Superplasticizers and Other Chemical Admixtures in Concrete

Superplasticizers and Other Chemical Admixtures in Concrete Proceedings Eleventh International Conference Ottawa, ON, Canada July 2015

V. Mohan Malhotra Pawan R. Gupta Terence C. Holland

SP-302

Eleventh International Conference on Superplasticizers and Other Chemical Admixtures in Concrete

Editors: V. Mohan Malhotra Pawan R. Gupta Terence C. Holland

SP-302

First printing, June 2015 Discussion is welcomed for all materials published in this issue and will appear ten months from this journal’s date if the discussion is received within four months of the paper’s print publication. Discussion of material received after specified dates will be considered individually for publication or private response. ACI Standards published in ACI Journals for public comment have discussion due dates printed with the Standard. The Institute is not responsible for the statements or opinions expressed in its publications. Institute publications are not able to, nor intended to, supplant individual training, responsibility, or judgment of the user, or the supplier, of the information presented. The papers in this volume have been reviewed under Institute publication procedures by individuals expert in the subject areas of the papers.

Copyright © 2015 AMERICAN CONCRETE INSTITUTE 38800 Country Club Dr. Farmington Hills, Michigan 48331 All rights reserved, including rights of reproduction and use in any form or by any means, including the making of copies by any photo process, or by any electronic or mechanical device, printed or written or oral, or recording for sound or visual reproduction or for use in any knowledge or retrieval system or device, unless permission in writing is obtained from the copyright proprietors.

Printed in the United States of America Editorial production: Carl R. Bischof Tiesha Elam Kaitlyn J. Hinman Ryan M. Jay Kelli R. Slayden ISBN-13: 978-1-942727-22-4

Preface In May 1978, the Canada Center for Mineral and Energy Technology (CANMET), in association with the American Concrete Institute (ACI) sponsored a three-day conference in Ottawa, ON, Canada, on the use of superplasticizers in concrete. Selected papers from the symposium were published as ACI SP-62. In 1981, CANMET, again in association with the ACI, sponsored a second three-day international conference in Ottawa on the use of superplasticizers in concrete. Proceedings of the conference were published as ACI SP-68. The purpose of the third international conference in Ottawa in 1989 was to review the progress made since the meetings in 1978 and 1981, and to bring together representatives of the chemical admixtures, cement, and concrete industries to exchange information and delineate new areas of needed research. The scope of this conference was expanded to include chemical admixtures other than superplasticizers. Proceedings of the conference were published as ACI SP-119. In October 1994, CANMET, in association with the ACI and several other organizations, sponsored the fourth conference in Montreal, QC, Canada. The objective of this conference was to bring attention to new developments in chemical admixtures since the last conference in 1998. Proceedings of the conference were published as ACI SP-148. In October 1997, the Committee for the Organization of CANMET/ACI International Conferences (ACI Council), in association with the ACI and several cement and concrete organizations in Italy, sponsored the fifth conference in Rome, Italy. The conference was aimed at transferring technology in the fast-moving field of chemical admixtures. Proceedings of the conference were published as ACI SP-173. In October 2000, the Committee of the Organization of CANMET/ACI International Conferences (ACI Council), in association with several organizations in Canada and France, sponsored the sixth conference in Nice, France. More than 50 papers from more than 20 countries were received and reviewed by the ACI review panel, and 37 were accepted for publication in the proceedings of the conference. The proceedings were published as ACI SP-195. In October 2003, the Committee for the Organization of CANMET/ACI International Conferences (ACI Council), in association with several organizations in Canada and Germany, sponsored the seventh conference in Berlin, Germany. The conference attracted more than 275 delegates and proceedings of the conference, consisting of 39 papers, were published as ACI SP-217. In October 2006, the Committee for the Organization of CANMET/ACI International Conferences (ACI) Council), sponsored the eight conference in Sorrento, Italy. More than 60 papers from more than 25 countries were received and peer reviewed by the CANMET/ ACI review panel in Budapest, and 36 were accepted for publication as ACI SP-239. In October 2009, the Committee for the Organization of International Conferences (COIC) (formerly CANMET/ACI International Conferences), sponsored the ninth ACI International Conference in Seville, Spain. More than 50 papers from more than 20 countries were received and peer reviewed, and 35 were accepted for publication in the proceedings of the conference. The proceedings were published as ACI SP-262. In October 28 to 31, 2012, COIC, sponsored the Tenth International Conference on Superplasticizers and Other Chemical Admixtures in Concrete in Prague, Czech Republic. More than 70 papers from the world over were peer reviewed, and 33 were accepted for

publication in the proceedings of the conference. The proceedings were published as ACI SP-288. In July 10-13, 2015, the COIC, in association with ACI, sponsored the Eleventh International Conference on Superplasticizers and Other Chemical Admixtures in Concrete in Ottawa, Canada. More than 60 papers from the world over were peer reviewed, and 28 were accepted for publication in the proceedings of the conference. The proceedings were published by ACI as SP-302. Also, additional papers were presented at the conference that were published in the Supplementary Papers Volume. Thanks are extended to members of the Technical Paper Review Panel that met in the Bahamas from October 3 to 11, 2014. Without their dedicated effort and hard work, it would not have been possible to publish the proceedings. Cooperation of the authors in accepting the reviewers’ suggestions and in revising the manuscripts accordingly is appreciated greatly. The help and assistance of Dr. Pawan R. Gupta and Prabha Gupta are acknowledged gratefully in the administrative work associated with the conference and processing of the manuscripts for both the conference proceedings (ACI SP-302) and the Supplementary Papers Volume. V. Mohan Malhotra Pawan R. Gupta Terence C. Holland Editors Eleventh International Conference on Superplasticizers and Other Chemical Admixtures July 12 to 15, 2015 Ottawa, Canada

Contents Preface 3 SP-302-01 11 Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes by Joseph J. Assaad and Salim E. Asseily INTRODUCTION 11 RESEARCH SIGNIFICANCE 12 EXPERIMENTAL INVESTIGATION 12 TEST RESULTS AND DISCUSSION 15 CONCLUSIONS 21 SP-302-02 25 A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques by Julia Pickelmann, Huiqun Li, Robert Baumann, and Johann Plank INTRODUCTION 25 RESEARCH SIGNIFICANCE 26 EXPERIMENTAL PROCEDURES 27 RESULTS AND DISCUSSION 29 CONCLUSION 34 SP-302-03 39 Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents by Wolfram Schmidt, Sarah Peters, and Hans-Carsten Kühne INTRODUCTION 39 RESEARCH SIGNIFICANCE 41 EXPERIMENTAL INVESTIGATION 41 EXPERIMENTAL RESULTS AND DISCUSSION 44 SUMMARY AND CONCLUSIONS 50

SP-302-04 53 New Additive to Enhance the Slump Retention by David Platel, Jean-Marc Suau, Clement Chosson, and Yves Matter INTRODUCTION 53 RESEARCH SIGNIFICANCE 54 EXPERIMENTAL AND ANALYTICAL INVESTIGATIONS 54 EXPERIMENTAL RESULTS AND DISCUSSION 56 FURTHER RESEARCH 59 CONCLUSIONS 60 SP-302-05 63 Synthesis of a Novel Superplasticizer Prepared from Brown Coal by Manuel Ilg and Johann Plank INTRODUCTION 63 RESEARCH SIGNIFICANCE 64 EXPERIMENTAL PROCEDURE 64 EXPERIMENTAL RESULTS AND DISCUSSION 68 CONCLUSIONS 74 SP-302-06 77 Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions by Lucia Ferrari and Pascal Boustingorry INTRODUCTION 77 RESEARCH SIGNIFICANCE 78 EXPERIMENTAL PROCEDURES 78 ANALYTICAL INVESTIGATION 80 COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS 81 EXPERIMENTAL RESULTS AND DISCUSSION 81 FURTHER RESEARCH 87 CONCLUSIONS 88

SP-302-07 93 Influence of Temperature and Retarder on Superplasticizer Performance by Karen Luke and Adrian Torres INTRODUCTION 93 RESEARCH SIGNIFICANCE 94 EXPERIMENTAL PROCEDURE 95 EXPERIMENTAL RESULTS AND DISCUSSION 97 SUMMARY AND CONCLUSIONS 109 SP-302-08 113 Properties of a New Type of Polycarboxylate Admixture for Concrete Using High Volume Blast Furnace Slag Cement by Shinji Tamaki, Kazuhide Saito, Kazuhisa Okada, Daiki Atarashi, and Etsuo Sakai INTRODUCTION 113 RESEARCH SIGNIFICANCE 114 EXPERIMENTAL PROCEDURE 114 EXPERIMENTAL RESULTS AND DISCUSSION 116 CONCLUSIONS 123 SP-302-09 125 Synthesis and Properties of High Solid-Content Polycarboxylate Superplasticizer by Yongwei Wang, Liya Wang, Yongsheng Liu, and Zepeng Chu INTRODUCTION 125 RESEARCH SIGNIFICANCE 126 EXPERIMENTAL INVESTIGATION 126 EXPERIMENTAL RESULTS AND DISCUSSION 128 CONCLUSIONS 131 SP-302-10 133 Mastering Flow Loss in Superplasticized Cementitious Systems by S. Mantellato, Q. Mehmeti, L. Ceni, M. Palacios, and R.J. Flatt INTRODUCTION 133 RESEARCH SIGNIFICANCE 134

EXPERIMENTAL PROCEDURE RESULTS DISCUSSION CONCLUSIONS

134 136 140 142

SP-302-11 145 Putting Concrete to Sleep and Waking It Up with Chemical Admixtures by L. Reiter, M. Palacios, T. Wangler, and R.J. Flatt INTRODUCTION 145 RESEARCH SIGNIFICANCE 146 EXPERIMENTAL INVESTIGATION 146 EXPERIMENTAL RESULTS 147 DISCUSSION 151 CONCLUSIONS 152 SP-302-12 155 A Simplified Preparation Method for PCEs Involving Macroradicals by Lei Lei and Johann Plank INTRODUCTION 155 RESEARCH SIGNIFICANCE 158 EXPERIMENTAL INVESTIGATION 158 EXPERIMENTAL RESULTS AND DISCUSSION 162 CONCLUSIONS 165 SP-302-13 169 Impact of Slags Contained in Blended Cement on Dispersing Effectiveness of PCEs by Ahmad Habbaba and Johann Plank INTRODUCTION 169 RESEARCH SIGNIFICANCE 170 EXPERIMENTAL INVESTIGATION 171 EXPERIMENTAL RESULTS AND DISCUSSION 173 CONCLUSIONS 178 SP-302-14 183 Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer by Xiao Liu, Ziming Wang, Jie Zhu, Ming Zhao, Wei Liu, and Dongjie Yin INTRODUCTION 183

RESEARCH SIGNIFICANCE 184 EXPERIMENTAL INVESTIGATION 184 EXPERIMENTAL RESULTS AND DISCUSSION 187 CONCLUSIONS 195 SP-302-15 199 Influence of Diester Content in Macromonomers on Performance of MPEG-Based PCEs by Johannes Paas, Maike W. Müller, and Johann Plank INTRODUCTION 199 RESEARCH SIGNIFICANCE 202 EXPERIMENTAL INVESTIGATION 202 EXPERIMENTAL RESULTS AND DISCUSSION 203 CONCLUSIONS 209 SP-302-16 211 Evidences about the Interactions between Grinding Aids and Cement Particles Surface by Valerio Antonio Paternò, Sara Ottoboni, Marco Goisis, and Paolo Gronchi INTRODUCTION 211 RESEARCH SIGNIFANCE 212 MATERIALS AND METHODS 212 DISCUSSION 217 FURTHER RESEARCH 223 CONCLUSION 223 SP-302-17 227 The Influence of C3A on the Dissolution Kinetics of Alite/ Gypsum Mixtures in the Presence of PCEs by Giorgio Ferrari, Vincenzo Russo, Massimo Dragoni, Gilberto Artioli, Maria Chiara Dalconi, Michele Secco, Leonardo Tamborrino, and Luca Valentini INTRODUCTION 227 RESEARCH SIGNIFICANCE 228 MATERIALS AND METHODS 229 RESULTS AND DISCUSSION 231 CONCLUSIONS 238

SP-302-18 243 Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers by Jinzhi Liu, Jiaping Liu, Yong Yang, Dongliang Zhou, and Qianping Ran INTRODUCTION 243 EXPERIMENTAL INVESTIGATION 244 RESULTS AND DISCUSSION 246 CONCLUSIONS 250 SP-302-19 253 Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption by Dirk Lowke and Christoph Gehlen INTRODUCTION 253 ELECTRIC DOUBLE LAYER AND ZETA POTENTIAL 254 MATERIALS AND METHODS 256 RESULTS AND DISCUSSION 257 CONCLUSIONS 263 SP-302-20 265 Optimization of the Structural Parameters and Properties of PCE Based on the Length of Grafted Side Chain by Zi-Ming Wang, Zi-Chen Lu, and Xiao Liu INTRODUCTION 265 RESEARCH SIGNIFICANCE 266 EXPERIMENTAL INVESTIGATION 266 RESULTS AND DISCUSSION 267 CONCLUSIONS 275 SP-302-21 279 Blended Antifreezing Admixture with Extreme Freezing-Point by Anatoly I. Vovk INTRODUCTION 279 RESEARCH SIGNIFICANCE 280 EXPERIMENTAL INVESTIGATION 281 EXPERIMENTAL RESULTS AND DISCUSSION 281 CONCLUSIONS 286

SP-302-22 289 Effect of Expansive Agents and Shrinkage Reducing Admixtures on the Performance of Fiber-Reinforced Mortars by M. Collepardi, V Corinaldesi, S. Monosi, and A. Nardinocchi INTRODUCTION 289 MATERIALS AND METHODS 290 RESULTS 291 CONCLUSIONS 295 SP-302-23 299 Interactions between Cements with Calcined Clay and Superplasticizers by Jens Herrmann and Jörg Rickert INTRODUCTION 299 RESEARCH SIGNIFICANCE 300 EXPERIMENTAL PROCEDURE 300 EXPERIMENTAL RESULTS AND DISCUSSION 305 SUMMARY AND CONCLUSIONS 310 SP-302-24 315 Modification of Fresh State Properties of Portland CementBased Mortars by Guar Gum Derivatives by Alexandre Govin, Marie-Claude Bartholin, and Philippe Grosseau INTRODUCTION 315 RESEARCH SIGNIFICANCE 316 MATERIALS AND METHODS 316 EXPERIMENTAL RESULTS 320 DISCUSSION 324 CONCLUSIONS 328 SP-302-25 333 Interaction of Polycarboxylatebased Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar with Claybearing Aggregates by Weishan Wang, Zuiliang Deng, Zhongjun Feng, Lefeng Fu, and Baicun Zheng INTRODUCTION 333 MATERIALS AND METODS 334

RESULTS AND DISCUSSIONS CONCLUSION

336 344

SP-302-26 349 Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases by Camille Nalet and André Nonat INTRODUCTION 349 RESEARCH SIGNIFICANCE 350 EXPERIMENTAL PROCEDURE 350 RESULTS AND DISCUSSIONS 351 FURTHER RESEARCH 357 CONCLUSIONS 357 SP-302-27 359 A New Accelerator Approach for Improved Strength Development by Franz Wombacher, Christian Bürge, Emmanuel Gallucci, Patrick Juilland, and Gilbert Mäder INTRODUCTION 359 RESEARCH SIGNIFICANCE 360 EXPERIMENTAL PROCEDURE 360 EXPERIMENTAL RESULTS AND DISCUSSION 362 CONCLUSIONS 367 SP-302-28 371 Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography by Christof Schroefl and Viktor Mechtcherine INTRODUCTION 371 RESEARCH SIGNIFICANCE 372 EXPERIMENTAL INVESTIGATION 372 EXPERIMENTAL RESULTS AND DISCUSSION 376 SUMMARY AND CONCLUSIONS 381

SP-302-29 387 Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with PolycarboxylateBased Superplasticizers by Serina Ng and Harald Justnes INTRODUCTION 387 RESEARCH SIGNIFICANCE 388 EXPERIMENTS AND METHODS 388 RESULTS AND DISCUSSION 390 CONCLUSIONS 398 SP-302-30 401 A Study on the Cement Compatibility of PCE Superplasticizers by A. Lange and J. Plank INTRODUCTION 401 RESEARCH SIGNIFICANCE 402 EXPERIMENTAL PROCEDURES 403 EXPERIMENTAL RESULTS AND DISCUSSION 407 CONCLUSIONS 412 FURTHER RESEARCH 413 SP-302-31 415 Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination by Hessam AzariJafari, Mohammad Shekarchi, Javad Berenjian, and Babak Ahmadi INTRODUCTION 415 RESEARCH SIGNIFICANCE 416 EXPERIMENTAL INVESTIGATION 416 EXPERIMENTAL RESULTS AND DISCUSSION 418 CONCLUSIONS 421 SP-302-32 425 Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF by Kazuki Matsuzawa, Daiki Atarashi, Masahiro Miyauchi, and Etsuo Sakai INTRODUCTION 425 RESEARCH SIGNIFICANCE 427 EXPERIMENTAL INVESTIGATION 427

EXPERIMENTAL DISCUSSION CONCLUSIONS

RESULTS

AND 429 432

SP-302-33 437 Cement Recycling System Using Sodium Gluconate by Daiki Atarashi, Yutaka Aikawa, Yuya Yoda, Masahiro Miyauchi, and Etsuo Sakai INTRODUCTION 437 RESEARCH SIGNIFICANCE 438 EXPERIMENTAL INVESTIGATION 439 EXPERIMENTAL RESULTS AND DISCUSSION 440 CONCLUSIONS 445 SP-302-34 449 The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete by Lucia Ferrari and Pascal Boustingorry INTRODUCTION 449 RESEARCH SIGNIFICANCE 450 EXPERIMENTAL INVESTIGATION AT THE GROUT SCALE 450 ANALYTICAL INVESTIGATION 452 COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS 453 EXPERIMENTAL RESULTS AND DISCUSSION 455 FURTHER RESEARCH 460 CONCLUSIONS 461 SP-302-35 463 Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency by Rachid Ait-Akbour, Christine Taviot-Guého, Fabrice Leroux, Pascal Boustingorry, and Frédéric Leising INTRODUCTION 463 RESEARCH SIGNIFICANCE 464 EXPERIMENTAL PROCEDURES 464

EXPERIMENTAL DISCUSSION CONCLUSIONS

RESULTS

AND 466 474

SP-302-36 477 Plasticizing Geopolymer-Type Auspensions: A Challenge by L. Nicoleau, M. Pulkin, and T. Mitkina INTRODUCTION 477 RESEARCH SIGNIFICANCE 478 EXPERIMENTAL SECTION 478 EXPERIMENTAL RESULTS AND DISCUSSION 479 CONCLUSIONS 487

SP-302-37 491 Admixture Concepts for the SubSaharan African Environment with Indigenous Raw Materials by Wolfram Schmidt, Nsesheye S. Msinjili, Herbert C. Uzoegbo, and John K. Makunza INTRODUCTION 491 RESEARCH SIGNIFICANCE 494 EXPERIMENTAL PROCEDURE 494 OBSERVATIONS 497 COMPARISON OF THE CONCEPT WITH DATA FROM PRACTICE 503 FURTHER RESEARCH 503 SUMMARY AND CONCLUSIONS 503 Index 507

SP-302-01

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes by Joseph J. Assaad and Salim E. Asseily The impact of clinker grinding aids (GAs) based on amine, glycol, or phenol on static yield stress (τ0) of cement pastes is not well understood. Results obtained from this project have shown that GA molecules remain active after the grinding process and provide variations in cement properties, whether in the fresh or hardened states. Flowability improved and τ0 decreased when the cement is ground using increased GA concentrations. This was attributed to the adsorption of these molecules onto the cement grains and saturation of surface charges, thus creating repulsive forces between neighboring particles. The decrease in τ0 was particularly pronounced when phenol-based GA was used, given the presence of polycarboxylate polymers that help dispersing cement particles upon mixing with water. Keywords: amine; cement; clinker; glycol; grinding aids; phenol; static yield stress. INTRODUCTION Grinding aids (GAs) are incorporated during comminution of clinker to reduce electrostatic forces and minimize agglomeration of cement grains.1-3 Because of their highly organic polar nature, such additives are preferentially adsorbed on surfaces formed by the fracture of electrovalent bonds (Ca-O and Si-O). For a given cement fineness, this helps in improving mill productivity and/or reducing grinding energy consumption. Typical dosage rates vary from 0.01% to 0.15% of the manufactured cement mass. After the grinding process, GAs may not preserve their original molecular structures; however, they do remain adsorbed onto the cement particles to provide variations of cement properties whether in the fresh or hardened states. The setting and hardening properties of cement containing GAs are well documented in literature. For instance, the hydration of C3A was accelerated in the presence of triethanolamine (TEA) due to the rapid formation of hexagonal aluminate hydrate and its transformation to a cubic form.4 Heren and Olmez found that the addition of increased ethanolamine concentrations alters cement hydration and leads to retardation in setting times in the order of TEA > diethanolamine (DEA) > monoethanolamine (MEA).5 Triisopropanolamine (TIPA), which is an amino-alcohol and belongs to the group of alkanolamines, was found to change hydration reactions and particularly increase cement strengths. Perez et al. reported that TIPA remains in the interstitial paste solution (not adsorbed to the cement surface, as the TEA) and form iron complexes 11

12 SP-302-01

to accelerate hydration of C3S and C4AF.6 Ichikawa et al. presented evidence that TIPA promotes hydration of limestone and densifies the interfacial transition zone (ITZ) between hydrated cement paste and sand or aggregate particles.7 Limited data exits in literature pertaining to the effect of GAs added during clinker grinding on rheology of cement-based materials. Aiad et al. found that viscosity of cement pastes is dependent on the type and dosage rate of ethanolamine used, whereby a decrease in viscosity was noticed following the sequence of TEA > poly-TEA > MEA.8 This was related to the number of O-H groups in the ethanolamine molecules that are adsorbed on the surface of cement grains, causing different repulsive forces and leading to variations in fluidity levels. However, it is important to note that the tests carried out by Aiad et al.8 cannot be conclusive as the ethanolamines were post-added to the cement (i.e., not added during the grinding process) at concentrations varying from 0.1% to 2% of cement weight (i.e., substantially higher than in real situations). Anna et al. compared the Z-potential of clinker containing TEA with others ground with polycarboxylate (PC) or poly-naphthalene sulphonate (PNS) concrete superplasticisers.9 They found that the TEA fluidifying mechanism for the dry cement system lies between the steric hindrance associated with PC polymers and electrostatic interaction of PNS with the positive charges of cement grains.9 While characterizing GAs and their impact on cement performance, Katsioti et al. noted an improvement in workability of cement pastes containing TIPA.10 This was related to the breaking down of cement agglomerates and balance modification between inter-particle forces. The objective of this paper is to assess the effect of GAs on variations in flow and static yield stress of cement pastes prepared with different water-to-cement ratios (w/c). Grinding tests were performed by adjusting the specific energy consumption (Ec) in order to maintain similar Blaine fineness. Amine, glycol, and phenol-based GAs were used at various concentrations. Relevant parameters including flowability, yield stress, water demand, setting time, and compressive strength were evaluated. RESEARCH SIGNIFICANCE Grinding aids are increasingly used during comminution of clinker to prevent cement particle attraction and re-agglomeration, thus resulting in clinker and energy savings that can both lead to reduced carbon dioxide (CO2) emissions. Results presented in this paper aim at assessing the effect of such additions on flow and yield stress of cement pastes. Such data can be of particular interest to cement manufacturers and concrete technologists as well as standardization committees dealing with specifications for GAs. EXPERIMENTAL INVESTIGATION Materials Industrial clinker used for the production of ASTM C150 Type I cement, ground granulated blast furnace slag meeting the requirements of ASTM C989 Grade 80, and gypsum materials were employed in this study. The C3S, C2S, C3A, and C4AF of clinker were 54.6%, 17.4%, 9.2%, and 13.7%, respectively. The slag activity index with cement at 28 days is 86.4%. The relative hardness of the clinker, slag, and gypsum determined according to the Mohs hardness scale were around 5.5, 6, and 2, respectively.

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  13

Figure 1–Photo of the grinding mill used for testing. Two commercially available GAs and a third one specially formulated for this study were tested. First, the amine-based GA is commonly used as a grinding aid and strength enhancer in the cement industry. It had 68% active chemicals when determined by the Karl Fischer method, and specific gravity and pH values of 1.09 and 7.2, respectively. The second GA is glycol-based composed by diethylene glycol (DEG) and monoethylene glycol (MEG) chemicals. It is commonly referred to as a grinding aid and pack-set inhibitor in the cement industry. Its active chemicals, pH, and specific gravity were 72%, 7.8, and 1.107, respectively. Finally, a specially formulated phenol-derivative GA with 70% active chemicals, pH of 5.8, and specific gravity of 1.09 was used. This latter GA was blended with 17.5% oleic fatty acids and 30% polycarboxylate ester molecules that are usually used in concrete admixtures to improve cement dispersion and workability during mixing. Production of cement used for testing—A 50-L (13.25-gal) grinding mill connected to an electric counter for monitoring Ec was used2,3 (Fig. 1). The mill’s drum diameter, width, and rotational speed were 400 mm (15.7 in.), 400 mm (15.7 in.), and 50 rpm, respectively. It contained a total of 80 kg (176 lb) steel balls among which 36 kg (79.2 lb) have 20-mm (0.7 in.) diameter and 44 kg (96.8 lb) have 30-mm (1.18 in.) diameter. Prior to grinding, the clinker, gypsum, and slag materials were crushed and sieved so that all particles were smaller than 10 mm (0.4 in.).

14 SP-302-01

All grinding tests were conducted using 7 kg (15.4 lb) of a mix composed of 90% clinker, 5% gypsum, and 5% slag. First, a mix ground without GA at 42 kWh/ton was tested and considered in this project as being the control cement; its Blaine fineness was 3460 cm2/g. Then, GAs were introduced at pre-selected concentrations varying from low to high levels; the Ec was adjusted accordingly in a way to maintain the fineness equivalent to the control cement, i.e. Blaine of 3460 ±100 cm2/g. The high GA levels were determined following previous studies,2,3 thus ensuring that ASTM C465 requirements for water demand, setting time, and compressive strength are satisfactorily fulfilled.11 The method for cement production consisted on grinding the materials for a certain elapsed time, stopping the mill, and sampling around 100 grams to check whether the Blaine was close to the targeted value. If not, additional grinding was performed. At the end of grinding, the temperature of the charge was found to increase from ambient (i.e., 23 ºC (73 ºF)) to around 37 ºC (99 ºF). Testing equipment and procedures Tests on powder cement—Following grinding, the cement fineness was determined using the Blaine apparatus, as per ASTM C204 Test Method, and by mechanical sieving on 106, 90, and 38 µm mesh openings. The R-90 and R-38 values given in this paper refer to the individual percentages retained on the 90-µm and 38-µm sieve, respectively. The residues on the 106-µm sieve were in the range of 0.2% to 3.5%, depending mostly on the Ec used (note that all particles retained on this later sieve were not included in the cement mix used for subsequent testing). Tests on cement pastes—All pastes were batched with a laboratory mixer using water cooled to a temperature of around 23 °C (73 ºF). Water was first introduced in the mixer followed gradually by the ground cement over 2 minutes. After a rest period of 30 s, the mixing was resumed for one additional minute. The ambient temperature and relative humidity during testing were maintained at 23 ±2 °C (73 ±36 ºF) and 55 ±5%, respectively. The water demand required to achieve normal consistency was determined following ASTM C187 Test Method. Using the same cement paste, the Vicat initial and final setting times were then determined as per ASTM C191 Test Method. The effect of GA types and concentrations on flow and rheological properties was evaluated using cement pastes prepared at 0.48 and 0.42 w/c. These w/c were selected in order to produce pastes with different consistency levels ranging from highly flowable to relatively cohesive. The flow was determined as the average diameter of the paste after spreading on a horizontal surface using an ASTM C230 mini-slump cone. A rotational viscometer connected to a data logger was used to evaluate static yield stress (τ0). The vane used consisted of four blades arranged at equal angles around the main shaft; it measured 24 mm (0.95 in.) in height and 12 mm (0.47 in.) in diameter. Right after mixing, the cement paste was poured in a cylindrical bowl and allowed to rest for one min prior to measuring the τ0 value. The testing protocol consisted on subjecting the paste to a very low rotational speed of 0.3 rpm and recording the changes in torque as a function of time. The τ0 was determined in accordance to Nguyen and Boger,12 by considering the maximum torque registered that indicates the initiation of flow. It is important to note that the total elapsed time from the initial contact of cement with water until the flow and τ0 measurements was around 10 minutes.

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  15

Table 1–Properties when cement is ground with amine-based GA Ec, kWh/ton 42 40.4 38.6 GA dosage, % of mass 0 0.03 0.06 Blaine, cm2/g 3460 3430 3495 R-90, % 8.6 8.1 8.2 R-38, % 28.3 25.8 22 Water demand, % 27.25 27.3 27.45 Final set, min 235 240 235 w/c = 0.48 Flow, mm 195 190 200 τ0, Pa 9.2 10 8.8 w/c = 0.42 Flow, mm 150 150 160 τ0, Pa 30.4 28.8 29.6 7-d compression, MPa 35.6 35.9 38.1 28-d compression, MPa 47.2 48.8 49.7 Notes: 1 cm2/g = 6.8 × 10-5 in2/lb; 1 MPa = 145 psi; 1 in. = 25.4 mm

37.1 0.09 3360 9.3 30.1 27.2 255 195 7.6 155 28 38 49.6

35.7 0.11 3405 9.8 28.6 27.5 275 200 7.2 160 28.8 39 52.1

34.8 0.13 3375 9.5 32.8 27.05 290 205 7.6 165 26.4 36.9 50.3

Table 2–Properties when cement is ground with glycol-based GA Ec, kWh/ton 42 40.7 39.2 38.4 GA dosage, % of mass 0 0.03 0.06 0.08 Blaine, cm2/g 3460 3415 3475 3520 R-90, % 8.6 7.8 8.4 9 R-38, % 28.3 25.2 24 29.1 Water demand, % 27.25 27.05 27.2 27.1 Final set, min 235 255 265 260 w/c = 0.48 Flow, mm 195 210 210 225 τ0, Pa 9.2 8.8 8 7.6 w/c = 0.42 Flow, mm 150 145 160 175 τ0, Pa 30.4 28.8 25.6 21.6 7-d compression, MPa 35.6 34.8 37 36.7 28-d compression, MPa 47.2 48 47 49.5 Notes: 1 cm2/g = 6.8 × 10-5 in2/lb; 1 MPa = 145 psi; 1 in. = 25.4 mm

37.5 0.1 3515 8.9 34.3 27.3 300 220 7.6 170 22.4 35.3 45

Tests on mortars—The compressive strength was determined as per ASTM C109 Test Method. The 50-mm (2-in.) cubes were cured in water until testing at 7 and 28 days. TEST RESULTS AND DISCUSSION The various cement properties determined following clinker grinding at fixed Blaine fineness of 3460 ±100 cm2/g using either amine, glycol, or phenol-based GA are summarized in Tables 1, 2, and 3, respectively. It is to be noted that several cement mixtures were ground two to three times to evaluate reproducibility of testing. Acceptable coefficients of variation (COV) were obtained; as these were equal to 3.8%, 5.1%, 4.6%, 7.4%, and 5.7% for the Blaine, R-38, water demand, setting time, and compressive strength, respectively. The COV increased to 8.8% for τ0 responses, given the variations in Blaine fineness that could lead to different restructuring rates during the rest period prior to shearing. Also, confinement and yielding of paste within the rotating vane impeller could lead to some

16 SP-302-01

Table 3–Properties when cement is ground with phenol-based GA Ec, kWh/ton 42 41.1 40.8 39.8 GA dosage, % of mass 0 0.02 0.03 0.06 Blaine, cm2/g 3460 3390 3475 3405 R-90, % 8.6 8 8.2 8 R-38, % 28.3 24.1 23 27.4 Water demand, % 27.25 27.3 27.1 26.75 Final set, min 235 250 230 215 w/c = 0.48 Flow, mm 195 205 220 235 τ0, Pa 9.2 8.4 8.2 6.5 w/c = 0.42 Flow, mm 150 160 165 180 τ0, Pa 30.4 28.9 26.4 21.6 7-d compression, MPa 35.6 37 36.8 37.2 28-d compression, MPa 47.2 48.1 45.6 47.1 Notes: 1 cm2/g = 6.8 × 10-5 in2/lb; 1 MPa = 145 psi; 1 in. = 25.4 mm

39 0.08 3420 9.1 30.2 26.4 205 250 4.6 190 20.7 34.4 43.5

38.5 0.09 3370 9.3 33 26.1 195 260 3.7 210 16.5 33 42.6

discrepancies in rheological measurements.12 The COV is taken as the ratio between standard deviation and mean values, multiplied by 100. Effect of GA on Ec values Regardless of GA type, the addition of increased concentration led to consecutively reduced Ec values. For example, Ec decreased from 42 kWh/ton for the control cement to 37.5 kWh/ton with the use of 0.1% glycol-based GA, corresponding to 10.7% reduction in energy consumption. Such decrease reached 34.8 and 38.5 kWh/ton with the use of 0.13% amine-based GA and 0.09% phenol-based GA, respectively (i.e., 17.1% and 8.3% energy decrease, respectively). This can normally be related to the GA molecules that are adsorbed onto newly fractured surface grains, thus attenuating electrostatic forces and improving cement fineness.2 It is to be noted that the decrease in Ec provided during the grinding process led to an increase in R-38 and R-90 values, particularly at high GA concentrations (Tables 1, 2, and 3). Hence, R-38 increased from 28.3% for the control cement to 32.8% and 34.3% when the amine or glycol-based GA were used, respectively, at a rate of 0.13% or 0.1%, respectively. This indicates that the sieve residues that are functions of the maximum particle size are directly affected by the amount of grinding energy.3 Nevertheless, data provided by the particle size analysis has shown an increased fraction of particles finer than around 5 μm for those later mixtures, which allowed maintaining the Blaine fineness within 3460 ± 100 cm2/g. For a given GA concentration, higher decreases in Ec were achieved with the use of amine-based GA. For example, at a rate of 0.06%, the targeted Blaine was ensured at Ec of 38.6, 39.2, and 39.8 kWh/ton for cement ground with amine, glycol, or phenol-based GA, respectively. Effect of GA on variations in water demand and setting time Water demand—The effect of incorporating various dosages of amine or glycol-based GA did not lead to remarkable variations in water demand, as compared to the 27.25% value

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  17

Figure 2–Variations in setting times for cement ground with various concentrations of amine, glycol, or phenol-based GA. determined on the control mixture (i.e., values given in Tables 1 and 2 remained within 27.25% ± 0.25%). Conversely, water demand necessary to achieve normal consistency considerably decreased with the use of phenol-based GA; this reduction reached 26.1% with the addition of 0.09% dosage rate (Table 3). This can essentially be attributed to the presence of polycarboxylate molecules that improve cement dispersion upon mixing with water. In other words, this indicates that the GA molecules added during grinding preserve, to a certain extent, their original functionality to effect variations in cement properties.3,13 Setting time—The variations in final setting time for cement ground with various concentrations of amine, glycol, or phenol-based GA are plotted in Fig. 2. In the case of amine and glycol-based GA, the setting time slightly increased at low to relatively moderate GA dosages, and then increased sharply at higher dosages. For example, the setting increased from 235 min for the control cement to 240 and 255 min when the amine-based GA was added at a rate of 0.03% and 0.09%, and then reached 290 min at a rate of 0.13%. Such increase was equal to 255, 260, and then 300 min when the glycol-based GA was used at dosages of 0.03%, 0.08%, and 0.1%, respectively. The delay in setting with increased GA concentrations is in complete agreement with other data published in literature. Teoreanu and Guslicov suggested the existence of an optimum GA concentration that can be correlated with the achievement of a continuous absorption monolayer onto the micro-fractured cement particles.14 Beyond such optimum, the adsorbed monolayer may partly blocks the hydration reactions and leads to retardation in setting times,5,14 just like what happens with the use of water reducers and superplasticizers in concrete mixtures. Unlike the variations obtained for cement ground with amine or glycol-based GA, the incorporation of phenol-based GA accelerated the setting times, as compared to the control mix (Fig. 2). Hence, a decrease in final setting from 235 to 205 and 195 min was measured for the control cement containing 0.08% and 0.09% GA, respectively. This can indirectly be related to the reduced water demand added to the cement paste that was used for testing normal consistency. The reduced w/c leads to increased cement hydration rates, thereby

18 SP-302-01

Figure 3–Typical variations in shear stress for 0.48 w/c cement pastes. shortening setting times.4 Additionally, phenol molecules are reported to remain in pore water and form insoluble complexes that react with calcium ions released from cement,15 thus possibly contributing in accelerated hydration reactions and setting times. Effect of GA on variations in τ0 values Typical plots of shear stress as a function of time determined on various cement pastes are shown in Fig. 3. All profiles exhibited an elastic linear region whereby the material resists shearing, until reaching a maximum shear stress indicating breakage of majority of bonds and yielding of structure. The maximum value required to initiate flow is considered as τ0. The variations of τ0 for tested cement pastes containing various GA types and concentrations at w/c of 0.42 and 0.48 are illustrated in Fig. 4 and 5, respectively (variations in flow are also shown). As can be seen, the τ0 appears to decrease when the cement is ground using increased GA concentrations. For example, at w/c of 0.42, a decrease from 30.4 Pa for the control cement to 28 and 26.4 Pa is noted when the amine-based GA was used at a rate of 0.09% and 0.13%, respectively. The corresponding flow improved from 150 to 155 and 165 mm, respectively (5.9 to 6.1 and 6.5 in., respectively) (Fig. 4). Given that Blaine fineness and water demand were almost similar for cement mixtures ground with aminebased GA, this indicates that the GA molecules remained active after the grinding process and resulted in certain variations in rheological properties. The τ0 decrease in presence of GA can be due to the adsorption of these molecules onto cement particles and saturation of surface charges, thus creating repulsive forces between neighboring cement and improving flowability.10,13 It is interesting to note that the decrease in τ0 was more pronounced when the glycol-based GA was used, as compared to the cement ground using amine-based GA. For example, at a dosage rate of 0.06%, τ0 decreased from 29.6 to 25.6 Pa for mixtures prepared at 0.42 w/c using cement ground with amine or glycol-based GA, respectively. When cement comes into contact with water, it is the aluminate phases (C3A and C4AF) that react first to form a gel based on complex sulfoaluminate hydrates. This gel exerts a barrier effect and governs the mass flow between the inner part of the cement grain and pore water, thus controlling the

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  19

Figure 4–Variations in τ0 for 0.42 w/c pastes made using cement ground with various concentrations of amine, glycol, or phenol-based GA (1 MPa = 145 psi; 1 in. = 25.4 mm).

