedoniagroup.com/brochure/27xx/2799smwe.pdf ) (Accessed May 21, 2013. 21. Cosmeceuticals, US Industry Study with Forecasts for 2015 & 2020, Study #2758, July 2011. http://www.freedoniagroup.com/brochure/27xx/2758smwe.pdf (Accessed May 21, 2013). 22. Moure A, Gullón P, Domínguez H, Parajó JC. Advances in the manufacture, purification and applications of xylo-oligosaccharides as food additives and nutraceuticals. Process Biochem 2006;41:1913–1923. 23. Arlorio M, Coïsson JD, Travaglia F, Varsaldi F, Miglio G, Lombardi G, Martelli A. Antioxidant and biological activity of phenolic pigments from Theobroma cacao hulls extracted with supercritical CO2. Food Res Int 2005;38:1009–1014. 24. Cacace JE, Mazza G. Extraction of anthocyanins and other phenolics from black currants with sulfured water. J Agr Food Chem 2002;50:5939–5946. 25. Muradoa MA, Montemayor MI, Cabo ML, Vázquez JA, González MP. Optimization of extraction and purification process of hyaluronic acid from fish eyeball. Food Bioprod Process 2012;90: 491–498. 26. Ferraro V, Cruz IB, Jorge RF, Malcata FX, Pintado ME, Castro PML. Valorisation of natural extracts from marine source focused on marine by-products: A review. Food Res Int 2010;43:2221–2233. 27. Andrade PB, Barbosa M, Matos RP, Lopes G, Vinholes J, Mouga T, Valentão P. Valuable compounds in macroalgae extracts. Food Chem 2013;138:1819–1828. 28. Saucier CT, Waterhouse AL. Synergetic activity of catechin and other antioxidants. J Agr Food Chem 1999;47:4491–4494.

308 

 12 Valorization of food processing streams for obtaining extracts enriched in BACS

29. Cacace JE, Mazza G. Mass transfer process during extraction of phenolic compounds from milled berries. J Food Eng 2003;59:379–389. 30. Shi J, Nawaz H, Pohorly J, Mittal G, Kakuda Y, Jiang Y. Extraction of polyphenolics from plant material for functional foods – engineering and technology. Food Rev Int 2005;21:139–166. 31. Inglett GE, Chen D. Antioxidant activity and phenolic content of air-classified corn bran. Cereal Chem 2011;88:36–40. 32. Bauer JL, Harbaum-Piayda B, Schwarz K. Phenolic compounds from hydrolyzed and extracted fiber-rich by-products. LWT - Food Sci Technol 2012;47:246–254. 33. Maier T, Göppert A, Kammerer DR, Schieber A, Carle R. Optimization of a process for enzyme-assisted pigment extraction from grape (Vitis vinifera L.) pomace. Eur Food Res Technol 2008;227:267–275. 34. Vatai T, Škerget M, Knez Z. Extraction of phenolic compounds from elder berry and different grape marc varieties using organic solvents and/or supercritical carbon dioxide. J Food Eng 2009;90: 246–254. 35. Metivier RP, Francis FJ, Clydesdale FM. Solvent extraction of anthocyanins from wine pomace. J Food Sci 1980;45:1099–1100. 36. Mendes RL, Nobre BP, Cardoso MT, Pereira AP, Palavra AF. Supercritical carbon dioxide extraction of compounds with pharmaceutical importance from microalgae. Inorg Chim Acta 2003;356: 328–334. 37. Avron M, BenAmotz A. Production of Glycerol, Carotenes and Algae Meal, US Patent 1980; 4199895-A. 38. Yang Z, Zhai W. Optimization of microwave-assisted extraction of anthocyanins from purple corn (Zea mays L.) cob and identification with HPLC-MS. Innov Food Sci Emerg 2010;11:470–476. 39. Vilkhu K, Mawson R, Simons L, Bates D. Applications and opportunities for ultrasound assisted extraction in the food industry – A review. Innov Food Sci Emerg 2008;9:161–169. 40. Brazinha C, Crespo JG. Membrane processing: Natural antioxidants from winemaking by-products. Filtr Separat 2010;47:32–35. 41. U.S. Food and Drug Administration (Accessed May 23, 2013, at http://www.fda.gov/food/ foodsafety/foodcontaminantsadulteration/default.htm). 42. Rapp A. Natural flavours of wine: correlation between instrumental analysis and sensory perception. J Anal Chem 1990;337:777–785. 43. Bocquet S, Viladomat FG, Nova CM, Sanchez J, Athès V, Souchon I. Membrane-based solvent extraction of aroma compounds: Choice of configurations of hollow fiber modules based on experiments and simulation. J Membrane Sci 2006;281:358–368. 44. Schäfer T, Crespo JG. Aroma recovery by organophilic pervaporation. In: Berger RG, ed. Flavours and Fragrances, Chemistry, Bioprocessing and Sustainability. Springer-Verlag, Berlin, Germany; 2007;427–437. 45. Schäfer T, Crespo JG. Extraction of Aromas from Active Fermentation Reactors by Pervaporation. In: Bélafi-Bakó K, Gubicza L, Mulder M, eds. Integration of Membrane Processes into Bioconversions. Kluwer Academic Publishers: New York; 2000;177–186. 46. Vane LM. A review of pervaporation for product recovery from biomass fermentation processes. J Chem Technol Biot 2005;80:603–629. 47. Torres BR, Aliakbarian B, Torre P, Perego P, Domínguez JM, Zilli M, Converti A. Vanillin bioproduction from alkaline hydrolyzate of corn cob by Escherichia coli JM109/pBB1, Enzyme Microb Tech 2009;44:154–158. 48. Schäfer T, Crespo JG. Mass transport phenomena during the recovery of volatile compounds by pervaporation. In: Barbosa-Cánovas G, Vélez-Ruíz J, Welti-Chanes J, eds. Transport Phenomena in Food Processing. CRC Press LLC: Boca Raton, FL; 2002;247–264. 49. Pereira CC, Ribeiro CP Jr, Nobrega R, Borges CP. Pervaporation recovery of volatile aroma compounds from fruit juices. J Membrane Sci 2006;274:1–23.

12.6 References 

 309

50. Schäfer T, Crespo JG. Vapour permeation and pervaporation. In: Afonso CN, Crespo JG, eds. Green Separation Processes. Wiley-VCH: Weinheim, Germany; 2005;271–289. 51. Baker RW, Wijmans JG, Athayde AL, Daniels JH, Le M. The effect of the concentration polarization on the separation of volatile organic compounds from water by pervaporation. J Membrane Sci 1997;137:159–172. 52. Brazinha C, Barbosa DS, Crespo JG. Sustainable recovery of pure natural vanillin from fermentation media in a single pervaporation step. Green Chem 2011;13:2197–2203. 53. Peshev D, Peeva LG, Peev G, Baptista IIR, Boam AT. Application of organic solvent nanofiltration for concentration of antioxidant extracts of rosemary (Rosmarinus officiallis L.). Chem Eng Res Des 2011;89:318–327.

13 Biocatalytic membrane reactors for the production of nutraceuticals Lidietta Giorno, Rosalinda Mazzei and Emma Piacentini 13.1 Introduction “One cannot think well, love well, sleep well, if one has not dined well.” Virginia Woolfe

The need for eating well is obvious. One may think therefore, that there is a clear reason for continuous improvement in food quality and food consumption. However, different analyses predict controversial tendencies in food safety and sustainability by 2030. From pessimistic views (food quality will decrease, globalization and large distribution will reduce food variety, processed and packaged food will be the most consumed food with few chances for consumer to affect the type of food available, health problems will arise from a wrong food consumption and sedentary life stile linked to more intellectual activity) to optimistic (more fresh, high-quality and safe food will be consumed, consumer awareness and demand will affect food production and processing, research and innovation in food and agro-food will contribute to improve health and lifestyle) vision. There is no doubt that advanced technologies to produce, process and enhance fresh food shelf-life at reasonable cost will play a crucial role in affecting market offer and consumer demand. In food processing, safe methods that do not damage food properties, with low energy and environmental impact are needed. Most food processing comprises mainly physical operations, including: – fluid flow (to create turbulence) – heat-transfer (cooling, refrigeration, freezing, heating) – mass transfer (which may or may not require phase transfer, e.g., distillation, gas absorption, crystallization, membrane processes, drying, evaporation) – mechanical separation (filtration, centrifugation, sedimentation, sieving) – size adjustment (size reduction: slicing, dicing, cutting; size increase: aggregation, agglomeration, gelation) – mixing (homogeneous blends of dry or liquid ingredients, e.g., solubilizing solids, preparing emulsions or foams, dry blending of ingredients such as for cakes, etc.) Biochemical operations are mostly involved in: – transformation and stabilization of food (e.g., for baked foods, brewing, dairy, fruit juices, wine, distilled alcoholic beverages, meat and fish) – production of bioactive molecules or organisms (protein, enzyme, yeast, bacteria, organic acids, etc.) – biopolymers.

312 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

Membrane technology is particularly suitable to implementing high-quality and safe food processing because of the following properties: – concentration and separation carried out without the use of heat – allowing innovative formulation strategies – the equipment is small, flexible and are easy to scale-up (they are enabling technologies and respond well to the PI strategy for a sustainable growth) – operating costs are low – the energy used is low (e.g., it can be decreased up to 90% compared to evaporation) – products are of high-quality – co-products are of high-quality – allowing innovative process design – recognized among the BAT (best available techniques) for stream water treatment. The combination of membrane operation with biochemical processes leads to unique systems, such as biocatalytic membrane reactors, recognized as highly precise, efficient and intensified systems able to promote sustainable production in the food sector [1, 2]. A significant number of studies on a laboratory level are present in the open literature. However, lack of either a predictive approach or standardized experimental methods and procedures limited their thorough understanding and elaboration of generalized relationships. The strong innovation potential of biocatalytic membrane reactors is still unexploited. Technological challenges include systems reproducibility and lifespan. Areas of major interest of biocatalytic membrane reactors (BMRs) application in food include the production of ingredients for functional foods, the higher quality and stability of liquid foods. Research development on functionalized membranes (for immobilization of biomolecules; for creating surfaces able to repulse cells and biomolecules, e.g., to control biofouling; for preparing intelligent packaging able to control the release of drugs when necessary and/or to detect the presence of harmful substances, etc.) is helping to advance the basic understanding of mechanisms and phenomena involved in BMR. Technological strategies in food applications include gentle operation to preserve organoleptic properties, stabilize food and beverages (to avoid additives), and recycling of by-products. Membrane bioreactors strongly contribute to these objectives. In particular, their applications include – preparation of new liquid food high nutritional milk and easy to digest – hydrolysis of pectins in fruit pulp – hydrolysis of limonin – malolactic fermentation – vegetal oil processing (olive oil, palm oil), for example, hydrolysis of triglycerides – oil enrichment with stable lipophilic antibacterial, antioxydants, etc. via hydrolytic processes using oil components (e.g., glucosidases, oleuropein, etc.) – polysaccharides hydrolysis

13.2 General aspects 

–

– – –

 313

the production of natural additives, nutriaceuticals (flavors) by bioprocessing as an alternative route to the chemical synthesis (products are natural-like and more pure, e.g., L-amino acids, L-carboxylic acids, etc.) production of optically pure enantiomers ester synthesis biopolymer synthesis.

13.2 General aspects Biocatalytic membrane reactors (BMRs) are intensified processes that allow simultaneous conversion and separation operation in a single unit. They are, in fact, systems able to optimally integrate and intensify chemical transformations and transport phenomena. The transformation is promoted by a catalyst of biological origin (“biocatalyst”) while the transport is governed by a membrane operation (i.e., by a DF acting through a micro-nano-structured porous or dense membrane) (Figure 13.1). Transport can be appropriately tuned to control reagent supply to the catalyst and/or product removal from the reaction site. Among the various membrane types available (Figure 13.2), most common ones applied in biochemical membrane reactors are made of polymeric materials, in asymmetric conformation and as flat-sheet and HF configuration. Ceramic membranes are more expensive than polymeric ones in the short term; however, in most cases they do become more effective in the long term, as they are more stable to cleaning solutions, temperature prohibitive for bacteria growth, can be stems-sterilized and have longer life cycles. Compared to ordinary chemical catalysts, catalysts from biological origin have higher selectivity, higher reaction rates, milder reaction conditions and greater

Biocatalyst a

a d

b

S

b c

d c

Biocatalytic membrane reactor

= S

C

d

c

b

Membrane

Enzyme

C

b

+

d c

P

P

Membrane Highly selectives thanks to specific interactions with reagents (or substrates)

A permselective barrier that regulates transport between two phases

Reaction and separation is promoted in the same unit

Figure 13.1: The combination of a biocatalyst with a membrane gives a biocatalytic membrane reactor

314 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

Membrane materials Polymers

Ceramics

Glass

Metals

Liquids

Membrane structures Symmetric

Homogeneous Cylindrical films pores

Asymmetric

Integral asymmetric

Porous skin layer

Composite structure

Homogeneous skin layer

Membrane configurations Flat-sheet

Tubular

Hollow fiber

Spiral-wound

Figure 13.2: Common membrane types classification

stereospecificity. Conversely, they are labile macromolecules, easy to deactivate. Their immobilization to the membrane significantly improves the macromolecular stability, not necessarily implying a reduction in their catalytic activity [3, 4]. Biocatalysts are mainly represented by enzymes and cells. Table 13.1 summarizes enzymes class and catalyzed reactions. They can be compartmentalized by the membrane in a well-defined physical and geometrical region (such as the lumen or shell zone) and immobilized on the membrane surface and/or within the membrane matrix. Figure 13.3 illustrates the case of cells and enzymes (visualized by confocal microscopy after complex with marked antibody) immobilized within the finger-like porous structure of asymmetric membranes. Common immobilized biocatalysts used for food processing and production include lactase (to hydrolyze beta-D-galactosidic linkage of lactose milk); glucose isomerase (to convert D-glucose to D-fructose); acylase (to produce L-aminoacids); E. Coli (to produce L-aspartic acid); Pseudomonas dacunahe (to produce L-alanine); Brevibacterium ammoniagenes (to produce L-malic acid); pectic enzymes (to hydrolyze pectins); thermolysin (to produce aspartame); lipase (to hydrolyze triglycerides); proteases (to hydrolyze caroteno-proteins); beta-glucosidase (to produce antioxidants). One of the major brakes of the application of BMR with immobilized enzyme on a large scale for bulk production is the enzyme deactivation during membrane cleaning to recover flux declined as a consequence of fouling. Therefore, either fouling control or non-conventional cleaning procedures are needed to drive this technique on wider application sectors. Because of these constraints, the development stage of the BMR is still at an emerging stage in a graph reporting research efforts vs. technology development (Figure 13.4).