Figure 5–Variations in τ0 for 0.48 w/c pastes made using cement ground with various concentrations of amine, glycol, or phenol-based GA (1 MPa = 145 psi; 1 in. = 25.4 mm). rheological behavior and hydration of the silicate phases.9 Given that TEA and TIPA have been identified to rapidly react with the aluminate phases,4,5 this may increase viscosity of the interstitial phase and formation of colloidal crystals between connected cement grains. This therefore explains the relative increase in τ0 measurements for pastes prepared with cement ground with amine-based GA, as compared to those containing glycol-based GA. The pastes prepared using cement ground with phenol-based GA exhibited the highest improvements in flowability and decrease in τ0. For example, at 0.42 w/c, the flow increased to 210 mm and τ0 decreased to 16.5 Pa with the addition of 0.09% phenol-based GA (Fig. 4). As already explained, this can be related to the polycarboxylate polymers present in this GA that help dispersing cement particles upon mixing with water.13 There-

20 SP-302-01

Figure 6–Relationships between τ0 and flow for all tested cement pastes (1 MPa = 145 psi; 1 in. = 25.4 mm).

Figure 7–Variations in compressive strength for mortars made using cement ground with various concentrations of amine, glycol, or phenol-based GA (1 MPa = 145 psi). fore, given that the tested pastes were prepared at given w/c of 0.42 or 0.48, this explains the improvement in flowability and corresponding decrease in τ0. Acceptable correlations exist between τ0 and flow values of all tested cement pastes, as shown in Fig. 6. The correlation coefficients (R2) were equal to 0.79 and 0.83 for pastes made with w/c of 0.42 and 0.48, respectively. Hence, the higher the flow, the lower τ0 responses. Effect of GA on compressive strength development The variations in compression strength determined after 7 and 28 days for mortars made with cement ground with various GA types and concentrations are plotted in Fig. 7. As can be seen, the increase in strength was much more pronounced for mortars prepared

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  21

using cement ground with amine-based GA, as compared to those containing the glycol or phenol-based GA. For example, the strength increased from 47.2 MPa (6.84 ksi) for the control mortar to 52.1 MPa (7.55 ksi) after 28 days for the mixture containing cement ground with 0.11% amine-based GA. For given Blaine, this can mainly be attributed to the presence of TIPA that strengthens the C-S-H compounds and densifies the interfacial transition zone between the cement paste and sand particles.3,6,7 At high GA concentrations, the hydration reactions can be altered by the adsorbed GA molecules,10,14 thus leading to decreased strength (Fig. 7). Such decrease was particularly pronounced in the case of mortars prepared with cement ground with glycol and phenolbased GA. Hence, at 0.1% rate of glycol-based GA, the compressive strength dropped to 45 MPa (6.52 ksi) after 28 days. A value of 42.6 MPa (6.17 ksi) was registered for mortar prepared using cement ground with phenol-based GA at a rate of 0.09%. In literature, the decline in strength for cement containing fatty acids was related to oxidization with dissolved oxygen of mixing water during the curing period, thus causing microscopic cracks in the mortar skeleton.16

•

•

•

•

• •

CONCLUSIONS Based on the above results, the following conclusions can be warranted: The incorporation of amine or glycol-based GA did not lead to remarkable variations in water demand, as compared to control mix. Conversely, water demand necessary to achieve normal consistency considerably decreased with the use of phenol-based GA, due to the presence of polycarboxylate molecules that improve cement dispersion. The setting time slightly increased at low to relatively moderate dosages of amine and glycol-based GA, and then increased sharply at higher dosages. Such delay in setting can be correlated to the GA adsorption onto the micro-fractured cement particles, thus partly blocking the hydration reactions and leading to retardation in setting times. Conversely, the use of phenol-based GA accelerated setting times, given the reduced w/c and creation of insoluble complexes that react with calcium ions released from the cement. The flowability improved and τ0 decreased when the cement is ground using increased concentrations of amine-based GA. For a given Blaine fineness and water demand, this was attributed to the adsorption of these molecules onto the cement particles and saturation of their surface charges, thus creating repulsive forces between neighboring and improving flowability. The decrease in τ0 was more pronounced when the glycol-based GA was used, as compared to the cement ground using amine-based GA. This was related to the presence of TEA and TIPA in this later GA that can rapidly react with the cement aluminate phases, thereby increasing viscosity of the solution and formation of colloidal crystals between connected cement grains. The pastes prepared using cement ground with phenol-based GA exhibited the highest decreases in τ0, given the presence of polycarboxylate polymers that help dispersing cement particles upon mixing with water and improving flowability. The increase in strength was particularly pronounced for mortars prepared using cement ground with amine-based GA, as compared to those containing glycol or phenol-based GA. This was attributed to the presence of TIPA that strengthens the

22 SP-302-01

C-S-H compounds and densifies the interfacial transition zone between the cement paste and sand particles. AUTHOR BIOS Joseph Assaad is Professor of civil engineering and R&D Manager at Holderchem Building Chemicals, Lebanon. He received his Ph.D in 2004 from Sherbrooke University, Québec, Canada. His research interests include grinding aids and strength enhancers for cement, rheology, formwork pressure, specialty concrete, injection grouts, repair systems, durability, and use of geotechnical equipment for testing cementitious-based materials modified with chemical and mineral additives. Salim Asseily is Managing Director at Holderchem Building Chemicals, Lebanon. He has been actively involved in the development of cement additives, concrete admixtures, ready-to-use mortars, and specialty chemicals. He received his BS in chemical engineering, MS in operation research, and MBA from Columbia University, New York, U.S.A. REFERENCES 1. Engelsen, C. J., “Quality improvers in cement making – State of the art,” SINTEF Building and Infrastructure, COIN Project Report 2-2008; ISBN 978–82–536–1071–9, pp. 1-24. 2. Assaad, J.; Asseily, S.; and Harb, J., “Effect of specific energy consumption on cement fineness incorporating amine and glycol-based grinding aids,” Materials and Structures, V. 42, No. 8, 2009, pp. 1077-1087. doi: 10.1617/s11527-008-9444-0 3. Assaad, J.; Asseily, S.; and Harb, J., “Effect of Grinding Aids on the Clinker Factor and Energy Consumption of Portland Cement,” Advances in Cement Research, V. 22, No. 1, 2010, pp. 29-36. doi: 10.1680/adcr.2008.22.1.29 4. Ramachandran, V. S., “Hydration of cement – Role of triethanolamine,” Cement and Concrete Research, V. 6, No. 5, 1976, pp. 623-631. doi: 10.1016/0008-8846(76)90026-0 5. Heren, Z., and Olmez, H., “The influence of ethanolamines on the hydration and mechanical properties of cement,” Cement and Concrete Research, V. 26, No. 5, 1996, pp. 701-705. doi: 10.1016/S0008-8846(96)85007-1 6. Perez, J.P., Nonat, A., Pourchet, S., Garrault, M., and Canevet, C., “Why TIPA leads to an increase in the mechanical properties of mortars whereas TEA does not,” ACI Materials Journal, SP217-38, Vol. 217, 2003, pp. 583-594. 7. Ichikawa, M.; Kanaya, M.; and Sano, S., “Effect of triisopropanolamine on hydration and strength development of cements with different character,” Proceedings of the 10th International Congress on the Chemistry of Cement, Ed. Amarkai A.B. and Goteborg A.B., Gothenburg, Sweden, 1997, 10 p. 8. Aiad, I.; Mohammed, A. A.; and Abo-El-Enein, S. A., “Rheological properties of cement pastes admixed with some alkanolamines,” Cement and Concrete Research, V. 33, No. 1, 2003, pp. 9-13. doi: 10.1016/S0008-8846(02)00911-0 9. Anna, B.; Tiziano, C.; Mariagrazia, G.; and Matteo, M., “Grinding aids: A study on their mechanism of action,” http://www.mapei.com/dam/Pdf/ConferencesGrinding.pdf, pp. 1-5.

Effect of Clinker Grinding Aids on Static Yield Stress of Cement Pastes  23

10. Katsioti, M.; Tsakiridis, P. E.; Giannatos, P.; Tsibouki, Z.; and Marinos, J., “Characterization of various cement grinding aids and their impact on grindability and cement performance,” Construction & Building Materials, V. 23, No. 5, 2009, pp. 1954-1959. doi: 10.1016/j.conbuildmat.2008.09.003 11. ASTM C465, “Standard specification for processing additions for use in the manufacture of hydraulic cements‬,” ASTM Int; 2010. Document number ASTM C465-10. 12. Nguyen, D. Q., and Boger, D. V., “Direct yield stress measurement with the vane method,” Journal of Rheology, V. 29, No. 3, 1985, pp. 335-347. doi: 10.1122/1.549794 13. Assaad, J., and Asseily, S., “Use of water reducers to improve grindability and performance of portland cement clinker,” ACI Materials Journal, V. 108, No. 6, 2001, pp. 619-627. 14. Teoreanu, I., and Guslicov, G., “Mechanisms and effects of additives from the dihydroxy-compound class on cement grinding,” Cement and Concrete Research, V. 29, No. 1, 1999, pp. 9-15. doi: 10.1016/S0008-8846(98)00180-X 15. Ervanne, H., and Hakanen, M., “Analysis of cement superplasticizers and grinding aids – A literature survey,” Working report 2007-15, Posiva, Finland, 2007, 85 p. 16. Albayrak, A. T.; Yasar, M.; Gurkaynak, M.; and Gurgey, I., “Investigation of the effects of fatty acids on the compressive strength of the concrete and the grindability of the cement,” Cement and Concrete Research, V. 35, No. 2, 2005, pp. 400-404. doi: 10.1016/j. cemconres.2004.07.031

24 SP-302-01

SP-302-02

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques by Julia Pickelmann, Huiqun Li, Robert Baumann, and Johann Plank The microstructure of MPEG-type polycarboxylate (PCE) copolymers, i.e. the distribution of side chains along the main chain was investigated via 13C NMR spectroscopy and the effect on the interaction with cement was determined. For this purpose, two series of polycarboxylate samples (one series synthesized by radical copolymerization, the other one via grafting/esterification) at molar ratios of –COO– to side chain of 2 to 10 were compared. The 13C NMR spectra suggest that the copolymerized PCEs possess a gradient-like distribution of side chains along the main chain while the grafted PCEs exhibit a statistical (random) repartition. Owed to those microstructural differences the grafted PCE copolymers show a tendency to adsorb in lower amount on cement. The reason is that in the copolymerized PCEs, large blocks of –COO– groups are present which exhibit high affinity to the surface of cement and therefore promote adsorption. Keywords: copolymerization; dispersing force; grafting; microstructure; polycarboxylates. INTRODUCTION Superplasticizers possessing PEO side chains which are connected with the backbone via methacrylic ester groups are known as PCEs of the first generation. They are widely used to improve the rheology of mortar and concrete or to reduce the water demand resulting in higher durability of the concrete. Generally, a major advantage of polycarboxylates is their variability in chemical composition and structure. This feature allows tailoring PCE molecules to specific needs in highly diverse applications. For example, PCE polymers exhibiting a long slump retention can be designed.1

25

26 SP-302-02

In the past, PCEs have been studied intensively with respect to the effects of their composition (e.g. side chain length, side chain density, anionic charge, hydrodynamic size of the polymer, etc.) on their dispersing force and the adsorbed amounts.2-6 However, only limited information on the correlation between the dispersing power in cement and the microstructure of PCEs is available. The reason is the complexity of the analysis of the microstructural composition of the polymer, especially the repartition of the side chains along the polymer trunk. Borget et al. were the first to propose an analytical method to elucidate the microstructure of polymethacrylic acid-ω-methoxy polyethylenglycol graft copolymers using 13C NMR spectroscopic measurements. According to them, the grafting process results in a copolymer with a random distribution of the side chains along the main chain.7 Rozzoni and Belotto also used 13C NMR spectroscopy to study the molecular configuration and tacticity of two copolymerized MPEG-type PCEs. They found a correlation between the tacticity of the copolymer and its macroscopic properties such as e.g. a lower critical solution temperature (LCST, cloud point).8 Still, the most interesting point is the correlation between the microstructure of the polymers and their behavior in cement. In a first attempt Pourchet et al. compared two well-defined polymers possessing a high side chain density but different repartition of the pendant groups.9 The first PCE sample exhibited a gradient distribution of the side chains while the second one possessed a random distribution. They found that the different microstructures significantly impacted the adsorption behavior of the polymers and their sulfate resistance. However, a major drawback of this study is that both copolymers were synthesized via RAFT copolymerization which is not performed in industrial manufacturing of PCEs. Therefore, at present no information is available which describes the relationship between the microstructure of commonly produced MPEG PCEs and their interaction with cement. Furthermore, no data on PCEs possessing low grafting density have ever been published. To close this gap, two series of MPEG-type PCEs were prepared employing the common industrial manufacturing methods of radical copolymerization and grafting (esterification) of ω-methoxy polyethylene glycol and a polymethacrylic acid (PMAA) backbone. Within each series, the molar ratios between the –COO– and the side chain bearing monomers were varied between 2 and 10 to represent the common PCE products used in actual applications. Their individual microstructures were determined by 13C NMR spectroscopic measurements and then related with their adsorption behavior in cement. RESEARCH SIGNIFICANCE The repartition of –COO– groups present in MPEG type PCEs synthesized via the two routes practiced in the industry (radical copolymerization and grafting/esterification) was determined analytically. It was found that copolymerized PCEs exhibit a gradient distribution of the –COO– groups and therefore generally adsorb in higher amounts on cement than grafted PCEs. The study reveals that different preparation methods for PCEs may result in different microstructures which can impact the interaction of those PCEs with cement. Furthermore, it provides evidence for the existence of different reactivity ratios for methacrylic acid and the macromonomer in radical copolymerization.

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  27 Table 1 – Phase composition of the CEM I 52.5 N sample, as determined by Q-XRD analysis.

C3Sm 52.0

C2Sm 27.6

C3Ac 4.4

C3Ao 3.6

C4AFo 4.3

Phase [wt. %] CaSO4. CaSO4. CaSO4 ½ H2O* 2 H2O* 2.1 0.7 0.4

Calcite 3.3

Quartz 0.8

Arcanite 0.5

LOI 0.01

* determined by thermogravimetry

EXPERIMENTAL PROCEDURES Cement A CEM I 52.5 N (‘Milke classic’, HeidelbergCement, Geseke plant, Germany) was used. Its phase composition as determined by Q-XRD employing Rietveld refinement is shown in Table 1. Furthermore, a mean particle size (d50 value, determined by laser granulometry) of 10.2 µm and a Blaine fineness of 3316 cm2/g (Helium pycnometry) were found for this sample. Synthesis of PCE samples PCE via copolymerization – For preparation, two solutions were prepared. Solution I contained water, the monomers methacrylic acid (MAA, Sigma Aldrich, Steinheim/ Germany) and polyethylene glycol methacrylate ester (PEG-MA 1000, Clariant, Burgkirchen/Germany), and the chain transfer agent mercapto propionic acid (MPA, abcr, Karlsruhe/Germany). Solution II constituted of an aqueous solution (100 mL) of the initiator sodium persulfate (SPS) which was kept at 0°C during the entire synthesis. In preparation, 50 mL of water were placed in a 500 mL five neck round bottom flask equipped with a reflux condenser and stirrer, flushed for 20 min with N2 and then heated to 85°C. Using two peristaltic pumps, solution I was added to the flask over 4 h and solution II over 5 h. After cooling the polymer solution to ambient, its pH value was adjusted to ~ 7 by adding NaOH (30 wt. %). The resulting PCE solutions exhibit a solid content of ~ 30 wt.% and are yellowish and slightly viscous. The individual quantities and molar ratios of all reactants used in the different syntheses are shown in Table 2. PCE via grafting (esterification) – Polymethacrylic acid (PMAA, Dow, Walsrode/ Germany) and ω-methoxy polyethylenglycol (PEG-M 1000, Clariant, Burgkirchen/ Germany) were placed in a 250 mL flask and then homogenized at 95°C. Esterification was initiated by gradually heating to 175°C under vacuum (0.1 mbar, 0.75 Torr) while water was distilled off and collected in a cooling trap. After ~ 6 h of reaction time, the final product was cooled to ambient, diluted with water to a solid content of ~ 30 wt. % and adjusted to pH ~ 7 by adding NaOH (30 wt. %). The resulting polymer solutions are colorless and slightly viscous. The quantities of raw materials and the molar ratios for all polymers prepared are shown in Table 2. Characterization of PCE samples Size exclusion chromatography – Molar masses (Mw, Mn), polydispersity index (PDI), hydrodynamic radius (Rh(z)) and conversion of all PCE samples were determined by size exclusion chromatography (SEC). PCE solutions (concentration 10 g/L) were injected into

28 SP-302-02

Table 2 – Quantities and molar ratios of reactants used in the preparation of copolymerized and grafted PCE samples molar ratio COO– to side chain 2 3 5 6 10

MAA 10.0 g 116 mmol 15.0 g 174 mmol 25.0 g 290 mmol 30.0 g 348 mmol 50.0 g 581 mmol

PCEs via copolymerization Macromonomer MPA water 58.08 g 1.23 g 42.0 g 58 mmol 12 mmol 58.08 g 1.64 g 45.0 g 58 mmol 15 mmol 58.08 g 2.07 g 51.0 g 58 mmol 23 mmol 58.08 g 2.71 g 54.0 g 58 mmol 23 mmol 58.08 g 4.52 g 66.0 g 58 mmol 43 mmol

PCEs via grafting SPS 1.38 g 5.8 mmol 1.22 g 5.2 mmol 2.77 g 11.6 mmol 3.23 g 13.6 mmol 5.07 g 21.3 mmol

PMAA 15.5 g 180 mmol 20.66 g 240 mmol 31.0 g 360 mmol 36.16 g 420 mmol 56.8 g 660 mmol

MPEG 60.0 g 60 mmol 60.0 g 60 mmol 60.0 g 60 mmol 60.0 g 60 mmol 60.0 g 60 mmol

a Waters 2695 Separation Module equipped with three UltrahydrogelTM columns (120, 250, 500) and a UltrahydrogelTM guard column (Waters, Eschborn/Germany) and a subsequent 3 angle static light scattering detector (mini Dawn, Wyatt Technology Corp., Santa Barbara, CA/USA). After separation, the polymer concentrations were monitored with a differential refractive index detector (RI 2414, Waters, Eschborn/Germany). Aqueous 0.1 M NaNO3 solution adjusted to pH = 12 with NaOH was used as an eluent at a flow rate of 1.0 mL/min. The value of dn/dc used to calculate Mw was 0.135 mL/g (value for polyethylene oxide).10 Specific anionic charge amount – The specific anionic charge of the polymers was determined in alkaline solution (pH = 12) utilizing a particle charge detector PCD 03 pH (Mütek Analytic, Herrsching, Germany). In this experiment, solutions of 0.2 g/L of the polymer samples were adjusted to pH = 12 with NaOH and titrated against an aqueous 0.162 g/L solution of polydiallyl dimethyl ammonium chloride (polyDADMAC) until charge neutralization (zero potential) was reached. From the amount of polyDADMAC consumed to reach a zero potential, the amount of negative charge per gram of polymer was calculated. Adsorbed amounts – Adsorption on cement was determined using the depletion method, i.e. the non-adsorbed amount of polymer remaining in solution at equilibrium was measured by analyzing the total organic carbon (TOC) content of the solution. In a typical experiment, 30 g of cement, 9.0 g of DI water and the amount of superplasticizer to be tested were filled into a 50 mL centrifuge tube, shaken in a wobbler (VWR International, Darmstadt, Germany) for 2 minutes at 2,400 rpm and then centrifuged for 10 minutes at 8,500 rpm. The supernatant was filtered through a 0.2 µm syringe filter and diluted with 0.1 M HCl. The TOC content of the solution was determined by combustion at 890°C on a High TOC II instrument (Elementar Analysensysteme, Hanau, Germany). The difference between the carbon content of the reference polymer and the TOC content of the supernatant reflects the adsorbed amount of superplasticizer. Measurements were generally repeated twice and the average was reported as adsorbed amount. 13 C NMR measurements – All 13C NMR spectra were recorded in D2O at 25°C and a polymer concentration of 30 mg/mL using a Bruker AV500cryo spectrometer operating at

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  29 500 MHz and equipped with a cryo magnet. In each measurement, 512 scans and a relaxation delay of 8 s were performed to obtain quantitative results. RESULTS AND DISCUSSION Characterization of synthesized PCE samples Two different synthesis methods (copolymerization and grafting) were selected to prepare PCE samples which were then compared with respect to their individual microstructures. Based on some of our previous experiments where representative samples were pulled during the copolymerization process and analyzed it was expected that the two methods will produce PCEs with different distribution of the side chains along the main chain. According to our earlier observations, MAA and MPEG-MA exhibit different reactivities, with the macromonomer reacting faster than methacrylic acid. This result differs from earlier findings of Smith and Klier.11 They reported that in D2O, MAA and PEG-MA show comparable reactivity ratios while in a 50/50 ethanol-D2O mixture, the macromonomer reacts faster, confirming our observation. Consequently, these discrepancies required clarification. Based on our conception of different reactivity ratios for the monomers in radical copolymerization it was expected that this method will produce PCEs with a gradient-like distribution of the –COO– group whereas from the grafting process, polymers possessing a statistical (random) distribution of the side chains should result. To investigate this point, PCE polymers possessing molar ratios of –COO– to side chain of 2, 3, 5, 6 and 10 were synthesized using both methods, free radical copolymerization and grafting technique. The general chemical structures of the two series of PCEs used in this study are shown in Figure 1. The synthesized polymer samples were characterized by SEC. The chromatograms (not shown here) confirmed high purity and uniformity for all polymers. Molar masses (Mw, Mn), PDI, hydrodynamic radii and conversion of the PCE samples are shown in Table 3. The data show that the grafted PCEs have a slightly more narrow molecular weight distribution and therefore lower PDIs in comparison to the copolymerized PCE samples. All copolymerized PCEs exhibit molecular weights in the range of 13,500 to 30,000 g/ mol whereas the molar masses of the grafted PCEs decrease as a result of the decreasing number of side chains at constant length of the PMAA backbone. For all PCE samples, the anionic charge amounts increase linearly with increasing MAA content. Interestingly, the specific anionic charge of grafted PCE samples consistently is slightly lower than those of the copolymerized samples. The reason behind this difference remained unclear, because the additional charge contribution provided by the chain transfer agent used in the copolymerization process should be almost negligible. Microstructural analysis via 13C NMR spectroscopy To determine the chemical shift of the 13C resonance signals representing the carboxylate and the ester carbon atoms, the spectra of pure PMMA and of a PCE without carboxylate groups holding only side chains designated as 23PC0 where captured first. From Figure 2 it is evident that the carbon atoms present in the carboxylate groups (–COO–) appear as a sharp signal at a chemical shift δ of ~187 ppm while the carbon atoms of the ester group (–COOR) occur as a broad signal in the range from ~ 176-181 ppm. According to literature,

30 SP-302-02

Figure 1 – General structures of the copolymerized (left) and grafted (right) MPEG type PCE samples prepared for this study. Table 3 – Molecular weights, polydispersity index, hydrodynamic radius, anionic charge 226 and conversion of the PCE samples prepared via copolymerization or grafting. Polymer sample 23PC2 G-23PC2 23PC3 G-23PC3 23PC5 G-23PC5 23PC6 G-23PC6 23PC10 G-23PC10 PMAA

Mw [g/ mol] 25,000 29,500 26,900 22,400 24,000 18,000 30,000 16,800 26,700 13,500 6,800

Mn [g/ mol] 12,600 14,900 11,900 11,800 12,200 9,700 11,900 8,800 13,100 7,300 5,700

PDI 2.0 2.0 2.3 1.9 2.0 1.9 2.5 1.9 2.0 1.8 1.2

Rh(z) [nm] 4.3 4.7 4.5 3.9 4.0 3.4 5.8 3.5 4.7 3.0 1.5

anionic charge in NaOH (pH = 12) [µeq/g] 2,200 1,600 2,800 2,200 3,600 3,100 4,400 3,700 11,300 9,700 14,500

conversion [%] 92 95 93 96 92 97 93 96 92 96 —

the carbon signals will shift depending on whether electron accepting or electron donating groups are present in the vicinity of these carbon atoms.8 For example, a high amount of acid groups in the vicinity of a carboxylate carbon atom causes a shift to a lower field (= higher value of chemical shift δ) while a high amount of electron donating ester groups next to a carboxylate group leads to a high field shift of the signal. Using this approach it can be determined whether a PCE molecule holds large blocks of –COO– groups, or whether a rather homogeneous (statistical) distribution of side chains along the main chain exists. As an example the complete 13C NMR spectrum of PCE sample G-23PC5 is presented in Figure 3. There, the section at δ = 176-187 ppm which holds the information about the repartition of –COO– groups in the PCE copolymer is shown enlarged. Thus, in further

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  31

Figure 2 – 13C NMR spectra of the reference polymers PMAA (top) and 23PC0 (bottom), measured in D2O analysis of all PCE samples only the range of chemical shifts between δ = 176-187 ppm signifying the carboxylate/ester carbon atoms was looked at. For analysis of the PCE samples, the carboxylate carbon atom signals were divided into two segments: carbon atoms mainly surrounded by –COO– carbons as represented in area (a) (see Figure 4), and carbon atoms surrounded by –COO– and ester or mainly ester carbon atoms in area (b). According to Figure 4 it becomes obvious that copolymerized PCEs generally contain larger segments of (a) than the grafted PCEs. This signifies that the copolymerized PCEs contain larger blocks of –COO– along the main chain, compared to the gradient-like distribution of comonomers in the copolymerized PCE samples which apparently is owed to the different reactivities of MAA and the macromonomer. Such disparaged incorporation of the monomers leads to blocks of methacrylic acid within the copolymer and therefore to increased intensity of the signals in range (a) (see Figure 4). In contrast to that, for the grafted polymers G-23PC2 to G-23PC5 this signal is either weak in intensity or completely missing which confirms a statistical distribution of –COO– and ester groups along the polymer trunk. However, for the PCE samples G-23PC6 and G-23PC10 possessing high charge density (= low grafting density) the difference between the carbon signals of grafted and copolymerized PCEs becomes smaller. Thus, for the corresponding copolymerized and grafted polymer samples, both the shape of the carboxylate signal and its chemical shift are comparable. This suggests that at high anionic charge density, the microstructures of copolymer-

32 SP-302-02

Figure 3 – Typical 13C NMR spectrum of PCE sample G-23PC5, measured in D2O at pH = 7; insert: magnification of the carbonyl region. ized and grafted PCE samples converge, as is schematically shown in Figure 5. Whereas at low anionic charge density, the copolymerized PCE samples still possess significant blocks of polymethacrylic acid and of the PEG polyester. Such structural disparities do not occur in the grafted PCE samples. In order to understand whether those microstructural differences translate into a different interaction with cement, adsorption isotherms of the PCE samples on cement were determined (see Figure 6). Note that under the experimental conditions chosen here the adsorption behavior of the PCE samples is mainly influenced by the type and reactivity of C3A (i.e. the surface area of the ettringite produced). For all PCE samples, the tendency as expected was observed: at increased charge density, the maximum adsorbed amount of polymer increases. When comparing the copolymerized and the grafted PCE polymers, however, it becomes obvious that the grafted copolymers consistently produce lower saturated adsorbed amounts than the copolymerized ones. Those variations appear to become smaller at increased charge density. For example, sample G-23PC2 adsorbs only half of the amount of 23PC2 while the maximum adsorbed amounts of samples G-23PC10 and 23PC10 are similar. This conversion of copolymerized and grafted copolymers exhibiting high charge density correlates well with the observations made in the 13C NMR spectra and the microstructural analysis (see Figure 4). To confirm the model that the adsorption tendency of PCE is owed to the amount of – COO– groups present, a PCE molecule without carboxylate groups (a homopolymer of the

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  33

Figure 4 – 13C NMR signals of the carboxylate and the ester carbon atoms of copolymerized (left) and grafted PCEs (right), recorded in D2O at pH = 7. Spectra of polymethacrylic acid (PMAA) and of a completely esterified polymer (23PC0) are shown for reference (bottom). macromonomer) was synthesized via radical copolymerization. When mixed into cement paste, no adsorption at all was detected thus confirming the strong influence of the –COO– blocks on the adsorption behavior of PCEs. Generally, a polymer with a gradient distribution of carboxylate groups is more likely to adsorb in a “tail” conformation, as the blocks of –COO– groups provide particularly strong electrostatic attraction to the surface of cement. However, such adsorbed conformation leads to only partial coverage of the surface of cement particles, as is schematically illustrated in Figure 7. Whereas, polymers exhibiting a statistical repartition of carboxylate groups can be expected to adsorb in a “loop” conformation whereby a larger area of the surface is covered (Figure 7). Consequently, there a lower amount of polymer should be required to achieve the same surface coverage as from a copolymer with a gradient-like

34 SP-302-02

Figure 5 - Schematic illustration of the microstructure of copolymerized and grafted PCE samples in dependence of their charge density.

Figure 6 - Adsorption isotherms on cement of PCE samples synthesized via copolymerization or grafting. microstructure. This effect can also explain the lower adsorbed amounts for the grafted PCEs. However, at increasing charge density the gradient structure converges into a more random one which is similar to that of the grafted PCE (Figure 5). Accordingly, both types of PCE (copolymerized and grafted) now adsorb in a “loop” conformation which results in quite comparable maximum adsorbed amounts. CONCLUSION MPEG-type PCE superplasticizers were synthesized via two different methods resulting in polymers possessing different microstructures. 13C NMR spectroscopic measurements

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  35

Figure 7 – Idealized illustration of the adsorbed conformations of PCE samples with gradient (left) and statistical (right) distribution of carboxylate groups along the trunk chain. revealed a gradient-like distribution of the side chains for the copolymerized PCEs and a statistical (more regular) distribution of the side chains for the grafted PCEs. With increasing charge density those structural differences become smaller which is also reflected in the interaction between the polymers and cement. At high side chain density, copolymerized PCEs possessing a gradient structure adsorb in higher amounts on cement due to enhanced accessibility of the charged parts. However, for PCE polymers exhibiting high charge density (= low side chain density), such differences in the maximum adsorbed amounts for the copolymerized and the grafted PCEs disappear. The study demonstrates that a microstructural analysis of individual PCE polymers is possible and that differences in the microstructure can affect the behavior of PCE superplasticizers in cement. AUTHOR BIOS Julia Pickelmann studied chemistry at TU München and currently works on her PhD thesis at the Chair for Construction Chemistry. Her focus is the correlation between the microstructure of PCEs and their dispersing effectiveness in cement. Huiqun Li studied Material Science at Beijing University of Technology and currently is hosted as a guest Ph.D. student at TUM’s Center For Advanced PCE Studies. Her research interests include synthesis and characterization of PCE polymers. Robert Baumann received a PhD degree in chemistry from Regensburg University in 1988 and then joined Dow Europe in Horgen (Switzerland) as development chemist. In

36 SP-302-02

his current role as R&D Fellow for Dow Construction Chemicals he is responsible for the strategic development of additives for the construction industry. Johann Plank is full Professor at the Institute of Inorganic Chemistry of Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemistry there. Research interests include cement chemistry, chemical admixtures, organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well cementing. ACKNOWLEDGEMENTS Julia Pickelmann wishes to thank Dow Europe for sponsoring her Ph.D. and Li Huiqun expresses her gratitude to “TUM Center For Advanced PCE Studies” for supporting her stay as a guest researcher. REFERENCES 1. Ramachandran, V. S., and Malhotra, V. M., Superplasticizers, Concrete Admixtures Handbook, (Second Edition), W. A. Publishing, Park Ridge, NJ, 1996. 2. Yamada, K.; Takahashi, T.; Hanehara, S.; and Matsuhisa, M., “Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer,” Cement and Concrete Research, V. 30, 2000, pp. 197-207. 3. Sakai, E.; Yamada, K.; and Ohta, A., “Molecular Structure and Dispersion-Adsorption Mechanisms of Comb-Type Superplasticizers Used in Japan,” Journal of Advanced Concrete Technology, V. 1, 2003, pp. 16-25. 4. Winnefeld, F.; Becker, S.; Pakusch, J.; and Götz, T., “Effects of the molecular architecture of comb-shaped superplasticizers on their performance in cementitious systems,” Cement and Concrete Composites, V. 29, 2007, pp. 251-262. 5. Plank, J.; Pöllmann, K.; Zouaoui, N.; Andres, P. R.; and Schaefer, C., “Synthesis and performance of methacrylic ester based polycarboxylate superplasticizers possessing hydroxy terminated poly(ethylene glycol) side chains,” Cement and Concrete Research, V. 38, 2008, pp. 1210-1216. 6. Lange, A.; Hirata, T.; and Plank, J., The Role of Non-Adsorbed PCE Molecules In Cement Dispersion: Experimental Evidence For a New Dispersion Mechanism, V. M. Malhotra (Ed.), 10th CANMET/ ACI Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, 2012, SP-288.30, 435-449. 7. Borget, P.; Galmiche, L.; Le Meins, J.-F.; and Lafuma, F., “Microstructural characterisation and behaviour in different salt solutions of sodium polymethacrylate-g-PEO comb copolymers,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 260, 2005, pp. 173-182. 8. Rozzoni, A., and Bellotto, M., Configurational NMR Study of Sodium Polymethacrilate-g PEO Comb Polymers, 14th International Congress on the Chemistry of Cement, Madrid, 2011 9. Pourchet, S.; Liautaud, S.; Rinaldi, D.; and Pochard, I., “Effect of the repartition of the PEG side chains on the adsorption and dispersion behaviors of PCP in presence of sulfate,” Cement and Concrete Research, V. 42, 2012, pp. 431-439.

A 13C NMR Spectroscopic Study on the Reparation of Acid and Ester Groups in MPEG Type PCEs Prepared via Radical Copolymerization and Grafting Techniques  37 10. Kawaguchi, S.; Aikaike, K.; Zhang, Z.-M.; Matsumoto, K.; and Ito, K., “Watersoluble bottlebrushes,” Int. J. Polym., V. 30, 1998, p. 1004 11. Smith, B. L., and Klier, J., “Determination of monomer reactivity ratios for copolymerizations of methacrylic acid with poly (ethylene glycol) monomethacrylate,” Journal of Applied Polymer Science, V. 68, 1998, pp. 1019-1025.

38 SP-302-02

SP-302-03

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents by Wolfram Schmidt, Sarah Peters, and Hans-Carsten Kühne Polysaccharides modify the rheological properties of cement based systems. Depending upon their chemistry, molecular architecture, and adsorption tendency, they have different modes of action. Some polysaccharides like diutan gum have strong effect on the fluid phase; others like starch strongly interact with particles. This paper presents effects of diutan gum and starches in presence of polycarboxylates. Rheometric investigations with varied particle volume fractions and increasing coarse aggregate diameters were conducted. The results show that starches have stronger influence on the rheology at high particle volume fractions than diutan gum. At lower particle volume fractions this trend is inverted. Experiments with aggregates sizes up to 16 mm (0.63 in.) indicate that stabilizing agent influences on the effects of aggregates on yield stress were small; however up to 1.0 mm (0.04 in.), a significant effect on the plastic viscosity could be observed, which levelled off at larger diameters. Keywords: diutan gum; starch; polysaccharides; polycarboxylate ether; volume fraction. INTRODUCTION Today, the fresh properties have become a characteristic item to designate various classes of concrete. For many concrete types, such as self-consolidating concrete (SCC) or ultrahigh performance concrete (UHPC), the rheological properties directly affect their performance at hardened state. The major driving force for innovations in concrete technology over the last three decades has been brought by the invention and steady enhancement of superplasticizers, since they allow uncoupling the workability properties from the water content.1 The main relevant rheological properties for cement based systems are the yield stress (τ0) and the plastic viscosity (ηpl), which describe the minimum shear stress that needs to be exceeded in order to make a system flow and the resistance of w flowable system against a shear deformation, respectively. However, sophisticated rheological properties in complex multi-phase systems like concrete come along with a high risk of failure as soon as the boundary framework changes. Since the function of superplasticizers depends upon the adsorption on the early cement hydration phases AFt and AFm, any influence that accelerates or slows down cement 39

40 SP-302-03

Fig. 1—Possible effects of diutan gum and starch based STA without and in presence of PCE. hydration at a very early point in time after water addition can automatically affect the rheological properties.2-4 Therefore, unexpected effects that affect the cement hydration at early age can yield both, unwanted stagnation or segregation. A stabilizing agent (STA) can help reduce unwanted loss of flowability or segregation. Therefore, their application has gained increasing interest in the scientific research.5-15 For the stabilization of cement based systems, polysaccharides such as cellulose, starch, or bio-gums (e.g. sphingans) are typically used. Depending on their molecular weight, radius of gyration, and ionic charges, they affect the rheology of cement based systems in different ways. The stabilizing mechanisms can be based on the water binding capacity,5,7 the viscosity modification of the pore solution,8,15 depletion forces,9,14 or effects induced by the sheer size of the polymers causing interactions with particles,8,9 which can have radii of gyration as large as 500 nm, or if still agglomerated even larger. For the understanding of the working mechanism of STAs, it is also of highest importance to take into account that superplasticizers can have a significant influence on their efficiency. These influences are shown in Fig. 1. It was found that in systems without superplasticizers, starch had only a tangible effect on the yield stress (τ0) at higher particle volume fractions, while diutan gum increased τ0 already without the presence of particles. Approximately 20% of the starch consists of amylose, a linear glucose chain, and about 80% consists of amylopectin, which is a huge branched molecule formed by glucose chains. The effect of the starch was mainly attributed to the effects of the amylopectin molecules, which interact with small solid particles like spacers but require interacting particles for activating their stabilizing effect. The effect of the diutan gum, which contains anionic groups in the backbone, was attributed to water absorption and thickening of the fluid phase as well as adsorption on cement and hydration phases. However, in the presence

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 41 Table 1—Stabilizing agent modifications used for the investigations and dosages required to achieve hPl of 3.5 mPa•s Diutan gum Abbreviation Dosage to achieve ηpl = 3.5 mPa•s in water [% by mass of water]

Starch with Starch with low degree of high degree of modification modification

Cationic starch

Anionic starch

DGUM

ST-low

ST-high

ST-cat

ST-an

0.04%

0.84%

1.3%

1.04%

0.44%

of superplasticizer, effects of STA adsorption retreated into the background and both STAs showed only negligible effects on τ0 but maintained a strong effect on the plastic viscosity ηpl.9,13 It is hence becoming obvious that for the understanding of STAs, it is necessary to distinguish between systems with and without superplasticizers as well as to take into account the particle volume fraction ΦP and possible interactions between STAs and particles. This study aims at understanding these influences for diutan gum and various starch modifications in limestone filler (LSF) and cement based flowable systems as well as to look at whether the aggregate size and volume fraction may have an effect on the efficiency of STAs. RESEARCH SIGNIFICANCE Today, the use of superplasticizers has become common practice for flowable concrete. However, flowable concrete systems often lack stability, and therefore may need to be amended by stabilizing admixtures, which are typically based on polysaccharides. Their mode of operation is not very well understood today. Being able to manipulate the rheology has become of utmost importance for bringing innovative concrete into practice. The present study investigates interactions of polysaccharides with superplasticizers, cement hydration, particle volume fraction, and the sand and aggregates. The goal is better understanding of the mode of operation of these materials. EXPERIMENTAL INVESTIGATION Materials In order to understand effects of different STA modifications, investigations were conducted on paste, mortar, and concrete matrices. The observations were focused on diutan gum and a variety of starches. Two starches were modified at different degrees with hydroxypropyl groups, another starch was modified with cationic groups, and another with anionic groups. The STAs that were evaluated are listed in Table 1. Their dosages were adjusted in a way that a plastic viscosity (ηpl) of 3.5 mPa•s in water was achieved. These required dosages to achieve this property are listed for each STA in Table 1. Paste investigations were conducted at varied particle volume fractions ΦP with LSF and with cement, respectively. Their specific gravities can be found in Table 2. For the LSF pastes, ΦP was varied between 0%, 33%, and 50%. For the cement pastes, ΦP was varied

42 SP-302-03

Table 2—Specific gravities of mixture constituents and mixture composition for SCC

Spec. Gravity [-] Net weight [kg/m3] Volume fraction [%]

Water

Cement

Lime stone filler

3.13

2.74

1.00

2.60

2.60

2.60

2.60

2.60

2.60

354

130

177

331

192

192

126

126

702

11.3

4.8

17.7

12.7

7.4

7.4

4.9

4.9

27.0

0.1-0.5 mm

Sand 0.5-1.0 1.0-2.0 mm mm

2.0-4.0 mm

Aggregate 4.0-8.0 8.0-16 mm mm

Fig. 2—Influence of the particle volume fraction on τ0 and ηpl of LSF and cement pastes. between 0%, 25%, and 40%. The reason is the stronger influence of ΦP on the increase of both yield stress (τ0) and ηpl in cement based systems, as can be found in 
Fig. 2. The influence of different dosages of polycarboxylate ether superplasticizer (PCE) was also investigated in LSF and cement pastes. These investigations were conducted at ΦP of 50% and 40% for LSF and cement pastes, respectively. The dosages of PCE were varied between 0% and 1% by mass of solid PCE related to the mass of solid particles. Mortar and concrete investigations were conducted on a mixture composition for SCC, which is shown in 
Table 2. The aim was to identify influences of ΦP and the maximum particle diameter. Therefore, rheometric investigations were conducted on paste, mortar and concrete with varied solid volume contents and maximum grain sizes according to Table 2. Pure systems without STA were compared with systems incorporating DGUM and ST-low. The STA dosages were as in Table 1 and the PCE dosage was 0.68% solids by the cement mass. The PCE in use was a low charge density PCE. The respective dosage provided a long flow retention at a slump flow value of 650 mm (25.6 in.) at 30 minutes after water addition.

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 43 Table 3—Measurement regime for rheometric paste investigations. Profile characteristic Time (max.) Shear rate [1/s] Purpose

Upward ramp 0 s - 30 s 0 to 73.5 -

Plateau 30 s - 60 s 73.5 -

Downward ramp 60 s- 240 s 73.5 to 0 Flow curve determination

Mixing of pastes and mortars For all mixes, the STAs were always first dissolved in a blender mixer in the water for the mixture addition. The mixing of the pastes and mortars was conducted in a mortar mixer. After a 30 s dry mixing phase, water including STA was added and mixed for 2 minutes. Then (if required) PCE was added and mixing was continued for another minute. In order to avoid peculiar initial effects induced by the hydration of ettringite and monosulfate, the experiments were conducted approximately at 10 minutes after the water addition. Paste rheometry For the investigations of the rheological properties of pastes, a Couette type viscosimeter (Schleibinger NT) was used in combination with a double gap cell with a network structured grid as shear body. The cell allows the measurement of pastes and mortar up to a grain size of 2 mm. Since shear forces affect the cement hydration and the adsorption of polymers, for cementitious systems it is of utmost importance to keep the measurement time as short as possible. The applied profile is shown in Table 3. It is considered to be a reasonable compromise between precision and compactness. For the conversion of the flow curves, a Bingham approximation was chosen. Though this does not take into account shear thinning effects that were observed to certain extent in some systems, it is in good agreement with the results for all ΦP and admixture dosages, and the comparison between the pastes is facilitated. Mortar rheometry For the rheometric mortar investigations in systems containing sand in the size fractions 0.1 mm to 4.0 mm (0.004 - 0.16 in.) the same rheometer was used as for the paste investigations. However, a mortar cell including a stirrer was used. The cell does not allow the conversion to fundamental units, but it allows determining an ordinate intercept and a slope of the measured torque, which can provide relative information on differences in yield stress and plastic viscosity. Concrete rheometry For concrete with aggregate sizes 4.0 mm to 16.0 mm (0.16 - 0.63 in.), a concrete rheometer (Rheometer-4SCC) was used. Like for the mortar equipment used in these investigations there is no reliable method for the conversion of torque and rotational velocity into shear stresses and shear rates, the results can only provide qualitative information on changes in yield stress and plastic viscosity. Therefore the measurements are referred to as G-Yield for qualitative observations of yield stress changes and H-Viscosity for qualitative changes of the plastic viscosity, respectively.

44 SP-302-03

Fig. 3—Influence of ΦP on τ0 of LSF and cement pastes with different STAs Paste slump flow tests In addition to the rheometric investigations, slump flow tests were conducted with varied dosages of PCE. In order to avoid segregation, they were conducted at a water volume, which just surrounds the particles and fills the voids in between but without any excess water. This water demand was determined according to the so called Puntke test.2,16,17 The slump flow tests were conducted 10, 20, and 30 minutes after the addition of water. Furthermore, paste slump flow tests were conducted at fixed PCE and STA dosages with varied cement volume fractions varying between 69.8% and 63.5%. Below 63.5% the cement pastes were no longer free from segregation at the used dosage of PCE. EXPERIMENTAL RESULTS AND DISCUSSION Paste rheometry at varied particle volume fraction The effect of the different STAs on τ0 depending upon ΦP can be observed for LSF and for cement pastes in Fig. 3. Since the differences in τ0 were strongly affected by the ΦP, the figure shows results normalized to reference mixtures without STA. The figure therefore shows the ratio of τ0 of a paste modified with STA in the dosage given in Table 1 and an identical paste without STA. Apart from the anionic starch ST-an, all STAs increase the τ0 of LSF pastes at ΦP = 33%. 
ST-an shows lower values for τ0 than the reference at 33% and at 50%, indicating that the anionic charges may interact with the LSF and have a plasticizing effect regardless of the ΦP. At 50%, the yield stress increasing effect of the other STAs is reduced significantly for all STAs. Only for the ST-low this reduction is less significant than for all other STAs. ST-cat and ST-high show lower τ0 at that solid volume fraction than the reference. In the cement pastes at low ΦP, all STAs apart from the cationic starch show higher τ0 than the reference, while this effect is reduced at higher particle solid volume fraction. An exception is DGUM. No effect of ΦP on how DGUM increases τ0 can be observed, while

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 45

Fig. 4—Influence of ΦP on ηpl of LSF and cement pastes with different STAs in LSF paste this drop is significant. The attractive forces in a LSF system are significantly lower than in a cement system. Therefore, the lower influence of ΦP on the effect of the DGUM in the cement system may be attributed to higher adsorption compared to the LSF system. The higher adsorption would withdraw polymers from the solution that would otherwise increase the solution’s viscosity. For ST-high, a similar drop of the yield stress effect can be observed in LSF and cement systems; however, the effect of this STA is much stronger in the cement system. For ST-low, which has a strong effect in limestone paste but not in cement paste, this observation is inverted. The reason for this effect could be that the higher ionic strength of the cement pore solution may negatively affect the performance of the slighty modified agent, e.g. by causing interactions within the huge amylopectin molecule or by linking amylose and amylopectin. The higher degree of modification of ST-high may more effectively prevent such interactions. A prominent observation is that the anionic and the cationic starch exactly show inverted behavior in LSF and in cement systems. Both show a plasticizing effect at high ΦP, indicating that they induce electrostatic repulsive forces. However, at low ΦP this plasticizing effect can only be observed for ST-an in limestone and for ST-cat in cement pastes. It is known that LSF in water attracts anionic polymers.2 It is therefore likely that in LSF pastes the adsorption of ST-an incorporates repulsive forces, while this adsorption is reduced in the cement system due to the presence of sulfates. Fig. 4 shows the effect of ΦP on ηpl. For the LSF system, it can be observed that apart from ST-an, all starch based STAs showed reduced effects of the STAs with increasing amount of ΦP. For DGUM at ΦP = 50%, an increasing effect can be observed, which is similar to the effect of the ST-an. Both STAs have anionic charges, so that this effect is likely to be attributed to their adsorption on particles. The strong effect of the ST-an on the viscosity increase relative to the reference system at ΦP = 33% indicates that the huge amylopectin molecule may bridge particles with increasing effect on ηpl. In this context, it is interesting

46 SP-302-03

Fig. 5—Influence of the PCE dosage on τ0 of LSF and cement pastes with different STAs to observe that at ΦP = 50% both anionic STAs show higher effect on ηpl, but as shown in Fig. 3, in terms of τ0 only the starch based agent had a plasticizing effect. It is assumable that this plasticizing effect may be induced by the adsorption of mainly the small amylose molecules, while the viscosity increase is caused rather by the huge amylopectin molecules and thus similar to the effect of the DGUM. In cement based systems, it can be observed that a ΦP of 25% causes reduced effectiveness of the STAs, which further reduces at 40%. At 40%, the plastic viscosities of all systems were lower than the reference. Paste rheometry at varied PCE dosage In order to observe effects of STAs in presence of superplasticizers, LSF and cement pastes were evaluated at ΦP of 50% and 40% for LSF and cement, respectively, at varied PCE dosages. Upon addition of PCE, τ0 was reduced significantly for all mixtures. Fig. 5 shows the ratios of τ0 with STA compared to systems without STA. For the LSF system, it is interesting to see that the two non-ionic STAs both do not distinguish greatly from the reference paste, regardless of the PCE dosage. The anionic admixtures, ST-an and DGUM, both show a horizontal curve up to a PCE dosage of 0.5% above which they generate a higher τ0 than the reference system. It is assumable that competitive adsorption between the STAs and the PCEs takes place causing less adsorption of PCE, which increases τ0. For the cationic starch, it can be observed that above a PCE dosage of 0.1% τ0 is always higher than in the reference system. In this case it is most likely that the cationic starch interacts with the anionic PCEs, which reduces their adsorption, which automatically increases τ0. In the cement system, up to a dosage of 0.5% the observations are similar to those of the limestone system with the only difference that the anionic starch ST-an does not have a different effect on τ0 than the other agents. At 1% PCE dosage, the observations are more complex. What is common to the LSF system is the observation that the influence of the non-ionic starches is small also at 1% of PCE addition. The effect of the cationic starch

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 47

Fig. 6—Influence of the PCE dosage on ηpl of LSF and cement pastes with different STAs is smaller than that of the other charged agents. In case of ST-cat τ0 is lower than that of the reference, which is different to the limestone filler system. Also τ0 of the paste with the anionic starch is lower than the reference. Only for the diutan gum STA a significantly higher τ0 can be observed at 1% PCE. The reason for this can be attributed to a competitive adsorption that reduces the effect of the PCE. However, it is interesting to see that the anionic starch shows an adverse effect. This may be caused by adsorption mainly of the amylose molecules, which may be responsible for a supplementary fluidizing effect as already shown in Fig. 3. The large amylopectin molecule may not be found adsorbed directly on the surfaces. This would also explain why the anionic and the cationic starch show similar behavior. The results for the ratio ηpl,STA/ηpl,Ref is shown in Fig. 6. It can be found that in the LSF system, all STAs maintained ηpl significantly higher than in the reference system. The ST-an, ST-cat and the ST-high became particularly effective at high PCE 0.5% PCE dosage, and this effect was maintained also at 1% PCE. In the cement system, all pastes with STAs had a lower ηpl than the reference. However, with increasing PCE dosage, they became increasingly more efficient in generating higher plastic viscosities than the references. At PCE dosages above 0.5% all STA modified systems apart from the system with ST-low showed higher plastic viscosities than the reference with an increasing trend for higher PCE dosages. The least effect could be observed for ST-low. Only at 1% PCE dosage a slightly viscosity enhancing effect can be observed. It can be assumed that compared to the adsorption of PCE, adsorption of STAs has a negligible effect on ηpl (different from τ0). Therefore, the yield stress enhancement compared to the reference is most likely induced mainly due to the pure presence of large polymers in the pore solution.

48 SP-302-03

Fig. 7—Influence of the presence of STA on the slump flow of cement pastes with different dosages of high charge PCE at ΦP = 0.6. The dosage of starch was selected to achieve a similar slump flow reduction at the PCE dosage that yields the maximum slump flow. Influence of STA on the effectiveness of PCE It was shown that the presence of PCE can have a significant influence on how effective stabilizing admixtures are . In order to better assess this interaction, flow tests on cement pastes were conducted with increasing PCE dosages. Due to the high powder particle fraction, the observed pastes do not show any flow without PCE addition. For each PCE dosage, the water content of the PCE is reduced from the total water content, hence, the flow is only induced by the presence and dosage of PCE. A more detailed description of the experiment can be found elsewhere.2 Without the presence of STA, the maximum flow of about 320 mm (12.6 in.) can be achieved at a PCE dosage of approximately 1% (Fig. 7), beyond which further addition of PCE does no longer contribute to a wider flow diameter. The flow initiation dosage is at approximately 0.2%. In a next step, it was observed how these values are changed by the addition of 0.1% DGUM, and it can be seen that the characteristic dosages for the flow initiation and maximum remain similar to the reference system without PCE. However, the maximum flow is significantly reduced to a value of approximately 250 mm (9.84 in.). In a further step, the dosage of ST-low was determined that causes a similar maximum flow as the system with DGUM. Based on the dosages listed in Table 1, the first attempts were conducted at significantly higher dosages than for the DGUM system. However, it was found that these systems were not flowable at all. Eventually, the dosage that showed performance like DGUM was only 0.06% of the starch. The fact that in water and low ΦP systems, the DGUM was required at significantly lower dosages than the starches for similar flow properties (Table 1). But since this is obviously inverted at very high ΦP (Fig. 7), it is recommended to test the influence of the ΦP. Fig. 8 shows results of slump flow values at 1.0% PCE and in the presence of 0.1% ST-low and DGUM dosage, respectively. This PCE dosage was the dosage that yielded the maximum flow for the reference systems in the tests shown in Fig. 7.

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 49

Fig. 8—Influence of the particle volume fraction of cement pastes on the effectiveness of starch based and diutan gum based STA.

Fig. 9—Influence of the maximum particle size diameter and the particle volume fraction on the effectiveness of stabilizing agents to affect the τ0. It can be clearly identified that indeed the starch addition causes a smaller slump flow than the diutan gum at a very high solid volume fraction of 69.8%. This effect is reduced with decreasing solid volume fraction. At a solid volume fraction of approximately 67%, both STAs have similar effects on the slump flow, while with decreasing ΦP the diutan gum becomes more and more efficient. It is hence obvious that ΦP has a very strong effect on the effectiveness of different STAs in flowable systems.

50 SP-302-03

Fig. 10—Influence of the maximum particle size diameter and the particle volume fraction on the effectiveness of stabilizing agents to affect the ηpl. Influence of cumulative addition of sand and aggregate fractions For the observation of the influence of the addition of aggregates, for the different dimensions of the aggregates different equipment was used. Both rheometers used, cannot derive adequate fundamental properties for yield stress and plastic viscosity and the results appear in different units. Therefore, here again, the ratios of G-yield with STA to the G-yield without (Fig. 9) as well as the ratio of the H-viscosity with STA to the H-viscosity without (Fig. 10) are shown. This way of representing has the benefit that effects of the aggregate size and the volume fraction on the mortar and concrete rheometry are neglected and the direct influence only of the presence of STA can be identified. For yield stress (Fig. 9) it can be observed that the effect of the starch based agent compared to the diutan gum was significantly higher. This observation is in contrast to Fig. 5, where in the presence of PCE the diutan gum showed stronger effects in presence of LSF and cement at a similar ΦP. Nevertheless, the pure pastes and the paste of the SCC from LSF and cement (Table 2) were different mixtures. Increasing aggregate volumes and maximum grain sizes slightly increase the effectivity of both STAs. For both STAs this effect is not very prominent, but clearly identifiable. Regarding the plastic viscosity, which is shown in Fig. 10, it can be observed that both STAs show similar trends. Up to sand size of about 1.0 mm (0.04 in.) it can be observed that the effect of STA on the plastic viscosity is reduced and then levels off still at higher value than the reference. This indicates that the maximum grain size and ΦP contribute significantly to the efficiency of STAs. SUMMARY AND CONCLUSIONS For the evaluation of the effectivity of stabilizing agents, the ionic strength of the pore solution, the presence of superplasticizers, and the ΦP have to be considered. The following conclusions can be drawn:

Effects of Particle Volume Fraction and Size on Polysaccharide Stabilizing Agents 51 • • • • • • • •

The performance of STAs in LSF and cementitious systems can vary greatly due to different ionic strengths. Charges incorporated in STAs can have a strong effect on their performance. For starch based STA supplementary effects have to be taken into account due to the difference between amylose and amylopectin. In the presence of PCE, STA has only little influence on the yield stress but a significant increase of the plastic viscosity can be achieved. As a function of ΦP the effectiveness of starch based STA is increased. Increasing sand and aggregate contents enhance the yield stress effect of STAs but reduce their viscosity effect. A strong effect of STAs on the viscosity can be observed up to a grain size of 1.0 mm (0.04 in.). As a consequence of the complex interactions, the performance of an STA in a liquid or paste system cannot be converted to concrete without further considerations.

AUTHOR BIOS Dr. Dipl.-Ing. Wolfram Schmidt is a researcher at the BAM Federal Institute for Materials Research and Testing in Berlin. He received a Dipl.-Ing. from RWTH Aachen and a PhD from TU Eindhoven. His research focuses on SCC, admixtures, and rheology. He is member of the RILEM committee 228-MPS and the fib task group 8.8. M.Eng. Sarah Peters is a researcher at the BAM Federal Institute for Materials Research and Testing in Berlin. She received a M.Eng from BHT Berlin in structural engineering. Her master thesis dealt with the use of stabilizing agents in different pastes. Her current research focuses on the application of lightweight granules, which are made from recycled masonry rubble. Dr.-Ing. H.-C. Kühne studied mineral processing at the TU Berlin. He received a Dr.-Ing. degree from the Technical University Hamburg-Harburg at the chair of building physics and construction materials. In 2003, he joined the Department for “Safety of Structures” at BAM and became head of the division for “Technology of Construction Materials” in 2011. REFERENCES 1. Schmidt, W.; Sonebi, M.; Brouwers, H. J. H.; Kühne, H.-C.; and Meng, B.Chemistry and Materials Research, V. 5, 2013, pp. 115-120. 2. Schmidt, W., Eindhoven University of Technology, 2014. 3. Schmidt, W.; Brouwers, H. J. H.; Kühne, H.-C.; and Meng, B., “Influences of superplasticizer modification and mixture composition on the performance of self-compacting concrete at varied ambient temperatures,” Cement and Concrete Composites, V. 49, 2014, pp. 111-126. doi: 10.1016/j.cemconcomp.2013.12.004 4. Plank, J., and Hirsch, C., “Impact of zeta potential of early cement hydration phases on superplasticizer adsorption,” Cement and Concrete Research, V. 37, No. 4, 2007, pp. 537-542. doi: 10.1016/j.cemconres.2007.01.007

52 SP-302-03

5. Khayat, K. H., “Viscosity-enhancing admixtures for cement-based materials — An overview,” Cement and Concrete Composites, V. 20, No. 2-3, 1998, pp. 171-188. doi: 10.1016/S0958-9465(98)80006-1 6. Pourchez, J.; Govin, A.; Grosseau, P.; Guyonnet, R.; Guilhot, B.; and Ruot, B., “Alkaline stability of cellulose ethers and impact of their degradation products on cement hydration,” Cement and Concrete Research, V. 36, No. 7, 2006, pp. 1252-1256. doi: 10.1016/j. cemconres.2006.03.028 7. Pourchez, J.; Peschard, A.; Grosseau, P.; Guyonnet, R.; Guilhot, B.; and Vallee, F., “HPMC and HEMC influence on cement hydration,” Cement and Concrete Research, V. 36, No. 2, 2006, pp. 288-294. doi: 10.1016/j.cemconres.2005.08.003 8. Marlière, C.; Mabrouk, E.; Faure, P.; Lamblet, M.; and Coussot, P., Advances in Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013. 9. W. Schmidt, H. J. H. Brouwers, H.-C. Kühne and B. Meng, Applied Rheology, 2013, 23. 10. Phyfferoen, A.; Monty, H.; Skaggs, B.; Sakata, N.; Yanai, S.; and Yoshizaki, M., First North American Conference on the Design and Use of Self-Consolidating Concrete, 2002. 11. Sonebi, M., “Rheological properties of grouts with viscosity modifying agents as diutan gum and welan gum incorporating pulverised fly ash,” Cement and Concrete Research, V. 36, No. 9, 2006, pp. 1609-1618. doi: 10.1016/j.cemconres.2006.05.016 12. Sonebi, M.; Schmidt, W.; and Khatib, J.Chemistry and Materials Research, V. 5, 2013, pp. 106-111. 13. Schmidt, W.; Brouwers, H. J. H.; Kuehne, H.-C.; and Meng, B., Tenth International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, Czech Republic, 2012. 14. Palacios, M.; Flatt, R.; Puertas, F.; and Sanchez-Herencia, A., 10th International Conference on Superplasticizers and Other Admixtures in Concrete, Prague, Czech Republic, 2012. 15. Hot, J., and Roussel, N., Fifth North American Conference on the Design and Use of Self-Consolidating Concrete, Chicago, USA, 2013. 16. Hunger, M., and Brouwers, H. J. H., “Flow analysis of water–powder mixtures: Application to specific surface area and shape factor,” Cement and Concrete Composites, V. 31, No. 1, 2009, pp. 39-59. doi: 10.1016/j.cemconcomp.2008.09.010 17. W. Puntke, beton, 2002, 2002, 242-248.

SP-302-04

New Additive to Enhance the Slump Retention by David Platel, Jean-Marc Suau, Clement Chosson, and Yves Matter For the Ready Mix Concrete mix design, the initial workability depends on the chemical composition of the Polycarboxylate Ether (PCE). Up to now the use of best PCE can achieve 2 hours of slump retention. However, a tendency to segregate is observed when over-dosage is made, due to water reduction capability of PCEs. An approach is to use also a combination of a water-reducing agent and a retarding agent which has the main disadvantage to delay the setting time and consequently the early strength of the concrete. This paper demonstrates the possibility to boost the performance of currently used PCEs. The new slump retention additive that we developed allows a significant increase of the slump retention while maintaining the initial fluidity without impacting the water reduction ability. The homogeneity of the concrete is also controlled by using this additive. On top of that, the combination between this new product and a standard water-reducting PCE is made at commonly used dosage. Keywords: polymer; slump; slump retention; water reduction; fluidity; cement; polycarboxylate ether; segregation. INTRODUCTION Since the development of the Polycarboxylate Ether (PCE), numerous studies1-12 have been carried out which demonstrated that they are essential components of the concrete admixtures when high fluidity and high strength are requested. For the retention of the workability for a long period of time, typically more than 120 minutes, several solutions based on the PCE technology have been developed for the ready-mix concrete industry such as the formulation of the PCE with a retarding agent, or the PCE including a release component13-16 or the new hyperbranched PCE using a cross-linked monomer.17 All these previous technologies have been found to be effective on the control of the slump retention but they also lead to some problems such as low early strength development or poor slump retention capacity when the dosage in PCE is limited as in certain concrete mix designs. In this paper, it is shown that the slump retention of the ready-mix concrete can be controlled by the use of Polycarboxylate Ethers (PCEs) having low water reduction capacity due to a poor dispersing ability. By combining macroscopic and microscopic 53

54 SP-302-04

Table 1–Physical properties and chemical compositions of cement Chemical composition (%)

SiO2

Al2O3

Fe2O3

CaO

MgO

SO3

K2O

Na2O

Na2Oeq

C3S

C2S

C3A

C4AF

Blaine fineness (cm2/g)

C1

18.8

6.0

4.1

62.7

1.1

3.5

0.59

0.45

0.84

63.1

15.0

8.5

13.4

3750

Table 2–Characteristics of the Polycarboxylate Ether

PCE 1 PCE 2 PCE 3

Function High Water Reducing Slump Retention Slump Retention

Brookfield viscosity @ 25°C (mPa.s)

Mw (g/mol)

Ip

rh (nm)

Solid content (%)

100 000

1.5

7

40

400

720 000

4.6

24

40

900

1 200 000

11.5

40

40

2 000

analysis techniques, the relationship between the molecular structure of the PCE and the slump retention control is clarified. RESEARCH SIGNIFICANCE Most of the studies regarding the slump retention properties of the PCE have been developed one unique chemical structure which controls two key parameters of the fresh concrete: the dispersibility and the slump retention. The right balance between these two parameters is very complex. For example a PCE with a high water reduction capability usually has a poor slump retention ability. This phenomenon is directly linked to the PCE dosage added in the concrete mix design. This paper demonstrates the possibility to independently control the slump retention of a fresh concrete by combining two PCEs: a high water-reducing agent and a slump retention agent. EXPERIMENTAL AND ANALYTICAL INVESTIGATIONS Cement The cement used for all tests was an Ordinary Portland Cement (OPC) or CEM I 52.5 N according to the European Standard EN 197-1. The physical properties and chemical analysis of the cement are listed in Table 1. Polymers Three Polycarboxylate Ether (PCE) were produced by free radical polymerization. The PCE 1 is a high water reducing agent and its main characteristics are described in Table 2. The PCE 2 & PCE 3 differ from the PCE 1 by their low water reducing capacity and by their high molecular weight (Mw) as described in Table 2. Gel Permeation Chromatography (GPC) The GPC analyses were performed using a chromatography system equipped with three columns (Ultrahydrogel Waters™) equilibrated at T = 55°C in an aqueous solution (1%

New Additive to Enhance the Slump Retention  55

Table 3–Concrete mixture proportions

Mix 1 Mix 2 Mix 3

W/C 0.69 0.54 0.67

Water reduction (%) 0% 22% 3%

Water 241 (406) 189 (318) 234 (394)

kg/m3 (lb/yd3) Cement Sand 350 (590) 860 (1449) 351 (590) 861 (1449) 352 (590) 862 (1449)

Gravel 990 (1517) 991 (1517) 992 (1517)

of KNO3) with a multi-detection system: a differential refractometer (Waters™) and a dual detector which combines a viscometer and a static light scattering detector at 90° (Malvern™). The molecular weight calibration was performed with only one standard from Polycal™ Malvern™. Samples were injected at a concentration of (2-3) x 10-3 g/ml, after filtration through a 0.2 µm filter (Millipore). Samples have been prepared following two different protocols. Concerning the Mw measured for the PCEs (cf. Table 2), the samples were diluted at a concentration of (2-3) x 10-3 g/ml in an aqueous solution at 1% of KNO3 . This method is refered to as “Reference”. For the second protocol, all the samples have been diluted at a concentration of (2-3) x 10-2 g/ml in different aqueous solutions at 1% of KNO3 which already contain a concentration from 6 to 10% in weight of the following salts: Na2SO4, K2SO4, CaCl2, CaO, NaOH and KOH. Then a sample of each previous preparation has been injected at a concentration of (2-3) x 10-3 g/ml after filtration through a 0.2 µm filter (Millipore) in order to measure the evolution of Mw in different salt conditions over time. Tests on concrete All the concrete formulations were produced according to the European Standard EN 480-1. Natural sand and coarse aggregates (max diameter 31.5 mm) were used in all the cases and the concrete mixture proportions are listed in Table 3. All the concrete mixtures were prepared with the same initial slump values (190 – 230 mm, measured 5 minutes after mixing the ingredients). Then slump values were measured at intervals of 45 minutes. Cylinder specimens of 160 mm in diameter and 320 mm in length were produced and investigated. All specimens have been prepared according to the European Standard EN 12930-2 in 160x320 mm steel cylinder molds. The strength properties were determined after 1, 7 and 28 days of curing at 20°C. Cement pastes For all the cement paste evaluations, the water cement ratio (w/c) was 0.38. It corresponds to a paste with the PCE 1 that has a spread diameter of about 180-200 mm.To ensure that all experiments can be easily compared, all mixtures were prepared using the same amount of materials (around 550g as shown in Table 4). Firstly, the PCE and the water were pre-mixed together then the cement was added within 30s into the solution, then a periodg of 30 s is given to the cement in order to be fully wet and finally hand mixing was carried-out for 2 minutes. Flow tests were performed 15 seconds after mixing on a ceramic tile having a smooth surface. The cylinder used had a diameter of 60mm and a height of 50mm. The measurements of the initial paste flow and those made every 30 minutes during 120 minutes afterwards allow to plot the flow of the different PCEs over time as shown in Figure 4.