13.2 General aspects 

 315

Table 13.1: Enzyme class and catalyzed reaction Enzyme class

Enzyme type

Reaction

Example

Oxidoreductase

Dehydrogenases, oxydases, peroxidases, reductases, monooxygenases, dioxygenases Esterases, glycosidases, peptidases, amidases Epimerases, cis trans isomerases, intramolecular trasferases

Transfer of electrons or hydrogen atoms from one molecule to another

Lactic acid dehydrogenase: oxidizes lactic acid (application in cofactor regeneration)

Hydrolysis

Lipase: hydrolysis of lipids

Tearrangement of atoms within a molecule

C-transferases, glycosyltransferases, aminotrasnferases, phosphotransferases C—C; C—O; C—N; C—S lyases

Moving a functional group from one molecule to another

Phosphoglucoisomerase: converts glucose 6-phosphate into fructose 6-phosphate Hexokinase: transfers phosphate form ATP to glucose

C—C; C—O; C—N; C—S lygases

Join two or more molecules

Hydrolases Isomerases

Transferases

Lyases

Ligases

(a)

Split a molecule in smaller components

Fructose 1,6-bisphophate aldolase: splits fructose bisphosphate into G3P and DHAP Acetyl-coA synthetase: combines acetate to Coenzyme A

(b)

Figure 13.3: (a) SEM photo illustrating bacteria cells; (b) confocal microscope illustrating enzyme immunolocalization in asymmetric polymeric membranes

So far, biochemical membrane reactors with immobilized enzymes are applied to niche high added-value production. The membrane bioreactor configuration where the membrane works as a separation unit combined with the bioconversion occurring in a bulk is already applied at a development stage, approaching to a mature stage.

316 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

Membrane processes applications Emerging/Exploratory stage

Membrane bioreactors (with membrane as a separation unit)

Technology performance

Academia

Biocatalytic membrane reactors (with membranes working as catalytic and separation unit)

Industry Maturing

Developing Technical service Development

Emerging Gap

• Waste water treatment (SMBR) • Production of functional ingredients • Production of nutraceuticals/pharmaceutical • Improvement of fluid processability • Improve product quality • Improve co-products quality • Transform deteriorative components

Exploratory

Basic

Research effort

Figure 13.4: Illustration of development stage of various membrane bioreactors configurations

Figure 13.4 also summarizes the major application in food sector of this membrane bioreactor configuration.

13.3 Applications Milk and whey are among the first applications of BMR in food processing. In particular, they are used for: – hydrolysis of lactose present in milk or cheese whey with the aim of reducing intolerance and allergy in people such as children and elderly, to use hydrolyzed compounds as nutraceuticals and food ingredients – hydrolysis of proteins with the aim of producing low molecular mass peptide to reduce intolerance, to produce baby food, bioactive peptides with high nutritional value, low bitterness and low antigenicity – hydrolysis of fats with the aim of obtaining food with low calorie content. Biocatalysts, such as beta-galactosidase, extracted from yeast or fungi, such as Kluyveromyces yeast and Aspergillus fungi, are used to hydrolyze lactose, as they are recognized being in the GRAS category. Digestive enzymes, such as proteases derived from microorganisms, are used to reduce the molecular weight of peptides. BMR are particularly useful in controlling the optimal hydrolysis degree, as excessive fragmentation with high content of amino acids causes bad a taste in the final formulation. The use of ultrafiltration membranes with tuned MWCO can assist the separation of peptides at a preferred size from the reaction environment, thus preventing its further degradation. Lipases are largely used to hydrolyze milk fat. The produced fatty acids are important for the flavor, texture and formulation of many dietary products, including soft

13.3 Applications 

 317

cheese, sauces, dressing, and snacks. Lipases have been immobilized on a large variety of membrane materials and configurations. The most promising are those made of hydrophobic polymers, as lipase is activated by interfacial interactions. Nevertheless, hydrophilic membranes used in multiphasic systems are also suitable. In fact, in this case the interfacial phenomena can be promoted by the organic phase, while the hydrophilic material is more stable as it is in general less swollen by the organic solvent. Lipases from Aspergillus niger, Candida rugosa, Mucor miehi, porcine pancreas are commonly used.

13.3.1 Starch sugars Starch is the major source of carbohydrate, consisting of a large number of glucose units linked by glycosidic bond. This polysaccharide is produced in all green plants and in the human diet is mainly obtained from cereals (wheat, corn, rice, etc.) and roots (cassava, potatoes, etc.). Enzymes that can hydrolyze starch include amylases, glucoamylases, amiloglucanases, pullulanases. Glucose isomerase is used to convert glucose into fructose, which is the sweetest of all naturally occurring carbohydrates. Membranes are largely used as a combined separation step to remove mono and disaccharides reaction products from polysaccharides substrates, as well as a support for the enzyme. The starch sugars widely used in food ingredients formulation include maltodextrin, glucose syrup, dextrose (i.e., glucose obtained from total starch hydrolysis), high fructose syrup.

13.3.2 Fruit juices processing Membrane bioreactors using pectinases are commonly applied to process fruit pulps. Pectinases hydrolyze pectins producing oligosaccharides. This is useful for juice liquefaction, to increase juice yield, in the production of alcohol-free juice, wine and cider. Furthermore, oligosaccharides can be used as liver lipid accumulation repressors and as antifungal and antimicrobial agents. Pectinases are usually added to the pulp to increase juice yield. Furthermore, the immobilization of pectinases on the membrane surface can both carry out the hydrolysis and help to control membrane fouling during ultrafiltration process, as they degrade pectins as they deposit on the membrane surface [5]. Pectines from Aspergillus niger are the most used because they are GRAS. They are composed by different hydrolytic functions, including polygalacturonase, polymethylgalacturonase, pectinesterase. Fractions enriched with polygalacturonase are usually used to produce olygosaccharides for baby food. Pectinases are also used in combination

318 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

with cellulase to improve the liquefaction efficiency. Pectinase has been immobilized by physical entrapment or adsorption on polymeric and inorganic membranes. Coimmobilization with alfa-amylase has been also reported.

13.3.3 Production of functional molecules and spices Plant cell cultures and enzymes are a great source of components that can transform exogenus substrates (including synthetic ones) into valuable food ingredients. Aromatic, steroid, coumarin, terpenoids are among examples that can be produced using enzymes of plant origin. Papain (able to hydrolyze peptide bonds), hydroxynitrile lyase (a spereoselective enzyme able to produce optically pure antipathogens agents), phenoloxidase (able to hydroxylate monophenols to catechols with regioselectivity), lipoxygenase (ironcontaining enzyme that catalyzes the dioxygenation of polyunsatured fatty acids in lipids) are among the most studied examples. Algae, already largely used in eastern countries, are gaining much attraction worldwide as a source of biomass and biotransformation for food ingredients production. This occurs thanks to the high content of proteins, vitamins, iodine, alginic acid, carragens, etc. (For more information see section 1.2.5.)

13.3.4 Fats and oils Lipases from various sources are used to produce high added-value lipid compounds, structured lipids with high added-value properties (such as omega-3 fatty acids, which have shown to lower risk of heart attacks), food ingredients with low calories, etc. Lipases are used to process dietary lipids (such as triglycerides, fats, and oils). They are a subclass of esterases, enzymes that split esters into acids and alcohol in the presence of water. As esters are soluble in the organic phase and acids in the water phase, and lipase is activated by interfacial phenomena, the use of this enzyme as immobilized in a membrane placed at the interface of the two immiscible phases has found great attention. The two phases are kept in contact and at the same time are separated by the membrane. Therefore, the system offers the possibility to achieve an extremely efficient and intensified process [6].

13.3.5 Alcoholic beverages Enzymes, such as laccase, beta-glucosidase, bacteria cells such as Leuconostoc oenos and yeast, are used to stabilize must and wine and to carry out alcoholic fermentation (i.e., conversion of sugars to alcohol and carbon dioxide). Oxidization of polyphenols, hydrolysis of antocyanes, conversion of malic acid to lactic acid (to adjust the pH of

13.3 Applications 

 319

white wines) are among the reactions carried out in this field. Membranes separation (for example, RO to remove water and concentrate the sugar content of the must; to microfiltrate wine before bottling; to carry out electrodyialsis to adjust the tartaric acid content, etc.) are applied in combination with the bioconversion occurring either in solution (where the enzymes/cells are freely suspended) or on the enzyme-loaded. The control of alcoholic fermentation is certainly among the most crucial aspects. In addition to in line control and monitoring essay, the use of membranes to promote mass transport of components needed to assist the alcoholic fermentation and/or to be removed to avoid reaction inhibition will certainly be aspects to be implemented in the future. Application in this field is more advanced than it is reported in the literature, as most data are proprietary to wine making factories [7]. Beer is an alcoholic beverage produced by fermentation of sugars coming from the saccharification of starch. After fermentation and maturation, the beer is clarified by microfiltration, which removes turbidity by retaining yeast and particles, mainly formed by aggregated proteins and polyphenols. Even though fouling is occurring during the operation, crossflow, back-flushing, and cleaning in place are effective procedures to control it and obtain efficient flux.

13.3.6 Water purification for food production Water is among the most important base for food nutrition. In addition to drinking water, water is needed for most food formulations, both liquid and non-liquid food. Nowadays, because of the anthropisation of most available land, the presence of trace contaminants is very often found in source water, including the “constituents of emerging concerns” (CECs). These include pharmaceuticals, personal care products, pesticides, caffein, and other trace organic molecules (Table 13.2). These molecules are extremely difficult to remove from water, because of their low molecular size and low concentration. Enzymes able to convert such molecules in other more easy to separate or less dangerous are likely to be applied in the future. Table 13.2: Compounds of emerging concern Hormones Naturally occurring

Synthetic

Estrone 17 α-Estradiol 17 β-Estradiol Estriol Testosterone Progesterone Cis-Androsterone

17 α-ethynylestradiol Mestranol 19-Norethisterone Equilenin Equilin Cholesterol 3 β-coprostanol carnivore Stigmastanol (Continued)

320 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

Table 13.2: Compounds of emerging concern (Continued) Pharmaceutically active compounds Analgesic/antinflammatory

Antibiotic

Calcium channel blocker

– Acetaminophen – Ibuprofen – Nabumetone – Oxaprozin – Naproxen – Tramadol – Propoxyphene

– Tetracycline – Erythromycin – Cefprozil – Clavulanic acid – Trimethoprim – Mupirocin – Clarithromycin – Penicillin – Sulfamethoxazole – Azithromycin – Cephalexin – Amoxicillin

– Diltiazem – Verapamil

Sunscreen – Benzophenone – Oxybenzone

Fragrance – 3-Phenylpropionate – Acetophenone – Diethyl 3-phenylpropionate – Galaxolide fragrance – Musk Ketone

Antidepressant – Bupropion – Sertraline – Nefazodone Diuretic – Triamterene – Hydrochlorothiazide

H2-receptor antagonist – Ranitidine – Cimetidine Others – Phenytoin (anticonvulsant) – Ipratropium bronchlodialator – Troglitazone (antidiabetic) – Pseudoephedrine (decongestan) – Atenolol (beta blocker) – Allopurinol (antigout)

Personal care products Surfactant – Alkylphenol ethoxylates – Nonlyphenol ethoxylates Shampoo 1,4 Dioxane Pesticides, herbicides, and insecticides Pesticides/insecticides – Lindane – Dichloro-diphenyltrichloroethane – Cis-Chloridane – Chlorpyrifos – Methyl parathion

Herbicide – Atrazine – Metolachlor – Alachlor

Industrial and household products Fossil fuel

Antioxidants

– Naphthalene – Phenanthrene

– 2,6-di-tert-Butylphenol – P-Nonylphenol – 5-Methyl-1H-benzotriazole – Nonylphenol monoethoxylate – Butylatedhydroxinaisole – Nonylphenol diethoxylate – Butylatedhydroxytoluene – Octylphenol monoethoxylate – 2,6-di-tert-Butyl-p– Octylphenol benzoquinone diethoxylate

– Anthracene – Fluoranthene – Pyrene

Detergent metableolite

Plasticizer – Bisphenol A – Bis(2-Ethylhexyl) adipate – Ethanol-2-butoxyphosphate – Bis(2-Ethylhexyl) phthalate – Diethylphthalate

(Continued)

13.4 Conclusions 

 321

Disinfection by-products Chlorination-based DBPS

Ozonation-based dbps

Others

– Chloroform – Dichloroacetic acid – Bromochloroacetonitrile – Dichloroacetic aldehyde – 1,1-dichloropropanon – 2-chlorophenol

– Formaldehyde – Glyoxal – Acetaldehyde

– NDEA – NDMA – NDPA – NDBA – NMEA – NPIP – NYPR

Metals

Microorganisms are also very useful to metabolize many organic and inorganic molecules; however, their use for degradation of antibiotic molecules must be considered with proper attention because of the risk that they can develop strains resistant to these biocide molecules and become extremely dangerous for the environment as well as for health. The combination of bioconversion of these contaminants assisted by enzyme and/or microorganisms combined with membrane operations will certainly make the technology more feasible from the economic point of view. It is expected that the application of membrane bioreactors in this field will grow significantly in the near future, becoming an important share of the membrane market. The use of immobilized enzymes to detect harmful substances and decontaminate them from water, soil and air is also a sector of growing interest.