56 SP-302-04

Table 4–Cement paste formulations

Paste 1

W/C 0.38

PCE dosage (% dry PCE/cement) 0.16%

g (lb) Water 150 (0,331)

Cement 400 (0,881)

Table 5–Water reduction of the PCE 1 & PCE 2 Concrete Mix Design Polymer PCE dosage (% dry PCE/cement) W/C Water Reduction DENSITY, kg/m3 (lb/yd3) Initial Slump Flow, mm (in.) Initial Air Content, %

Mix 1 None

Mix 2 PCE 1

Mix 2 PCE 2

Mix 3 PCE 2

/

0,21

0,21

0,21

0,69 0%

0,54 22%

0,54 22%

0,67 3%

2287 (3855)

2301 (3878)

2281 (3845)

2296 (3870)

205 (8.1)

190 (7.5)

45 (1.8)

190 (7.5)

1.7

3.3

2.2

2.4

EXPERIMENTAL RESULTS AND DISCUSSION Water reduction capacity The data in Table 5 show the water reduction capacity of the PCE 1 and PCE 2 which have been realized according to the European Standards. The PCE 1 is a high water reducing agent and in these conditions, its water reduction ability is 22%. In the same conditions and at the same dosage, the PCE 2 showed clearly a weaker dispersing ability which is measured by a poor initial slump of 45 mm instead of 190 mm for the PCE 1. The water reduction of the PCE 2 at this polymer dosage was only 3% which clearly demonstrated that PCE 2 has no water reduction capability. This evaluation has also been carriedout on the PCE 3 and the water reduction was close to 1.5%. The high molecular weights (Mw) of the PCE 2 and the PCE 3 have a strong impact on their dispersing capabilities by comparison with the PCE 1 which has a lower Mw of 100 000 g/mol. Slump retention and early strength development Even if the PCE 1 has a strong water reduction capacity, data in Figure 1 show also the weakness of the PCE 1 regarding the slump retention behavior. By using the concrete mixture proportions (Mix 2) described in Table 3, the PCE 1 has a polymer dosage limited to 0.21% of dry PCE/cement weight in order to reach an initial slump of 190 mm and a good homogeneous concrete formulation. Above this polymer dosage, side effects appear such as over-fluidity, bleeding, segregation, which typically lead to the production of heterogeneous concrete materials. For example, with 0.24% of PCE1 per cement weight instead of 0.21%, a severe segregation appears. In Figure 1, the Mix 2 made with combinations of the PCE 1 and 0.1% of the PCE 2 or 0.1% of the PCE 3 show very good slump retentions. By increasing the polymer dosage in PCE and by using the combination of a high water reducing PCE and a slump retention PCE, no bleeding or no segregation have been observed for the two concretes made with the PCE 2 or with the PCE 3.

New Additive to Enhance the Slump Retention  57

Fig. 1–Slump retention values of the PCE 1 (0.21%) combined or not with a Slump Retention PCE.

Fig. 2–Compressive strength development of the PCE 1 (0.21%) combined or not with a Slump Retention PCE. In Figure 2, the measurement of the early age strength after one day and of the compressive strength at 28 days show that there is no retarding effect when using the PCE 2 or the PCE 3, and there is no retarding effect by increasing the polymer dosage in this concrete mix design. The incrementation of the slump retention PCE into a concrete mix design formulated with a high water-reducing polymer allow to control the fluidity over time without any side effects on the compressive strength development. Cement pastes In order to understand the mechanism of such slump retention PCEs, some cement paste formulations have been studied according to the proportions detailed in Table 4. The use of cement pastes as models instead of concrete mixture formulations allowed us to assess the behavior of the PCEs over time at the microscale. In Figure 3, the cement paste prepared with the PCE 1 alone showed a fluidity loss as it has been already observed on the concrete

58 SP-302-04

Fig. 3–Paste flow values of the PCE 1 (0.21%) combined or not with a Slump Retention PCE.

Fig. 4–Paste flow values of the PCE 1, PCE 2 & PCE 3. tests. By adding the PCE 2 or the PCE 3 in combination with the PCE 1, no fluidity loss has been observed for the two cement paste formulations. The fluidity retention behaviors are similar for the formulations with the PCE 2 or the PCE 3 on the cement pastes or on the concrete mixture formulations. All these results confirm that model cement pastes can be used as a simplified model for the concrete formulation for this study. Figure 4 shows some very interesting behavior of the PCEs. The PCE 1 has the best initial fluidity and then a slump loss. The initial fluidity of the PCEs are related to their molecular weight (Mw), consequently, the higher the initial fluidity, the lower the Mw. Nevertheless, the paste flow evolutions are completely different between the PCEs which increase for the PCE 2 or for the PCE 3 and which decreases for the PCE 1.The fluidity of the PCE 2 is above the PCE 1 before 30 minutes and then is stable for the next 120 minutes. For the PCE 3, the evolution of the paste flow is fast for 30 minutes like for the PCE 2, and then increases slowly until 120 minutes. The paste fluidity of the PCE 3 is higher than that of PCE 1 just before 90 minutes. These observations on the cement pastes demonstrate a

New Additive to Enhance the Slump Retention  59

Fig. 5–Mw values of the PCE 3 in different salt conditions after storage at room temperature. change of dispersing ability of the PCE 2 and the PCE 3 over time which can be related to a modification of the polymer structure. Molecular weight evolution The molecular weight is one of the main characteristics which describs the Polycarboxylate Ether. Several studies11,12,18 have already been carried-out on the interactions between the PCEs and the salts present in the cement or pore solution composition. Different ions such as sulfates, calcium, sodium, potassium or the pH are well known to affect the polymer structure and conformation of the PCEs. In Figure 5, the Mw of the PCE 3 as measured by Gel Permeation Chromatography in an aqueous solution of KNO3 at 1wt% is around 1 200 000 g/mol and referred to named as “Reference”. Then the Mw values of PCE 3 measured in different salt solutions such Na2SO4, K2SO4, CaCl2, CaO, NaOH and KOH show important variations. For Na2SO4, K2SO4 and CaCl2, the Mw values are similar to the one measured in the standard GPC conditions. In basic pH range, the decrease of the Mw is very significant during the workability period. These different measurements confirm the change of polymer architecture of the PCE 3. In Figure 6, the Mw of PCE 1 does not change in basic pH range on the contrary to the PCE 2 which shows a strong variation of its Mw from 720 000 to 320 000 g/mol. FURTHER RESEARCH The results shown in this study explains the interactions between different kinds of polymer structures and their behaviors in a cement paste or concrete formulation. All the experimental tests have been carried-out with an Ordinary Portland Cement. Consequently, further developments have to be done on blended cements with various ionic concentrations in order to establish if the modifications of the polymer structures are similar.

60 SP-302-04

Fig. 6–Comparison of the Mw values for the PCE 1, PCE 2 & PCE 3 in standard and basic pH conditions CONCLUSIONS According to the different experimental investigations carried-out in this study, the high molecular weight PCEs are able to control the slump retention from the cement paste to the concrete mixutre. No increase of fluidity, no retarding effect and no bleeding have been observed by using a combination of a slump retention PCE and a standard high waterreducing PCE. AUTHOR BIOS David Platel is a R&D Manager of the Construction Laboratory at the Coatex SAS, France. He received his PhD from UPMC in Paris, France. His research interests include the development of new building chemicals and especially the chemistry & physicochemistry of the polymers in hydraulic binder formulations. Jean-Mars Suau is a Research Manager and is in charge of the Synthesis and Pilot Laboratory at Coatex SAS, France. His research interests include the Controlled Radical Polymerization, the Hydrophobically Alkali Swellable Emulsion, the Hydrophobically Ethoxylated Urethane, and the Polycarboxylate Ether. All theses polymers are water soluble or water compatible and they are rheological additives dedicated to industrial areas such as Construction, Paint, Paper, Cosmetic… Clement Chosson is a R&D Manager of the Analysis and the Quality Control Laboratories at the Coatex SAS, France. He received his MS from Claude Bernard University in Lyon, France. His research interests include the advanced characterization of polymers. Yves Matter is a Research Scientist at Coatex SAS. He received his PhD from Karlsruhe Institute of Technology. His research interests include polymer synthesis and physicochemical characterization. He is in charge of the development of new macromolecules used as concrete admixures.

New Additive to Enhance the Slump Retention  61

ACKNOWLEDGMENTS The authors wish to express their gratitude and sincere appreciation to Benoit Magny, R&D Director of Coatex group, for helpful support and discussions. rh

NOTATION = hydrodynamic radius

REFERENCES 1. Kirby, G. H., and Lewis, J. A., “Comb polymer architecture effects on the rheological property evolution of concentrated cement suspensions,” Journal of the American Ceramic Society, V. 87, No. 9, 2004, pp. 1643-1652. doi: 10.1111/j.1551-2916.2004.01643.x 2. Yamada, K., and Hanehara, S., “Working mechanism of polycarboxylate superplasticizer considering the chemical structure and cement characteristics,” Proceedings of the 11th International Congress on the Chemistry of Cement. 2003. 3. Magarotto, R.; Torresan, I.; and Zeminian, N., “Influence of the molecular weight of polycarboxylate ether superplasticizers on the rheological properties of fresh cement pastes, mortar and concrete,” Proceedings of the 11th International Congress on the Chemistry of Cement. 2003. 4. Flatt, R. J., “Polymeric dispersants in concrete,” Polymers in particulate systems properties and applications, Hackley, V.A., Somansundaran, P., and Lewis, J.A., Editors. 200, p. 247-294. 5. Yoshioka, K. et al., “Adsorption characteristics of superplasticizers on cement component minerals,” Cement and Concrete Research, V. 2056, 2002, pp. 1-7. 6. Yamada, K.; Ozu, H.; and Yano, M., “Prevention of incompatibility phenomena between cement and superplasticizer by blending several types of polycarboxylates polymers,” Proceedings of 1st fib congress, 2002, 143: p. 16-26. 7. Yamada, K., and Hanehara, S., “Interaction mechanism of cement and superplasticizers - The roles of polymer adsorption and ionic conditions of aqueous phase,” Concrete Science and Engineering, V. 3, 2001, pp. 135-145. 8. Platel, D., “Impact of polymer architecture on superplasticizer efficiency,” Proceedings of 9th ACI Conference. 2009, p 381-394. 9. Yamada, K., “A summary of important characteristics of cement and superplasticizer,” Proceedings of 9th ACI Conference. 2009, p 85-96. 10. Flatt, R. J. et al., “Polymer physics and superplasticizers,” Proceedings of 9th ACI Conference. 2009, p 113-122. 11. Flatt, R. J. et al., “The role of adsorption energy in the sulfate-polycarboxylate competition,” Proceedings of 9th ACI Conference. 2009, p 153-164. 12. Zimmermann, J. et al., “Effect of polymer structure on the sulfate-polycarboxylate competition,” Proceedings of 9th ACI Conference. 2009, p 165-175. 13. Göller, F, et al., “The relationship between retention stability and chemical structure,” Proceedings of 9th ACI Conference. 2009, p 249-260. 14. Izumi, T. et al., “A new hybrid type superplasticizer,” Proceedings of 7th ACI Conference. 2003 – supplementary papers, p 67-81. 15. Cerulli, T. et al., “Superplasticizers for extending workability,” Proceedings of 8th ACI Conference. 2006 – supplementary papers, p 263-277.

62 SP-302-04

16. Hamada, D. et al., “Development of slump-loss controlling agent with minimal setting retardation,” Proceedings of 7th ACI Conference. 2003, p 127-142. 17. Hamada, D. et al., “Development of new Superplasticizer providing ultimate workability,” Proceedings of 8th ACI Conference. 2006, p 31-50. 18. Borget, P.; Galmiche, L.; Le Meins, J.-F.; and Lafuma, F., “Microstructural characterisation and behaviour in different salt solutions of sodium polymethacrylate-g-PEO comb copolymers,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 260, No. 1-3, 2005, pp. 173-182. doi: 10.1016/j.colsurfa.2005.03.008

SP-302-05

Synthesis of a Novel Superplasticizer Prepared from Brown Coal by Manuel Ilg and Johann Plank Polycondensates and polycarboxylates are known to be effective superplasticizers. Here, the synthesis of a novel, brown coal based superplasticizer by grafting acrylic acid (AA) and 2-acrylamido-2-tert.butyl sulfonic acid (ATBS) onto the alkali soluble components of brown coal (lignite) as backbone using free radical copolymerization technique is described. Furthermore, an ATBS-acrylic acid copolymer was synthesized to investigate the influence of the graft chains on the performance of the lignite copolymer. Successful grafting was confirmed by size exclusion chromatography (SEC) and comparison of the adsorbed layer thicknesses of the brown coal substrate and the grafted product. The dispersing performance of the graft copolymer was probed via mini slump tests and compared with that of BNS. Additionally, slump flow retention and sulfate tolerance were determined. It was found that the graft copolymer possesses higher dispersion effectiveness than BNS and exhibits high sulfate tolerance. Keywords: admixture; adsorbed layer thickness; brown coal; graft copolymer; lignite; superplasticizer. INTRODUCTION Superplasticizers are an important part of modern concrete technology to improve the workability of fresh concrete. With the help of these admixtures the water-to-cement ratio can be reduced, thus obtaining cementitious building materials with higher strength and durability.1,2 Although polycondensates and polycarboxylates represent very effective polymers, new structures and concepts are still investigated.3 In recent years, natural polymers became increasingly popular as a starting material for the synthesis of new admixtures.4 A main reason is that natural polymers offer novel structural motifs and often bear a high diversity of functional groups. In literature, a number of examples already exist where through functionalization and modification of natural polymers, novel superplasticizers have been synthesized.5,6 Ordinary brown coal (often referred to as lignite) also presents a promising candidate for the development of new superplasticizers, because of its global abundance and its low cost. Brown coal is a combustible sedimentary rock that was built million years ago from dead plant material through a process called coalification.7 In flooded areas such as swamps 63

64 SP-302-05

the main constituents of plant residues, lignin and cellulose, were decomposed gradually to humic substances (peat). Burial by sediments and subsequent tectonic pressure over millions of years led to an increase of the temperature causing compaction of the decomposition products. In this process, peat was successively metamorphosed into brown coal under release of water and gases such as methane and carbon dioxide.7 In this study, a caustic extract of brown coal containing humic and fulvic acids as main constituents was utilized.8-10 Such alkali extracts are commercially available and widely used in different industries including agriculture (as fertilizer)10 and in oil well drilling as drilling fluid (the so-called “lignite” mud).11 However, up to date no brown coal based concrete superplasticizer has been developed. Here, we report on the synthesis of a novel superplasticizer prepared by grafting acrylic acid and ATBS monomers onto the alkali soluble components of brown coal. Generally, humic and fulvic acids possess a large number of functional groups (i.e. aliphatic/aromatic carboxylate, carbonyl, amino, phenolic and ketone groups) which present a perfect substrate for graft copolymerization reactions.12,13 The synthesized brown coal-ATBS-acrylic acid graft copolymer was characterized with respect to its molecular properties (Mw, Mn) and its anionic charge amount. Its dispersing effectiveness and slump retention behavior was compared with that of an industrial grade β-naphthalene sulfonate (BNS) sample. Additionally, the sulfate tolerance of the graft copolymer was studied and heat flow calorimetry measurements were performed to assess its effect on cement hydration. To prove successful grafting, the adsorbed layer thicknesses of the brown coal substrate and the graft copolymer were measured via dynamic light scattering utilizing cationic polystyrene nanoparticles as adsorbent.14 Finally, the working mechanism of the superplasticizer was clarified via zeta potential measurements. RESEARCH SIGNIFICANCE A novel brown coal based superplasticizer has been synthesized using graft copolymerization technique. From a commercial brown coal the alkali soluble fraction (mainly humic and fulvic acids) was extracted and utilized as backbone. Up to date, coal and natural gas are the only readily available alternatives to petroleum. Therefore, the different constituents of brown coal could be an alternative feedstock in the production of superplasticizers to save the more precious resources of petroleum. The overall goal of this study was to validate new structural motifs and to offer a synthetic route to highly economical, brown coal based superplasticizers. EXPERIMENTAL PROCEDURE Materials Brown coal – The used brown coal was mined in the Lusatia brown coal mining district near Cottbus, Germany. Chemicals – 2-Acrylamdio-2-tert.butyl sulfonic acid (ATBS), acrylic acid (AA), sodium hydroxide (NaOH), sodium peroxodisulfate (Na2S2O8), sodium pyrosulfite (Na2S2O5) and EDTA were used as per obtained. Industrial superplasticizer sample – A spray dried powder BNS superplasticizer was used as an industrial grade reference sample.

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  65

Table 1 – Phase composition of the CEM I 52.5 N sample determined by Q-XRD using Rietveld refinement. Phase C3S, m C2S, m C3A, c C3A, o C4AF, o Free Lime (Franke) Periclase (MgO) Anhydrite Hemihydrate* Dihydrate* Calcite Quartz Arcanite (K2SO4)

wt.% 54.14 26.63 3.28 4.26 2.45 0.10 0.03 2.64 1.21 0.02 3.61 1.16 0.46

* determined by thermogravimetry

Fig. 1 – Scheme for extraction of alkali soluble components from brown coal. Cement – The cement used for this study was an ordinary Portland cement CEM I 52.5 N. Its phase composition as determined by X-ray diffraction and subsequent Rietveld refinement is illustrated in Table 1. Its specific surface area was 3,583 cm2/g (Blaine method) and its particle size (d50 value) was 11.5 µm [45.3 · 10-5 in.] (laser granulometer). The density as obtained by helium pycnometry was 3.15 g/cm3. Extraction of brown coal The individual steps performed in the extraction process are schematically presented in Fig. 1. First, chunks of brown coal were coarsely crushed with a hammer and the residue was sieved to < 250 µm [9.84 · 10-3 in.] to obtain a brown coal powder exhibiting a narrow

66 SP-302-05

particle size distribution and a high specific surface area. For extraction of the alkali soluble components, 70 g [2.47 oz.] of the brown coal powder were mixed with 700 mL [23.7 fl. oz.] of 0.5 M NaOH placed in a 1 L [33.8 fl. oz.] round-bottom flask.9 The mixture was refluxed for three hours at 90 °C [194 °F] under constant stirring. Next, the dark solution was cooled to ambient temperature and centrifuged two times for 10 minutes at 8,500 rpm. The supernatant was separated from the residue and freeze dried, yielding 21.7 g [0.765 oz.] of a black solid (theor. yield 31%). It is noteworthy that many coal producers offer such extracts in liquid or powder form designated as “caustic lignite”. Synthesis of brown coal-ATBS-acrylic acid graft copolymer The novel brown coal-ATBS-acrylic acid graft copolymer was synthesized by aqueous free radical copolymerization using sodium peroxodisulfate as initiator. Here, 2-acrylamido-2-tert.butyl sulfonic acid (ATBS) and acrylic acid were grafted onto the extracted alkali soluble components (i.e. humic and fulvic acid) of brown coal. The molar ratio between ATBS and acrylic acid was 1: 0.15 and the weight ratio between the brown coal and the grafted monomers was 20: 80 (wt/wt). In a five necked, 1 L [33.8 fl. oz.] roundbottom flask equipped with stirrer, thermometer, reflux condenser and inlet for N2 gas, 13.2 g [0.466 oz.] of the dry alkali soluble components were dissolved in 206 mL [6.97 fl. oz.] DI water. Thereafter, 8.8 g [0.31 oz.] NaOH pellets were added to adjust a pH of 12 and then cooled to 18 °C [64 °F]. Next, 50.0 g [1.76 oz.] ATBS (241 mmol, 1.0 eq) were dissolved stepwise and the temperature was kept constantly under 25 °C [77 °F] to avoid homopolymerization of ATBS. 2.60 g [0.092 oz.] acrylic acid (36.1 mmol, 0.15 eq), 600 mg [0.021 oz.] EDTA and 1.00 g [0.035 oz.] of an organo-modified polysiloxane defoamer were added and the mixture was purged with N2 for 1 h at room temperature. After heating to 50 °C [122 °F] the first portion of sodium peroxodisulfate initiator was added (8 g [0.282 oz.]) and the reaction mixture was stirred for 50 minutes at this temperature. The second part of initiator was added (8 g [0.282 oz.]) and the polymerization continued for additional 70 minutes at 50 °C [122 °F]. Then the temperature was increased to 60 °C [140 °F] and kept there for 1 h. Finally, the reaction flask was heated to 80 °C [176 °F], 3.60 g [0.127 oz.] of sodium pyrosulfite were added to quench remaining radicals and stirring continued for 1 h at 80 °C [176 °F]. The solution was cooled to room temperature to obtain a viscous, dark brownish polymer solution with a solid content of 29 wt.% and a pH of 2.5. The polymer solution was used without any further purification. Synthesis of ATBS-acrylic acid copolymer An ATBS-acrylic acid copolymer was synthesized according to the procedure described above except that no alkali soluble components were present. Here, a pale yellowish, 27.6 wt.% aqueous solution with low viscosity and a pH of 2.5 was obtained. This polymer was used for comparison. Characterization of polymers Size exclusion chromatography – Molecular weights (Mw and Mn) and polymer radii (Rh(z) and Rg(z)) of all synthesized polymers were determined by size exclusion chromatography. The instrument is equipped with a RI detector and an 18 angle dynamic light scattering detector. Polymer solutions exhibiting a concentration of 2 g/L were prepared

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  67

for the SEC analysis. The polymers were separated on a precolumn and two columns using 0.2 M NaNO3 solution (adjusted with NaOH to pH 9) as an eluent at a flow rate of 1.0 mL/ min. The value of dn/dc used to calculate Mw and Mn was 0.218 mL/g (value for lignin).15 Anionic charge amount of the polymers – The anionic charge of the polymers was determined via polyelectrolyte titration using a particle charge detector. 0.001 M cationic polydiallyl dimethyl ammoniumchloride (polyDADMAC) solution was employed as titrator. Polymer solutions with a concentration of 0.1 g/L were prepared in DI water, in 0.1 M NaOH and in cement pore solution (CPS). Cement pore solution was freshly prepared by vacuum filtration of neat cement slurries using a water-to-cement ratio of 0.455. In a typical experiment, 10 mL [0.34 fl. oz.] of the polymer solution were pipetted into a PTFE cylinder with an oscillating PTFE piston in the center. The dissolved polymers can adsorb via Van der Waals forces on the surface of the cylinder and the piston. Because of the oscillating movement of the piston, counter ions are removed from the immobilized polymers and a streaming current results which can be measured by two platinum electrodes located within the PTFE cylinder. The polyDADMAC solution was titrated until the isoelectronic point was reached. For every polymer sample the measurement was repeated three times and the values were averaged. From the consumption of polyDADMAC the amount of negative charge per gram of polymer was calculated. Heat flow calorimetry – To investigate the influence of the synthesized polymers on cement hydration, isothermal heat flow calorimetric measurements were carried out. There, 4 g [0.141 oz.] cement were filled into 20 mL [0.68 fl. oz.] glass ampoules and mixed with the respective amount of aqueous polymer solution to obtain a water-to-cement ratio of 0.455. The ampoules were sealed, homogenized for 1 min in a wobbler and then placed into the isothermal conduction calorimeter. Data logging was continued until heat evolution from the hydration reaction subsided completely. Performance of the synthesized polymers Mini slump test – The dispersing effectiveness of the synthesized polymers was assessed utilizing a mini slump test following in principle DIN EN 1015, but with some modifications. At first, the water-to-cement ratio required to reach a slump flow of 18 ± 0.5 cm [7.1 ± 0.2 in.] was established for the cement paste without polymers. At this specific waterto-cement ratio, the dosage of the polymers was determined to attain a spread flow of 26 ± 0.5 cm [10.2 ± 0.2 in.]. In a typical experiment, the superplasticizer was dissolved in the required amount of mixing water placed in a porcelain cup. The amount of water contained in the polymer solution was subtracted from the amount of mixing water. 300 g [10.6 oz.] of cement were added to the mixing water over a period of 1 min, then rested for 1 min and subsequently were stirred manually for 2 min with a spoon. Immediately after the end of stirring, the cement slurry was poured into a Vicat cone (height 40 mm [1.57 in.], top diameter 70 mm [2.76 in.], bottom diameter 80 mm [3.15 in.]) placed on a glass plate, filled to the brim and the cone was lifted vertically. The resulting paste spread was measured twice, the second measurement being perpendicular to the first one and averaged to obtain the slump flow value. Time dependent mini slump test – Development of dispersing performance over time was investigated via time dependent mini slump testing. Here, 400 g [14.1 oz.] cement were mixed with the required amount of water and polymer to achieve an initial slump flow

68 SP-302-05

of 26 ± 0.5 cm. The procedure was the same as described above. After each measurement the cement paste was transferred back into the porcelain cup and covered with a wet towel to avoid desiccation. Prior to each measurement the cement paste was vigorously stirred for two minutes. Measurements were conducted every 15 minutes over a total period of 120 minutes. Adsorbed layer thickness The adsorbed layer thickness of the polymers was captured by dynamic light scattering. Here, monodisperse polystyrene nanoparticles exhibiting an average particle size of 75.5 ± 0.5 nm [2.97 · 10-6 in.] were utilized as adsorbent. Starting from a 150 mg/L stock solution of each polymer in 0.1 M NaOH, different concentrations were prepared by dilution (diluent 0.1 M NaOH). Prior to the measurements, the solutions were filtered through a 0.2 µm [7.9 · 10-6 in.] filter to remove undesired dust particles that can disturb measurements because of their high scattering intensity. Next, 50 µL [1.7 · 10-3 fl. oz.] of a suspension of cationic polystyrene nanoparticles (for preparation see14) were added and sonicated for 5 min. The polymer solution was filled into a glass cuvette and then placed in the instrument. Every measurement was repeated 150 times per sample and the average value was calculated. Each test consisted of a 10 s light scattering run taken at a temperature of 25 °C [77 °F]. Measurements were continued at increasing polymer concentrations until a stable, final value was reached that was regarded as the point of saturated adsorption. The adsorbed layer thickness was calculated using equation 1.

adsorbed layer thickness nm = dads nm - dpolystyrene [nm]2 (1)

where dads represents the particle size of the polystyrene nanoparticle holding adsorbed polymer, and dpolystyrene represents the particle size of the native polystyrene particle. Zeta potential measurement Zeta potential measurements were performed at room temperature on an electro acoustic spectrometer. This instrument measures a vibration current induced by an acoustic wave which causes the aqueous phase to move relative to the cement particles. From that, a potential difference results which can be measured and designated as zeta potential. Immediately after mixing, the freshly prepared cement slurries holding the respective dosage of the polymer samples required for a slump flow of 26 ± 0.5 cm were filled into the cup of the instrument and measured under continuous stirring for a total period of 30 min. EXPERIMENTAL RESULTS AND DISCUSSION Characterization of brown coal-ATBS-acrylic acid graft copolymer The synthesized graft copolymer and the ATBS-acrylic acid copolymer were characterized using SEC. The molar masses and polymer radii are listed in Table 2. According to these data, the brown coal-ATBS-acrylic acid graft copolymer exhibits a significantly higher molar mass (Mw ~ 443,300 g/mol) than the ATBS-acrylic acid copolymer (Mw ~ 183,300 g/mol). This result is a first indication that grafting was successful. Furthermore it was observed that the hydrodynamic and gyration radii increased from 14.5 nm [5.71 · 10-7 in.] and 23.9 nm [9.41 · 10-7 in.] respectively for the ATBS-acrylic acid copolymer to

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  69

Table 2 – Molar masses, polydispersity index (PDI) and polymer radii of the synthesized polymers and of BNS as reference superplasticizer sample. Polymer Brown coal-ATBS-AA ATBS-AA copolymer BNS

Mw [g/mol] 443,300 183,300 140,000*

Mn [g/mol] 216,100 93,460 -

PDI 2.0 2.0 -

Rg(z) [nm] 37.2 23.9 -

Rh(z) [nm] 22.1 14.5 -

* = batch measurement

Fig. 2 – SEC spectrum of the brown coal-ATBS-AA graft copolymer. Table 3 – Anionic charge amounts of the polymer samples in DI water, CPS and 0.1 M NaOH solution. Polymer Brown coal-ATBS-AA ATBS-AA copolymer BNS

DI water [µeq/g] 3,347 4,241 3,643

CPS [µeq/g] 3,964 3,974 2,713

0.1 M NaOH [µeq/g] 4,765 4,855 3,989

22.1 nm [8.70 · 10-7 in.] and 37.2 nm [1.46 · 10-6 in.] resp. after the grafting process. In comparison, BNS possesses a much lower molar mass (Mw ~ 140,000 g/mol) than the graft copolymer. In summary, the graft copolymerization process produces a relatively homogeneous graft copolymer that exhibits a quite narrow polydispersity index of ~ 2.0 (see SEC spectrum of graft copolymer in Fig. 2). Next, the anionic charge amounts of the polymers were determined in DI water, cement pore solution (CPS) and in 0.1 M NaOH solution. The results are summarized in Table 3. Generally, the graft copolymer possesses a highly anionic character (charge amount e.g. in CPS 3,964 µeq/g), owed to the carboxylate and sulfonate groups present in the graft chains. Such high anionic charge promotes adsorption of the graft copolymer onto positively charged surfaces of cement. The ATBS-acrylic acid copolymer exhibits an even higher anionic charge while BNS possesses a much lower anionic charge than the brown coal based graft copolymer and the ATBS-acrylic acid copolymer.

70 SP-302-05

Fig. 3 – Proposed chemical structure of the brown coal-ATBS-AA graft copolymer. For the chemical structure of the graft copolymer, the model as follows was developed: Humic acid which presents the main component in the alkali extract possesses numerous functional groups (i. e. phenolic and carboxylate groups) in its structure which provide perfect docking sites for the grafting reaction. After addition of an initiator (e.g. sodium peroxodisulfate), macroradicals can form through abstraction of hydrogen from these functional groups. Consequently, ATBS and acrylic acid monomers can be grafted onto these free radical sites. Simultaneously, side chain propagation can continue. As a result, a graft copolymer is formed that is composed of a humic acid backbone and grafted ATBS-coacrylic acid side chains. A structural model of the graft copolymer is proposed in Fig. 3. Note that for humic acid, only model structures exist16 due to complexity of composition and abundant natural variants. Generally, humic acid contains a number of condensed aromatic rings, and thus attains a relatively stiff, linear conformation. In contrast to this, the ATBS-co-acrylic acid graft chains exhibit high conformational flexibility and are coiled in solution. These differences in solution conformation were confirmed by the Burchard parameters (ratio of Rg(z)/Rh(z)). There, a value of 1.7 was obtained for the graft copolymer which represents a linear, stretched random coil.17 Cement dispersion The dispersing performance of the brown coal based graft copolymer was determined using a mini slump test. Here, the dosages were established to reach a paste flow of 26 ± 0.5 cm. The water-to-cement ratio of the neat cement paste without superplasticizer was set to produce a slump flow of 18 ± 0.5 cm (w/c ratio of 0.455). This water-to-cement

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  71

Fig. 4 – Time dependent development of the slump flow of a cement slurry (w/c = 0.455) containing 0.21% bwoc of brown coal-ATBS-AA graft copolymer, 0.19% bwoc of ATBS-AA copolymer and 0.3% bwoc of BNS respectively. value was applied for all measurements. For the brown coal-ATBS-acrylic acid graft copolymer a dosage of 0.21% bwoc was required to reach the desired slump flow of 26 ± 0.5 cm. Thus, the novel graft copolymer was even more effective than a commercial BNS reference sample (dosage 0.30% bwoc). Additional mini slump testing carried out for the ATBS-acrylic acid copolymer revealed that this polymer was slightly better (dosage 0.19% bwoc) than the graft copolymer itself. The results suggest that the dispersing effect mainly originates from the ATBS-co-acrylic acid graft chains. This was confirmed further via mini slump tests, evidencing that the alkaline brown coal extract does not disperse cement. Furthermore, the time dependent slump loss behavior of a cement paste prepared at a water-to-cement ratio of 0.455 containing the polymers at dosages required for a slump flow of 26 ± 0.5 cm was investigated. The results of these measurements are displayed in Fig. 4. Starting from an initial slump flow of 26 ± 0.5 cm, fluidity was monitored every 15 min for a period of 120 min. According to Fig. 4, the dispersing performance of the brown coal-ATBS-acrylic acid graft copolymer quickly decreases in the first 30 min (slump flow 21.4 cm [8.43 in.]). Afterwards, the decrease becomes very slow. In contrast, the ATBSacrylic acid copolymer exhibits better slump retention and behaves more similar than the commercial BNS superplasticizer. Sulfate tolerance of graft copolymer The presence of alkali sulfates (i.e. K2SO4, Na2SO4) in cement can have a significant impact on the dispersing performance of superplasticizers. This phenomenon (the so-called “sulfate effect”) has been observed mainly for polycarboxylates. In literature, two mechanisms are discussed for the negative impact of sulfate on PCEs: competitive adsorption between sulfate ions and PCEs18 and shrinkage of PCE molecules.19 To investigate whether the brown coal based graft copolymer is also affected by sulfate ions, mini slump tests at increasing additions of sodium sulfate were carried out. The graft copolymer was applied at a dosage of 0.21% bwoc (slump flow 26 ± 0.5 cm). The results are illustrated in Fig. 5.

72 SP-302-05

Fig. 5 – Effect of different Na2SO4 dosages on the slump flow of cement pastes (w/c = 0.455) holding different polymer samples. There, it was observed that the graft copolymer is not much affected by the presence of different concentrations of sulfate, thus confirming high sulfate tolerance for the grafted product. In contrast, the ATBS-acrylic acid copolymer shows strong sensitivity to sulfate. At increasing Na2SO4 additions, its initial slump flow of 26.3 cm [10.4 in.] quickly drops to 20.9 cm [8.22 in.]. Furthermore, a physical blend of 20 wt.% brown coal extract and 80 wt.% of the ATBS-acrylic acid copolymer (as present in the graft copolymer) was tested. It was found that this physical mixture exhibits the same poor sulfate tolerance than the ATBS-acrylic acid copolymer, thus providing further evidence that in fact a chemical reaction had occurred between the brown coal substrate and the monomers. A potential explanation for the different behaviors of the brown coal-ATBS-acrylic acid graft copolymer and the ATBS-acrylic acid copolymer are the different molar masses of the polymers. The graft copolymer possesses a higher Mw than the copolymer, so its adsorption is mainly driven by entropic effects (desorption of a huge amount of ions and water molecules adsorbed on the surface of cement), thus producing a high Gibbs energy of adsorption.20 Whereas sulfate ions, whose energy of adsorption results from an enthalpic contribution only and is comparatively low cannot displace already adsorbed polymers from the surface of cement. Adsorbed layer thickness Additional experiments were carried out to prove successful grafting of ATBS and acrylic acid monomers onto the alkaline extract of brown coal. For this purpose, the adsorbed layer thicknesses of the graft copolymer, the ATBS-acrylic acid copolymer and the brown coal extract were measured and compared. Monodisperse, spherical cationic polystyrene nanoparticles were taken as adsorbent and layer thicknesses were determined using dynamic light scattering. This method allows facile determination of the adsorbed layer thickness of negatively charged polyelectrolytes at high pH conditions such as in cement pore solution.14 Layer thicknesses were measured as a function of polymer concentration until the point of saturated adsorption was reached. The results are displayed in Fig. 6. The alkaline extract of brown coal reaches the point of saturated adsorption at an adsorbed layer thickness of ~ 2.5 nm [9.8 · 10-8 in.] only whereas the brown coal-ATBS-acrylic acid graft

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  73

Fig. 6 – Concentration-dependent adsorbed layer thicknesses of the brown coal-ATBS-AA graft copolymer, of ATBS-AA copolymer, of the alkaline brown coal extract and BNS. polymer exhibits a substantially higher layer thickness of ~ 6.4 nm [2.5 · 10-7 in.]. Contrary to this, the ATBS-acrylic acid copolymer produces an adsorbed layer thickness of ~ 1.7 nm [6.7 · 10-8 in.] only. These values signify that grafting of the monomers onto the brown coal substrate has indeed occurred, and that the graft product possesses pendants of ATBS-coacrylic acid. However, BNS shows a very low adsorbed layer thickness of only ~ 0.3 nm. Effect on cement hydration The influence of the superplasticizers on cement hydration was tested by means of isothermal heat flow calorimetry. The heat evolution from cement hydration was monitored for cement pastes prepared at a water-to-cement ratio of 0.455 holding 0.25% bwoc of the brown coal-ATBS-acrylic acid graft copolymer, of ATBS-acrylic acid copolymer and of BNS respectively. The results are illustrated in Fig. 7. It was observed that the brown coalATBS-acrylic acid graft copolymer causes very minor retardation, apparently caused by the ATBS-co-acrylic acid pendant groups. Mechanism of dispersion Adsorption of superplasticizers on cement particles can be tracked via zeta potential measurements. Here, the zeta potential of cement slurries holding different polymer samples was measured. Concentrations of the polymer samples were those required for a slump flow of 26 ± 0.5 cm. The neat cement paste exhibited a slightly negative zeta potential of – 3.3 mV. After addition of the polymer samples, the zeta potentials of all cement slurries decreased to similar values: brown coal-ATBS-acrylic acid graft copolymer – 28.0 mV; ATBS-acrylic acid copolymer – 27.3 mV and BNS – 29.3 mV. These results signify that the brown coal-ATBS-acrylic acid graft copolymer adsorbs and also achieves dispersion through an electrostatic repulsion effect.

74 SP-302-05

Fig. 7 – Isothermal heat flow calorimetry of cement slurries (w/c = 0.455) holding 0.25% bwoc of brown coal-ATBSAA graft copolymer, ATBS-AA copolymer and BNS respectively. CONCLUSIONS An alkali brown coal extract holding humic and fulvic acids was successfully used as substrate for the synthesis of a brown coal based superplasticizer. Acrylic acid and ATBS were successfully grafted onto the extracted lignite, as was confirmed via SEC analysis and measurements of the adsorbed layer thicknesses of the graft copolymer and an ATBSacrylic acid copolymer prepared under comparable conditions. A structural model for the novel graft copolymer suggests that humic/fulvic acid backbones hold grafted chains of ATBS-acrylic acid copolymer. The synthesized brown coal-ATBS-acrylic acid graft copolymer presents an effective cement dispersant which is superior over industrial BNS and exhibits excellent sulfate tolerance. Its working mechanism relies on a combination of a strong electrostatic and a minor steric effect. The study shows that caustic lignite extracts present interesting and low-cost starting materials for the development of novel superplasticizers holding unique structural motifs. Future research should focus on the replacement of the ATBS monomer by less expensive alternatives such as e.g. itaconic, methacrylic, styrene sulfonic, allyloxy hydroxy propyl sulfonic acid or esters thereof. Due to the chemical variability of brown coal and the number of natural variants, different brown coal types should be tested as well. We consider our work as an initial study which may stimulate further ideas to exploit the potential of this concept, especially in countries where more sophisticated monomers are not available. AUTHOR BIOS Manuel Ilg studied chemistry and received his B.Sc. and M.Sc. from Technische Universität München. He is currently a Ph.D. student at the Chair for Construction Chemistry in Garching where he works on new structural concepts for superplasticizers.

Synthesis of a Novel Superplasticizer Prepared from Brown Coal  75

Johann Plank is a full Professor at the Institute of Inorganic Chemistry of Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemistry there. His research interests include cement chemistry, concrete admixtures, organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well cementing. ACKNOWLEDGEMENTS The authors would like to thank Vatenfall Europe Mining AG for providing the brown coal samples and Lubrizol for the supply of the ATBS monomer. REFERENCES 1. Ramachandran, V. S.; Malhotra, V. M.; Jolicoeur, C.; and Spiratos, N., “Superplasticizers: Properties and applications in concrete”, CANMET, Ottawa, Canada, 1998. 2. Spiratos, N., Page, M., Mailvaganam, N. P., Malhotra, V. M., Jolicoeur, C., “Superplasticizers for Concrete: Fundamentals, Technology, and Practice”, Supplementary Cementing Materials for Sustainable Development, Ottawa, Canada, 2003. 3. Plank, J.September 2012 , ““PCE Superplasticizers – Chemistry, Applications and Perspectives”, 18. ibausil,” Weimar, Germany, V. 1, pp. 91-102. 4. Plank, J.2003 , “Applications of Biopolymers in Construction Engineering,” Biopolymers, V. 10, pp. 29-95. 5. Lv, S.; Gao, R.; Cao, Q.; Li, D.; and Duan, J.2012 , “Preparation and characterization of poly-carboxymethyl-β-cyclodextrin superplasticizer,” Cement and Concrete Research, V. 42, No. 10, pp. 1356-1361. doi: 10.1016/j.cemconres.2012.06.006 6. Zhang, D. F.; Ju, B. Z.; Zhang, S. F.; He, L.; and Yang, J. Z.2007 , “The study on the dispersing mechanism of starch sulfonates as a water-reducing agent for cement,” Carbohydrate Polymers, V. 70, No. 4, pp. 363-368. doi: 10.1016/j.carbpol.2007.04.024 7. Hatcher, P. G., and Clifford, D. J.1997 , “The organic geochemistry of coal: from plant materials to coal,” Organic Geochemistry, V. 27, No. 5/6, pp. 251-274. doi: 10.1016/ S0146-6380(97)00051-X 8. Ashida, R.; Morimoto, M.; Makino, Y.; Umemoto, S.; Nakagawa, H.; Miura, K.; Saito, K.; and Kato, K.2009 , “Fractionation of brown coal by sequential high temperature solvent extraction,” Fuel, V. 88, No. 8, pp. 1485-1490. doi: 10.1016/j.fuel.2008.12.003 9. Pang, L. S. K.; Vassallo, A. M.; and Wilson, M. A.1990 , “Chemistry of alkali extraction of brown coals – I. Kinetics, characterization and implications to coalification,” Organic Geochemistry, V. 16, No. 4-6, pp. 853-864. doi: 10.1016/0146-6380(90)90122-G 10. Garcia, D.; Cegarra, J.; Abad, M.; and Fornes, F.1993 , “Effects of the extractants on the characteristics of humic fertilizer obtained from lignite,” Bioresource Technology, V. 43, No. 3, pp. 221-225. doi: 10.1016/0960-8524(93)90034-9 11. Caenn, R., and Chillingar, G. V.1996 , “Drilling fluids: state of the art,” Journal of Petroleum Science Engineering, V. 14, No. 3-4, pp. 221-230. doi: 10.1016/0920-4105(95)00051-8 12. Pehlivan, E., and Arslan, G.2006 , “Comparison of adsorption capacity of young brown coals and humic acids prepared from different coal mines in Anatolia,” Journal of Hazardous Materials, V. B138, No. 2, pp. 401-408. doi: 10.1016/j.jhazmat.2006.05.063 13. Krol-Domanska, K., and Smolinska, B.2012 , “Advantages of lignite addition in purification process of soil polluted by heavy metals,” Biotechnology and Food Sciences, V. 76, No. 1, pp. 51-58.

76 SP-302-05

14. Tiemeyer, C.; Lange, A.; and Plank, J.2014 , “Determination of the adsorbed layer thickness of functional anionic polymers utilizing chemically modified polystyrene nanoparticles,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 456, pp. 139-145. doi: 10.1016/j.colsurfa.2014.05.014 15. Gupta, P. R., and Goring, D. A. I.1960 , “Physicochemical Studies of Alkali Lignins III: Size and Shape of the Macromolecule,” Canadian Journal of Chemistry, V. 38, No. 2, pp. 270-279. doi: 10.1139/v60-036 16. Stevenson, F. J., “Humus Chemistry: Genesis, Composition, Reactions”, John Wiley & Sons, New York, USA, 2nd Edition, 1994. 17. Burchard, W.1983 , “Static and dynamic light scattering from branched polymers and biopolymers,” Advances in Polymer Science, V. 48, pp. 1-124. doi: 10.1007/3-540-12030-0_1 18. Han, S., and Plank, J.2013 , “Mechanistic study on the effect of sulfate ions on polycarboxylate superplasticizers in cement,” Advances in Cement Research, V. 25, No. 4, pp. 200-207. doi: 10.1680/adcr.12.00002 19. Yamada, K.; Ogawa, S.; and Hanehara, S.2001 , “Controlling of the adsorption and dispersing force of polycarboxylate-type superplasticizer by sulfate ion concentration in aqueous phase,” Cement and Concrete Research, V. 31, No. 3, pp. 375-383. doi: 10.1016/ S0008-8846(00)00503-2 20. Plank, J.; Sachsenhauser, B.; and de Reese, J.2010 , “Experimental determination of the thermodynamic parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers,” Cement and Concrete Research, V. 40, No. 5, pp. 699-709. doi: 10.1016/j.cemconres.2009.12.002

SP-302-06

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions by Lucia Ferrari and Pascal Boustingorry Cementitious suspensions all feature common flow characteristics when their flow curve is observed. When plotted as shear stress vs shear rate a minimum in stress is observed towards low shear rates which may be related to the hydration of cement. When plotted as apparent viscosity versus shear rate a minimum often appears towards high shear rates, beyond which the suspension enters a shear-thickening regime the origin of which remains unclear. In between these two limits of shear rate, the expected shear-thinning behaviour takes place, where apparent viscosity may be linked to a shear-rate-dependent degree of suspension flocculation. The present paper aims at shedding some light onto the origins of those features in the context of mix design and superplasticizer technology. Keywords: pumping; civil engineering concrete; viscosity; superplasticizer; phosphonate. INTRODUCTION Whereas the rheological behaviour of cementitious suspensions has often been discussed under the framework of the Bingham model,1-3 thorough observations at the steady-state and under no-slip conditions indicate more complex behaviour. It has been reported that high dosages of superplasticizer may lead to a shear-thickening behaviour4-7 when shear rate increases, which is quite well simulated by the Herschel-Bulkley model, but with no satisfactory explanation of the underlying mechanism.3,5 Low shear rate regimes have attracted less attention, but some studies have shown that the ageing nature of a suspension may lead to a noticeable change in the flow curves with the occurrence of an increasing stress branch at low strain rates.8,9 Between these two limits, the material features the often reported shear-thinning behaviour. The present work aims at showing that cementitious compositions all feature such characteristics, though in various intensities and ranges, depending on the mixture proportions

77

78 SP-302-06

Table 1–Grout Mix #1 – French mix Component Type V Cement 0/0.160 mm Siliceous filler Silica Fume 0/0.315 mm sand

Weight (g) 220,42 117,57 18,25 43,76

Weight (oz) 7,7751 oz 4,1472 oz 0,6437 oz 1,5436 oz

and the superplasticizer technology and dosage. More specifically the differences between a phosphonate based and a polycarboxylate ether (PCE)-based admixture will be described. RESEARCH SIGNIFICANCE This work points out some peculiar features of the flow of cementitious materials especially at low and high shear rates where, respectively, concrete placing and pumping are concerned. The influence of PCE and phosphonate superplasticizer technologies on such properties is studied and the advantages and drawbacks of each are discussed within a theoretical framework allowing an interpretation in terms of suspension microstructure. EXPERIMENTAL PROCEDURES Materials A CEM I 52.5N PM-ES Portland Cement (approximately equivalent to a Type V cement), silica fume, a 0/0.160 mm and a 0/0.315 mm sand were used. A PCE and a phosphonate based superplasticizer were used as aqueous solutions of 30% by weight of polymer in tap water with the addition of a suitable defoamer to prevent air entrainment. The phosphonate superplasticizer is a linear poly(ethylene oxide) diphosphonate bearing roughly 45 ethylene oxide units. The average structure of the PCE is a comb-shaped molecule of about 50 total methacrylic units out of which 20 bear 23-unit poly(ethylene oxide) grafts (40% grafting ratio). Equivalent grout mix design The concrete mix-design was scaled down through the use of an approach inspired by multiscale studies previously published.10 It relies upon applying a cutoff to the concrete grading curve at an arbitrary particle size—in our case, 315 µm. After normalizing to 100% passing, a target grading curve is obtained which is then matched as closely as possible by a blend of the binders and fine sands. This methodology may be considered as a way to simulate the grout surrounding the largest aggregates in the concrete, while allowing working in a rheometer where the sample is sheared in a very small gap, usually of the order of several millimeters. Two mixture proportions were used in the study: the dry composition of Grout Mix #1 is summarized in Table 1 while the different water amounts used are displayed in Table 2. This grout mixture was computed according to the procedure described above, based upon a civil engineering concrete composition from a French jobsite. The proportions of the second Grout Mix #2, with a very low water-to-binder ratio are provided in Table 3. This exceptional design simulates a civil engineering concrete from an Asian jobsite, and will illustrate quite extreme flow features.

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 79 Table 2–Water amounts for the investigated solid volume fractions for Grout mix #1 Target φ value 0.555 0.580 0.606 0.655

Water amount (ml) 144.21 ml 129.79 ml 116.81 ml 94.62 ml

Water amount (fl oz) 4.88 fl oz 4.39 fl oz 3.95 fl oz 3.20 fl oz

Table 3-Grout Mix #2 – Asian mix Component Type V Cement Fly Ash Silica Fume 0/0.315 Sand Total Water

Weight (g) 272.36 44.63 24.85 58.15 71.01

Weight (oz) 9.61 oz 1.57 oz 0.88 oz 2.05 oz 2.51 oz

Procedures Water and admixtures were weighed in a Krups YY8506FD mixer bowl; the dry powders were added during the first 30 seconds of mixing at speed 1 with a leaf-shaped blade. The mixing speed was increased to speed 7 for 1 minute and then stopped for 30 seconds (to scrape the sides of the bowl) before applying a last mixing stage of 1 minute at speed 7. The sample was then loaded in the cylinder bob of a Kinexus Pro rheometer (Malvern Instruments, U.K.) equipped with a vane geometry (4 mm side gap width, 1 mm bottom gap). For applied shear rate measurements, the procedure started five minutes after the beginning of mixing with a pre-shear at 200 s-1 during one minute, followed by logarithmic shear rate steps from 200 to 0.1 s-1. Each stress data point was sampled after the steady state was reached whenever possible in order to build the flow curve. At the same time, minislump tests at 5 minutes were performed (cone dimensions: upper diameter 18 mm – 0.71 in, lower diameter 36 mm-1.42 in, height 54 mm-2.13 in equipped with a pneumatic lifting fork for reproducible results). For applied stress protocols, the only changes are a pre-shear at 100 Pa, the stress step program being logarithmic between 100 and 0.1 Pa. After the flow curve measurement, the structure is reset to zero with an oscillating shear period during one minute with a strain amplitude of 100% and a 1 Hz frequency. Then a constant stress of 6.5 Pa is applied to the material in order to observe the structure buildup close to rest through the increase of viscosity with time. This applied stress value was chosen to match the stress applied by the weight of the largest aggregate in the system, according to the following rough calculation:

σ0 ≈

∆ρgd / 2 (1) 10.5

Δρ: Specific weight difference between the falling aggregate and the suspending fluidg: gravity constant, 9.81 m/s2d: particle diameter

80 SP-302-06

Eq. 1 yields approximately 6.5 Pa for a diameter of 16 mm, a paste density of 1800 kg/ m3 (112.4 lb/ft3) and an aggregate density of 2600 kg/m3 (162.3 lb/ft3). This last section of material testing are called ‘ageing stage’ in the rest of the paper. ANALYTICAL INVESTIGATION Since suspension structure in the flow regime is to be discussed, a model developed explicitly with structure level and solid volume fraction parameters was chosen. The discussion is based on the works of Quemada11-14 who considered the suspension as a dispersion of so-called Structural Units (SU), aggregates of grains each containing a number fraction S of individual particles, S being a function of shear rate and time. The resulting equation for the apparent viscosity follows a Krieger-Dougherty-like law, the solid volume fraction being replaced by an effective volume fraction11: −2



 φeff  η = ηF  1 − (2) φ m  

ϕeff: effective volume fractionϕm: maximum packing fractionηF: suspending fluid viscosity This effective volume fraction may be defined as a function of S and the compactness C of the SUs:

ϕeff = (1 + CS)ϕ

(3)

ϕ: Solid volume fractionC: SU compactness The last step is to define a behaviour law for S by stating that its value results from a dynamic equilibrium between aggregation and deflocculation by shear. When the system reaches a steady state, an expression for S is given:

Seq =

S0 + S∞ θ (4) 1+ θ

θ: function of shear rate which may take several forms depending on the complexity of the phenomena to model. In its simplest form (shear-thinning), it reads:

θ = (tc γ ) p (5)

tc: characteristic time for deflocculation by shear γ : steady shear ratep: exponent for deflocculation by shear, lower than 1 and shown experimentally to often take values close to ½. This theoretical development provides a shear rate dependence of the effective volume fraction through Eq. 3, Eq. 4 and Eq. 5. In its simplest form the model requires no less than five individual parameters, namely C, S0, S∞, tc and p given that ϕ and ϕm are suspension parameters determined by the chosen water/solid ratio and the dry material particle size distribution. A simpler analysis may

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 81

Fig. 1- Example of the general shape of flow curves be performed though, by a simple transformation of the data according to the method described below. COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS From flow curve data, apparent viscosity as a function of shear rate may be extracted. The first level of analysis is quite straightforward—from Eq. 1 comes:

φeff φm

 η( γ )  = 1−    ηf 

−1/ 2

(6)

This transformation allows plotting the ratio of the effective volume fraction to the maximum packing fraction as a function of shear rate—this allows an observation of the change in relative structure under shear. In the present paper it was chosen not to extend the analysis into further stages since data provided by Eq. 6 already describes the microstructural evolution of the suspension. EXPERIMENTAL RESULTS AND DISCUSSION General features of the flow of cementitious suspensions An illustration of typical experimental flow curves obtained as close to steady state as possible is provided in Fig. 1. On Fig. 1(a) was plotted the measured stress vs the applied shear rate; the shape of the curve is counter-intuitive in the sense that the general consensus stands upon an ever-increasing curve when shear rate increases. In numerous cases during this study, such U-shaped curves are obtained, the minimum of which defines a critical stress σc and critical shear rate γ c. The origin of such a shape will be discussed below but has also been reported in the past8,15 and is discussed more precisely in the paper SP-089 of the present conference.

82 SP-302-06

Fig. 1(b) shows the same data plotted in terms of apparent viscosity vs shear rate. After the expected shear-thinning behaviour (decreasing viscosity branch), a viscosity minimum ηmin is also observed, beyond which the suspension enters a shear-thickening regime. The critical shear rate defining the onset of this behaviour was noted γ st , the ‘st’ index standing for ‘shear thickening’. As already stated, these features are quite ubiquitous among the cement suspensions studied by the authors of this paper, the only differences from one system to another being the values obtained for the four critical parameters described above. Low shear limit: the influence of suspension ageing As previously stated, Fig. 1 shows an unexpected non-monotonous trend in shear stress vs shear rate. Some previous work seems to point at the influence of ageing or thixotropy; an analysis by Roussel et al8 shows that the introduction of a time-dependence component in a flow curve equation leads to U-shaped flow curves (see also paper SP-089). A more general theoretical approach developed by Picard et al15 also concludes that non monotonous curves are expected. Both approaches rely upon the coupling of a general flow equation with a differential equation providing a time-dependence of an underlying parameter (structure level for Roussel et al., fluidity for Picard et al.) showing that structure evolution through time is involved. A closer look to the raw data recorded here allows observing such an influence. As a matter of fact, all data points below γ c appear to be flagged with a ‘non-steady state alarm’ by the rheometer software. This means that the data points were recorded despite the fact that no steady torque value was achieved for the requested shear rate. Fig. 2 shows the transient data in terms of steady-state index, a value supposed to reach one when the steady state is achieved. Fig. 2(b) shows that when the applied shear rate is higher than the critical shear rate, steady flow is achieved in mere seconds whereas an applied rate lower than the critical value induces an oscillating behaviour with an increasing amplitude as seen on Fig. 2(a), preventing the rheometer to establish a steady flow. Roussel et al8 explain that below critical stress or rate no steady flow may indeed be achieved anymore and the suspension ages (i.e. its structure level increases), though slower than at complete rest. It may be argued that since no homogeneous or steady flow is possible anymore, the critical stress acts as an apparent yield stress. Some further insight about this interesting flow feature may be brought up by a stress step protocol applied to a freshly mixed batch of the same suspension. Fig. 3 shows the flow curves obtained by applying a decreasing stress step program, then an increasing stress step program with a steady-state condition. The obvious observation is that the flow curves do not have the same shape as the applied shear rate curve. At relatively high applied stress where flow is established all data sets are consistent. The decreasing stress stage shows a critical stress below which shear rate suddenly decreases by almost five decades to reach very low values in the range 10-5-10-4 s-1. The subsequent increasing stress protocol features the same rate jump, but for a higher stress. There appears a hysteresis loop around a low value of stress that is a result of the competition between shear and ageing, as supported by the work by Picard et al,15 quoted below:

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 83

Fig. 2- Steady state index during the transient periods right after the setting of a new shear rate value. (a) below critical shear rate. (b) above critical shear rate.

Fig. 3- Decreasing then increasing stress steps applied to the suspension near the critical stress. Shear rate jumps over several decades are observed for different applied stresses depending on the shear history. ‘When [stress] is imposed, the flow curve jumps in a hysteretic fashion between two branches: one that corresponds to no flow but for a wall layer in the vicinity of the walls; the other corresponds to a fully fluidized situation. When the [global shear rate] is imposed, shear banding can occur, as well as sometimes a stick-slip like oscillating behaviour at small shear rate that corresponds to a localized oscillation of the fluidity.’ This description based upon the assumption of a time-depending evolution is then totally consistent with the observations presented in this paper. It was shown previously9 and is shown in paper SP-089 in the same book that superplasticizer nature and dosage both influence the values of critical stresses and shear rate, with a noticeable effect on concrete flow during casting.

84 SP-302-06

Fig. 4- Influence of solid volume fraction on the shape of flow curves. Grout mix #1. PCE superplasticizer with a 0.65% dosage. Here the influence of suspension solid volume fraction φ was studied by gradually decreasing the amount of water in the mix. Fig. 4 shows that an increase of solid volume fraction at a constant PCE superplasticizer dosage of 0.65% by weight of binder yields both an increase of apparent viscosity throughout the observed shear rate range and an increase in critical shear rate and stress, a sign of a faster ageing. Ageing itself was studied as described above under a constant stress of 6.5 Pa. The results are plotted in Fig. 5, which shows for φ = 0.555 a rather slow viscosity increase through time, with a sharper slope towards 30 min corresponding to gelling. For φ = 0.580, the flow period before gelling is shorter while the upper two φ values yield gelling in less than a minute. There seems to be a connection between the values of σc and γ cand the ageing kinetics of the suspension since they feature the same trend. It may then be stated that suspension ageing is responsible for the flow behaviour at low shear rates, the above parameters defining a boundary between homogeneous flow and heterogeneous or impossible flow. This contributes to showing how suspension ageing through hydration is closely connected to the low-shear-rate flow characteristics. Steady flow regime: effective volume fraction flow curves, admixture robustness As stated above, the flow regime is composed of a shear-thinning domain followed by a shear-thickening regime beyond a shear rate γ st . It seemed interesting to compare flow curves of both superplasticizers at equal workability, which is the purpose of Fig. 6. In terms of viscosity, there appears a crossover between the diphosphonate and the PCE technology at around 4 s-1, beyond which shear thickening is more intense for the PCE. Fig. 7 corresponds to the same data transformed into the ratio of effective volume fraction to maximum packing fraction through the use of

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 85

Fig. 5- Ageing curves in rotational protocol as a function of solid volume fraction. PCE superplasticizer. 0.65% by weight of total binder.

Fig. 6- Comparison of flow curves at 5 min obtained with both technologies at equivalent workability Eq. 6. It may be observed that the phosphonate technology indeed allows a lower structure level of the suspension, thus a higher amount of deflocculation, beyond 10 s-1, i.e. for shear rates corresponding to pumping or mixing. Interestingly, shear thickening is here interpreted as an increase of structure with an increase of shear rate, i.e. shear is here considered as a flocculation mechanism for example through the formation of hydroclusters.12,13 Shear thickening may interfere with operations at high shear rate, i.e. pumping, and these results show the advantage of a phosphonate technology over a PCE technology on this property. The tradeoff is a much higher dosage, thus a higher cost and a possible influence on the early strength.

86 SP-302-06

Fig. 7- Flow curves of Fig. 6 transformed according to Eq. 6. Grout mix #1. φ = 0.555. Another important admixture property is its robustness to the water amount. Any inaccuracy in the weighing of water or the moisture measurement of aggregates may lead to a variation in initial workability. Water reduction at constant dosage was investigated, leading to an increase of the solid volume fraction φ. The data recorded at 5 min, expressed in terms of relative structure through Eq. 6, are shown in Fig. 8. There is an obvious difference between the two technologies; while the phosphonate is able to keep a consistent structure level in the 0.1-10 s-1 range despite the increase in φ, the PCE is unable to do the same and the structure degree increases with φ at all shear rates, leading to a loss of workability. It may then be concluded that the phosphonate admixture is more robust to water variations, the tradeoff being once again a much higher dosage. Another interesting feature of the phosphonate experiments is that the shear thickening effect seems to intensify when φ increases, showing that this phenomenon depends on the solid volume fraction. This feature will be further investigated in the next section. High shear limit: possible mechanisms at the origin of shear thickening The first mechanism that may explain shear thickening is the rise of inertial forces16,17 and this was investigated here by following the approach used by Brown and Jaeger16 who defined a suspension Reynolds number under the form:

Re =

ρl γd 2 (7) ηmin

ρl: liquid specific gravityd: particle diameterηmin: minimum viscosity defined on Fig. 1. Even by taking the diameter of the largest particle (315 µm), at the highest shear rate investigated (200 s-1), the Reynolds number of the suspension is no higher than 5.10-3. This shows that inertial effects may be neglected in the system. The high shear rate limit was investigated on Grout #2 the advantage of which is to require almost the same dosage for both technologies (2.0% for the PCE, 2.5% for the

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 87

Fig. 8- Relative structure levels of suspensions (Eq. 6) with a variable volume fraction phosphonate) at equal workability. This difference from Grout #1 comes primarily from a much lower water amount and the presence of fly ash in Grout #2. The phosphonate robustness may be stressed out again with a small dosage difference between grouts at the expense of a high overall dosage with respect to the PCE. Grout #2 features a much more intense shear thickening regime than Grout #1 as shown in Fig. 9, though the phosphonate superplasticizer still seems to mitigate shear thickening when compared to the PCE. Indeed the apparent structure level computed from Eq. 6 on Fig.9(b) shows obviously that the phosphonate induces a much lower degree of structure in the range 10 – 200 s-1, though shear thickening still appears towards 100 s-1. A final insight may be given about the possible origin of shear thickening. Though the vane geometry is not properly designed for such a purpose, normal force was measured during the experiments. Fig. 10 represents the evolution of the normal force exerted by the suspension onto the vane versus the recorded shear stress. A very sharp increase of normal force (up to 0.14 N) may be observed at high stress in the case of the PCE superplasticizer while for the phosphonate the normal force remains lower than 0.025 N in the same stress range. It was previously shown16-19 that the rise of contact forces between particles, especially of frictional nature, are often involved whenever such a coupling between shear and normal stresses is observed. This means that the occurrence of frictional regimes may be a cause of shear thickening in the observed suspensions. If such is the case, it would mean that the phosphonate adsorbed layer is able to mitigate friction between particles while the PCE adsorbed layer is less able to do so, leading to a sharper rise of friction, normal stresses and shear viscosity. This would be obviously linked to the structure of the adsorbed layers, which would be dependent upon the polymer structure. FURTHER RESEARCH If the low shear rate regime may be well understood in the framework of suspension ageing, there still remains some work to better understand the underlying mechanisms of

88 SP-302-06

Fig. 9- Comparison between apparent viscosity and apparent suspension structure on both technologies. Grout mix #2.