13.4 Conclusions The increasing interest toward the development of membrane bioreactor technology in the food industry demonstrates the potential of this technology. However, few examples are reported at industrial scale about the application of membrane bioreactors, such as the lactose and pectins hydrolysis in milk and juice production and in waste water treatment. The main configuration used is when the membrane works as a separation unit, while the reaction is carried out in a separate vessel, while waste water treatment was the main research field of application. The success in water treatment was guided from the restrictive law in the development of sustainable process with low economic and environmental impact. Progress and integration between membrane biotechnology as well as process design and modeling will promote further development.

322 

 13 Biocatalytic membrane reactors for the production of nutraceuticals

13.5 References 1. Giorno L, Drioli E. Biocatalytic membrane reactors: applications and perspectives. Trends Biotechnol 2000;18:339–349. 2. Mazzei R, Drioli E, Giorno L. Biocatalytic membranes and membrane bioreactors. In: Drioli E, Giorno L, eds. Comprehensive Membrane Science and Engineering. Elsevier Science: NY; 2010, Vol. 3, 195–212. 3. Giorno L, D’Amore E, Mazzei R, Piacentini E, Zhang J, Drioli E, Cassano R, Picci N. An innovative approach to improve the performance of a two separate phase enzyme membrane reactor by immobilizing lipase in presence of emulsion. J Membrane Sci 2007;295:95–101. 4. Mazzei R, Drioli E, Giorno L. Enzyme membrane reactor with heterogenized beta-glucosidase to obtain phytotherapic compound: Optimization study. J Membrane Sci 2012;390–391:121–129. 5. Alkorta I, Garbisu C, Llama MJ, Serra JL. Industrial applications of pectic enzymes: a review. Process Biochem 1998;33:21–28. 6. Malcata FX, Reyes HR, Garcia HS, Hill Jr CG, Amundson CH. Immobilized lipase reactors for modification of fats and oils – A review. J Am Oil Chem Soc 1990;67:890–910. 7. Kourkoutas Y, Bekatorou A, Marchant R, Banat IM, Koutinas AA. Immobilization technologies and support materials suitable in alcohol beverages production: a review. Food Microbiol 2004;21: 377–397.

14 Membrane emulsification in integrated processes for innovative food Catherine Charcosset 14.1 Introduction Membrane emulsification (ME) [1–4] has received increasing attention over the last 20 years as an alternative to other methods of emulsification, such as high pressure homogenizers, ultrasound homogenizers and rotor/stator systems, including stirred vessels, colloid mills or toothed disc dispersing machines. In the dispersing zone of these machines, high shear stresses are applied to deform and disrupt large droplets. Therefore, high energy inputs are required and shear-sensitive ingredients such as proteins or starches may lose functional properties. In a typical ME setup, the dispersed phase is pressed through the pores of a microporous membrane, while the continuous phase flows along the membrane surface. Droplets grow at pore openings until they reach a certain size and detach. Surfactant molecules in the continuous phase stabilize the newly formed interface, to prevent droplet coalescence immediately after formation. The distinguishing feature is that the resulting droplet size is controlled primarily by the choice of the membrane and not by the generation of turbulent droplet break-up. The apparent shear stress is lower than in classical emulsification systems, because small droplets are directly formed by permeation of the dispersed phase through the micropores, instead of disruption of large droplets in zones of high energy density. In addition to the possibility of using shear-sensitive ingredients, emulsions with narrow droplet size distributions can be produced. Furthermore, ME processes allow the production of emulsions at lower energy input (104–106 J/m3) compared to conventional mechanical methods (106–108 J/m3) [5]. Over the years, an increasing number of ME applications have been reported, including preparation of emulsions [water-in-oil (w/o) or oil-in-water (o/w)], multiple emulsions and colloidal dispersions such as microspheres, microcapsules, nanospheres, nanocapsules, liposomes, colloidosomes, aerated gels, etc. [6]. Some of these applications have been reported for food purposes, including preparation of dairy products, food complements and others [7, 8]. As membrane processes are well integrated in the food industry with traditional processes such as UF, MF, RO and electrodialysis (ED), it is believed that ME could find its own place and be part of these integrated processes. The purpose of this chapter is to give some general backgrounds on ME and its potential association to other membrane processes in the food industry. The first part presents some general backgrounds on ME including configurations, membranes and parameters, the second part deals with applications of ME such as emulsions,

324 

 14 Membrane emulsification in integrated processes for innovative food

multiples emulsions, and colloidal dispersions; finally, the third part discusses integration of ME in integrated processes for food applications, including beverage and dairy products.

14.2 Membrane emulsification The emulsions or colloidal dispersions obtained are characterized by their size, size distribution, zeta potential and stability vs. storage. The flowrate through the membrane (or flux) is also an important parameter. A decreasing flowrate vs. time indicates membrane fouling.

14.2.1 Configurations A schematic picture of a typical ME in a crossflow configuration setup is shown in Figures 14.1 and 14.2. The system incorporates a tubular MF membrane, a pump, a feed vessel, and a pressurized (N2) oil container. The dispersed phase is pumped under gas pressure through the pores of the membrane into the continuous phase, which circulates through the membrane device. The membrane should not be wetted with the dispersed phase. Therefore, at the beginning of the experiment, the membrane is wetted with the continuous phase, i.e., a hydrophilic membrane for o/w emulsions is wetted with the water phase and a hydrophobic membrane for w/o emulsions is wetted with the oil phase. At the end of the experiment, the membrane is cleaned, using an appropriate solution, until the pure water flux is restored to its original value. In conventional direct ME, fine droplets are formed at the membrane-continuous phase interface by pressing the disperse phase through the membrane. In order to ensure a regular droplet detachment from the pore outlets, shear stress is generated at the membrane-continuous phase interface by recirculating the continuous phase Dispersed phase permeation under applied pressure Membrane Tangential flow of the continuous phase

Figure 14.1: Schematic diagram of the membrane emulsification process

Droplets/or particles

14.2 Membrane emulsification 

 325

M Pressurized vessel dispersed phase M

Stirrer

M Membrane module

N2 bottle

Figure 14.2: Typical experimental set-up for the membrane emulsification process M: manometer

using a pump or by agitation in a stirring vessel [7]. The rate of mixing should be high enough to provide the required tangential shear on the membrane surface, but not too excessive to induce further droplet break-up. Premix ME is another configuration of membrane emulsification. A pre-emulsion with a large droplet size is passed through the porous membrane into the continuous phase, instead of directly passing the oil or water. The droplets of the pre-emulsion are disrupted into fine droplets during their permeation through the membrane. For similar mean pore sizes, the mean droplet size resulting from premix membrane emulsification is smaller than in direct ME, which is often an advantage [9]. Repeating the processes with the same membrane results in smaller mean droplet size, narrower droplet size distribution, and long-term physical stability. Premix ME has been reviewed recently by focusing on droplet formation mechanisms (localized shear, interfacial tension, steric hindrance between droplets), and process parameters (membrane properties, transmembrane pressure, continuous phase viscosity, and number of homogenization cycles) [10]. Premix ME is increasingly reported for the preparation of single emulsions, multiple emulsions, gel microbeads, and polymer microspheres. The ME process may also be carried out in a dead-end mode without tangential flow of the continuous phase [11], or in a stirred cell configuration (Figure 14.3). A stirred cell is not a common device for ME, because it is usually believed that a uniform shear field at the membrane surface is required for the generation of uniform droplets. The stirred cell with a varying radial shear field at the surface of a flat disc membrane could produce uniform droplets of paraffin wax and refined sunflower oil [12]. In this study, microengineered flat disc membranes were used, on top of which a paddle blade stirrer was operated to induce surface shear. These configurations are particularly suited for the preparation of small amounts of emulsions, microcapsules, and microparticles loaded with high values chemicals.

326 

 14 Membrane emulsification in integrated processes for innovative food

Dead-end membrane emulsification

Cross-flow membrane emulsification

Premix membrane emulsification in dead-end

Stirring, rotating or vibrating membrane emulsification

Membrane Figure 14.3: Various configurations for membrane emulsification

Other systems use a moving membrane, in which the droplet detachment from the pore outlets is obtained by rotation or vibration of the membrane within the stationary continuous phase. Droplets can be spontaneously detached from the pore outlets at small disperse phase fluxes, particularly in the presence of fast adsorbing emulsifiers in the continuous phase and for a pronounced noncircular cross section of the pores. These configurations have the advantage of eliminating external pumps to circulate the continuous phase. This is particularly attractive for coarse emulsion droplets or fragile particulate products, as their structure can be easily destroyed during their circulation inside the pump. Rotating membrane devices were tested to increase the performances of the ME process, especially to increase the flux of the dispersed phase through the membrane [13, 14]. With the rotation speed (higher wall shear stress), the particle size decreases.

14.2.2 Membranes The most commonly used membrane for the preparation of emulsions is the Shirasu porous glass (SPG) membrane (Ise Chemical Co., Japan), because of its narrow pores size distribution and tubular shape [15]. The SPG membrane is synthesized from CaOAl2O3-B2O3-SiO2 type glass, which is made from “Shirasu”, a Japanese volcanic ash. The SPG membrane has uniform cylindrical interconnected micropores, a wide spectrum of available mean pore sizes (0.05–30 μm), and a high porosity (50–60 %).

14.2 Membrane emulsification 

 327

In addition to SPG membranes, o/w emulsions were successfully prepared using silicon and silicon nitride microsieves membranes (Aquamarijn Microfiltration BV, The Netherlands) [16]. These are made by photolithographic treatment of a silicon wafer and subsequent etching, or electrochemical metal deposition on a skeleton in an electrolysis bath, respectively. These membranes have interesting properties, such as a smooth and flat surface, a very low membrane resistance and narrow pores size distribution. Different pore geometries (circular, square, slit shaped), pore size, pore edges and membrane porosities are available. Flat or tubular metal membranes with monosized circular pores distributed in a highly regular array are also available (Micropore Technologies Ltd., UK) [17]. The membranes are chemically treated to make the surface hydrophilic for the preparation of o/w emulsions. Polycarbonate track-etch membranes (Millipore Corp.) with a very narrow pore size distribution were also tested for the preparation of particles [18]. Other commercial MF membranes are attractive because of their availability in very large surface area, and their high flux through the membrane pores, such as ceramic aluminium oxide (α-Al2O3) membranes (Membraflow, Germany) [19], α-alumina and zirconia-coated membranes (SCT, France) [20], and polytetrafluoroethylene (PTFE) membranes (Advantec Tokyo Ltd., Japan [21] and Goretex Co. Ltd., Japan [22]). W/o emulsions were also successfully prepared using, for example, macroporous silica glass membranes [23], polytetrafluoroethylene (PTFE) membranes [24], and polyamide HFs membranes [25].

14.2.3 Influence of parameters The major factors influencing ME include membrane parameters, phase parameters and__process parameters. Several authors have shown that the __average droplet diameter, dp  , increases with the average membrane pore diameter, dp  , by a linear relationship, for given operating conditions: __

__

dd  = c dp 

(14.1)

where c is a constant. For SPG membranes, values of c are typically in the range 2–10. This range was explained by differences in operating conditions, and by the type of SPG membrane used [26]. For membranes other than SPG, the values reported for c are higher, typically 3–50. Monodispersed emulsions can be produced if the membrane pore size distribution is sufficiently narrow. The porosity of the membrane surface is also an important parameter for the ME process because it determines the distance between two adjacent pores. This distance is critical to ensure that two adjacent droplets do not come sufficiently close to allow contact with each other, which may lead to coalescence.

328 

 14 Membrane emulsification in integrated processes for innovative food

Droplets formed at the membrane/continuous phase interface detach under the shear stress of the continuous phase. The characteristic parameter of the flowing continuous phase is the crossflow velocity or the wall shear stress. It is shown that the droplet size becomes smaller as the wall shear stress increases and that the influence is greater for small wall shear stresses [20, 27]. The effect of the wall shear stress on reducing droplet size is dependent on the membrane pores size, being more effective for smaller membrane pores size [19]. Surfactants play two main roles in the formation of an emulsion. Firstly, they lowered the interfacial tension between oil and water. This facilitated droplet distribution and in case of membranes lowers the minimum emulsification pressure. Secondly, surfactants stabilize the droplets against coalescence and/or aggregation. The influence of the type of surfactant in the ME process has been studied by several authors [27, 28]. A proper choice of surfactant type and amount is crucial for a successful emulsification. As with other emulsification methods, the Bancroft rule applies: “The phase in which a surfactant is more soluble constitutes the continuous phase”. As a consequence, the type of emulsion (o/w or w/o) is dictated by the emulsifier and the emulsifier should be soluble in the continuous phase. The hydrophilelipophile balance (HLB) number is a measure of the ratio of the hydrophilic and lipophilic groups of the surfactant molecule. Non-ionic surfactants have HLB numbers in the range 0–20. HLB numbers > 10 have an affinity for water (hydrophilic) and HLB number < 10 have an affinity for oil (lipophilic). The HLB number can be used in order to select the surfactant for a given emulsion. A water continuous phase system benefits from a surfactant with a high HLB number and an oil continuous phase system from a surfactant with a low HLB number. The viscosity of the dispersed phase has also an important effect on the ME performance. According to Darcy’s law, the dispersed flux is inversely proportional to the dispersed phase viscosity. If the dispersed phase viscosity is higher, then the dispersed flux will be lower, and as a consequence the droplet diameter will be large compared to the mean pore diameter.

14.3 Applications A very large range of emulsions and particles were prepared using the ME technique. Some of these emulsions and particles are presented in Figure 14.4 and include simple emulsions, multiple emulsions, and particles such as polymeric, lipid nano and microparticles, nano and microcapsules, and microbubbles.

14.3.1 Simple emulsions An emulsion is a suspension of one phase in another in which it is immiscible. There are two main types of simple emulsion for food applications [29]. In o/w emulsions,

14.3 Applications 

o/w emulsions Oil

Water

Microbubbles Air void

Solid shell

w/o emulsions Water

Oil

w1 /o/w2 emulsions W1 W2

Polymeric or solid lipid micro nano/particles Polymer or solid lipid

 329

Oil

Microcapsules

Oil

Polymeric shell

Figure 14.4: Examples of emulsions and particles prepared by membrane emulsification

droplets of oil are suspended in an aqueous continuous phase. They exist in many forms (mayonnaises, cream liqueurs, creamers, whippable toppings, ice cream mixes). Their properties can be controlled by varying both the surfactants used and the components present in the aqueous phase. W/ o emulsions include butter, margarines, and fat-based spreads. Their stability depend more on the properties of the fat or oil and the surfactant used than in the properties of the aqueous phase. Therefore, there are fewer parameters that can be varied to control their properties. Numerous studies have been realized on the preparation of emulsions using ME. The emulsions are characterized by their mean droplet size, size distribution, and stability vs. storage. The influence of the previously listed parameters is usually investigated to define an optimum preparation. Various preparations of food emulsions by ME have been reported; however, these studies are often limited to model emulsions and not real ones for the food industry. For example, Suzuki et al. [30] prepared o/w and w/o food emulsions from corn oil; Joscelyne and Trägårdh [20] investigated the preparation of o/w emulsions consisting of vegetable oil as the dispersed phase and skimmed milk as the dispersion medium; Sotoyama et al. [31] optimized the preparation of o/w food emulsions from corn oil.

14.3.2 Multiple emulsions Multiple emulsions (or double emulsions) are “emulsions of emulsions”, e.g., water droplets-in-oil droplets-in-water (w/o/w) or oil droplets-in-water droplets-in-oil (o/w/o) [32]. Multiple emulsions may offer interesting structural properties but not

330 

 14 Membrane emulsification in integrated processes for innovative food

been used often as yet for several reasons [33]. Firstly, they are more complicated to produce compared with simple emulsions, as they need to be made by two successive homogenization steps, which need to be more strictly controlled than a traditional single homogenization. In addition, the prepared double emulsions are susceptible to breakdown (to become simple emulsions) during processing. This may be avoided by gelling the interior water droplets, either completely or at the interior oil-water interfacial layer. The primary emulsion may be produced by means of a conventional method or by ME [34]. The mild conditions of ME are especially useful for the second emulsification step in order to prevent rupture of the double emulsion droplets, which might even lead to inversion into a single o/w emulsion. In contrary to conventional emulsification methods, it becomes possible to produce small and monodisperse droplets without using high shear stresses that cause escape of the internal droplets.

14.3.3 Encapsulation Microcapsules are composed of a polymer wall (coat or shell), and an active ingredient refereed to as “core” or “nucleus” [35, 36]. The active ingredient may be food additive, a medicament, a biocide, or an adhesive. A food additive may impart texture or bulk, or it may play a functional role in terms of nutritional value, food preparation or preservation. Examples of functional ingredients include processing aids (leavening agents and enzymes), preservatives (acids and salts), fortifiers (vitamins and minerals), flavors (natural and synthetic), and spices. Among these ingredients, the use of microencapsulated flavors is the most widely established. Microencapsulation prevents (or minimizes) aroma evaporation during processing, transport, storage and cooking. ME has been reported for the preparation of a very large range of colloidal dispersions with size ranging from a few nanometers (nanoparticles, nanocapsules) to some micrometers (microparticles and microcapsules) [6]. For example, microspheres and microcapsules were prepared by combining the ME process and subsequent polymerization or solvent evaporation, to prepare a large range of polymeric microspheres such as poly(styrene)-poly(methyl methacrylate) [37], polyurethane urea [28], magnetite [38], and titanium dioxide microcapsules [39]. Solid lipid particles and microcapsules were prepared using a high melting point lipid as the encapsulating material, followed by cooling of the preparation at room temperature [40, 41]. Polymeric nanoparticles were obtained successfully by the nanoprecipitation or the interfacial polymerization methods [42]. Colloidal dispersions are characterized by their mean size and size distribution, zeta potential, stability during storage, final properties such as encapsulation efficiency and functional ingredient delivery. These studies investigated parameters including the influence of membrane properties (pore size, pore shape, active surface, etc), device parameters (applied pressure, crossflow velocity in case of a crossflow configuration, stirring steep for the rotating device, etc.), formulation of

14.4 Integrated processes 

 331

both dispersed and continuous phase (polymer and solvent used, amount, etc.), and temperature on the properties of the colloidal dispersions produced by ME.

14.3.4 Aerated food gels Food gels are soft solids containing a high amount of an aqueous phase (i.e., > 80 %) [43]. Gel-like structures are present among most high-moisture processed foods: jellies, yogurt, processed meats, etc. Air is also a component of several food products, usually present as a dispersed phase of bubbles or pores within a matrix. Air bubbles are also abundant structural elements in solid food foams, e.g., bread, cakes, aerated chocolate bars and meringue, in semi-solid foams such as whipped cream or mousses and in beverages like milkshakes. Aerated food gels were produced recently by membrane foaming [44]. The dispersed phase (gas) is pressed through the pores of a tubular membrane into the continuous phase. The bubbles formed are covered with surface-active substances of the continuous phase and they are detached from the membrane surface by the shear forces exerted by the phase flowing along the membrane surface. Bals and Kulozik [44] investigated the influence of pore size, foaming temperature and viscosity of the continuous phase on the properties of foams produced by membrane foaming. An important factor is that the added amount of gas must be stabilized as completely as possible in the foam. Raising the foaming temperature increased the quantity of stabilized gas. The whey proteins then diffused faster to the bubble surfaces and stabilized these by unfolding and networking reactions to prevent the coalescence of the bubbles. A dynamically enhanced membrane foaming process consists of two cylinders: the inner cylinder rotated, the outer cylinder being fixed has been also tested [45]. The membrane can either be mounted to the inner or outer cylinder. The gas is pressed through the membrane and is detached as small bubbles by the flow shear stress. The technique improved significantly the foam microstructure, i.e., smaller mean bubble sizes and narrower size distributions were achieved.

14.4 Integrated processes The following section gives some examples of potential applications of ME and integrated membrane processes in some food emulsion fields: beverage and dairy products.

14.4.1 Beverages Since the late 1990s, the beverage industry has provided new products, such as flavored tea, flavored water, juice drinks, and dairy-based juice drinks [46]. Citrus flavor

332 

 14 Membrane emulsification in integrated processes for innovative food

products are based primarily on the essential oils from the peel of the fruits (e.g., orange oils, lemon oils, etc.), and are not water miscible. The first method to incorporate these flavors into soft drinks is to separate out the water soluble fraction of the flavor from the essential oils by extraction and distillation. The second method involves converting the oil into a water-dispersible “beverage emulsion”. Beverage emulsions are different from other food emulsions in that they are consumed in a highly diluted form, rather than in their original concentrate form. They are prepared as an emulsion concentrate, which is usually diluted several hundred times in a sugar solution to produce the final beverages. Two-stage homogenizers are usually used to prepared beverage emulsions to ensure a very small average particle size and very uniform particle size distribution. The emulsions in both the concentrate and diluted forms must be stable for at least 6 months, as required by the beverage industry. Several studies have investigated some of the major factors influencing the formation and stability of beverage emulsions containing orange and lemon oils, including the preparation method used, emulsifier type, droplet charge, addition of antioxidants, physical state of the oil phase and oil composition [47]. ME may be a suitable alternative to classical homogenization for the preparation of beverage emulsions. The essential oil can be converted in a water-dispersible emulsion using ME with appropriate production conditions (membrane characteristics and process parameters). Membrane fouling may be reduced by using new systems such as rotating device or pulsed flow systems. Premix emulsification could be also a possible alternative as several passes through the membrane can improve the average particle size and size distribution. So far, integrated membrane processes have been developed mainly for fruit juice concentration. For example, Galaverna et  al. [48] reported the production of concentrated blood orange juice according to the following scheme: an initial clarification of freshly squeezed juice by UF; the clarified juice was successively concentrated by two consecutive processes: first RO, used as a preconcentration technique (up to 25–30°Brix), then OD, up to a final concentration of approximately 60°Brix. The integrated membrane process was presented as a valuable alternative to obtain high-quality concentrated juice, as the final product showed a very high antioxidant activity and a very high amount of natural bioactive components. ME could find its place in integrated membrane processes for beverage production, particularly for beverage emulsion preparation.

14.4.2 Dairy products Dairy emulsions include w/o emulsions such as butter, margarines, and fat-based spreads which are products in themselves [49]. Other dairy emulsions include yogurts, processed cheeses, and other gelled systems containing emulsions droplets, which

14.4 Integrated processes 

 333

participate in forming the structures of more complex products. Other components of the food (proteins, polysaccharides) form a matrix in which the fat globules are trapped or with which they interact. The preparation of emulsions with reconstituted milk is usually done by high pressure homogenization. In such cases the molecular or ultrastructural status of the milk components (casein micelles, whey proteins, and free milk fat globule membranes in buttermilk) may be changed. The structures and composition of the proteins at the fat surface play an important role in determining the functional properties of recombined milks. ME could be therefore a suitable alternative to other emulsification process. Using SPG membranes, Scherze et al. [50] and Muschiolik et al. [51] prepared o/w emulsions with liquid butter fat or sunflower oil as the dispersed phase and a continuous phase containing milk proteins. The emulsions obtained were characterized by particle size distribution, creaming behavior and protein adsorption at the dispersed phase. The advantage of ME was shown to be the low-shear forces on the physicochemical and molecular properties of the proteins. Ceramic membranes were also used to produce o/w emulsions consisting of vegetable oil as the dispersed phase and skim milk as the dispersion medium [20]. Katoh et  al. [52] prepared a low fat spread with a fat content of 25 % (v/v). They showed that the dispersed phase flux was increased 100 times using a hydrophilic membrane pretreated by immersion in the oil phase, and that the ME process was suitable for preparation of large-scale w/o food emulsions. ME can be used also as a suitable process for microcapsules production. There has been considerable recent interest in encapsulation of valuable compounds in milk products. For example, the encapsulation of probiotic bacteria such as Lactobacillus and Bifidobacterium will be more resistant to the stressful conditions of the stomach and the upper intestine, which both contain bile under highly acidic conditions. Using a microporous glass (MPG) membrane, Song et  al. [53] prepared microcapsules containing viable cells (Lactobacillus casei) with a narrow particle size distribution. For artificial gastric acid and bile, the viable count of encapsulated cells was constant through the incubation time, while the count of non-encapsulated cells was significantly decreased. A storage stability test at different temperatures resulted in a viability of encapsulated cells 3–5 log cycles higher than the viability of nonencapsulated cells. Membrane operations are widely applied throughout the milk and dairy processing chains – milk reception, cheese making, whey protein concentration, fractionation of protein hydrolysates, waste stream purification and effluents recycling and treatment [54]. Integrated membrane processes have been reported for example by Kelly et al. [55] who developed a series of large pilot plant membrane separation systems based on MF, UF and ED for the separation and fractionation of milk and whey components. ME could then be added to other integrated membrane process in the dairy industry.