Fig. 10- Normal force applied on the geometry as a function of stress. Grout mix #2. the shear-thickening regime. If inertia may be ruled out, hydrocluster formation or the rise of contact or frictional forces at high shear rates may still be involved. Superplasticizer technology seems to have an influence on the phenomenon, but the exact molecular parameters controlling the performance are still to be determined. CONCLUSIONS This paper is dedicated to seldom-studied features of the flow of cementitious materials, beyond the often described shear thinning or Bingham behaviours. It was first shown that low-shear-rate regimes may lead to a discontinuous behaviour with a critical value of applied rate or stress below which no steady flow may occur. Previous work points at the influence of the ageing of the suspensions, through cement hydration, that interferes with

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 89 shear. The diphosphonate and the PCE technologies both behave in a similar manner in this regime. Beyond the main shear thinning regime, there appears a high shear rate regime of viscosity increase similar to what some authors define as continuous shear thickening.17,25 This is where the studied technologies differ, the phosphonate superplasticizer producing much lower viscosities in the 10-200 s-1 shear rate range. Elements from the literature and some experimental facts allow arguing that the underlying mechanism is related to the way the adsorbed phosphonate polymers mitigate contact forces between particles in a more efficient manner than the PCE, at the expense of its water reducing ability, which seems noticeably lower. AUTHOR BIOS Dr Lucia Ferrari is the Physical Chemistry manager in the main research and development laboratory of CHRYSO in France. She received her PhD from the Technische Universität München (Germany) after she completed her PhD work with the EMPA in Dübendorf (Switzerland) under the supervision of Dr Frank Winnefeld and Pr. Dr. Johann Plank. Dr Pascal Boustingorry is the Head of the Interface Physical Chemistry Team in the main research and development laboratory of CHRYSO in France. He received his PhD from the INP Grenoble along with the Ecole des Mines in Saint Etienne (France). Their main research interests are the interaction of organic molecules with cement suspensions and the links between superplasticizer chemical architecture and flow properties of building materials. REFERENCES 1. S, G., “S, G. “Determination of Bingham Parameters of Fresh Portland Cement Concrete Using Concrete Shear Box,” Bonfring International Journal of Industrial Engineering and Management Science, V. 2, No. 4, 2012, pp. 84-90. doi: 10.9756/BIJIEMS.1620 2. Wallevik, J. E., “Relationship between the Bingham parameters and slump,” Cement and Concrete Research, V. 36, No. 7, 2006, pp. 1214-1221. doi: 10.1016/j. cemconres.2006.03.001 3. Wallevik, O. H., and Wallevik, J. E., “Rheology as a tool in concrete science: The use of rheographs and workability boxes,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1279-1288. doi: 10.1016/j.cemconres.2011.01.009 4. Cyr, M.; Legrand, C.; and Mouret, M., “Study of the shear thickening effect of superplasticizers on the rheological behaviour of cement pastes containing or not mineral additives,” Cement and Concrete Research, V. 30, No. 9, 2000, pp. 1477-1483. doi: 10.1016/ S0008-8846(00)00330-6 5. Feys, D.; Verhoeven, R.; and De Schutter, G., “Why is fresh self-compacting concrete shear thickening?,” Cement and Concrete Research, V. 39, No. 6, 2009, pp. 510-523. doi: 10.1016/j.cemconres.2009.03.004 6. Raghavan, S. R., and Khan, S. A., “Shear-thickening response of fumed silica suspensions under steady and oscillatory shear,” Journal of Colloid and Interface Science, V. 185, No. 1, 1997, pp. 57-67. doi: 10.1006/jcis.1996.4581

90 SP-302-06

7. Toussaint, F.; Roy, C.; and Jézéquel, P. H., “Reducing shear thickening of cementbased suspensions,” Rheologica Acta, V. 48, No. 8, 2009, pp. 883-895. doi: 10.1007/ s00397-009-0362-z 8. Roussel, N.; Le Roy, R.; and Coussot, P., “Thixotropy modelling at local and macroscopic scales,” Journal of Non-Newtonian Fluid Mechanics, V. 117, No. 2-3, 2004, pp. 85-95. doi: 10.1016/j.jnnfm.2004.01.001 9. Ferrari, L.; Boustingorry, P.; Pineaud, A.; and Bonafous, L., From cement grout to concrete scale: a study of superplasticizer-design-controlled thixotropy to match SCC application requirements. in Rheology and processing of Construction Materials – 7th RILEM International Conference on Self-Compacting Concrete and 1st RILEM International Conference on Rheology and Processing of Construction Materials 285–292 (RILEM Publications, 2013). 10. Toutou, Z., and Roussel, N., “Multi scale experimental study of concrete rheology: from water scale to gravel scale,” Materials and Structures, V. 39, 2006, pp. 167-176. 11. Quemada, D. Rheological modelling of complex fluids. I. The concept of effective volume fraction revisited. Eur. Phys. J. AP 1, 119–127 12. Quemada, D. Rheological modelling of complex fluids: II. Shear thickening behavior due to shear induced flocculation. Eur. Phys. J. AP 2, 175–181 13. Quemada, D. Rheological modelling of complex fluids: III. Dilatant behavior of stabilized suspensions. Eur. Phys. J. AP 3, 309–320 14. Quemada, D. Rheological modelling of complex fluids: IV: Thixotropic and” thixoelastic”behaviour. Start-up and stress relaxation, creep tests and hysteresis cycles. Eur. Phys. J. AP 5, 191–207 15. Picard, G.; Ajdari, A.; Bocquet, L.; and Lequeux, F., “Simple model for heterogeneous flows of yield stress fluids,” Physical Review E: Statistical, Nonlinear, and Soft Matter Physics, V. 66, No. 5, 2002, p. 051501 doi: 10.1103/PhysRevE.66.051501 16. Brown, E. & Jaeger, H. M. The role of dilation and confining stresses in shear thickening of dense suspensions. Journal of Rheology (1978-present) 56, 875–923 (2012). 17. Brown, E. et al., “Generality of shear thickening in dense suspensions,” Nature Materials, V. 9, 2010, pp. 220-224. 18. Seto, R.; Mari, R.; Morris, J. F.; and Denn, M. M., “Discontinuous shear thickening of frictional hard-sphere suspensions,” Physical Review Letters, V. 111, No. 21, 2013, p. 218301 doi: 10.1103/PhysRevLett.111.218301 19. Brown, E., “Friction’s Role in Shear Thickening,” Physics, V. 6, 2013, p. 125 doi: 10.1103/Physics.6.125 20. de Gennes, P. G., “Conformations of polymers attached to an interface,” Macromolecules, V. 13, No. 5, 1980, pp. 1069-1075. doi: 10.1021/ma60077a009 21. Alexander, S., “Adsorption of chain molecules with a polar head a scaling description,” Journal of Physics, V. 38, No. 8, 1977, pp. 983-987. doi: 10.1051/ jphys:01977003808098300 22. Fleer, G. J., “Polymers at interfaces and in colloidal dispersions,” Advances in Colloid and Interface Science, V. 159, No. 2, 2010, pp. 99-116. doi: 10.1016/j.cis.2010.04.004 23. Fleer, G. J., and Skvortsov, A. M., “Reconciling lattice and continuum models for polymers at interfaces,” The Journal of Chemical Physics, V. 136, No. 13, 2012, p. 134707 doi: 10.1063/1.3693515

Influence of Superplasticizers on the Flocculation Degree of Cement Suspensions 91 24. Lyklema, H.; Stuart, M. C.; and Leermakers, F., “Gerard Fleer,” Advances in Colloid and Interface Science, V. 159, No. 2, 2010, pp. 95-98. doi: 10.1016/j.cis.2010.06.008 25. Fall, A., Lemaıtre, A., Bertrand, F., Bonn, D. & Ovarlez, G. Continuous and discontinuous shear thickening in granular suspensions. Physical Review Letters (2010). tables and figures List of Tables:

92 SP-302-06

SP-302-07

Influence of Temperature and Retarder on Superplasticizer Performance by Karen Luke and Adrian Torres Superplasticizers are often used in conjunction with other additives and this can produce either an adverse or synergistic effect on rheology and setting properties of cementitious systems. These effects can be enhanced when temperatures are increased due to environmental changes or induced temperature as in hydrothermal curing. This research focuses on the compatibilities of different types of superplasticizer either sulfonated naphthalene or polycarboxylate based in combination with a lignosulphonate or hydroxycarboxylic acid type retarder. Rheological measurements were made using a rotational viscometer at temperatures from 25°C (77°F) to 120°C (248°F) under pressure, and plastic viscosity and yield point determined based on the Bingham Plastic model though in almost all cases it was noted that the Power Law or more so the Herschel-Buckley model gives a better fit. Zeta potential was used to characterize particle surface interactions to understand synergy of additive combinations. Setting properties, investigated using conduction calorimetry, were observed to be dominated by retarder response. Keywords: Calorimeter; polycarboxylate; polynaphthalene; sulfonated acetone formaldehyde; rheology; retarder; temperature; zeta-potential. INTRODUCTION Superplasticizer technology and utilization has advanced notably since its early development (Hattori, K., 1978) in the late 1960’s. There is an abundance of literature on the use of superplasticizer as a high performance concrete (Aitcin, P.-C., 1998), with compressive strengths nowadays exceeding 100 MPa (14,503 psi), and also on high-workability concrete or more recently self-compacting concrete (Walraven J.C., 2010). The benefits of high performance concrete are well known where water to binder ratios as low as 0.2 can be achieved without reducing the consistency giving low permeability, high compressive strength, elastic modulus and flexural strength, as well as improved abrasion resistance and durability. Self-compacting concrete technology has advanced considerably with the invention of modern superplasticizers based on polycarboxylate polymers and coupled with advances in the use of stabilizing additives (Schmidt, W., 2014). Superplasticizers are often used in combination with other additives (Michaux, M., et al., 1986, 1986a) and the 93

94 SP-302-07

effects, adverse of synergistic, are notably influenced by temperature either due to environmental factors or hydrothermal conditions as obtained in cementing of oil and gas wells (Nelson, E.B. et al., 2006). Although publications on superplasticizers in well cementing are limited they are used in a variety of applications to improve mixability and/or flow characteristics of the cement slurry (mixture of cement, water and additives). Superplasticizer addition along with particle packing methodology is used in designing low density, 1020 kgm-3 (63.7 lb/ft3) to about 1450 kgm-3 (90.5 lb/ft3), slurries that are also required to provide high strength and low permeability, achieved by using a high solid fraction consisting of a light weight extender. In normal density slurries, 1700 kgm-3 (106.1 lb/ft3) to 1900 kgm-3 (118.6 lb/ft3), superplasticizers are used to improve slurry mixability and reduce friction pressures, particularly where small annular spaces may be encountered and high friction pressures may cause fracture to weaker formations. Densities above 2000 kgm-3 (124.9 lb/ft3) where water to solid ratios are less than 0.4, superplasticizers are used to give flowable slurries. Such slurries are used for abandonment plugs that are required to have very low permeabilities or for whipstock plugs used for directional drilling where the cement slurry should have adequate time for placement, develop strength rapidly and be stronger than the surrounding formation. Superplasticizers are also often used in combination with fluid loss additives, where they improve cement particle dispersion and aids in optimal fluid loss performance. In addition many fluid loss additives are viscosifying and the addition of superplasticizer aids with mixability and flowability. Rheological parameters, obtained on a coaxial rotational viscometer (API RP 10B, 2010), are used to provide predictions of friction pressure and slurry flow properties through pipes, casing and annulus using appropriate mathematical relationships (Guillot, D., 2006). Current technology has limited rheological test conditions for cement slurries to 80°C (176°F) and atmospheric pressure whereas under well conditions the slurry can be exposed to temperatures up to 320°C (608°F) and pressures of 280 MPa (40,610 psi). In addition, testing under conditions specified by API RP 10B at 60°C (140°F) to 80°C (176°F) the slurry can potentially viscosify due to evaporation of the aqueous water phase during the conditioning phase. Recent developments in rotational viscometers allows cement slurry rheological parameters to be measured up to 316°C (600°F) and 207 MPa (30,023 psi) and eliminates potential issues of aqueous phase evaporation of the slurry. This study provides a preliminary investigation using the high temperature, high pressure (HTHP) rotational viscometer to determine in the first instance the effect of three commercial superplasticizers on the rheological parameters from 25°C (77°F) to 120°C (248°F) and 10.0 MPa (1450 psi) in combination with two different retarder types and an antifoam additive. Additional testing was also performed using zeta potential measurements and calorimetry to further elucidate on the effects of the superplasticizer, retarder and antifoam combinations. RESEARCH SIGNIFICANCE Rheological data obtained to date for determining flow properties and friction pressures for well applications are typically measured at atmospheric pressure and at temperatures in the range of 25°C (77°F) to 80°C (176°F). At 60 °C (140°F) to 80°C (176°F) there is potential for viscosification of the slurry due to evaporation during the conditioning period of the test. Development of a HTHP rotational viscometer that can measure rheologies at high

Influence of Temperature and Retarder on Superplasticizer Performance  95

temperatures and pressures is believed to give more accurate values on friction pressures and slurry flow properties of the cement slurry under well conditions. The significance of this research is in elucidating the rheological properties of different types of superplasticizer when used in combination with retarder and antifoam at temperature and pressure more realistic of actual well conditions. EXPERIMENTAL PROCEDURE Rheological experiments were carried out on three different types of superplasticizer, a sulfonated naphthalene formaldyehyde condensate, a sulfonated acetone formaldehyde condensate and a carboxylated polymer, at temperatures from 25°C (77°F) to 120°C (248°F) in a high temperature, high pressure (HPHT) rotational viscometer. Two different retarders were used, one for low temperature less than 100°C (212°F) and the second for high temperature equal to or greater than 100°C (212°F). Two different retarders were used as in field applications in this temperature range the reaction rate of cement has an exponential effect on retarder response. The result is that retarders effective below 100°C (212°F) are ineffective above 100°C (212°F) and effective above 100 °C (212°F) too powerful below 100°C (212°F) even at low concentrations. In addition small changes in the concentration of HC-R below 100°C (212°F) can cause large variations in thickening time. This in field applications can lead to premature set in the casing, or long waiting on cement to set times that can result in costly rig time and in laboratory testing can cause variable data and non-reproducible results. An antifoam additive, common in most well slurries to minimize foaming during mixing (Nelson, E.B. et al. 2006), was included in all tests. Correlation of data was made in relation to Bingham Plastic, Power Law and Herschel-Buckley fluid models. Additional measurements were also made using calorimeter and zeta-potential to better define the performance of the different types of superplasticizer, retarder combinations. Materials A commercially available Class G cement conforming to API Specifications (API Specification 10A, 2011) was used in this study. Physical, chemical and mineralogical data are provided in Table 1. Three different chemical categories of superplasticizers were investigated, a sulfonated polynaphthalene formaldehyde condensate (PNS), a sulfonated acetone formaldehyde condensate (PAS) and a polycarboxylate polymer (PC). The superplasticizers were in powder form, with the PNS and PAS powders containing about 10% sodium sulfate non-reacted material from the manufacturing process whereas the PC powder consisted of 20% by weight of polymer and 80% by weight of an added inert material. As such the PNS and PAS powders contained 90% polymer component while the PC contained only 20% polymer component. Retarders used were, a lignosulfonate (LS-R) for temperatures less than 100°C (212°F) and a hydroxycarboxylic acid (HC-R) for temperatures at and above 100°C (212°F). A glycol polymer antifoam (PG-A) was also used in the slurry blends in order to minimize foaming during blending. All the additives were commercially available. Reversed osmosis (RO) water was used for all testing.

96 SP-302-07

Table 1–Physical and chemical composition of Class G cement Class G Specific gravity Blaine fineness, cm2/g Loss of ignition, % SiO2% CaO % Al2O3% Fe2O3% MgO % SO3% Na2O % K2O % Na2Oe % Phase mineralogy C3S C2S C3A C4AF

3.14 2919 1.14 20.69 62.25 3.54 4.29 2.70 2.97 0.32 0.52 0.67 57.74 15.84 2.12 13.03

Table 2 – Slurry compositions Temperature °C (°F)

Slurry Components

20 – 80 (77-176)

100 – 120 (212-248)

1

2

Blend 3

4

5

Cement Superplasticizer (% bwoc) LS-R (% bwoc) PG-A (% bwoc)

100 0.0 0.5 0.2

100 0.2 0.5 0.2

100 0.4 0.5 0.2

100 0.6 0.5 0.2

100 0.8 0.5 0.2

Cement Superplasticizer (% bwoc) HC-R (% bwoc) PG-A (% bwoc)

6 100 0.0 0.5 0.2

7 100 0.04 0.5 0.2

Blend 8 100 0.1 0.5 0.2

9 100 0.2 0.5 0.2

10 100 0.4 0.5 0.2

Test Methods A dry blend of additives and cement was mixed with water at a water to cement ratio of 0.44, in a Waring blender according to API Recommended Practice (API RP 10B-2, 2010) to give a 600 mL slurry. Additives were added as a percentage by weight of cement (% bwoc). The compositions of the blends are provided in Table 2. All slurries had a density of 1904 kgm-3 (118.6 lb/ft3) and yield of 0.757 m3/t (0.012 ft3/lb). Aliquots of the blended slurry were then taken and used for either rheological or calorimetric studies. Rheological studies The slurry after blending was conditioned in an atmospheric consistometer at 25°C (77°F) for 20 minutes. An aliquot of approximately 150 mL was then placed in the measuring cup

Influence of Temperature and Retarder on Superplasticizer Performance  97

of a Chandler 7500 HPHT rotational viscometer. Samples were heated to the required temperature, 40°C (104°F), 60 (140°F), 80 (176°F), 100°C (212°F), or 120°C (248°F) under a pressure of 10 MPa (1,450 psi). Ramp time to reach the required temperature and pressure was in all cases 45 minutes. At the selected temperature data were recorded during both a ramp- up and ramp-down at 3, 6, 100, 200, and 300 RPM with the speed held for one minute at each RPM setting. The total time from blending to the first measurement was an average time of 70 minutes. Calorimetric studies Approximately 2.5 mL of the 600 mL slurry was syringed into a pyrex flat bottomed tube (6 mm diameter x 5 cm), and placed into a high pressure stainless steel Setaram C80 calorimeter cell supplied by Setaram Instrumentation, France. The cell was sealed and placed into the calorimeter simultaneously with the reference cell. The reference sample used for tests at 80°C (176°F) and below, consisted of a LS-R retarded Class G cement previously cured at 80°C (176°F) for 24 hours on the assumption that the thermal capacity matched, as closely as possible, the actual experiments (Aukett and Bensted, 1992). At temperatures greater than or equal to 100°C (212°F), the reference sample was a HC-R retarded Class G cement cured at 120°C (248°F) for 24 hours. Nitrogen was applied to each cell at a pressure of 7.5 MPa (1,087 psi) using a high pressure gas control panel. Heat was applied according to API schedules (API RP 10B, 2010) for appropriate temperature and then maintained at temperature for 24 hours. Ramp time for 25°C (77°F) to 40°C (104°F) and 25°C (77°F) to 120°C (248°F) was 22 minutes, and 60 minutes respectively. The benefit of using a ramp simulating well conditions versus isothermal conditions has been discussed (Aukett, P.N. and Bensted, J. 1992). Zeta Potential Zeta potential of dilute suspensions, 200 mgL-1, solid material in RO water or pore solution extracted from cement slurry after approximately 70 minutes hydration, was measured in a Horiba, Nano-particle analyzer SZ 100. The solid material consisted of Class G cement or Class G cement plus 0.5% bwoc LS-R or HC-R and 0.2% bwoc PG-A consistent with slurry compositions used for rheology and calorimeter studies. PNS, PAS or PC was added at a concentration of 0.4% bwoc to the LS-R cement suspensions and 0.1% bwoc to the HC-R cement suspensions and zeta potential data obtained after approximately 1 hour. EXPERIMENTAL RESULTS AND DISCUSSION Data analysis and rheological models Bingham Plastic, Power Law and Herschel-Bulkley mathematical fluid models are commonly used to define the flow properties of well cement slurries. Of the three models the Bingham Plastic model as described in equations (1) and (2) has been the most widely used.

τ = τy + μp γ when τ > τy (1)



γ = 0 when τ ≤ τy (2)

98 SP-302-07

where: τ = shear stressτy = yield stressµp = plastic viscosityγ = is the shear rate However, the shear-stress/shear-rate relationship appears more consistent with the Power Law model or more the Herschel-Bulkley model as illustrated in equation (3) that also includes a yield stress component, and is common in concentrated suspensions such as obtained in cement slurries.

τ = τy + kγn when τ > τy (3)

where: τ = shear stressτy = yield stressγ = is the shear ratek = consistency indexn = Powerlaw index When n is less than 1, Herschel-Bulkley fluids are shear thinning, while n greater than 1 is shear thickening. At n equals 1 the Herschel-Bulkley model reduces to the Bingham Plastic model. At present there is no defined rational on whether to use the Herschel-Bulkley or Bingham Plastic model other than choosing the model that appears to give the best fit to the rheological data. Shear stress/shear rate plots were determined for all slurries and compared with the Bingham Plastic or Herschel-Bulkley models. Figure 1 shows the shear stress/shear rate plots obtained from the rheological data of selected slurries and illustrates how the best fit model, Bingham Plastic or Herschel-Bulkley, is selected in the current study. Figure 1a and b shows the data from the shear stress/shear rate plots for the 0.2% bwoc PAS slurry at 25°C (77°F), 10 MPa (1,450 psi) illustrating that the best fit in this case is the Herschel-Bulkley model. Figure 1c and d, shows data from 0.5% bwoc PNS at 80°C (176 °F), 10 MPa (1,450 psi) where either the Bingham Plastic or Herschel-Bulkley model give similar fit. Figure 1e and f, 0.4% bwoc PAS at 40°C (104°F), 10 MPa (1450 psi), provides an example where n equals 1 and the Herschel-Bulkley model reduces to the Bingham Plastic model and as would be expected both plots give the same linear slope and yield stress. The change in the shear stress/shear rate relationship and hence choice of model for the different slurries can result from shear effects, pressure and/or temperature effects, additive chemistry and/or additive concentration. Rheological profiles and effect of superplasticizer at 25-80°C (77-176°F) The rheological data obtained for the superplasticizers investigated with the LS-R retarder in combination with the antifoam PG-A at 25°C (77°F) to 80°C (176°F), is provided in Table 3, Table 4, Table 5, and Table 6. Data shows that slurries with low concentrations of superplasticizer, 0.2% bwoc or less, follow the Herschel-Bulkley model and notably have an n value significantly less than 1.0 and is indicated by the shaded sections in the tables. As concentrations increase above 0.2% bwoc superplasticizer slurries behave either according to the Bingham Plastic or Hershel-Bulkley models as illustrated in Fig 1 c and d, and notably as n increases from about 0.6 to around 1.0. In a number of cases such as for 0.5% bwoc PNS at 40°C (104°F) a value of n greater than 1, is observed. This is even more evident at 60°C (140 °F), not only for the PNS but also PAS indicating shear thickening behavior of the slurry, Figure2. The explanation for this behavior, remains unclear. It is noted that for the 0.6% bwoc PNS at 60°C (104°F) slurry with n equal to 1.186 that both the Bingham Plastic and Herschel-Bulkley models give a good fit to the data, Figure 2 a and b whereas with the 0.5% bwoc PNS at 40°C (104°F) and n equals 1.274 it is clear that the Herschel-Bulkley model is a better fit, Figure 2d and e. At values of n above 1.274

Influence of Temperature and Retarder on Superplasticizer Performance  99

Fig 1.Shear stress/shear rate plots showing a)Bingham Plastic model, b)Herschel-Bulkley model(n = 0.077) for 0.2% bwoc PNS slurry at 25°C(77°F), c)Bingham Plastic model, d)Herschel-Bulkley model (n = 0.500) for 0.5% bwoc PNS at 80°C(176°F)and e) Bingham Plastic model, f)Herschel-Bulkley model(n = 1.01) for 0.4% bwoc PAS at 40°C(104°F). the fit to the Herschel-Bulkley model was even more evident. Yield stress values for the slurries having n less than 1, and using the Herschel-Bulkley model, generally tend to be lower than those calculated from the Bingham Plastic model whereas for n approximately equal to 1.0, values are similar. It should be noted, however, with complex fluids such as cement slurries in-homogeneities can occur due to particle settling, slip or time-dependent behavior that can create conditions for erroneous measurement particularly at low end shear rates and makes model fitting problematic (Sairam, P.K.S., et al., 2012). Values taken for rheological studies were an average of the ramp-up and ramp-down readings. These are in general very close, though it was noted in the present study that values increased significantly during ramp-down, for all superplasticizers, at 80°C (176°F) showing an

100 SP-302-07

Table 3–Rheological properties at 25 °C (77°F) and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.2 (0.18)* 0.4 (0.36) 0.5 (0.45) 0.6 (0.54) 1.0 (0.90) 0.0 0.2 (0.18) 0.4 (0.36) 0.6 (0.54) 0.8 (0.72) 0.0 0.2 (0.04) 0.4 (0.08) 0.6 (0.12) 0.8 (0.16)

Bingham Plastic PV (cP) YP (Pa) 38.0 23.3 19.6 23.0 13.8 12.8 16.4 4.42 16.4 0.83 14.3 0.80 38.0 23.3 19.6 21.8 14.1 1.62 17.2 1.07 17.6 0.24 38.0 23.3 17.8 9.13 15.2 3.02 13.0 1.54 19.1 1.03

n 0.161 0.077 1.116 0.817 0.825 0.850 0.161 0.304 0.730 0.827 0.915 0.161 0.905 0.776 0.818 0.869

Herschel-Bulkley Yield stress (Pa) K (Pasn) 12.72 17.0 19.12 15.0 0.006 12.9 0.048 4.15 0.047 0.57 0.035 0.61 12.72 17.0 1.890 17.5 0.072 1.22 0.048 0.80 0.023 0.00 12.72 17.0 0.031 8.99 0.059 2.69 0.039 1.32 0.041 0.82

* () – polymer component

Table 4–Rheological properties at 40 °C (104°F) and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.2 (0.18)* 0.4 (0.36) 0.5 (0.45) 0.6 (0.54) 1.0 (0.90) 0.0 0.2 (0.18) 0.4 (0.36) 0.6 (0.54) 0.8 (0.72) 0.0 0.2 (0.04) 0.4 (0.08) 0.6 (0.12) 0.8 (0.16)

Bingham Plastic PV (cP) YP (Pa) 48.6 26.0 40.9 39.2 17.2 23.3 10.0 10.6 10.5 2.74 10.6 1.91 48.6 26.0 18.2 27.6 14.1 4.73 15.0 1.63 16.0 0.57 48.6 26.0 21.7 17.8 11.9 5.77 9.50 2.37 11.9 2.42

n 0.163 0.105 0.412 1.274 0.803 0.613 0.163 0.202 1.010 0.731 0.847 0.163 0.167 0.916 0.744 0.916

Herschel-Bulkley k Yield stress Pa) 17.05 18.0 29.69 17.5 0.302 22.2 0.002 10.8 0.035 2.54 0.116 1.38 17.05 18.0 4.050 20.9 0.013 4.75 0.077 1.21 0.40 0.35 17.05 18.0 6.78 7.58 0.019 5.69 0.046 2.07 0.019 2.34

* () – polymer component

approximately 20% difference at the lower shear rates. It is assumed that this relates to gelation effect where PNS > PC > PAS as indicated by the apparent plastic viscosity and yield point values. This was also confirmed by the thickening time (API RP 10B-2 (2010)

Influence of Temperature and Retarder on Superplasticizer Performance  101

Table 5–Rheological properties at 60 °C (176°F)and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.2 (0.18)* 0.4 (0.36) 0.5 (0.45) 0.6 (0.54) 1.0 (0.90) 0.0 0.2 (0.18) 0.4 (0.36) 0.6 (0.54) 0.8 (0.72) 0.0 0.2 (0.04) 0.4 (0.08) 0.6 (0.12) 0.8 (0.16)

Bingham Plastic PV (cP) YP (Pa) 83.9 51.8 50.5 48.5 3.8 38.0 2.6 32.2 4.6 14.3 10.2 4.39 83.9 51.8 23.1 32.7 6.7 9.46 9.0 3.47 15.9 2.17 83.9 51.8 33-3 25.6 6.3 10.2 7.4 5.50 7.6 3.60

n 0.149 0.106 1.290 0.880 1.186 1.240 0.149 0.523 1.326 0.778 0.743 0.149 0.121 0.817 1.005 0.979

Herschel-Bulkley Yield stress (Pa) k Pasn) 38.83 40.0 36.58 38.0 0.007 38.1 0.006 32.1 0.001 14.4 0.002 4.54 38.83 40.0 0.493 30.2 0.001 9.58 0.035 3.26 0.767 1.73 38.83 40.0 18.77 21.0 0.020 10.0 0.007 5.49 0.008 3.57

* () – polymer component

Table 6–Rheological properties at 80 °C (176°F) and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.2 (0.18)* 0.4 (0.36) 0.5 (0.45) 0.6 (0.54) 0.0 0.2 (0.18) 0.4 (0.36) 0.6 (0.54) 0.8 (0.72) 0.0 0.2 (0.04) 0.4 (0.08) 0.6 (0.12) 0.8 (0.16)

Bingham Plastic PV (cP) YP (Pa) 117.0 88.9 95.3 66.8 75.5 68.2 24.9 48.6 2.0 18.0 116.4 89.2 39.5 44.4 3.9 14.5 5.0 13.8 5.7 12.4 116.1 89.3 58.6 47.4 17.8 28.3 17.5 17.7 9.6 19.1

n 0.174 0.118 0.106 0.500 1.014 0.174 0.241 0.533 1.322 1.312 0.168 0.113 1.484 1.529 0.477

Herschel-Bulkley Yield stress (Pa) k (Pasn) 33.65 70.0 52.23 50.0 51.82 37.5 43.70 46.0 0.002 18.0 33.46 41.3 6.252 33.0 0.073 14.2 0.001 13.9 0.001 12.5 35.50 75.0 35.70 39.0 0.001 28.7 0.0004 18.7 0.276 17.9

* () – polymer component

plots, Figure3, where at 70 minutes, Bearden units of consistency (Bc) values were, 20 Bc, 15 Bc and 11 Bc for PNS, PC and PAS respectively. These plots also indicated different profiles for each superplasticizers in building up consistency to 70 Bc where the slurry is

102 SP-302-07

Fig 2. Shear stress/shear rate plots showing a)Bingham Plastic model, b)Herschel-Bulkley model (n = 1.186) for 0.6% bwoc PNS slurry at 60°C(140°F), and c)Bingham Plastic model, d)Herschel-Bulkley model (n = 1.274) for 0.5% bwoc PNS at 40°C(104°F). considered no longer pumpable. Bc unit is a dimensionless quantity with no direct conversion factor to the more common units of viscosity. Effect of superplasticizer concentration on rheology The trends observed on increasing superplasticizer concentration and illustrated in Figure4 are consistent with that observed previously (Michaux, M., et al. 1986). Notably there is, in a number of cases, a minimum concentration below which there is only a minimal effect on plastic viscosity and yield point at 0.2% bwoc or below, suggesting incomplete interaction with the cement grain surfaces or modification of the surface charge. Optimum values for dispersions, in all cases, are around 0.3 to 0.4% bwoc for the PAS and PC and 0.5% bwoc for the PNS. At concentrations above this there is little change in PV until around 0.8% bwoc, where there is slight increase, though YP continues to fall. Increase in PV can be attributed to too high a concentration of superplasticizer in the aqueous solution

Influence of Temperature and Retarder on Superplasticizer Performance  103

Figure 3 – Thickening time curves for LS-R retarded cement with superplasticizer at 80°C(176°F)

Figure 4– Concentration effect on a) plastic viscosity and b) yield point that causes bridging between particles giving rise to particle agglomeration and segregation that is detrimental to cement slurry stability. Although the PC superplasticizer used in this study shows similar results to the PNS and PAS, it composes only 20% of the polycarboxylate polymer component compared to 90% polymer component for the PNS and PAS. On this basis the concentration of polycarboxylate polymer if used by itself and not diluted with inert material would give a value of around 0.04% bwoc as the minimum concentration required, 0.08% bwoc as the optimal concentration and over-dispersion occurring at 0.16% bwoc. Impact of temperature on the rheology of superplasticized retarded slurries Rheological data of PNS, PAS and PC superplasticizers on Class G cement slurries containing LS-R retarder and antifoam GP-A at temperatures up to and including 80°C (176°F) have been discussed based of rheological models, Table 3, Table 4, Table 5, and Table 6. A closer investigation on the rheological data, Figure 5, shows that there are,

104 SP-302-07

Figure5–Temperature effect on plastic viscosity a) 40°C, c) 60°C, e) 80°C and yield point b) 40°C, d) 60°C and f) 80°C. however, notable differences on the behavior of the different superplasticizers. Plastic viscosity and yield values increase with temperature in the absence of superplasticizer increases from 38 cP and to 117 cP and 23.3 Pa to 88.9 Pa respectively over the 25°C (77°F) to 80°C (176°F) temperature range. Addition of superplasticizer shows a decrease in both plastic viscosity, Figure 5a, c and e and yield point, Figure 5b, d and f. The PAS and PC superplasticizers show similar trends with the PAS showing a greater rate of decrease in plastic viscosity and yield point compared to PC as concentration increases, with optimal values being around 0.3 to 0.4% bwoc. The higher yield point, 28.3 Pa of the PC, and lower plastic viscosity of 17.8 cP may have potential to minimize the effect of thermal thinning that occurs at high temperature. Although the PC produces trends similar to PAS it is at a much lower dosage given that the PC has only 20% polymer component compared to the PAS which has 90% polymer component. The PNS superplasticizer by comparison shows a somewhat different trend that becomes more evident with increase in temperature. PNS

Influence of Temperature and Retarder on Superplasticizer Performance  105

Table 7–Rheological properties at 100 °C (212°F) and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.1 (0.09)* 0.2 (0.18) 0.4 (0.36) 0.0 0.06 (0.054) 0.1 (0.09) 0.2 (0.18) 0.4 (0.36) 0.0 0.04 (0.008) 0.1 (0.02) 0.2 (0..04) 0.4 (0.16)

Bingham Plastic PV (cP) YP (Pa) 26.6 8.2 19.4 4.2 11.6 1.9 6-6 1.0 26.6 8.2 24.9 5.7 20.0 2.3 12.9 1.6 26.6 28.0 17.4 10.4 12.2

8.2 9.8 4.8 2.3 2.3

n 0.334 0.455 0.959 1.008 0.334 0.454 0.675 0.749 0.334 0.243 0.410 0.825 0.815

Herschel-Bulkley Yield stress (Pa) k (Pasn) 2.009 3.6 0.610 2.3 0.014 1.9 0.006 1.0 2.009 3.6 0.791 3.2 0-147 1.5 0.060 1.2 2.009 4.363 0.751 0.030 0.037

3.6 1.8 2.7 2.1 2.1

* () – polymer component

requires a minimal concentration before there is an obvious decrease in plastic viscosity or yield point and this value appears to change with temperature being around 0.3% bwoc at 40°C (104°F) to 60°C (140°F) and 0.4% bwoc at 80°C (176°F) and with minimal or optimum values obtained at about 0.6%. The effects observed both in rheology and thickening time can be related to an acceleration of the aluminate phase hydration resulting in a precursor ettringite gel that increases viscosity and in addition retards silicate hydration by coating on the cement grains (Luke K. and Aitcin, P.-C., 1991). Clearly a higher concentration of PNS is required in comparison to PAS and even more to PC to achieve the same rheological results with LS-R retarder and PG-A antifoaming agent. Rheological profiles and effect of superplasticizer at 100 -120°C (212 – 248°F) The data for PNS, PAS and PC on the rheological properties are given in Table 7 and Table 8. Clearly values are much lower than those observed with the LS-R retarded systems and there appears to be little difference in the rheological performance at either 100°C (212°F) or 120°C (248°F) and irrespective of the superplasticizer used. Trends are however similar to those obtained at lower temperatures with slight to no effect at lower concentrations, and over-dispersion at higher concentrations giving rise to particle segregation and free water formation over time consistent with the low yield stresses observed. Optimum concentration of superplasticizer to decrease the plastic viscosity and yield point in this case occurs in the range of 0.1 to 0.2% bwoc. However, the low rheological values obtained in the cement slurry without superplasticizer are more than sufficient to provide good slurry flowability, and is a result predominantly of the HC-R retarder. LS-R and HC-R although used primarily for retarding cement setting both have secondary effects of acting as dispersant. As evidenced by the data the HC-R has a much stronger dispersing effect than the LS-R and in the concentrations required to provide adequate pumping time at 100°C (212°F)

106 SP-302-07

Table 8–Rheological properties at 120 °C (248°F) and 10 MPa (1,450 psi) Superplasticizer (% bwoc)

PNS

PAS

PC

0.0 0.1 (0.09)* 0.2 (0.18) 0.4 (0.36) 0.0 0.06 (0.054) 0.1 (0.09) 0.2 (0.18) 0.4 (0.36) 0.0 0.04 (0.008) 0.1 (0.02) 0.2 (0..04) 0.4 (0.16)

Bingham Plastic PV (cP) YP (Pa) 25.4 7.1 23.9 6.5 15.6 1.9 11.7 1.5 25.4 7.1 24.9 6.4 20.0 3.4 12.9 2.0 25.4 24.1 19.2 16.3 11.5

7.1 5.7 4.5 3.3 3.4

n 0.584 0.629 0.764 1.208 0.584 0.395 0.534 0.907 0.584 0.602 0.669 0.617 0.581

Herschel-Bulkley Yield stress (Pa) k (Pasn) 0.338 5.7 0.236 5.4 0.066 1.5 0.003 1.7 0.338 5.7 0.342 2.9 0.371 1.9 0.025 1.8 0.338 0.287 0.148 0.177 0.157

5.7 4.4 3.7 2.4 2.8

* () – polymer component

and 120°C (248°F), 0.5% bwoc, it also dominates rheological properties in relation to the superplasticizers. This is consistent with previous observations using sodium gluconate in combination with polycarboxylate superplasticizer which allowed a reduction of approximately 50% of the dosage of the superplasticizer (Plank J. et al. 2009). It is also of note that in this case the slurries tend to follow the Herschel-Bulkley model more closely than the Bingham plastic model, at most superplasticizer concentrations used particularly at 120°C (248°F) as illustrated for selected samples in Figure 6. Figure 6 a) and b) show the shear stress/shear rate plots comparing the Bingham Plastic model with the Herschel-Bulkley model for 0.06% bwoc PC at 120°C (248°F), and Figure 6 c) and d) show the comparison at 0.4% bwoc PC at 120°C (248°F) illustrating the Herschel-Bulkley model as giving the best fit. Figure 6 e) and f) for 0.2% bwoc PNS at 120°C (248°F) shows that the HerschelBulkley model gives the better fit with n equals 0.764 although the Bingham Plastic model is relatively close. Superplasticizer effect on calorimetric and zeta potential of Class G slurry Variations in the heat evolution of the LS-R retarded Class G cement with and without PNS, PAS and CF at 40°C (104°F) is shown in Figure 7. The retarded Class G cement shows a heat flow curve typical of cements cured at 25°C (77°F) where the induction period, second hydration peak from formation of C-S-H and CH from hydration of the silicate phases and then a third peak observed as a shoulder occurring after the silicate hydration and attributed to ettringite (AFt) to monosulfate (AFm) conversion (Ghose, A. and Pratt, P.L., 1981). The initial peak due to wetting and initial dissolution can be observed as the decline of the peak since the main reaction has occurred during the mixing procedure that is performed outside the calorimeter. In addition there is an endothermic peak that occurs during temperature ramp-up and is attributed to non-equilibrium conditions within the calorimeter (Luke, K., 2011). The PNS shows a significant increase in the induction period retarding the hydration

Influence of Temperature and Retarder on Superplasticizer Performance  107

Figure 6 - Shear stress/shear rate plots showing a)Bingham Plastic model, b)HerschelBulkley model for 0.06% bwoc PC slurry at 120°C(248°F), c)Bingham Plastic model, d)Herschel-Bulkley model for 0.4% bwoc PC at 120°C(248°F)and e) Bingham Plastic model, f)Herschel-Bulkley model for 0.2% bwoc PNS at 120°C(248°F).