334 

 14 Membrane emulsification in integrated processes for innovative food

14.5 Conclusions ME is an alternative to other emulsification processes, such as high pressure homogenizers, ultrasound homogenizers and rotor/stator systems (stirred vessels, colloid mills or toothed disc dispersing machines). ME can be used for the preparation of simple emulsions (w/o and o/w), double emulsions, and colloidal dispersions such as microparticles and microcapsules. As membrane processes such as UF and MF are widely used in the food industry, ME could find its place as a part of integrated membrane processes. Drawbacks of ME are often attributed to membrane fouling, which leads to a decrease of the flux rate through the membrane vs. time. Periodic cleaning is then necessary, which increases the overall time and cost of the preparation [8]. This negative effect is expected to be limited by recent developments in ME related to flow configuration with rotating flow or pulsated flow, and/or to membrane developments with microsieve or corrugated membranes.

14.6 References 1. Joscelyne SM, Trägårdh G. Membrane emulsification—a literature review. J Membrane Sci 2000;169:107–117. 2. Gijsbersten–Abrahamse AJ, van der Padt A, Boom RM. Status of crossflow membrane emulsification and outlook for industrial application. J Membrane Sci 2004;230:149–159. 3. Piacentini E, Figoli A, Giorno L, Drioli E. Membrane emulsification. In: Drioli E, Giorno L., eds. Comprehensive Membrane Science and Engineering. Elsevier; Oxford, UK: 2010; 47–75. 4. Charcosset C. Membranes for the preparation of emulsions and particles. In: Charcosset C, ed. Membrane Processes in Biotechnology and Pharmaceutics. Elsevier; Oxford, UK: 2012; 213–238. 5. Altenbach-Rehm J, Suzuki K, Schubert H. Production of O/W-emulsions with narrow droplet size distribution by repeated premix membrane emulsification, 3ième Congrès Mondial de l’Emulsion, 24–27 September 2002, Lyon, France. 6. Vladisavljević GT, Williams RA. Recent developments in manufacturing emulsions and particulate products using membranes. Adv Colloid Interfac 2005;113:1–20. 7. Charcosset C. Preparation of emulsions and particles by membrane emulsification for the food processing industry. J Food Eng 2009;92:241–249. 8. Ribeiro HS, Janssen JJM, Kobayashi I, Nakajima M. Membrane emulsification for food applications. In: Peinemann KV, Nunes SP, Giorno L, eds. Membrane Technology. Vol. 3: Membranes for Food Applications. Wiley-VCH: Weinheim, Germany; 2010; 129–166. 9. Vladisavljević GT, Shimizu M, Nakashima T. Preparation of monodisperse multiple emulsions at high production rates by multi-stage premix membrane emulsification. J Membrane Sci 2004;244:97–106. 10. Nazir A, Schroën K, Boom R. Premix emulsification: A review. J Membrane Sci 2010; 362:1–11. 11. Kukizaki M, Goto M. Preparation and characterization of a new asymmetric type of Shirasu porous glass (SPG) membrane used for membrane emulsification. J Membrane Sci 2007;299:190–199.

14.6 References 

 335

12. Stillwell MT, Holdich RG, Kosvintsev SR, Gasparini G, Cumming IW. Stirred cell membrane emulsification and factors influencing dispersion drop size and uniformity. Ind Eng Chem Res 2007;46:965–972. 13. Aryanti N, Hou R, Williams RA. Performance of a rotating membrane emulsifier for production of coarse droplets. J Membrane Sci 2009;326:9–18. 14. Aryanti N, Williams RA, Hou R, Vladisavljević GT. Performance of rotating membrane emulsification for o/w production. Desalination 2006;200:572–574. 15. Nakashima T, Shimizu M, Kukizaki M. Membrane emulsification by microporous glass. Key Eng Mater 1991;61–62:513–516. 16. Zhu J, Barrow D. Analysis of droplet size during crossflow membrane emulsification using stationary and vibrating micromachined silicon nitride membranes. J Membrane Sci 2005;261:136–144. 17. Gasparini G, Kosvintsev SR, Stillwell MT, Holdich RG. Preparation and characterization of PLGA particles for subcutaneous controlled drug release by membrane emulsification. Colloid Surface B 2008;61:199–207. 18. Tangirala R, Revanur R, Russell TP, Emrick T. Sizing nanoparticle-covered droplets by extrusion through track-etch membranes. Langmuir 2007;23:965–969. 19. Schröder V, Schubert H. Production of emulsions using microporous, ceramic membranes. Colloid Surface A 1999;152:103–109. 20. Joscelyne SM, Trägårdh G. Food emulsions using membrane emulsification: conditions for producing small droplets. J Food Eng 1999;39:59–64. 21. Kanichi S, Yuko O, Yoshio H. Properties of solid fat O/W emulsions prepared by membrane emulsification method combined with pre-emulsification. 3ième Congrès Mondial de l’Emulsion, 24–27 September 2002, Lyon, France. 22. Yamazaki N, Yuyama H, Nagai M, Ma GH, Omi S. A comparison of membrane emulsification obtained using SPG (Shirasu Porous Glass) and PTFE poly(tetrafluoroethylene) membranes. J Disper Sci Technol 2002;23:279–292. 23. Fuchigami T, Toki M, Nakanishi K. Membrane emulsification using sol-gel derived macroporous silica glass. J Sol-Gel Sci Techn 2000;19:337–341. 24. Yamazaki N, Naganuma K, Nagai M, Ma GH, Omi S. Preparation of w/o (water-in-oil) emulsions using a PTFE (polytetrafluoroethylene) membrane – A new emulsification device. J Disper Sci Technol 2003;24:249–257. 25. Giorno L, Li N, Drioli E. Preparation of oil-in-water emulsions using polyamide 10 kDa hollow fiber membrane. J Membrane Sci 2003;217:173–180. 26. Omi S. Preparation of monodisperse microspheres using the Shirasu porous glass emulsification technique. Colloid Surface A 1996;109:97–107. 27. Kobayashi I, Yasuno M, Iwamoto S, Shono A, Satoh K, Nakajima M. Microscopic observation of emulsion droplet formation from a polycarbonate membrane. Colloid Surface A 2002;207: 185–196. 28. Yuyama H, Watanabe T, Ma GH, Nagai M, Omi S. Preparation and analysis of uniform emulsion droplets using SPG membrane emulsification technique. Colloid Surface A 2000;168:159–174. 29. Friberg SE, Larsson K, Sjblöm J. Food Emulsions, 4th edn. Marcel Dekker: NY, USA; 2004. 30. Suzuki K, Fujiki I, Hagura Y. Preparation of corn oil/water and water/corn oil emulsions using PTFE membranes. Food Sci Technol 1998;4:164–167. 31. Sotoyama K, Asano Y, Ihara K, Takahashi K, Doi, K. Water/Oil emulsions prepared by the membrane emulsification method and their stability. J Food Sci 1999;64:211–215. 32. Muschiolik G. Multiple emulsions for food use. Curr Opin Colloid In 2007;12:213–220. 33. Dalgleish DG. Food emulsions – their structures and structure-forming properties. Food Hydrocolloid 2006;20:415–422.

336 

 14 Membrane emulsification in integrated processes for innovative food

34. van der Graaf S, Schroën CGPH, Boom RM. Preparation of double emulsions by membrane emulsification–a review. J Membrane Sci 2005;251:7–15. 35. Vilstrup P. Microencapsulation of Food Ingredients. Leatherhead Food RA Publishing: Surrey; 2001. 36. Forssell P, Partanen R, Poutanen K. Microencapsulation-better performance of food ingredients. Food Sci Tech 2006;20:18–20. 37. Ma GH, Nagai M, Omi S. Effect of lauryl alcohol on morphology of uniform polystyrene-poly (methyl methacrylate) composite microspheres prepared by porous glass membrane emulsification technique. J Colloid Interf Sci 1999;219:110–128. 38. Omi S, Senba T, Nagai M, Ma GH. Morphology development of 10-μm scale polymer particles prepared by SPG emulsification and suspension polymerization. J Appl Polym Sci 2001;79:2200–2220. 39. Supsakulchai A, Ma GH, Nagai M, Omi S. Uniform titanium dioxide (TiO2) microcapsules prepared by glass membrane emulsification with subsequent solvent evaporation. J Microencapsul 2002;19:425–449. 40. Kukizaki M, Goto M. Preparation and evaluation of uniformly sized solid lipid microcapsules using membrane emulsification. Colloid Surface A 2007;293:87–94. 41. D’Oria C, Charcosset C, Barresi A, Fessi H. Preparation of solid lipid particles by membrane emulsification: Influence of process parameters. Colloid Surface A 2009;338:114–118. 42. Khayata N, Abdelwahed W, Chehna MF, Charcosset C, Fessi H. Preparation of vitamin E loaded nanocapsules by the nanoprecipitation method: from laboratory scale to large scale using a membrane contactor. Int J Pharmaceut 2012;423:419–427. 43. Zúñiga RN, Aguilera JM. Aerated food gels: fabrication and potential applications. Trends Food Sci Tech 2008;19:176–187. 44. Bals A, Kulozik U. The influence of pore size, the foaming temperature and the viscosity of the continous phase on the properties of foams produced by membrane foaming. J Membrane Sci 2003;220:5–11. 45. Müller-Fischer N, Bleuler H, Windhab EJ. Dynamically enhanced membrane foaming. Chem Eng Sci 2007;62:4409–4419. 46. Tan CT. Beverage emulsions. In: Food Emulsions, Friberg SE, Larsson K, Sjöblom J, eds. 4th edn. Marcel Dekker: NY, USA; 2004; 485–524. 47. Rao J, McClements DJ. Impact of lemon oil composition on formation and stability of model food and beverage emulsions. Food Chem 2012;134:749–757. 48. Galaverna G, Di Silvestro G, Cassano A, Sforza S, Dossena A, Drioli E, Marchelli R. A new integrated membrane process for the production of concentrated blood orange juice: Effect on bioactive compounds and antioxidant activity. Food Chem 2008;106:1021–1030. 49. Dalgleish DG. Food emulsions: Their structures and properties. In: Friberg SE, Larsson K, Sjöblom J, eds. Food Emulsions, 4th edn. Marcel Dekker: NY; 2004; 1–44. 50. Scherze I, Marzilger K, Muschiolik G. Emulsification using micro porous glass (MPG): surface behaviour of milk proteins. Colloid Surface B 1999;12:213–221. 51. Muschiolik G, Dräger S, Scherze I, Rawel HM, Stang M. Protein–stabilized emulsions prepared by the micro-porous glass method. In: Dickinson E., ed. Food Colloids: Proteins, lipids and polysaccharides. Royal Society of Chemistry: Cambridge; 1997; 393–400. 52. Katoh R, Asano Y, Furuya A, Sotoyama K, Tomita M. Preparation of food emulsions using a membrane emulsification system. J Membrane Sci 1996;113:131–135. 53. Song SH, Cho YH, Park J. Microencapsulation of Lactobacillus casei YIT 9018 using a microporous glass membrane emulsification system. J Food Sci 2003;68:195–200.

14.6 References 

 337

54. Daufin G, Escudier JP, Carrère H, Bérot S, Fillaudeau L, Decloux M. Recent and emerging applications of membrane processes in the food and dairy industry. Trans IChemE 2001;79: 89–102. 55. Kelly PM, Kelly J, Mehra R, Oldfield DJ, Raggett E, O’Kennedy BT. Implementation of integrated membrane processes for pilot scale development of fractionated milk components. Lait 2000;80:139–153.

15 Electrodialysis in integrated processes for food applications Hélène Roux-de Balmann 15.1 Introduction Electrodialysis (ED) was firstly used in the 1960s in Japan for the production of drinking water from desalinated sea water, with the simultaneous production of sodium chloride, in order to tackle the scarcity of soft water. Since that time, ED systems and applications have changed considerably. Following the development of RO, which is the dominant technology for the production of drinking water from sea water desalination, ED moved to other applications, especially in the field of the food industry, where it can be integrated at different stages of the processes, in combination with other technologies, e.g., ion-exchange or nanofiltration. Other ED processes than conventional ED, such as bipolar-membrane ED, were also developed with the first industrial installations in the late 1990s. The main reason for implementing ED, like other membrane operations, is that it is expected to reduce the environmental impact of the processes as well as to improve the valorization of by-products. In this chapter, ED operations are first introduced. The membranes and systems used as well as the mass transfer mechanisms and their modeling are summarized. Then, some examples of processes used at the industrial scale for food applications, including ED operations, are presented and discussed. Obviously, many other applications are currently in operation. Moreover, because of the development of improved materials (membranes, spacers, etc.) and the increasing knowledge of the mechanisms involved in such processes and the increasing know-how of the suppliers, ED will be used for increasing applications in the future. The search and need for more and more sustainable processes is also a driver of that expansion.

15.2 Principle of electrodialysis 15.2.1 Conventional electrodialysis (EDC) 15.2.1.1 Membranes and stacks Conventional electrodialysis (EDC) uses ion-exchange membranes, Cation-exchange membranes (CEM) and anion-exchange membranes (AEM). These membranes allow a selective transfer of ions depending on their charge, i.e., transfer of cations and anions through CEM and AEM, respectively. The chemistry of these membranes is very close to that of ion-exchange resins. CEM have negative functional groups, such as SO3–. Once in a solution, these groups are in equilibrium with a cation, such as Na+

340 

 15 Electrodialysis in integrated processes for food applications

e.g., which is the membrane counter ion. Ion-exchange membrane (IEM) preparation and characterization are depicted into details in a recent review paper [1]. The transfer of ions takes place by a process of ion-exchange from one site to the other in the membrane material between the migrating specie and the membrane counter ion. Depending on the objective of the ED operation, these membranes are arranged in different configurations, like two-compartments or three-compartment configurations. The most common one is the two-compartment configuration, depicted on Figure 15.1, using an alternance of AEM and CEM membranes. The ED stack is composed of repeated AEM/CEM units placed between two electrodes. Under the effect of the voltage, cations migrating towards the cathode, transfer through the CEM membrane but are stopped by the adjacent AEM. In the same manner, anions migrating towards the anode transfer through the AEM but are stopped by the adjacent CEM. Following this selective migration of charged species, we obtain two solutions, respectively depleted “diluate” and enriched “concentrate” compared to the feed one. EDC is then an operation that realizes the simultaneous demineralization and concentration of feed solutions. IEM have an intrinsic selectivity, as different ions can have different mobilities through the membrane, depending on their charge and/or size. For example, monovalent ions generally have higher transfer rates than divalent ions. This selectivity can be tuned according to the membrane properties and selective IEM are available in the current market. The concentration or demineralization can be selective and provide a solution in which the ionic composition is different from that in the feed [2]. In the same manner, as neutral solutes do not migrate in the system, ED can be used to demineralize solutions containing neutral organic compounds while achieving a high organic yield. It was reported that the organic yield increases with the demineralization rate but also depends on the ionic composition [3].

Diluate demineralized MX (C  C 0 )

Concentrate concentrated solution MX (C  C 0 )

AEM CEM

AEM CEM AEM X

Anode 

X M

X M

M

Cathode 

Salt solution MX (C0 )

Figure 15.1: Conventional electrodialysis. Two-compartments configuration for the concentration/ demineralization of a salt solution AEM, anion-exchange membrane; CEM, cation-exchange membrane.

15.2 Principle of electrodialysis 

 341

15.2.1.2 Transfer mechanisms and modeling The migration of charged species, which is the main objective of ED, is accompanied by some other phenomena, the combination of which fixes the limits of the performances achievable. Because of the concentration gradient across the membranes, a diffusion flux takes place directed towards the decreasing concentration, i.e., from the concentrate to the diluate. Simultaneously, a solvent flux is established in the opposite direction, because of the osmosis phenomenon. Then, the charged species migrating through the membranes carry a shell of water molecules constituting their hydration layer. As a result, the solute flux is accompanied by a water flux, which is called “electroosmosis”, both being oriented in the same direction. The solute and solvent flux contributions are summarized on Figure 15.2. For charged species, the transfer coming from the current, i.e., migration and electroosmosis, is predominant. Thus, diffusion and osmosis can be neglected. On the contrary, for uncharged species, like sugars or other organic solutes for example, the only contribution comes from the diffusion, depending on the concentration gradient across the membrane. This transfer is thus expected to increase during the ED operation. In some situations, especially with low solute concentrations where diffusion is very low, it was demonstrated that an additional contribution can take place, coming from the convection because of the electroosmotic flux [4]. For any specie i, the mass balance gives the expressions of the concentrations in the compartments as function of the solvent, JV, and solute, Ji, fluxes, assuming that they remain constant during the operation [5]. These fluxes can be obtained from the variations of the solute mass and of the volume in the concentrate vs. time, as they correspond to the slope of the straight. The following expression is obtained for the  C variation of the concentration in the concentrate, Ci  (t), vs. time, in a batch mode: C

 C Ci  (t)

CEM

AEM

m0 + Jit = ________     C    V0  + JVt

(15.1)

CEM

JV electroosmosis JV osmosis Cathode 

Anode 

Ji migration Ji diffusion Diluate

Concentrate

Figure 15.2: Different contributions to the solvent, JV, and solute, Ji, fluxes involved in electrodialysis AEM, anion-exchange membrane; CEM, cation-exchange membrane.

342 

 15 Electrodialysis in integrated processes for food applications

c

c

in which mo  and v o  are the initial solute mass and volume in the concentrate. Because of the co-transport of water carried by the migrating ions, resulting in the solvent flux JV, the concentration does not vary linearly vs. time but reaches a plateau value, which is the maximum concentration that can be achieved. It was demonstrated that this maximum value can be calculated using the following relationship:  max

Ci 

J = __   i   JV

(15.2)

It is independent of the current and of the initial solute concentration but depends on the solute properties as well as on the membrane properties [5]. It is as high as the transport of water through the membranes is low. Then, higher concentrations are expected to be reached with less permeable, i.e., more reticulated, membranes or with less hydrated ions. At the microscopic level, it is possible to show, using the boundary layers or polarization concentration theory, the existence of a particular value of the current density for which the concentration at the membrane/solution interface in the diluate tends to zero [6]. This value, called the “limiting current density”, ilim, can be expressed according to the operating conditions using the following relationship (considering the transfer of a cation through a CEM as an example):  +

 +

C   ⋅ D ⋅ F C D  ⋅ F ⋅ k i lim  = ⎯⎯⎯⎯⎯⎯   +D +   =   ⎯⎯⎯⎯    +     (tm  − t ) ⋅ δ (tm  − t+)

(15.3)

 +

in which CD  is the concentration in the diluate, δ the boundary layer thickness, D the + diffusion coefficient, F the Faradic constant and t+ and tm   the cation transport number in the solution and in the membrane, respectively. This value is very important as it fixes the type of ions carrying the current. Below the limiting current, the current is carried by the ions initially present in the solution. On the contrary, for values above the limiting current, as the ion concentration tends to zero at the membrane interface, the current is carried by the H+ and OH– ions provided by the water dissociation. As this value is quite difficult to obtain from the former equation because of the parameters involved, it is generally determined experimentally. Another macroscopic parameter that can be used to evaluate the ED efficiency is the Faradic yield, η, (sometimes also called the current yield), defined as the fraction of current carried by the migrating ions. It is provided by the following equation: Σk zk.jk.F η = ⎯⎯⎯⎯       i

(15.4)

in which zk and jk are the valence and the flux of the corresponding k species, i is the current density and F is the Faradic constant.

15.2 Principle of electrodialysis 

 343

15.2.2 Electrodialysis with bipolar membranes (EDBM) 15.2.2.1 Membranes and stacks A bipolar-membrane (BPM) is composed of three adjacent layers, one cation-exchange face, CE, one anion-exchange face, AE, and an hydrophilic interface acting as a junction (Figure 15.3). Under specific current conditions, i.e., above limiting current density as defined in previous section, the electrical current can be carried by the only H+ and OH- ions coming from the water dissociation. Facing the CE and AE layers towards the cathode and anode, respectively, the H+ and OH– ions can be released into the solutions. In such conditions, the use of a BPM membrane enables a physical titration, i.e., the production of acid and base without the addition of chemicals. BPM are used in different configurations, in association with IEM, CEM or AEM, for different types of operations. The first configuration, for which BPM was developed at the industrial scale in the mid-1990s, is the three-compartment configuration, associating the BPM with AEM and CEM (Figure 15.4). It is used to produce acid and base from the corresponding salt, i.e., NaOH and HCl from a NaCl solution, for example. The salt solution is fed into a central compartment. Under the current, anions and cations migrate through the corresponding AEM and CEM membranes, to form acid and base solutions combining with the OH– and H+ ions produced by the BPM. A depleted salt solution is simultaneously obtained at the outlet of the salt compartment. The second configuration is the two-compartment configuration, associating a BPM with an AEM or a CEM. This latter is represented on Figure 15.5, in the case of a BPM/CEM configuration. This configuration is used to convert a weak acid salt, AcNa, into its acidic form, HAc, in the acidic compartment, i.e., where the H+ ions provided by the BPM are released. This conversion comes from the following equilibrium: nH(t) · nAc(t) AcNA + H2O → NaOH + HAc with the equilibrium (acidic) constant KA = ⎯⎯⎯⎯⎯       V(t) · nHAc(t) At the same time, a stream of base, i.e., NaOH for a sodium salt, is produced in the base compartment, where the OH– ions provided by the BPM are released, thanks to the migration of the weak acid counter ion through the CEM.

Anion exchange layer Anode  OH

H 2O H  OH

Cation exchange layer

H

Cathode 

Junction layer

Figure 15.3: Schematic drawing of a bipolar-membrane

344 

 15 Electrodialysis in integrated processes for food applications

Salt solution MX (CC0 ) Acid HX

Base MOH BPM OH

AEM CEM BPM

H

Anode 

H

OH

X

Cathode 

M

H 2O Salt solution MX (C0 ) Figure 15.4: Electrodialysis with bipolar-membrane. EDBM three-compartments configuration used for salt splitting AEM, anion-exchange membrane; BPM, bipolar-membrane; CEM, cation-exchange membrane.

Converted acid HAc

Base NaOH

BPM OH Anode 

MEC

H

BPM OH

H Cathode 

Ac Na

H 2O Acid salt solution AcNa Figure 15.5: Electrodialysis with bipolar-membrane. EDBM two-compartments configuration used for the conversion of a weak acid BPM, bipolar-membrane; CEM, cation-exchange membrane.

15.2.2.1.1 Transfer mechanisms and modeling The transfer mechanisms are identical to those described for EDC, except that the water dissociation equilibrium takes place inside the BPM. In the three-compartment configuration, the transfer of ions through the IEM is accompanied by a water flux, caused by electroosmosis. Then, in either the acid or base compartment, there is a concentration upper limit that cannot be overpassed, which is fixed by the following equations, which are similarly to equation (15.2): AEM

JCl−  acid C max   = ⎯⎯⎯  AEM   JV 

(15.5)

15.3 Food applications 

Concentrate (base)

COH

 345

Diluate (acid)

pH pKA Time (h)

Time (h)

Figure 15.6: Conversion of a weak acid by EDBM. Typical variations of the pH in the diluate (acidic compartment) and of the concentration of base in the concentrate (base compartment) CEM

JNa+  base C max  = ⎯⎯⎯  CEM   JV 

(15.6)

Again, these maximum concentrations depend on the ion and membrane properties. In the case of the two-compartment configuration, for the same reason, the concentration that can be reached in the base compartment is restricted to the following: CEM

JNa+  base C max  = ⎯⎯⎯  CEM   JV 

(15.7)

This maximum value can reach 2 M in usual conditions. But concerning the acidification in the acid compartment, there is not theoretical restriction and a complete conversion can be reached, following the replacement of the Na+ ions, migrating through the CEM, by the H+ ions produced by the BPM. The conversion factor, τ, can be obtained by the following relationship, knowing the equilibrium constant KA: 1    τ(t) = ⎯⎯⎯⎯⎯⎯⎯   1 + 10(pH(t)−pKA)

(15.8)

Typical variations of the concentration of base in the concentrate and of the pH in the acid compartment (titration curve) are illustrated in Figure 15.6.

15.3 Food applications 15.3.1 Whey In the dairy industry, whey is obtained from the manufacturing of cheese. Considered first as a waste, mainly used as a dry powder to feed animals, it is now considered as a valuable co-product or resource from which different fractions can be obtained. Membrane operations, combined with other technologies, are widely used at many stages in the processing of whey. UF is currently used to produce whey protein concentrate. Desalination of whey is also an important step for many applications. It can be done using different technologies, such as ED, nanofiltration (NF) and ion-exchange resins

346 

 15 Electrodialysis in integrated processes for food applications

Demineralization rate (%)

Sweet whey (6% TS)

Preconcentrated sweet whey (20% TS)

90

IEX 

60

20

NF

IEX 

ED 

NF

NF

IEX  ED

ED

Figure 15.7: Integrated processes for the demineralization of sweet whey: influence of the whey composition and of the demineralization rate on the process configuration ED, electrodialysis; IEX, ion-exchange resins; NF, nanofiltration; TS, total solids content of the whey.

(IEX). Depending on the composition of the whey, sweet vs. acid whey for example, as well as on use of the end product, the demineralization process has to be tailored. Competing technologies can be compared one by one, as was done for ED and IEX for example, considering the cumulative energy demand, including that of the process itself but also that coming from the need of chemicals or from the treatment of the wastes [7]. In the conditions investigated, i.e., the demineralization of sweet whey with a total solid (TS) content of 20%, it was concluded that a prior step by NF should be used to concentrate and partially demineralize the whey, especially to remove the monovalent ions. Then, for further desalination down to a 90% demineralization rate, ED is preferred as the cumulative energy demand of ED is significantly lower than that of IEX. In addition to that, the effluent volume as well as the total salt discharge is much lower when IEX are used. Moreover, the optimum process becomes more and more complex to reach increasing demineralization rates. Thus, it generally requires the combination of different technologies. However, even if no general conclusion can be drawn concerning the best combination to reach a given demineralization rate, it is possible to provide some general guidelines, as schematically represented in Figure 15.7.