108 SP-302-07

Figure 7 – Heat flow curves showing the effect of superplasticizers at 40°C (104°F) from 7.5 hours to 10 hours. The increased heat flow indicates though a faster reaction rate for the silicate reaction and more notably the conversion of AFt to AFm which appears to occur at about the same time (bimodal peak). The PAS causes a decrease in the induction period from 7.5 hours to 6.5 hours and a notable increase in the rate of the reaction though in this case the AFt to AFm conversion is less pronounced. The PC appears to have the least effect with perhaps a slight increase in the rate of reaction compared to the retarded Class G without superplasticizer. At 120°C (248°F) the curves differ from that obtained at 40°C (104°F), the induction period is shorter based on the different retarder used and also shows a different heat flow curve typical of retarded cement slurries at temperatures of 90°C (194°F) and above (Luke, K., 2011). In this case the decline in the initial hydration peak is clearly evident for the HC-R retarded cement whereas the superplactized cements show a peak maxima. However it is not possible to interpret the data at this point as the calorimeter is in the temperature ramp-up process and is not in equilibrium. Equilibrium is only established when temperature has been reached and the heat flow value is around 0 mW which coincides with the end of the endothermic peak. This is notably longer at higher temperatures as ramp-up time is longer. An additional peak observed before the silicate hydration peak is attributed to aluminate phase hydration as previously indicated by XRD (Luke, K., 2011). The addition of superplasticizer to the HC-R retarded Class G cement at a temperature of 120°C (248°F) shows minimal effect with perhaps a slight acceleration with PNS and retardation in the induction period with PC, Figure8. The rate of reaction and the peaks observed are the same with or without the superplasticizer indicating at higher temperature the effect of the retarder is more dominant. Zeta potential values of the PNS, PAS, and PC on the retarded cement systems in both RO water and in pore solution extracted from a Class G cement hydrated for approximately 70 minutes equivalent to time of rheology testing is given in Table 9. The high negative value of the retarded cement system in RO water without superplasticizer is consistent with that determined previously, -17 to -15 mV (Nägele, E., 1986) at short hydration times

Influence of Temperature and Retarder on Superplasticizer Performance  109

Figure 8 – Heat flow curves showing the effect of superplasticizers at 120°C (248°F) indicative of the retardation of reaction by the LS-R and HC-R. Addition of superplasticizer notably decreases the value to even higher negative values consistent with deflocculation of the particles. However, there was no differentiation in zeta potential of the PNS, PAS or PC. Alkali is known to influence the zeta potential both based on pH value and on the ions present, K+ or Na+ (Nägele, E., 1986). In order to simulate the solution to that of the cement in terms of pH and ionic composition the pore solution of the Class G was extracted after approximately 70 minutes of hydration. Zeta potential values as expected were significantly less negative than obtained with the RO and is attributed to the fact that the increased ionic strength compresses the diffuse double layer (Nägele, E., 1987). Values were close to zero and consistent with those obtained on well cement slurries (Hodne, H. and Saasen, A., 2000). In this case variations in value were noted with the different superplasticizers where the PNS and PAS give negative values and the PC gives a positive value and the validity of the data is being further investigated. SUMMARY AND CONCLUSIONS This preliminary study investigated the effect of three different types of superplasticizer on the rheological characteristics of two different retarded cement systems that also included an antifoam additive at 25°C (77°F) to 120°C (248°F). The Herschel-Bulkley model give the best fit for all slurries studied though where the value of n ranged from about 0.5 to about 1.2 the Bingham Plastic model was also noted to give a good fit to the shear stress/shear rate plots of the data. This was particularly more notable as n approached 1, consistent with the fact that the Herschel-Bulkley model reduces to the Bingham Plastic model n = 1. PNS and PAS, in some cases, showed shear thickening effects, n greater than 1, though there was no correlation to superplasticizer concentration or temperature. Optimal concentration of superplasticizer was around 0.3 to 0.4% bwoc for the PAS and PC superplasticizers, though given that the PC is only 20% active the optimal concentration for pure PC is closer to 0.16% bwoc and for PNS the least effective it is 0.5% bwoc. The PNS was also observed to have a lesser effect in controlling plastic viscosity and yield

110 SP-302-07

Table 9 – Zeta Potential of PNS, PAS and PC on Retarded Class G slurries Retarder

Aqueous phase RO water

LS-R Pore solution

RO water HC-R Pore solution

SP No SP PNS PAS PC No SP PNS PAS PC No SP PNS PAS PC No SP PNS PAS PC

Zeta Potential (mV) -16.6 -22.5 -22.5 -23.2 1.9 -2.9 -5.5 1.8 -17.0 -24.0 -18.7 -22.2 -4.5 -0.3 -1.2 1.4

point in comparison to PAS and PC superplasticizers at higher temperature and this may be attributed to acceleration of the aluminate phase hydration and formation of a precursor ettringite gel. Calorimetric results showed that the PNS had a significant retarding effect compared to the PAS which was slightly accelerating and the PC which had little effect except to slightly increase reaction rate of the silicate phases. At temperatures of 100°C (212°F) and 120°C (248°F) the slurries followed a Herschel-Buckley model irrespective of the concentration of superplasticizer used. The Bingham Plastic model showed a good fit to the shear stress/shear rate plot of the data only at n values above about 0.8. Plastic viscosity and yield point values were notably lower than at temperatures of 20°C (77°F) to 80°C (176°F) and was dominated by the HC-R retarder which also shows strong dispersing properties. This was also confirmed by calorimetry which showed minimal effect of the superplasticizers. Zeta potential showed minimal differences between superplasticizers in RO water irrespective of the retarder used though there were some slight differences in cement pore solution in that PC had a positive value compared the negative value of the PNS and PAS. These are being further investigated. AUTHOR BIOS ACI member Karen Luke is Principal Research Advisor - Cement at the R&D Centre, Trican Well Service, Calgary, Alberta, Canada. She received her PhD from the University of Aberdeen, Scotland, UK. Her primary research interests are in the chemistry of cement, cement additive interactions, durability of cement, hydrothermal phase equilibria of well cements, well cement performance and mechanical properties. Adrian Torres is research Technical Specialist at the R&D Centre, Trican Well Service, Calgary, Alberta, Canada. He received his BSc at the University of Calgary, Alberta, Canada. His current interests are cement additive interactions as well as enhancing the mechanical properties of oil well cement.

Influence of Temperature and Retarder on Superplasticizer Performance  111

REFERENCES Aitcin, P.-C., (1998), High Performance Concrete, Ed. Aitcin, P.-C., E & FN Spon Publisher, 591 pp. API RP 10B-2 (2010), Recommended Practice for Testing Well Cements, Washington, DC:API. Aukett, P. N., and Bensted, J.1992 , “Application of Heat Flow Calorimentry to the Study of Oilwell Cements,” Journal of Thermal Analysis, V. 38, No. 4, pp. 701-707. doi: 10.1007/BF01979399 Ghose, A., and Pratt, P. L., (1981), “Studies of the Hydration Reactions and Microstructure of Cement-Flyash Pastes,” Materials Research Society, Effects of Flyash Incorporation in Cement and Concrete, Proceedings, Symposium N, Annual Meeting Nov 16-19, pp. 82-91. Guillot, D., (2006), “Rheology and Flow of Well Cement Slurries,” In Well Cementing, Eds. E.B. Nelson and D. Guillot, Schlumberger Publisher, pp. 93 - 189. Hattori, K.1978 , “Experiences with Mighty Superplasticizer in Japan,” ACI Publication, V. SP-62, pp. 37-61. Hodne, H., and Saasen, A.2000 , “The Effect of Cement Zeta Potential and Slurry Conductivity on the Consistency of Oilwell Cement Slurries,” Cement and Concrete Research, V. 30, No. 11, pp. 1767-1772. doi: 10.1016/S0008-8846(00)00417-8 Luke, K., (2011), “Conduction Calorimetry and X-Ray Diffraction Investigation of Cement Retardation at 70 – 120°C,” 13th International Congress on the Chemistry of Cement, Madrid Jul 3-8, pp 6. Luke, K., and Aitcin, P.-C.1991 , “Effect of Superplasticizer on Ettringite Formation,” Ceramic Transactions on Advances in Cementitious Materials, V. 16, pp. 147-166. Michaux, M., and Defosse, C.1986 , “Oilwell Cement Slurries – Pt.1: Microstructural Approach of their Rheology,” Cement and Concrete Research, V. 16, No. 1, pp. 23-30. doi: 10.1016/0008-8846(86)90064-5 Michaux, M.; Oberste-Padtberg, R.; and Defosse, C.1986a, “Oilwell Cement Slurries – Pt.2: Adsorption Behaviour of Dispersants,” Cement and Concrete Research, V. 16, No. 6, pp. 921-930. doi: 10.1016/0008-8846(86)90016-5 Nägele, E.1986 , “The Zeta-Potential of Cement Part II: Effect of pH Value,” Cement and Concrete Research, V. 16, No. 6, pp. 853-863. doi: 10.1016/0008-8846(86)90008-6 Nägele, E.1987 , “The Zeta-Potential of Cement Part III: The Non-Equilibrium Double Layer on Cement,” Cement and Concrete Research, V. 17, pp. 573-580. Nelson, E. B.; Michaux, M.; and Drochon, B., (2006), “Cement Additives and Mechanisms of Action,” In Well Cementing, Eds. E.B. Nelson and D. Guillot, Schlumberger Publisher, pp. 71-80. Plank, J., Schröfl, C. and Gruber, M., (2009), “Use of Supplemental Agent to Improve Flowability of Ultra-High-Performance Concrete,” SP-262-1, Superplasticizers and Other Chemical Admixtures in Concrete, pp. 1-16. Sairam, P. K. S.; Morgan, R.; and Pangu, G., (2012), “A Combined Mixer Design with Helical Blades to Probe Rheology of Complex Oilfiled Slurries and Pastes,” Paper SPE 159112, Presented at SPETT 2012 Energy Conference and Exhibition, Port of Span, Trinidad, June 11-13.

112 SP-302-07

Schmidt, W., (2014), Design Concepts for the Robustness Improvement of SelfCompacting Concrete – Effects of Admixtures and Mixture Components on the Rheology and Early Hydration at Varying Temperatures,” PhD Thesis, Eindhoven University of Technology, The Netherlands, pp. 308. Specification, A. P. I., 10A (2011), Specification for Cements and Materials for Well Cementing, Washington DC:API Walraven, J. C., (2010), “Self Compacting Concrete: Properties, Development and Code Recommendations,” 6th International RILEM Symposium on Self-Compacting Concrete and 4th North American Conference on the Design and Use of SCC, Montreal, Canada, pp. 25-44.

SP-302-08

Properties of a New Type of Polycarboxylate Admixture for Concrete Using High Volume Blast Furnace Slag Cement by Shinji Tamaki, Kazuhide Saito, Kazuhisa Okada, Daiki Atarashi, and Etsuo Sakai Several studies have been pursued in Japan on developing concrete using high volume blast-furnace slag cement for reducing CO2 emissions arising from calcination of cement. However, when using high volume blast-furnace slag cement, various problems are encountered, such as decreased fluidity retention ability caused by the reduction of admixture dosage and decreased strength enhancement. In this paper, the authors focus on the adsorption properties of polycarboxylate ether superplasticizers and the properties of hardened concrete that incorporates a component of high volume blast-furnace slag cement, and discuss the development of a new type of superplasticizer through molecular design and optimization of the admixture composition. The admixture improved the fluidity and properties of hardened concrete using slag cement containing more than 60% blastfurnace slag. Keywords: adsorption; blast-furnace slag; blast-furnace slag cement; chemical admixture; fluidity retention ability. INTRODUCTION Presently, 50–60 million tons of cement are produced per year in Japan, and CO2 emissions during cement production, including emissions separated and discharged from the limestone mineral, account for 3%–4% of the total CO2 emissions in Japan. Attempts to reduce CO2 emissions derived from cement production have been ongoing for some time. Industrial byproducts, such as blast-furnace slag (BFS) and fly ash (FA), have been used quite effectively, and blended cements comprising mixtures of these byproducts with ordinary Portland cement (OPC) are normalized in Japanese Industrial Standards (JIS). Blended cement tends to possess low early strength, high drying shrinkage, and high carbonation rate. Therefore, both BFS and FA cements are typically used in the Type B formulation. The types of blended cement defined in JIS are shown in Table 1. 113

114 SP-302-08

Table 1—Standards of blended cement (JIS)

BFS cement FA cement

Replacement ratio, % Type B 30 < BFS ≤ 60 10 < FA ≤ 20

Type A 5 < BFS ≤ 30 5 < FA ≤ 10

Type C 60 < BFS ≤ 70 20 < FA ≤ 30

* BFS cement: JIS R 5211 **FA cement: JIS R 5213

Table 2—Physical and chemical properties of OPC and BFS Code OPC BFS4000 BFS6000

Binder type Ordinary Portland cement Blast-furnace slag Blast-furnace slag

Blaine Density, fineness, g/cm3 cm2/g SiO2

Chemical composition, % Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O

3.16

3310

21.56

4.68

2.98

65.63 1.30

1.30

0.33

0.39

2.91

4570

33.60

14.80

0.29

42.30 6.30

—

0.21

0.35

2.91

6130

33.70

14.50

0.42

42.60 6.10

—

0.19

0.38

To significantly reduce the CO2 emissions, the authors have been developing a high volume BFS cement (denoted as Energy CO2 minimum; ECM) containing more than 60% BFS by weight.1 To improve low early strength and reduce drying shrinkage, we have adjusted the cement composition; to address the high carbonation rate, we have selected the concrete structures in addition to adjusting the cement composition. Furthermore, for practical use, the chemical admixture for concrete derived from ECM requires a high fluidity retention ability that is caused by a reduction in the admixture dosage compared with that when using OPC or low-volume-blended cement (for example, BFS cement Type B). The focus of this paper is the adsorption of polycarboxylate ether (PCEs) superplasticizers in ECM and the resulting fluidity, and the development of a new type of superplasticizer through molecular design and optimization of the admixture composition. RESEARCH SIGNIFICANCE The authors report on a new admixture for ECM containing more than 60% blast-furnace slag. The new admixture ameliorates the problems encountered when using ECM, such as the concerns associated with the fluidity and properties of the hardened concrete. ECM greatly reduces CO2 emissions from cement production by contributing to the expansion of the application range of BFS cements and by furthering efforts toward environmental preservation. EXPERIMENTAL PROCEDURE Materials Two types of BFS with fineness of 4570 cm2/g (hereafter BFS4000) and 6130 cm2/g (hereafter BFS6000), were used. In addition to BFS, OPC and ECM were used. The physical and chemical properties of OPC and BFS are listed in Table 2. ECM was blended in the ratio of OPC: BFS: anhydrite (fineness 3890 cm2/g) = 30: 63: 7. ECM containing

Properties of a New Type of Polycarboxylate Admixture for Concrete Using High Volume Blast Furnace Slag Cement  115

Fig. 1―Molecular structure of superplasticizer Table 3—Mortar mixture proportions W/B 0.4

Target air, % 2.0

W 300 [10.6]

Weight, g [oz.] B 1000 [35.3]

S 1500 [52.9]

* Water-to-binder ratio (W/B), Water (W), Binder (B), and Fine aggregate (S)

BFS4000 is denoted as ECM4000, and that containing BFS6000 is denoted as ECM6000. These blends were designed to improve early strength and reduce drying shrinkage by increasing the amount of SO3 relative to that used in general-use BFS cement.2,3 The fine aggregate (S) comprised land sand with saturated surface dry (SSD) density of 2.58 g/cm3 and fineness modulus of 2.87. The coarse aggregate (G) was crushed stone with SSD density of 2.68 g/cm3 and solid volume of 60%. Chemical admixtures The superplasticizers used in this study were PCEs with methoxy polyethylene glycol graft chains. A commercially available product (hereafter PCE) for the OPC concrete was used as reference. Two types of polymers (M23 and M9) of different side chain lengths and methacrylic acid (MAA) to polymer ratio were synthesized and used in the development of the admixtures for ECM. For M23, the side chains comprised 23 mol of ethylene oxide, and the molar ratio of MAA to polymer was 69%. For M9, the side chains comprised 9 mol of ethylene oxide, and the molar ratio of MAA to polymer was 40%. In addition, in this work, the concentration of the superplasticizer added was converted to a solid component. The molecular structure of the superplasticizer is shown in Fig. 1. The superplasticizers were characterized by gel permeation chromatography. Sodium gluconate (C6H11O7Na; GLNa) was used as the slump retention agent. Mortar test Table 3 shows the mortar mixture proportions. The binder consisted of OPC and BFS4000 in different ratios. The mortar test was conducted in accordance with JIS R 5201; the extent of mortar flow on removal of the cone was measured. Paste test Paste specimens were obtained at a water-to-cement ratio (W/C) of 0.32 by mixing for 5 min, allowing the mixture to stand for 4 min, and mixing again for 1 min. The paste flow

116 SP-302-08

Table 4—Mixture proportions of concrete W/C 0.5 0.4

Cement ECM6000

Target slump, cm [in.]

Target air, %

18±2.5 [7.1±1.0]

4.5±1.0

Unit weight, kg/m3 [lb/yd3] W C S G 160 [270] 320 [539] 854 [1439] 951 [1603] 160 [270] 400 [674] 740 [1247] 997 [1680]

* Water (W), Cement (C), Fine aggregate (S), and Coarse aggregate (G)

was measured in accordance with JASS15M-103; the extent of paste flow on removal of the cone was measured. Adsorption of superplasticizer Paste mixed in the same ratio as that for the paste flow measurement was separated into the solid and liquid phases by a centrifuge operating at 5000 rpm for 10 min. The amount of organic carbon in the liquid phase was subsequently measured by a total organic carbon analyzer (TOC5050A, Shimadzu Corporation), and the adsorbed superplasticizer amount per unit mass was calculated. In addition, the hydration of the paste was stopped using a large quantity of acetone, and drying under reduced pressure ( 99%, Merck Schuchardt OHG, Hohenbrunn, Germany), ω-methoxy poly(ethylene glycol) (Polyglycol M350, Polyglycol M500, Polyglycol M2000) (> 99% purity, Clariant Produkte Deutschland GmbH, Burgkirchen, Germany), H2SO4 (98% purity, SIGMA-ALDRICH CHEMIE, Steinheim, Germany), tertbutyl peroxybenzoate (98% purity, SIGMA-ALDRICH CHEMIE, Steinheim, Germany) were used as is without any further purification. Reference sample —a self-synthesized polycarboxylate superplasticizer possessing high grafting density, designated 45PC2, was used for comparison. This PCE was prepared from methacrylic acid and ω-methoxy poly(ethylene oxide) methacrylate ester via aqueous free radical copolymerization reaction. The molar ratios of the monomers were 2:1, and the number of ethylene oxide units in the side chain was 45. The synthesis process is described in detail in.4 Cement — A CEM I 52.5 N (Milke®classic, HeidelbergCement, Geseke plant) was used. Its phase composition as determined by XRD is presented in Table 1. The average particle size (d50 value, determined by laser granulometry) was found at 11.5 µm. Its density was 3.153 g/cm3 (Helium pycnometry).

A Simplified Preparation Method for PCEs Involving Macroradicals  5

Synthesis of PCE Graft Copolymers MPEG 350 - graft - MPEG 350 mono maleate/maleic anhydride — 7.5 g (76 mmol) of maleic anhydride, 25.2 g of MPEG 350 (72 mmol) and 0.15 g of 98 wt.% H2SO4 were placed in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3 hours at 140 °C under constant stirring. Thereafter, 15 g (153 mmol) of maleic anhydride were dissolved in 20.125 g (58 mmol) of MPEG 350. Furthermore, 1.2 g (6 mmol) of tert-butyl peroxybenzoate initiator were also added to this homogeneous mixture. Using a peristaltic pump, this solution was fed continuously into the vessel over a period of 60 min while maintaining the temperature at 140 °C. The viscosity of the mixture increased during the synthesis procedure. When the addition was complete, the mixture was stirred for another 60 min at 140 °C. After that, the mixture was cooled to ~ 70 °C and ~ 65 g of water were added. Then the pH value was adjusted to 7 by addition of 30 wt.% aqueous NaOH solution. The final product was designated as MPEG 350 - g - MPEG 350 mono ester/maleic anhydride. It presented a reddish, 34.3 wt.% aqueous solution which was used without further purification. MPEG 500 - graft - MPEG 500 mono maleate/maleic anhydride — 7.5 g (76 mmol) of maleic anhydride, 36 g of MPEG 500 (72 mmol) and 0.15 g of 98 wt.% H2SO4 were placed in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3 hours at 140 °C under constant stirring. Thereafter, 15 g (153 mmol) of maleic anhydride were dissolved in 28.8 g (58 mmol) of MPEG 500, then 1.2 g (6 mmol) of tert-butyl peroxybenzoate initiator were added to this homogeneous mixture. Using a peristaltic pump, this solution was fed continuously into the vessel over a period of 60 min while maintaining the temperature at 140 °C. Also here, viscosity of the mixture increased during the synthesis process. All further steps were the same as for the MPEG 350 graft copolymer. The final product was designated as MPEG 500 - g - MPEG 500 mono ester/maleic anhydride. It presented a reddish, 43.8 wt.% aqueous solution which was used without further purification. MPEG 2000 - graft - MPEG 2000 mono maleate/maleic anhydride — 1.875 g (19 mmol) of maleic anhydride, 36 g of MPEG 2000 (18 mmol) and 0.15 g of 98 wt.% H2SO4 were placed in a 250 mL three-neck round bottom flask equipped with a stirrer and heated for 3 hours at 140 °C under constant stirring. Thereafter, 3.75 g (38 mmol) of maleic anhydride were dissolved in 7.2 g (3.6 mmol) of MPEG 2000. Furthermore, 0.6 g (3 mmol) of tertbutyl peroxybenzoate initiator were added to this homogeneous mixture. Using a peristaltic pump, this solution was fed continuously into the vessel over a period of 60 min while maintaining the temperature at 140 °C. The rest of the preparation followed the method as described above. The final product was designated as MPEG 2000 - g - MPEG 2000 mono ester/maleic anhydride. It presented a reddish, 43.7 wt.% aqueous solution which was used without further purification. Characterization of PCE samples Molar masses (Mw, Mn) and polydispersity index (PDI) of the PCE samples were determined utilizing size exclusion chromatography (SEC). Additionally, the anionic charge density was measured. Size exclusion chromatography (SEC) — PCE solutions containing 10 g/L of the polymer were prepared for SEC analysis. Measurement was performed on a Waters 2695

6 SP-302-12

Separation Module equipped with three Ultrahydrogel™ columns (120, 250, 500) and an Ultrahydrogel™ guard column from Waters, Eschborn, Germany, and a subsequent 3 angle static light scattering detector (“mini Dawn” from Wyatt Technology Corp., Santa Barbara, CA, USA). The polymer concentration was monitored with a differential refractive index detector (RI 2414, Waters, Eschborn, Germany). Aqueous 0.1 N NaNO3 solution adjusted to pH 12 with NaOH was used as an eluent at a flow rate of 1.0 mL/min. From the SEC measurements, the polydispersity index (PDI) and the molar masses (Mw and Mn) were obtained. The value of dn/dc used to calculate Mw and Mn was 0.135 mL/g (value for polyethylene oxide).8 Specific anionic charge density — The specific anionic charge densities of the polymers were determined employing a particle charge detector PCD 03 pH (Mütek Analytic, Herrsching, Germany). This instrument allows the experimental determination of the anionic charge of polymers in solution. Here, 0.2 g/L of the polymers were dissolved in DI water and titrated against an aqueous 0.34 g/L solution of poly-diallyl dimethyl ammonium chloride (polyDADMAC) until charge neutralization (zero potential) was reached. From the amount of polyDADMAC consumed to reach a zero potential, the amount of negative charge per gram of polymer was calculated.9 Calorimetry — To detect a potential retarding effect of the novel PCE samples, timedependent heat evolution during cement hydration was monitored. There, 4 g of cement were filled into 10 mL glass ampoules, mixed at 21 °C with the appropriate amount of aqueous polymer solution (w/c = 0.3), shaken for 1 min in a wobbler and then placed in an isothermal heat conduction calorimeter (TAMair, Thermometric, Järfälla, Sweden) to monitor the heat flow of the hydration reaction. Data logging was continued for 7 days. Dispersing performance in cement For determination of the paste flow, a “mini slump test” modified from DIN EN 1015 was utilized and carried out as follows: First, the water-to-cement (w/c) ratio of the paste without polymer was set to 0.3. At this w/c ratio, the dosages of the PCE samples required to reach a spread of 26±0.5 cm were determined. Generally, the polymer was dissolved in the required amount of mixing water placed in a porcelain cup. When aqueous polymer solutions were used, then the amount of water contained in the polymer solution was subtracted from the amount of mixing water. In preparation, 300 g of cement were added to the mixing water and agitated manually for 1 minute utilizing a spoon, then rested for 1 minute without stirring and were again stirred for 2 minutes. After the stirring, the cement paste was immediately poured into a Vicat cone (height 40 mm, top diameter 70 mm, bottom diameter 80 mm) placed on a glass plate and the cone was removed vertically. The resulting spread of the paste was measured twice, the second measurement being in a 90 ° angle to the first and averaged to give the final spread value. For the time-dependent flow behavior of the paste, 400 g of cement were mixed with 120 mL of DI water as described in the procedure above. After each measurement, the slurry was transferred back into the cup and covered with a wet towel in order to avoid drying. Before each subsequent measurement, the paste was stirred again for 2 minutes. Measurements were taken every 15 minutes over a total period of 120 minutes.

A Simplified Preparation Method for PCEs Involving Macroradicals  7

Fig. 3 – SEC spectra of the three synthesized graft polymers EXPERIMENTAL RESULTS AND DISCUSSION Properties of synthesized PCE polymers In this novel synthesis process, maleic anhydride as well as MPEG mono maleate ester were grafted onto MPEG macroradicals serving as backbone of the PCE polymer. This way, a comb polymer is formed whereby along an MPEG trunk chain, maleic anhydride and MPEG mono maleate pendants are randomly arranged, as will be shown in the mechanistic part of this paper. The SEC spectra of the synthesized PCE polymers are displayed in Fig. 3. According to them, the conversion of the monomers was 71 - 77%. It should be mentioned here that so

8 SP-302-12

Table 2 – Molar masses and polydispersity index (PDI) of the synthesized polymers and of a comparative commercial PCE sample Polymer MPEG 350 - g - MPEG 350 mono ester/maleic anhydride MPEG 500 - g - MPEG 500 mono ester/maleic anhydride MPEG 2000 - g - MPEG 2000 mono ester/maleic anhydride Industrial PCE

Mw (g/mol) 9,100

Mn (g/mol) 3,349

PDI (Mw/Mn) 2.7

Yield 77.3%

6,848

2,946

2.3

72.6%

16,500

6,163

2.7

71.3%

63,840

28,650

2.2

88.0%

Table 3 – Specific anionic charge amount of superplasticizer samples tested in pH 12.5 aqueous solution

Fluid System pH 12.5 aqueous

MPEG 350 - g - MPEG 350 mono ester/maleic anhydride 3,239

Specific anionic charge amount [μeq/g] MPEG 500 - g - MPEG MPEG 2000 - g - MPEG 500 mono ester/maleic 2000 mono ester/maleic anhydride anhydride 1,249

1,551

45 PC 2 1,033

far we have not attempted to optimize the yield of polymer. Molar composition of the polymers was confirmed by 1H NMR spectroscopy whereby the peak areas of the individual protons were integrated (spectra not shown here). The characteristic molecular properties of the synthesized polymers are presented in Table 2. According to the SEC data, the synthesized PCE polymers exhibit relatively low molecular weights (Mw, Mn), compared to a commercial benchmark MPEG PCE product prepared from MPEG - MAA macromonomer via free radical copolymerization. The polydispersity index of the samples lies at ~ 2.5 which indicates a fairly narrow molecular weight distribution. Next, the specific anionic charge amounts of the synthesized polymers and of the industrial polymer were determined in DI water. The results are exhibited in Table 3. From the data, it is obvious that at increasing side chain length of the PCE polymers, the anionic charge density decreases as expected. Cement dispersion ‘Mini slump’test — To probe into the dispersing effectiveness of the newly synthesized PCE samples, the dosages required to achieve a cement paste spread of 26±0.5 cm were determined. For this test, a cement paste prepared at a w/c ratio of 0.3 was employed. According to Fig. 4, the newly synthesized polymers work as effective cement dispersants at low w/c ratios. Their effectiveness increases with increased length of the MPEG utilized in the synthesis. Time - dependent slump loss behavior — Next, time - dependant slump loss behavior of the superplasticizer samples was measured over a period of two hours. In this test, cement pastes prepared at a w/c ratio of 0.3 treated with the superplasticizer dosages as shown in Fig. 4 were employed. The results are displayed in Fig. 5. According to this data,

A Simplified Preparation Method for PCEs Involving Macroradicals  9

Fig. 4 – Dosages of synthesized graft copolymers required to achieve a paste spread of 26±0.5 cm (w/c = 0.3)

Fig. 5 – Slump loss behavior of cement pastes (w/c=0.3) containing the synthesized graft copolymers the synthesized graft polymers maintain workability of the cement paste for a short time period only. Generally, polymer MPEG 2000 - g - MPEG 2000 mono ester/maleic anhydride seems to retain fluidity slightly better than the polymers prepared from the short chain MPEGs (MPEG 350 and MPEG 500). This trend is owed to the specific molecular architecture of the PCE polymers, because PCE molecules with longer side chains are more coiled, adsorb slower and thus can provide better slump retention over time. Impact on cement hydration — To detect a potential retarding effect of the synthesized graft polymers, time - dependent heat evolvement from a cement paste (w/c ratio = 0.3) holding different polymer samples was monitored. The results are exhibited in Fig. 6. There, it is demonstrated that the synthesized graft polymers retard cement hydration, possibly due to the excessive amount of maleic anhydride used in the synthesis which

10 SP-302-12

Fig. 6 – Time-dependent heat evolvement from cement paste (w/c = 0.3) holding the synthesized graft polymers

Fig. 7 – Esterification of maleic anhydride with MPEG to yield a mono maleate ester partially did not incorporate into the PCE (see SEC spectra in Fig. 3). Again, the polymer containing MPEG 2000 performs better than the two polymers with shorter pendants. Grafting Mechanism via Macroradicals Stepwise analysis of the cascade of reactions occurring during this novel synthesis method revealed a mechanism as follows: In the first step, a mono maleate ester is produced via esterification reaction from ω-methoxy polyethylene glycol and maleic anhydride, as is shown in Fig. 7.

A Simplified Preparation Method for PCEs Involving Macroradicals  11

In the subsequent grafting process, the peroxide initiator at first undergoes thermal homolytic cleavage to form radicals RO• (see Fig. 8). In a second step, ω-methoxy polyethylene glycol (MPEG) macroradicals are formed through a chain transfer reaction with the radicals RO• whereby hydrogen is abstracted from MPEG. It should be noted that in previous literature, formation of such macroradicals from peroxide initiators has been well documented for polyethylene and polypropylene and successful grafting of maleic anhydride onto these macroradicals is described there as well.10-14 Subsequently, maleic anhydride and the unsaturated MPEG mono maleate ester can graft onto these macroradicals by forming a comb - shaped PCE structure which again presents a macroradical, see step (3) in Fig. 8. This reaction is favoured by the strong electron attracting properties of the double bond of maleic anhydride and the mono maleate ester. Through continued chain transfer reactions as shown in step (4) of Fig. 8, new MPEG macroradicals are formed which then can undergo the same sequence of grafting reaction as before, thus leading to an MPEG trunk chain which holds multiple maleic acid and MPEG maleate pendant groups. Termination occurs when two ω-methoxy polyethylene glycol macroradicals combine into the final product. The proposed chemical structure of the final graft copolymer derived from this reaction mechanism is displayed in Fig. 9. It should be noted here that due to the complexity of the reaction, the final product likely consists of a mixture of several polymers exhibiting different degrees of grafting (as is also the case in MPEG PCE synthesis via free radical copolymerization using macromonomers, but this fact is often ignored), thus the structure presented in Fig. 9 represents only one of several possibilities. Another possibility requiring consideration is the formation of poly(maleic acid) homopolymer. However, the SEC spectra do not provide evidence of its presence. Also, according to literature the homopolymerization of maleic anhydride is inhibited by organic electron donors such as in oxygen or sulfur containing compounds.12,13 Consequently, this side reaction was ruled out. Another consideration was the formation of an α - olefin occurring in the initial esterification step from dehydration of MPEG with sulfuric acid. Such α – olefin could then undergo copolymerization. This possibility was checked via 1H NMR spectroscopy of the esterification product. However, no signals characteristic for α - olefinic protons could be detected. CONCLUSIONS A new preparation method for PCEs including a simplified process is presented as an alternative to the common free radical copolymerization and esterification processes. Here, the polymerization mechanism relies on the formation of ω-methoxy polyethylene glycol macroradicals which are generated by hydrogen abstraction from the MPEG. Subsequently, maleic anhydride as well as MPEG - mono maleate ester are grafted onto such macroradicals yielding a comb type PCE polymer with carboxylic and EO pendants arranged along the MPEG main chain. SEC data show that the synthesized graft polymers exhibit relatively low molecular weights (Mw, Mn). Performance tests indicate that this new type of PCE can disperse cement well although it was synthesized from simple raw materials. Future work should focus on further optimization of molar mass and conversion of raw materials to capture the full potential of this novel PCE chemistry.