15.3.2 Sugar and beverages industry Membrane operations are widely used in the sugar industry for different purposes. Indeed, with the increasing cost challenges and environmental concerns the sugar yield is to be maximized and this can be done using membrane technologies. A typical example is the removal of melassigenic ions, i.e., inorganic cations such as K+ and Ca2+. As these ions prevent sugar being recovered by crystallization, their extraction from the molasses can improve the sugar yield. IEX and ED are well established technologies for this. ED can be used to remove K+ and Na+ ions but only in

15.3 Food applications 

Mother liquor (Na, K )

Thin juice

Electrodialysis

 347

Brine (Ca2)

IEX Cationic (Na, K form)

IEX Cationic (Ca2form)

Demineralized thin juice Mother liquor (without Na, K ) Figure 15.8: Combined process including electrodialysis and ion-exchange resins for the simultaneous desalination of thin juice and removal of melassigenic ions from mother liquor

controlled operating ranges regarding the temperature and the liquor concentration to prevent contamination and guarantee the membranes and stacks life span [8]. Specific technological improvements, especially concerning the spacers and the anionic exchange membrane, have developed recently in order to extend this operating range and to increase the economy of the ED process. Conversely, the removal of melassigenic ions can be also done using IEX in the Mg2+ form but the problem is the amount of Mg2+ required for the resins regeneration. To tackle that problem, it was considered to use Ca2+ resins instead of Mg2+ resins and to use the Ca2+ ions naturally present in the thin juice following the liming operations upstream in the process, but the Ca2+ concentration was too low. An interesting process, depicted in Figure 15.8, was thus proposed in which ED is combined with IEX [9]. In this process, the Ca2+ enriched brine (concentrate) obtained during the ED treatment of the thin juice is used to regenerate the IEX in the suitable Ca2+ form for the removal of melassigenic ions from the mother liquor. Many applications of electrodialysis operations can be found in the beverage industry. One is the deacidification of fruit juices, ED being one of the different possible technologies, together with precipitation and IEX, for example. These processes have been compared in the case of the deacidification of a clarified passion fruit juice [10]. Two ED processes were investigated, a three-compartment AEM/ AEM/CEM configuration (EDC3) and a two-compartment AEM/BPM configuration (EDBM2). It was concluded that ED was the most appropriate technology, particularly EDBM2. Indeed, the results showed that the physico-chemical and sensorial properties of the juice were satisfactory, contrary to those obtained after IEX treatment. Conversely, EDBM2 does not require any chemicals and produces a citric acid solution that can be used to compensate the higher energy cost compared to some other processes. Different combinations were further tested for the deacidification of a pulpy juice by EDBM, as depicted on Figure 15.9 [11]. Two clarification

348 

 15 Electrodialysis in integrated processes for food applications

Pulpy juice

Enzymatic treatment

EDBM2C

Centrifugation

Microfiltration

EDBM2C

EDBM2C

Clarification

Deacidification

Deacidified pulpy juice Figure 15.9: Production of a deacidified pulpy juice. Different process routes. EDBM2C, electrodialysis with bipolar membranes two-compartments AEM/BPM

technologies, – centrifugation and MF were considered – the juice being previously submitted to an enzymatic treatment. The pulp extracted during the clarification step can be recycled after the deacidification to obtain a pulpy deacidified juice. EDBM can also be used to acidify cloudy juices, in order to inactivate enzymes responsible for juice browning and opalescence instability. Compared to the direct acidification using HCl for example, the EDBM process avoids the juice dilution and the salty aftertaste. The pH adjustment can be optimized for the enzyme activation. To complete the treatment, the pH further needs to be increased at its initial value while maintaining the enzyme inactivation. Different combinations were tested for the treatment of a cloudy apple juice [12]. On one hand, acidification and pH readjustment steps were performed by EDBM in a BP/AEM configuration, using different base solutions, like KCl and HCl. On the other hand, the combination of the EDBM treatment followed by a mild heat treatment was investigated. It was concluded that this combination is an efficient stabilization process. Effluents and by-products of many food processing procedures contain valuable compounds that can be recovered using ED operations. For example, the concentration and purification of malate contained in beverage industry waste water was investigated [13]. It was shown that a high malate recovery, up to 85%, can be achieved by EDC, together with a five-fold concentration, while removing most of the divalent cations and citrate ions.

15.3.3 Wine As described above for other beverages, EDBM can be used to adjust the pH, and very fine tuning can be obtained without the addition of chemicals. The pH can be reduced

15.3 Food applications 

 349

by a BPM/CEM configuration, in a range between 0.1 and 0.3 for wine and grape must and up to 0.9 for grape juice. As only the cations (mainly K+) are removed, flavors and color are perfectly preserved. But the main application of ED is probably the tartaric stabilization of wine. The treatment consists of the removal of the potassium bitartrate ions to reach a concentration low enough to prevent any further precipitation during storage. Different methods can be used to stabilize wines [14]. The most traditional one is the cold treatment, in which the wine is cooled down to reach precipitation of tartrate. The crystals are then removed to get the stabilized wine. This cold treatment is a time- and energyconsuming process. Then, other technologies were investigated. IEX resins can be used, but only the cationic ones are authorized for wine treatment. The treated wine has to be mixed with untreated wine in order to obtain an equilibrated product. The mixing ratio, i.e., the rate of treated vs. untreated wine, that is generally in the range 10–20%, needs to be adapted. The automation of the process is thus difficult. Moreover, IEX treatment can produce slight changes in color and sensorial properties. In spite of this, it is effective for almost all wines and less time- and energy-consuming than the cold treatment. ED is also an interesting technology as the deionization rate can be tuned by adjusting the current density and it can be monitored according to the electrical conductivity. Moreover the variation of the pH, alcoholic strength, total and volatile acidity during the ED treatment are very low. Finally, it was observed that this treatment can also simultaneously reduce the sulfate ions content, which is interesting for wines previously acidified with gypsum. Figure 15.10 shows an integrated process including an ED step. The wine is first clarified using a microfiltration system, in order to ensure the microbiological stability and the brilliance. The MF retentate is recycled in order to maximize the wine recovery. The filtrate is then fed into a two-compartment ED stack operating in a “feed-and-bleed” mode in order to be stabilized. A tank located on the MF permeate is used to mix the ED-treated wine, i.e., the ED diluate, with the untreated wine to

MF

(1) EDC

Diluate Concentrate

Wine

Retentate Permeate

Stabilized wine

RO Figure 15.10: Integrated processes for the stabilization of wine EDC, conventional electrodialysis; MF, microfiltration; RO: reverse osmosis; (1) mixing tank for wine adjustment.

350 

 15 Electrodialysis in integrated processes for food applications

reach the appropriate composition. An RO system is mounted in the ED concentrate loop in order to increase the water recovery, i.e., to decrease the water consumption of the overall process. In some conditions, it is possible to reduce the water consumption down to 5% of the treated wine volume.

15.3.4 Organic acids Organic acids (OA) are used in different industries, and their use in food applications are very important. They are mainly produced by fermentation. Fermentations are generally carried out in alkaline or neutral pH conditions for productivity purposes, in order to maintain as low as possible the concentration of the acid, which is an inhibitor of the fermentation. Then, the organic acid is produced after the fermentation as an anion, for example, lactate. The counter ion is the cation of the base used for the pH regulation, i.e., Na+ for a regulation with caustic soda or NH4+ for a regulation with ammonia. After the fermentation, the broth has to be treated in order to get the organic acid in a suitable form regarding its final use. This is done using a combination of different operations, named downstream processes, including clarification, extraction and purification. Extensive research was devoted to the production of lactic acid from fermentation [15]. But the outcomes can be also applied to other OA. The main drawback of conventional downstream processes, generally based on precipitation, is the production of large amount of effluents with a high salt content. In order to reduce this environmental impact, other processes, including membrane operations, were thus investigated. One of these is depicted on Figure 15.11.

Recycling (diluate)

Fermentation

MF

Removal divalent cations

EDC

Water

Effluents

EDBM

Purification

Effluents Recycling (base stream)

Organic acid

Sugars yeast nutriments water Figure 15.11: Integrated process including ED for the production of organic acids from fermentation broths EDC, conventional electrodialysis; EDBM, electrodialysis with bipolar membranes CEM/BPM; MF, microfiltration.

15.4 References 

 351

The fermentation broth is first clarified using crossflow MF [16]. Then, the divalent cations are removed in order to prevent any scaling during the further EDBM step. This is generally done using IEX but NF can also be considered [17]. The organic acid salt is then concentrated by EDC. This concentration is generally accompanied by a purification, i.e., a removal of the residual sugars. The diluate can be recycled to the fermentation. The conversion of the organic salt into its neutral form is next carried out by EDBM, using a BPM/CEM configuration (Figure 15.5). The acidic stream contains the converted acid. As previously pointed out, the conversion can be tuned with respect to the pH profile and very high conversion factors can be reached. The base stream can be recycled to the fermentation for controlling the pH or used for the regeneration of the IEX. Final purification can be made using anionic IEX to remove, if necessary, the remaining traces of inorganic anions.

15.4 References 1. Nagarale RK, Gohil GS, Shahi VK. Recent developments on ion-exchange membranes and electro-membrane processes. Adv Colloid Interfac 2006;119:97–130. 2. Zhang Y, Van der Bruggen B, Pinoy, Meesschaert B. Separation of nutrient ions and organic compounds from salts in RO concentrates by standard and monovalent selective ion-exchange membranes used in electrodialysis. J Membrane Sci 2009;332:104–112. 3. Singlande E, Roux-de Balmann H, Lefebvre X, Sperandio M. Improvement of the treatment of salted liquid waste by integrated electrodialysis upstream biological treatment. Desalination 2006;199:64–66. 4. Borges F, Roux-de Balmann H, Guardani R. Investigation of the mass transfer processes during the desalination of water containing phenol and sodium chloride by electrodialysis. J Membrane Sci 2008;325:130–138. 5. Bailly M, Roux-de Balmann H, Aimar P, Lutin F, Cheryan M. Production processes of fermented organic acids targeted around membrane operations: design of the concentration step by conventional electrodialysis. J Membrane Sci 2001;191:129–142. 6. Rautenbach R, Albrecht R. Membrane Processes. Wiley Interscience: New York; 1989. 7. Greiter M, Novalin S, Wendland M, Kulbe KD, Fischer J. Electrodialysis versus ion exchange: comparison of the cumulative energy demand by means of two applications. J Membrane Sci 2004;233:11–19. 8. Elmidaoui A, Chay L, Tahaikt M, Menkouchi Sahli MA, Taky M, Tiyal F, Khalidi A, Alaoui Hafidi R. Demineralsation for beet sugar solutions using an electrodialysis pilot plant to reduce melassigenic ions. Desalination 2006;189:209–2014. 9. Zhu S, Yu S, Fu X, Chai X, Zhu L, Yang X. A modified Quentin method for desalting thin juice and mother liquor B in beet sugar industry. Desalination 2001;208:214–220. 10. Vera E, Ruales J, Dornier M, Sandeaux J, Sandeaux R, Pourcelly G. Deacidification of clarified passion fruit juice using different configurations of electrodialysis. J Chem Technol Biot 2003;78:918–925. 11. Vera E, Sandeaux J, Persin F, Pourcelly G, Dornier M, Ruales J. Deacidification of passion fruit juice by electrodialysis with bipolar membrane after different treatments. J Food Eng 2009;90:67–73. 12. Lam Quoc A, Mondor M, Lamarche F, Ippersiel D, Bazinet L, Makhlouf J. Effect of combination of electrodialysis with bipolar membranes and mild heat treatment on the browning and opalescence stability of couldy apple juice. Food Res Int 2006;39:755–760.

352 

 15 Electrodialysis in integrated processes for food applications

13. Lameloise ML, Matinier H, Fargues C. Concentration and purification of malate ion from beverage industry waste water using electrodialysis with homopolar membranes. J Membrane Sci 2009;343:73–81. 14. Lasanta C, Gomez J. Tartrate stabilization of wines. Trends Food Sci Tech 2012;28:52–59. 15. Joglekar HG, Raman I, Babu S, Kulkarni BD, Joshi A. Comparative assessment of downstream processing options for lactic acid. Sep Sci Technol 2006;52:1–17. 16. Carrère H, Blaszkow F, Roux-de Balmann H. Modeling of the clarification of lactic acid fermentation broth by cross flow microfiltration. J Membrane Sci 2001;91:129–142. 17. Bouchoux A, Roux-de Balmann H, Lutin F. Investigation of nanofiltration as a purification step for lactic acid production processes based on conventional and bipolar electrodialysis operations. Sep Purif Technol 2006;52:266–273.

Index acetone 16–17, 19 acetone-butanol-ethanol (ABE) 12 acid whey 135–136 analysis of mixture properties (AMP) 34, 41 analysis of pure component properties (APCP) 34, 41 analysis of reaction (AR) 34, 41 antioxidant activity 257, 259 antioxidants 16 anthocyanins 74–75, 99–100, 269, 298 apply extended Kremser method (AKM) 34, 41 aroma compounds 66, 69, 72, 74, 89, 93, 97, 102–103, 300 ascorbic acid 99, 107 back-flushing 150 bactofugation 114 back-pulsing 150 back-washing 150 beet sugar 163–172 – demineralization 172, 346–347 – preconcentration of thin juice 173–175 – press water 165–167 – raw juice purification 167–172 beverage emulsions 330–332 bicinchoninic acid 259 bioactive peptides 143–144, 234–235 bioalcohol 11–12 biocatalytic membrane reactors (BMRs) 312–316 – alcoholic beverages 318–319 – fats and oils 318 – fruit juice processing 317 – functional molecules 318 – starch sugars 317 – water purification 319–321 biologically active compounds (BAC) 272, 295 – fractionation/concentration 300–306 – market 295–297 – production 297–300 biological oxygen demand (BOD5) 133, 138 biofouling 312 biofuel 1, 3, 11–13 biomass 1–4, 10, 11, 13, 21 biorefinery 1–2, 11–14 – membranes in biorefinery 4–10, 20–21

bovine serum albumin (BSA) 133, 141–142, 259 brewing industry 183, 185 brewing process 185–194 – beer clarification 188–190 – beer dealcoholization 190–193 – beer from tank bottoms 193–194 – fermentation 187 – filtration 187 – filtration in the lautering process 187–188 – lautering 186–188 – malting 186 – mashing 186 – maturation 187 – milling 186 – packaging 187 – wort boiling 186 – wort clarification 187 brine – flow rate 67 – concentration 67 building blocks 28–29, 31–32 – simultaneous phenomena (SPB) 33, 40, 48–49 butanol 12 cake filtration 64 cane sugar 163–164, 175–182 – concentration of clarified juice 181 – decolorization of remelted sugar 182–183 – molasses treatment 181–182 – raw juice purification 176–181 canola 15 β-carotene 81 capillary electrophoresis (CE) 271–272 carotenoids 18, 81 casein 113–15, 117, 121–125, 135, 143 cation transport number 342 centrifugation 114, 148, 212–215 cheese 113, 121 – mascarpone 125–127 – quark-type 122–123 citrus juice processing – clarification 89–92 – concentration 93–102 – debittering 92–93

354 

 Index

– recovery of aroma compounds 102–103 – treatment of by-products 103–107 citrus juices – bergamot 66, 106–107 – blood orange 64, 75, 90, 98–101 – frozen concentrated orange juice (FCOJ) 87 – lemon 92, 94 – mosambi 61–63, 90 – not-from-concentrate (NFC) orange juice 87, 89 – orange 69–70, 91–94, 97, 102 – refrigerated orange juice from concentrate (RECON) 87–89 citrus peel 105 – oil extraction 17–18 citrus waste 17 clotted cream cheese 125 concentration polarization 6, 15, 90, 301 continuous stirred-tank reactor (CSTR) 45 cow colostrum 142–143 critical flux 150 cross-flow filtration 5 cross-rotation filtration 5–6 cryoconcentration 65 cutting 10 dairy emulsions 332–333 Darcy’s law 238 decimal reduction (DR) factor 116 decomposition-based solution 29, 31 degree of deionization (DD) 154 degumming 14–16 deodistillate 304–305 development of superstructure (DS) 34, 41 diafiltration 123, 139, 144, 167, 191–192, 207–208, 222, 224, 226, 236, 247–249, 251–254 dialysis 191 diatomaceous earth filtration 60, 89, 148, 188 disk filter systems – duble shaft 7 – single shaft 6 driving force method 43 dynamic filtration 151–152 electrodialysis (ED) 136–137, 152–156, 172, 182, 239–244, 339–342 – bipolar membranes 156, 241–243, 249–254, 343–345, 347–348, 351 – integrated ED–UF process 243–244, 258–261 – monopolar membranes 239–241

electromagnetic pulses (PEF) 14 emulsifying activity index 248–249, 256 emulsion stability index 248–249, 256 energy transfer phenomena 32 ethanol 12, 16, 19 extended shelf-life (ESL) milk 114–115 extended Kremser method 43 extrusion 10 evaporation flux 68, 71, 97–98 Faradic constant 342 fish oil 18 flavanols 269 flavonoids 105–107, 269 fluid milk stabilization 113–119 foam capacity 249 forward osmosis 104, 273, 275, 306 free fatty acids 15 free lactose milk 119 fruit juices – acerola 67 – apple 63–64, 66, 71–72 – blackcurrant 66, 74 – cactus pear 79–80 – kiwi 62, 69, 78–79 – melon 62, 81 – pear 66 – pineapple 61–62, 65, 68 – pomegranate 62, 75, 77 – red 74–78 fruit juice processing – clarification 59–65 – concentration 65–71, 281–288 – acidification/deacidification 347–348 – integrated membrane operations 71–81 gas chromatography (GC) 271 goodies 18–20 grinding 10 growth factors 144 heating 10 heptane 16 hexane 15–16 high performance liquid chromatography (HPLC) 271 hyaluronic acid 297 hydrophile–lipophile balance (HLB) number 328

Index 

hydroxycinnamic acids 99–101 hydroxytyrosol 19, 303 ideal whey 135 immunoglobulins 133, 141–143 ion-exchange resins (IEX) 34–347, 349, 351 isoelectric precipitation 233–234, 245, 254 isoflavones 245–246 isopropanol (IPOH) 45, 53–54 isopropyl acetate (IPAc) 45–47, 51–54 isothermal membrane distillation 67–69 knowledge base search (KS) 34, 41 knowledge base synthesis approach (KBS) 29, 34–39, 42, 51 knowledge base tool 42 lactic acid bacteria (LAB) starter 122 α-lactalbumin 133, 141–142 lactoferrin 133, 141 β-lactoglobulin 133, 141–142 lactose hydrolysis 119–120 lecithin 15–16 liming 170 limitation/bottleneck analysis (LBSA) 34, 41 limiting current density 342 limonene 296 limonin 92–93 limonoate A-ring lattone (LARL) 92–93 low-alcohol beer 190–193 MMV process 122 membrane-assisted process intensification 26–27 – synthesis/design concepts 28–33 – synthesis/design workflow 34–40 – synthesis/design sub-algorithms 41–44 membrane distillation (MD) 69–74, 95, 97, 99, 102, 173, 181, 195, 282, 284, 289, 306 – air gap (AGMD) 70 – direct contact (DCMD) 70, 74–75, 287, 289 – sweep–gas (SGMD) 70, 284 – vacuum (VMD) 70, 74, 281, 284, 286 membrane contactor (MC) 159–160, 195 membrane emulsification (ME) 323–326 – aerated food gels 331 – applications 328–333 – cross–flow 325–326 – dead–end 325–326

 355

– direct 324 – encapsulation 330–331 – membranes 326–327 – multiple emulsions 329–330 – premix 325 – simple emulsions 328–329 membrane evaporation 67–69 membrane foaming 331 membrane fouling 5, 7, 9–10, 12–13, 15–16, 62–66, 78, 90–91, 104–107, 118, 124, 139–141, 145, 148–150, 158, 163, 167, 172–173, 179, 190, 219, 237–238, 244–245, 250–254, 259, 261, 273, 289, 300, 314, 317, 319, 324, 332, 334 membrane osmotic distillation 78, 284 membrane reactors 119 membranes – anionic-exchange 153, 156, 239, 242–244, 339–340, 343–344 – asymmetric 10.5 – cation-exchange 153, 156, 239, 242–244, 339–340, 343–344 – cellulose acetate 93, 104, 167, 173 – ceramic 60, 105, 90–91, 98, 151, 178, 182–183, 193, 333 – hydrophilic 149 – hydrophobic 67, 69, 95, 149, 159 – microporous glass (MPG) 333 – organophilic 12 – polyamide (PA) 60, 104, 106 – poly-di-metyl-syloxane (PDMS) 12, 103, 193, 301 – poly(ether ether ketone) (PEEK WC) 63, 91 – polyethersulfone (PES) 61, 92, 106, 140, 149, 188–189, 208, 216, 251 – polyoctylmethylsiloxane (POMS) 301 – polypiperazine amide 106 – polypropylene (PP) 60, 68–70, 74, 81, 101–102 – polysulfone (PS) 60, 92, 166 – polytetrafluoroethylene (PTFE) 102, 327 – poly(1-trimethylsilyl-1-propyne) (PTMS) 301 – polyvinylidene fluoride (PVDF) 60, 90–93, 97, 102, 124 – rotating 6–7, 151, 326 – Shirasu porous glass (SPG) 326, 333 – stainless steel 169, 178, 190 – symmetric 237 – vibrational 7–10, 151

356 

 Index

microalgae 13–14 microcapsules 330, 333 microfiltration (MF) 59–65, 72, 74, 89–92, 97–98, 104–105, 114–120, 122–124, 127, 135, 141–144, 147–152, 160, 169–171, 175–185, 188–190, 195, 207–208, 212–215, 281, 283, 286–287 mild extraction techniques 2–4 – cold aqueous acid extraction 3 – enzymatic extraction 3 – milling and pressing 3, 10 – pulsed electric field 4 – supercritical CO2 4 – ultrasound 3 milk processing – fluid milk 114–120 – cheese milk 121–127 milk protein concentrate (MPC) 124–125 mixing 32 model-based search (MBS) 34, 41 model library 42 must concentrate (MC) 157 nanoparticles 330 nanofiltration (NF) 65–66, 74–75, 106–107, 122–123, 127, 136–138, 143–144, 156–158, 166, 171–174, 207–208, 220–222, 258–260, 273–282, 286–288, 302–306 oil algae 11–12 oleuropein 19–20 oligosaccharides 245–246 olive oil 17 – waste 19 omega fatty acids 18 – emulsions 120 operating process window (OPW) 34, 41 organic acids 124–125 organic solvent nanofiltration (OSN) 273 osmotic distillation (OD) 67–69, 72, 74–75, 77–81, 97–101, 159–160, 173, 195, 273–274, 281–286, 289 oxygen radical absorbance capacity (ORAC) 259, 261 pasteurized milk 114–119 pectin 104

permeate flux 8, 61–64, 66, 70–71, 90–94, 97, 104–106, 117, 127, 149, 150–151, 158, 166, 207–208, 212, 214–217, 219–220, 222, 224, 236–238, 245, 252–254 pervaporation (PV) 12, 47, 72, 78, 102, 103, 191, 193, 195, 300–302 phase-change 32 phase-contact 32 phase-separation 32 phase-transition 32 phenomena-based synthesis (PBS) 29, 34–39, 42, 53 phytic acid 245–247, 251–252 phytosterols 16, 18 phospholipids 15–16 plasmin 118 plasminogen 118 polyphenols 19–20, 66, 74–75, 106, 149, 269–270 – separation/concentration 270–272 – concentration by membrane processes 272–281 pore blocking – complete 64, 91–92 – intermediate 215 – internal 64 – partial 64, 215 – standard 215 potassium hydrogenotartrate (KHT) 152 press liquor 103, 105 proanthocyanidins 269 process intensification 25–27, 71, 108, 195 rapeseed 15 reaction 32 rectified must concentrate (RMC) 157 resistance – fouling 63, 150, 238 – membrane 63, 238 – reversible polarized layer 63, 150, 238 – semi–reversible polarized layer 63 – total 63, 149 resistance-in-series model 63 response surface methodology 106, 193, 209–213 reverse osmosis (RO) 66–67, 72–75, 93–95, 103–104, 121–122, 135–136, 143, 157–158, 166, 173–175, 191–193, 195, 208, 273, 279, 281–283, 285–287 road-side grass 10–11 roasting 10

Index 

rotary drum filtration 152 rotating dynamic crossflow filter 151 selection of phenomena (SoP) 34, 41 serum proteins 138–143 simulated milk ultrafiltrate (SMUF) 117 solvent recovery 16–17 somatic cells (SC) 118–119 soybean products – soy bioactive peptides 234–235, 254–261 – soy protein hydrolysate 255–258 – soy protein isolates 233–235, 244–256 – soy protein products 235–244 Stevia rebaudiana Bertoni 201–202 steviol glycosides 202–207 steviol glycosides extraction 205–207 – chelating agents 206 – chromatographic separation 206 – hot water extraction 208–212 – ion-exchange 205 – microwave-assisted extraction 206 – solvent extraction 205 – supercritical fluid extraction (SCFE) 206–207 – ultrasonic extraction 206 stevioside purity 216–217 stevioside selectivity 216–217 stream dividing 32 sunflower 15, 17 sweet whey 135, 139 tartrates 152–156 thermodynamic insights 42–43, 47 thin layer chromatography 271 thin juice 165, 173–175, 347 toasting 16 tocopherols 16, 18–19 total antioxidant activity (TAA) 72, 75, 77–79, 90–91, 98–99, 107 transmembrane pressure 63, 90, 118, 124, 175, 238 tyrosol 19–20 ultrafiltration (UF) 59–65, 72, 78–80, 89–93, 97–101, 104–106, 119, 121–127, 139–140, 142–144, 156–157, 167–171, 175–185, 195, 207–208, 215–222, 224–226, 235–238, 244–249, 251–261, 273–278, 280–288, 305

 357

– crossflow 237 – dead-end 237 uniform transmembrane pressure (UTP) 116–117, 124 unit operation-based synthesis (UBS) 29, 34–39, 42, 53 vanillin 296, 302 vapor-liquid interface 67 vapor permeation (VP) 12, 300–302 vapor transport 67, 70 vegetable oils and fats 14–20 vibratory shear-enhanced processing (V-SEP) 8–9 volume concentration factor (VCF) 115, 117, 121–123, 125, 167, 178, 220, 222 volume concentration ratio (VCR) 236, 248, 259 volume reduction factor (VRF) 61–64, 137 volume reduction ratio (VRR) 136, 138–139, 222, 251–253 wax removal 17–18 whey composition 133–135 whey processing – concentration and demineralization  135–138, 346 – concentration of serum proteins 138–141 – fractionation of serum proteins 141–143 – value-added products 143–144 whey proteins 133 whey protein concentrate (WPC) 139–140, 345 whey protein isolate (WPI) 140–141 whey separation 113 wine making – clarification of must and wines 148–152 – gas control 160 – must concentration 157–158 – reduction of must sugars 156–157 – wine dealcoholization 158–160 – wine stabilization 152–256, 348–350 wine polysaccharides 149 winterisation 17 xylo–oligosaccharides 297