12 SP-302-12

Fig. 8 – Cascade of reactions involved in the grafting of maleic anhydride and MPEG mono maleate ester onto a MPEG macroradical

A Simplified Preparation Method for PCEs Involving Macroradicals  13

Fig. 9 – Proposed chemical structure for the newly synthesized MPEG - g - MPEG mono ester/maleic anhydride PCE polymers AUTHOR BIOS Lei Lei is a Ph.D. student at the Chair for Construction Chemistry at Technische Universität München, Germany. Her research focuses on the impact of different clay minerals on the performance of polycarboxylate - based superplasticizers and new routes to PCE synthesis. Johann Plank is full Professor at the Institute of Inorganic Chemistry of Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemistry there. His research interests include cement chemistry, chemical admixtures, organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well cementing. ACKNOWLEDGMENTS L. Lei wishes to thank the Jürgen Manchot Foundation for generously providing a scholarship to finance this research at TU München. REFERENCES 1. Petit, J. Y.; Wirquin, E.; Khayat, K. H.; and Vanhove, Y., “Coupled effect of temperature and superplasticizer on rheological properties of SCC mortar”, 5th International RILEM Symposium on Self-Compacting Concrete, RILEM Publications SARL, Ghent, Belgium, 2007, pp. 1099 – 1104. 2. Plank, J., Schroefl, C., Gruber, M., Lesti, M., Sieber, R., “Effectiveness of Polycarboxylate Superplasticizers in Ultra-High Strength Concrete: The Importance of PCE compatibility with Silica Fume”, Journal of Advanced Concrete Technology, V. 7, No.1, 5-7, 2009, pp. 5 - 12. 3. Plank, J., ““PCE Superplasticizers – Chemistry, Applications and Perspectives”, 18. IBAUSIL,” Weimar, V. 1, 2012, pp. 91-102. 4. Plank, J.; Pöllmann, K.; Zouaoui, N.; Andres, P. R.; and Schaefer, C., “Synthesis and performance of methacrylic ester based polycarboxylate superplasticizers possessing

14 SP-302-12

hydroxy terminated poly(ethylene glycol) side chains,” Cement and Concrete Research, V. 38, No. 10, 2008, pp. 1210-1216. doi: 10.1016/j.cemconres.2008.01.007 5. Wang, Z. M.; Xu, Y.; Wu, H.; Liu, X.; Zheng, F. Y.; Li, H. Q.; Cui, S. P.; Lan, M. Z.; and Wang, Y. L., “A Room Temperature Synthesis Method for Polycarboxylate Superplasticizer” CN patent 101974135 B, 2013, assigned to Beijing University of Technology. 6. Guicquero, J. P.; Maitrasse, P.; Mosquet, M. A.; and Sers, A., “A water soluble or water dispersible dispersing agent”, FR Patent 2,776,285, 1999, assigned to Chryso. 7. Hirata, T.; Yuasa, T.; and Nagare, K., “Cement admixture and cement composition”, US patent 6,166,122, 2000, assigned to Nippon Shokubai. 8. Teresa, M.; Laguna, R.; Medrano, R.; Plana, M. P.; and Tarazona, M. P., “Polymer characterization by size-exclusion chromatography with multiple detection,” Journal of Chromatography. A, V. 919, No. 1, 2001, pp. 13-19. doi: 10.1016/S0021-9673(01)00802-0 9. Plank, J., and Sachsenhauser, B., “Experimental determination of the effective anionic charge density of polycarboxylate superplasticizers in cement pore solution,” Cement and Concrete Research, V. 39, No. 1, 2009, pp. 1-5. doi: 10.1016/j.cemconres.2008.09.001 10. Ghaemy, M., and Roohina, S., “Grafting of Maleic Anhydride on Polyethylene in a Homogeneous Medium in the Presence of Radical Initiators,” Iranian Polymer Journal, V. 12, 2003, pp. 21-29. 11. Gaylord, N. G.; Mehta, R.; Kumar, V.; and Tazi, M., “High density polyethylene-gmaleic anhydride preparation in presence of electron donors,” Journal of Applied Polymer Science, V. 38, No. 2, 1989, pp. 359-371. doi: 10.1002/app.1989.070380217 12. Gaylord, N. G., and Mehta, R., “Peroxide-catalyzed grafting of maleic anhydride on to molten polyethylene in the presence of polar organic compounds,” Journal of Polymer Science. Part A, Polymer Chemistry, V. 26, No. 4, 1988, pp. 1189-1198. doi: 10.1002/ pola.1988.080260419 13. Cheng, Q.; Lu, Z.; and Byrne, H., “Synthesis of maleic anhydride grafted polypropylene-butadiene copolymer and its application in PP/OMMT/SBS composite as compatibilizer,” Journal of Applied Polymer Science, V. 114, No. 3, 2009, pp. 1820-182. doi: 10.1002/app.30678 14. Yin, J. H.; Shi, D.; and Yang, J. H. et al., “Functionalization of isotactic polypropylene with maleic anhydride by reactive extrusion: mechanism of melt grafting,” Polymer, V. 42, No. 13, 2001, pp. 5549-5557. doi: 10.1016/S0032-3861(01)00069-6

SP-302-13

Impact of Slags Contained in Blended Cement on Dispersing Effectiveness of PCEs by Ahmad Habbaba and Johann Plank In portland slag cements (PSC), different slag compositions can produce variations in workability due to the disparity in the surface chemistry of the slags. Here, the surface chemistry of different PSCs dispersed in water was studied in the absence and presence of polycarboxylate (PCE) superplasticizers. Six PSCs were prepared by mixing portland cement with 30 or 70 wt.% of three slags. As PCEs, two copolymers based on methacrylic acid–co–ω–methoxy poly(ethylene glycol) methacrylate ester were employed. It was found that the slags sequester ions from the pore solution, namely Ca2+ and SO42- ions forming an electrical double layer on the slag surface. Zeta potential measurements confirmed that different slags can exhibit different surface charges which can strongly affect PCE adsorption. The differences in the amounts of PCEs adsorbed result in different dosages required to achieve comparable dispersion. Generally, all slag cements tested required less PCE to achieve the same fluidity as with neat cement. Keywords: adsorption; anionic charge amount; polycarboxylate; pore solution; portland slag cement; superplasticizer; zeta potential. INTRODUCTION Cement is considered to present the largest industrially made product in the world. Due to its calcination process at 1450 °C (2,642 °F), the manufacturing of cement consumes a huge amount of energy which is associated with a significant emission of greenhouse gas (7% of the world’s total CO2 emission).1 To overcome this environmental problem, portland cement is blended with other supplementary cementitious materials (SCMs) to produce composite cements (e.g. CEM II and CEM III). Blast furnace slag obtained from the iron and steel industry belongs to the group of SCMs which are blended with portland cement clinker and gypsum to produce portland slag cement (PSC).2 Nowadays, PSC is widely recognized as an eco-friendly binder because it can replace a large amount of clinker (up to 80%) and thus reduce CO2 emission from cement production. In PSC, slag not only acts as a mere filler or aggregate, but it can also enhance the properties of hardened concrete by improving sulfate and chloride resistance. This effect is owed to the low content of C3A in 169

170 SP-302-13

PSC as well as a decrease in the permeability against water and different types of ions.3-6 It has also been reported that compared to the non-blended portland cement, dispersants show higher fluidizing properties in PSC.7 Polycarboxylate (PCE) based superplasticizers are applied in the construction industry to produce highly flowable concrete possessing a low water-to-cement ratio (w/c). Generally, PCEs are comb-shaped copolymers which consist of a negatively charged backbone comprised of carboxylate groups, and uncharged graft chains, mainly polyethylene glycols. The charged backbone of PCE can adsorb onto the surface of hydrated cement particles in three different possible conformations (train, loop or tail), while the non-adsorbed side chains freely protrude from the cement surface into the pore solution.8 Therefore, the anionic charge amount of a PCE polymer presents a main factor guiding its adsorption behavior and thus its dispersing performance. The higher the anionic charge amount of a PCE molecule, the higher its adsorbed amount. The mechanism behind the dispersing effect of a PCE is based on a combination of steric hindrance and electrostatic repulsive forces between the cementitious particles.9-11 In previous works by the authors, the surface chemistry of slag dispersed in deionized water and synthetic cement pore solution (SCPS) has been studied extensively.12,13 The major observation was that in slag suspensions, strong interaction occurs between the surface of slag and the ions present in the cement pore solution. First, at the high pH value of the suspension, the surface of slag becomes negatively charged as a result of deprotonation of silanol groups. This negatively charged surface then adsorbs calcium ions, thus forming a layer of Ca2+ on the surface of slag which in turn attracts sulfate ions present in the pore solution, hence forming a second ion layer consisting of sulfate ions. Therefore, when slag is dispersed in SCPS, the apparent surface charge of slag is negative, because of the high sulfate content. Further deprotonation of silanol groups present on slag occurs over time, and more calcium and sulfate ions are adsorbed. These processes alter the zeta potential of slag slurries after a certain time period (~ 3 h) until a stable state of equilibrium is reached. Furthermore, interaction between slag and PCE superplasticizers was studied as well.13 It was found that a competitive adsorption between the polymers and sulfate ions for positively charged sites present on the slag surface occurs. Sufficiently anionic PCE molecules occupy adsorption sites on slag while PCEs of low anionic character cannot compete with SO42- and hence do not adsorb in large amounts on the surface of slag. The study here continues the previous work from the authors on pure slags and now investigates the interaction between PCEs and different portland slag cements (PSCs). At first, the electrical surface charge of PSC samples dispersed in water was determined using a zeta potential instrument. Next, the concentrations of Ca2+ and SO42- ions contained in the pore solutions were determined and compared with those occurring in neat cement pore solutions. Furthermore, the physico-chemical interactions between PCE and PSC were studied via adsorption and zeta potential measurements. Based on these experimental results, the mechanism behind the different dispersing effect of two chemically different PCE polymers added to PSC pastes comprised of slags from different sources will be discussed. RESEARCH SIGNIFICANCE Previous studies on the interaction between slag cements and polycarboxylates have considered slag as an inert component which has no influence on this interaction. There-

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  171 Table 1 – Oxide compositions and properties of cement and GGBFS samples studied Oxide content (wt. %) SiO2 CaO Al2O3 MgO TiO2 K2O Na2O Fe2O3 Mn3O4 SO3 SrO ZrO2 BaO P2O5 Spec. surface area (Blaine) [cm2/g (ft2/oz)] d50 value [μm]* Density [g/cm3 (oz/in3)]

CEM I 52.5N 23.56 67.76 3.58 0.53 0.24 0.73 0.02 1.29 0.04 2.61 0.20 0.00 0.05 0.18

Slag S1 35.9 42.8 11.4 6.44 0.82 0.33 0.27 0.45 0.28 2.40 0.09 0.03 0.13 0.00

Slag S2 36.3 36.4 11.5 11.50 0.78 0.66 0.34 0.26 0.22 2.57 0.09 0.03 0.19 0.00

Slag S3 38.6 38.6 12.4 6.40 0.82 0.53 0.45 0.46 0.26 1.55 0.09 0.03 0.00 0.00

3400 (102)

4000 (120)

3480 (104.4)

4080 (122.4)

11.52 3.16 (1.83)

9.53 2.86 (1.65)

10.19 2.91 (1.68)

9.25 2.91 (1.68)

*conversion factor: 1 µm = 3.937 × 10-6 in

fore, the impact of different types of slag on this interaction was ignored. Various slags can result in considerable differences in workability when incorporated into cement. This effect is due to the disparity in the surface chemistry of different slag samples which results from their different ability to adsorb calcium ions. The anionic charge amount of a PCE polymer and the thickness of the calcium layer present the key factors for the dispersing effect of PCE in PSC. EXPERIMENTAL INVESTIGATION Materials A commercial portland cement sample (CEM I 52.5 N) was used as base cement to prepare a total of six blends whereby 30 or 70 wt.% of the cement were replaced with either slag S1, slag S2 or slag S3. The three ground granulated blast furnace slag (GGBFS) samples were from different sources in Germany. Table 1 lists the oxide composition (XRF), specific surface area (Blaine instrument) and particle size distribution (d50 value; laser granulometer) of the neat cement sample and the three slag samples. Two PCE superplasticizers (denominated as 45PC1.5 and 45PC6) were synthesized and used in the tests. Free aqueous radical copolymerization was employed to synthesize the copolymers using sodium peroxodisulfate as initiator and methallyl sulfonic acid as chain transfer agent. A detailed description of the synthesis process has been published in previous work.14 These PCEs are based on a copolymer of methacrylic acid and ω–methoxy poly(ethylene glycol) methacrylate ester (so-called MPEG type PCE). Their general chemical structure is displayed in Fig. 1. In the designations 45PC1.5 and 45PC6, “45” refers to

172 SP-302-13

Fig. 1 – Chemical structure of the synthesized PCE samples Table 2 – Characteristic properties of the synthesized PCE polymers Copolymer 45PC1.5 45PC6

Molar ratio methacrylic acid:ester   1.5   6.0

Side chain nEO 45 45

MW (g/mol)* 196,300 222,300

Mn (g/mol)* 51,900 52,340

Polydispersity index (MW/Mn)    3.8    4.2

Hydrodynamic radius Rh(avg) (nm)**    8.7    10.4

*conversion factor: 1 g/mol = 0.035 oz/mol; **conversion factor: 1 nm = 3.937 × 10-8 in

the number of ethylene oxide units (nEO) present in the side chain, whereas “1.5” and “6” refer to the molar ratio between methacrylic acid and the MPEG methacrylate ester. The characteristic properties of the PCE polymers are presented in Table 2. Methods Rheological properties – Cement pastes were prepared and their flow properties were determined using a ‘mini slump’ test according to DIN EN 1015. The w/c ratios of the cement pastes were selected such as to produce a flow value (spread) of 18 ± 0.5 cm (7.1 ± 0.2 in) in the absence of PCE polymer (Table 3). In measurements incorporating PCE, the polymer was added to the mixing water, and its dosage was adjusted to produce a flow value of 26 ± 0.5 cm (10.2 ± 0.2 in). The test was carried out as follows: In a porcelain cup, over 1 min
300 g (10.58 oz) of PSC were added to the specific amount of mixing water as given in Table 3, then left to soak for 1 min which was followed by manual stirring with a spoon for 2 min. Immediately after the end of stirring, the slurry was poured into a Vicat cone [height 40 mm (1.57 in), top diameter 70 mm (2.76 in), bottom diameter 80 mm (3.15 in)] placed on a glass plate and filled to the brim. The cone was removed vertically and the resulting spread of the paste (= diameter of slurry cake) was taken as flow value of the slurry. The diameter was measured twice perpendicularly, and the two values were aver-

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  173 Table 3 – Water to cement ratios determined via “mini slump test” to achieve 18±0.5 cm spread Sample w/c ratio Cal. w/c*

CEM I 52.5 N 0.505 0.505

30% S1 0.57 0.81

30% S2 0.55 0.79

30% S3 0.52 0.74

70% S1 0.60 2.00

70% S2 0.60 2.00

70% S3 0.55 1.83

* calculated as a ratio between the cement part present in PSC and water

aged to give the slump spread value. Each test was repeated three times, and the average was reported as slump flow value (the margin of error was ± 3%). Generally, the amount of water introduced with the PCE solution was subtracted from the amount of mixing water to maintain comparable w/c ratios. The dosage of PCEs is expressed on a dry mass basis and is stated in % by weight of cement (% bwoc). Ion concentrations in pore solutions – From these cement pastes, 10 mL (0.338 fl oz) were taken in 20 min intervals over a total period of 180 min, and then were centrifuged for 10 min (8,500 rpm) and diluted with 0.1N HCl to avoid precipitation of calcium carbonate. The ion concentrations were obtained from an atomic absorption spectroscope. Additionally, sulfate concentrations were quantified by utilizing ion chromatography. Electrokinetic properties of blended cements – Electrokinetic properties were measured using Electroacoustic Spectrometer. The highly solids loaded suspensions used in this work require an electroacoustic instrument to obtain zeta potential values which are representative of the conditions occurring in actual concrete.15 Zeta potential values were measured during the dropwise addition of aqueous solutions of the copolymers (concentration 10 wt. %, pH=7) to the cement pastes, and the zeta potentials of the slurries were recorded as a function of PCE concentration. PCE adsorption on PSC – Polymer adsorption was determined according to the depletion method. Different dosages of PCE copolymer were added to the individual cement pastes (w/c ratios as listed in Table 3), stirred for 2 min, centrifuged for 10 min (8,500 rpm) and then diluted with 0.1 N HCl to remove inorganic carbonates and to prevent dissolution of carbon dioxide in the alkaline solution. A High TOC II apparatus was employed to determine the total organic carbon content in the supernatants. The adsorbed amount of PCE was calculated by subtracting the concentration of PCE found in the supernatant from the initial PCE concentration used prior to contact with PSC. EXPERIMENTAL RESULTS AND DISCUSSION Zeta potential of PSC pastes In previous publications where the authors studied neat slag suspensions, it was found that in alkaline solution the surface of slag is generally negatively charged as a result of deprotonation of the silanol groups present on the surface.12,13 This charged surface attracts counter ions from the pore solution which then adsorb and form an electrical double layer on the surface of slag. The thickness of this layer mainly depends on two parameters: the type of slag (i.e. the number of silanol groups) and the ionic strength of the pore solution. Therefore, different slag samples exhibit different zeta potential values. Over time, the zeta potential values change due to partial hydration of the slag whereby some ions are released into the pore solution. These results explain the fundamental behavior of slag in cement pore solution and were taken as basis for the following experiments with the slag cement samples.

174 SP-302-13

Fig. 2 – Time-dependent zeta potential of CEM I 52.5 N and PSC paste (w/c ratios as shown in Table 3) For all 30% PSC samples (i.e. cement containing 30% slag), the initial zeta potential values were negative (~ - 4.5 mV), as is shown in Fig. 2, while the 70% PSC samples consistently exhibited positive initial zeta potential values (e.g. +3.5, +1.6 and +1.5 mV for 70% S1, 70% S2 and 70% S3, respectively). When considering the initial zeta potential value for the neat cement slurry (- 6 mV), it seems that the addition of slag to cement changes the zeta potential to less negative or even positive values, depending on the amount of slag added. Furthermore, the PSC samples containing the same amount of slag (30% or 70%) still exhibit slightly different initial zeta potential values, signifying that variations in the chemical composition of the slags cause different behaviors. For the slag samples studied here, the zeta potential values of the resulting PSC blends follow the trend S1 > S2 ~ S3 which infers a higher positive charge to the blend incorporating S1 than S2 or S3. These results are in agreement with previous results on pure slag.12,13 Over time, the surface charges of the PSCs increased to more positive values and stabilized after ~ 3 h (Fig. 2). The results indicate that the zeta potential values of PSCs are strongly related to the amount and type of slag added. Ion Concentrations in PSC Pore Solution The time - dependent evolution of Ca2+ and SO42- concentrations present in the pore solutions of all PSC slurries was measured. These measurements were performed to determine the impact of each slag on the ionic strength of the pore solutions. For comparison, neat portland cement slurries prepared at the same w/c ratios as the PSCs (see Table 3) were also analyzed. Their w/c ratios were calculated assuming that slag constitutes an inert component in the PSC samples. When compared with the neat portland cement slurries, lower concentrations of Ca2+ and SO42- ions in the pore solutions were observed for all PSC samples. For the 30% PSC samples, the slags seem to consume significant amounts of Ca2+ (Fig. 3 –top). This trend was even stronger for PSCs possessing high slag contents (70% PSC) (Fig. 3 – bottom). In all PSC pastes the changes in Ca2+ concentrations over time followed almost the same

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  175

Fig. 3 – Time-dependent evolution of Ca2+ concentrations in CEM I 52.5 N and PSC pore solutions (conversion factor:1 g/L = 0.001 oz/fl oz) trend found for neat portland cement pastes. This effect is due to the continuous release of Ca2+ ions from the cement at the same rate as the slag consumes Ca2+. On the other hand, the 30% PSCs consume considerable amounts of sulfate ions from the pore solution, compared to the related neat portland cement samples (Fig. 4 – top). The sulfate ions released from the slag samples were not significant because of the substantially higher concentration of sulfate present in cement. Contrary to this, the sulfate concentrations occurring in the pore solutions of 70% PSCs are lower or comparable to those from the neat cement (Fig. 4 – bottom). According to these data, slag S2 releases the highest amount of SO42- ions into the pore solution while slag S3 produces a medium and slag S1 the lowest concentration of sulfate ions in the pore solution. The same trend was observed before for the pure slag samples.13 Note that the concentrations of SO42- ions in the pore solutions of the 70% PSCs is lower than that of the 30% PSCs. This signifies that in the 30% PSC samples, the negative sulfate layer adsorbed onto the positive sites of the slag surface is thicker. It explains why the 30% PSCs exhibit negative zeta potential values while the 70% PSCs show positive zeta potentials (see Fig. 2). In general, the presence of slag in blended cements significantly changes the ionic strength of the pore solution, compared to the neat cement. Furthermore, the pore solution compositions of PSC samples containing the same amount of different slags can differ considerably as a result of variations in the chemical composition of the slags. This finding confirms that in such composite cements, slag is not an inert constituent.

176 SP-302-13

Fig. 4 – Time-dependent evolution of SO42- concentrations in CEM I 52.5 N and PSC pore solutions (conversion factor:1 g/L = 0.001 oz/fl oz) Dispersing effectiveness of PCEs in PSC Similar to what has been reported before from another research group,7 PCEs show higher fluidizing effect in PSC compared to the non-blended portland cement. Here, for the 70% PSC pastes, enhanced flow properties were recorded compared to the 30% PSCs. However, the dosage of PCEs required to achieve the same paste spread of PSC samples incorporating the same amount of different slags varied significantly. Generally, in CEM I 52.5 N as well as PSC pastes significantly lower dosages of PCE polymer 45PC6 (0.012 – 0.06% bwos) were required to achieve the same paste spread than with 45PC1.5 (0.09 – 0.55% bwos), as is shown in Fig. 5. The reason behind this result is the high anionic charge amount of polymer 45PC6 [~ 1200 µeq/g (34286 µeq/oz)]. Obviously, PCE polymer 45PC1.5 which is a polycarboxylate typically used in ready-mix concrete to provide extended slump life presents a less effective dispersant. To clarify the mechanism behind this effect, the adsorption of both polymers on the PSCs was compared. Adsorption of PCE on PSC For the polymers 45PC1.5 and 45PC6, adsorption isotherms in PSC pastes were produced using TOC measurement. In general, for both PCEs the adsorbed amounts increase with dosage until they reach a saturation point (Langmuir type isotherm; see Fig. 6). The saturated adsorbed amounts are the lowest for the cement blended with slag S3 and the highest for cements which contain slag S1.

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  177

Fig. 5 – Dosages of polycarboxylate superplasticizers required to obtain a target slump flow of 26±0.5 cm (10.2±0.2 in) for PSC pastes (w/c ratios as shown in Table 3) Comparable adsorbed amounts of PCE 45PC1.5 were found for all 30% PSCs (see Fig. 6 – top left). The same trend has been observed before for the pure slags dispersed in SCPS.13 Apparently, adsorption of 45PC1.5 is driven by a gain in entropy and not by electrostatic attractive forces, as in 30% blended slag systems its adsorption is independent of the different positive surface charges of the slag.16 Furthermore, the low adsorbed amount explains why 45PC1.5 is not an effective dispersant, as was shown in Fig. 5. In general, for the 70% PSCs the adsorbed amounts of 45PC6 and 45PC1.5 follow the order: Slag S1 > S2 > S3. This trend correlates well with the zeta potential results presented in Fig. 2: PSCs containing slag S1 exhibit the most positive surface charge while those blended with slag S3 show the lowest. The adsorbed amounts of PCE polymer 45PC6 on the blended cements were consistently lower than those on the neat cement (Fig. 6). This explains why sample CEM I 52.5 N required a higher dosage of this polymer to achieve the 26 cm (10.2 in) spread compared to the same cement blended with slag (Fig. 5). The reason behind this effect is that cement particles adsorb higher amounts of this polymer [3 mg/g (21 gr/lb)] than pure slag [0.4 – 1.8 mg/g (2.8 – 12.6 gr/lb)].12,13 Apparently, PCE polymer 45PC6 exhibits a higher affinity for cement than for slag. Influence of PCE on Zeta potential of PSC slurry The evolution of zeta potential during dropwise addition of the PCE solutions to the PSC pastes is shown in Fig. 7. When PCE is added to the PSC slurries, a shift in zeta potential values towards the isoelectric point (IEP) occurs which confirms that adsorption of PCE does indeed take place. The reason behind such shift to the IEP is the steric effect of the polyethylene oxide (PEO) side chains contained in the PCEs. They move the shear plane of the zeta potential further away from the particle surface. At such distance, the zeta potential approaches a value close to zero.17

178 SP-302-13

Fig. 6 – Adsorption isotherms for PCE polymer 45PC1.5 (top) and 45PC6 (bottom) on PSC (w/c ratios as shown in Table 3) (conversion factor: 1 mg/g = 7.015 gr/lb)

Fig. 7 – Effect of dosage of PCE polymers 45PC1.5 (left) and 45PC6 (right) on zeta potential of PSC pastes (w/c ratios as shown in Table 3) CONCLUSIONS When blended with cement, various slags are known to produce different results with respect to workability of the concrete or mortar. This is due to the disparity in the surface chemistry of the slag samples which results from their dissimilar ability to sequester calcium and sulfate ions from the pore solution. Partial replacement of cement by slag alters the ionic strength of the pore solution, and this change varies depending on the composition of the slag. In PSC pastes, at first significant amounts of calcium ions are

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  179

Fig. 8 – Schematic illustration of the electrochemical double layer existing in the equilibrium state on the surface of the slag samples in PSC suspension, and the consequences for the adsorption behavior of PCE polymers on these slags released from the cement into the pore solution which are partially taken up by the negatively charged slag surface. The amount of calcium consumed by a slag sample depends on the absolute value of its negative surface charge which can vary. Slags having thick layers of Ca2+ ions on their surface adsorb higher quantities of PCE and thus require increased dosages of superplasticizers to achieve the same fluidity as slags possessing a thin Ca2+ ion layer. This mechanism is schematically summarized in Fig. 8. This study demonstrates that in blended cements, strong interaction between polycarboxylate superplasticizers and the surfaces of both slag and portland cement occurs. Apparently, slag is not inert towards PCE. In portland slag cements, normally lower polycarboxylate dosages are required compared to the pure cement. Furthermore, when using slags from different sources, the dosages of superplasticizer required to establish comparable

180 SP-302-13

flowability can vary significantly. Accordingly, different qualities of slag may impact the economies of portland slag cements in different ways. AUTHOR BIOS Ahmad Habbaba is a Postdoctoral researcher at the Chair for Construction Chemistry, Technische Universität München, Germany. He studied chemistry and received his BS, MS and Diploma degree from Aleppo University (Syria) and a Ph.D. degree from TU München. His research interests include interaction of admixtures with cement and supplementary cementitious materials (SCMs). Johann Plank is full Professor at the Institute of Inorganic Chemistry of Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemistry there. His research interests include cement chemistry, concrete admixtures, organic-inorganic composite and nano materials, concrete, dry-mix mortars and oil well cementing. ACKNOWLEDGMENTS A. Habbaba wishes to express his gratitude and sincere appreciation to the Chair for Construction Chemistry for financing his postdoctoral researches. REFERENCES 1. Malhotra, V. M., “Sustainability issues and concrete technology” Paper presented at 7th International Symposium on Cement and Concrete (ISCC), Jinan (China), May 9 - 12, 2010. 2. Moranville, M., “Cement made from Blast furnace slag”, in: ‘Lea’s Chemistry of Cement and Concrete’, 4th edition, Peter Hewlett, Elsevier Science & Technology Books, 2004, pp 637-678. 3. Frearson, J. P. H., “Sulphate resistance of combination of Portland cement and ground granulated blast furnace slag”, Proceedings, 2nd International Conference on Fly Ash, Silica Fume, Slag and Natural Pozzolans in Concrete, Madrid, Spain, ACI, SP-91, Detroit, MI, USA, 1986, pp. 1495–1524. 4. Osborne, G. J., “Durability of Portland blast-furnace slag cement concrete,” Cement and Concrete Composites, V. 21, No. 1, 1999, pp. 11-21. doi: 10.1016/S0958-9465(98)00032-8 5. Lang, E., “Blast furnace cements” in: Structure and performance of cements, 2nd ed. Bensted, J. and Barnes, P., Spon Press, 2002, pp. 310-325. 6. Fu, X.; Hou, W.; Yang, C.; Li, D.; and Wu, X., “Studies on high-strength slag and fly ash compound cement,” Cement and Concrete Research, V. 30, No. 8, 2000, pp. 12391243. doi: 10.1016/S0008-8846(00)00312-4 7. Palacios, M.; Puertas, F.; Bowen, P.; and Houst, Y. F., “Effect of PCs superplasticizers on the rheological properties and hydration process of slag-blended cement pastes,” Journal of Materials Science, V. 44, No. 10, 2009, pp. 2714-2723. doi: 10.1007/s10853-009-3356-4 8. Yamada, K.; Takahashi, T.; Hanehara, S.; and Matsuhisa, M., “Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer,” Cement and Concrete Research, V. 30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6

Impact of Slags Contained in Blended Cements on Dispersing Effectiveness of PCEs  181 9. Ramachandran, V. S., and Malhotra, V. M., “Superplasticizers”, in: Ramachandran, V. S., editor, Concrete Admixtures Handbook, 2nd ed, New Jersey, USA, Noyes publications, 1996, pp. 410-517. 10. Sakai, E.; Kang, J. K.; and Daimon, M., “Action mechanisms of comb-type superplasticizers containing grafted polyethylene oxide chains”, in: Malhotra, V. M. editor, 6th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Nice, France, SP-195, CANMET/ACI, 2000, pp. 75-90. 11. Blask, O., and Honert, D., “The electrostatic potential of highly filled cement suspensions containing various superplasticizers”, in: Malhotra, V. M., editor, 7th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Berlin, Germany, SP-217, CANMET/ACI, 2003, pp. 87-101. 12. Habbaba, A., and Plank, J., “Interaction Between Polycarboxylate Superplasticizers and Amorphous Ground Granulated Blast Furnace Slag,” Journal of the American Ceramic Society, V. 93, No. 9, 09 2010, pp. 2857-2863. doi: 10.1111/j.1551-2916.2010.03755.x 13. Habbaba, A., and Plank, J., “Surface Chemistry of Ground Granulated Blast Furnace Slag in Cement Pore Solution and Its Impact on the Effectiveness of Polycarboxylate Superplasticizers,” Journal of the American Ceramic Society, V. 95, No. 2, 2012, pp. 768-775. doi: 10.1111/j.1551-2916.2011.04968.x 14. Plank, J.; Pöllmann, K.; Zouaoui, N.; Andres, P. R.; and Schaefer, C., “Synthesis and performance of methacrylic ester based polycarboxylate superplasticizers possessing hydroxy terminated poly(ethylene glycol) side chain,” Cement and Concrete Research, V. 38, No. 10, 2008, pp. 1210-1216. doi: 10.1016/j.cemconres.2008.01.007 15. Dukhin, A. S., and Goetz, P. J., “Acoustic and Electroacoustic Spectroscopy,” Langmuir, V. 12, No. 18, 1996, pp. 4336-4344. doi: 10.1021/la951086q 16. Plank, J., and Sachsenhauser, B., “Experimental determination of the thermodynamic parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers”. V. M. Malhotra (Ed.) 9th CANMET/ACI Conference on superplasticizers and other chemical admixtures in concrete (supplementary papers), ACI, Seville, 2009, pp. 87-102. 17. Plank, J., and Sachsenhauser, B., “Impact of molecular structure on zeta potential and adsorption conformation of α-allyl-ω-methoxypolyethylene glycol - maleic anhydride superplasticizers,” Journal of Advanced Concrete Technology, V. 4, No. 2, 2006, pp. 233-239. doi: 10.3151/jact.4.233

182 SP-302-13

SP-302-14

Preparation and Characterization of StarShaped Polycarboxylate Superplasticizer by Xiao Liu, Ziming Wang, Jie Zhu, Ming Zhao, Wei Liu, and Dongjie Yin A polycarboxylate superplasticizer (PCE) with a novel star-shaped structure was prepared through copolymerization of acrylic acid (AA), isobutenyl polyethylene glycol (IPEG), and star-shaped polymerizable active center by an esterification between polyol and AA. In the first esterification step, the esterification rate reached more than 95% with the catalyst/polyol ratio of 0.07:1, inhibitor/AA ratio of 0.04:1 (or 0.011:1), water-carrying agent dosage of 70g and esterification time of 7 hours. In the second polymerization step, the highest fluidity of cement paste was achieved at the initiator/AA/IPEG ratio of 0.28: 3.3: 1. Infrared spectroscopy (IR) and 1H Nuclear magnetic resonance (1H NMR) measurements were used for structural characterization, and the spectral results confirmed the product’s star-shaped structure. Furthermore, this star-shaped PCE exhibited higher energy efficiency than the conventional comb-shaped PCE, indicated by its excellent paste fluidity and adsorption behavior in cement paste. Keywords: polycarboxylate superplasticizer; star-shaped; molecular design; adsorption; fluidity. INTRODUCTION In recent years, with the developing technology of construction engineering, the demand for increased concrete quality has become greater. Polycarboxylate superplasticizer (PCE) invented in 1981 as a novel class of superplasticizers clearly represented a milestone technology in construction engineering, due to its advantages of low dosage, high slump retention and water-reducing capability. PCE is a key chemical admixture which can be added to concrete mixtures to improve workability1 and prepare high-performance concrete, and has rapid development and growth in construction engineering.2,3 The main function of PCE is to enable a decrease in the water-cement ratio (W/C) without losing fluidity of the cement paste, which results in higher strength and better durability of concrete.4,5 The properties of the PCE polymer are closely related to its structure, and thus the structural innovation of PCE is very important to promote the development of new PCE technologies. At present, the structure of common PCE is a comb-shape structure, and thus it has exhibited a unique mechanism and excellent performance in cement paste or concrete systems up to 183

184 SP-302-14

now. As the modification of new molecular structures becomes a research focus, polymers with branched structures have received greater attention because of their unique properties and potential application prospects.6,7 The branched or star-shaped polymer has particular characteristics including low viscosity, more side chains and functional ends compared to conventional comb polymers. Therefore, a PCE with this structure is likely to bring higher adsorption efficiency and wider coverage on the surfaces of cement particles.8 According to this research idea, a star-shaped PCE is synthesized by means of the structure innovation, which relies on the theory of molecular design to achieve some unique characteristics. This type of PCE may exhibit higher adsorption efficiency and excellent paste fluidity by virtue of its special structure. Only a few researchers have reported the synthesis and properties of polymers with branched or star-shaped structures which were usually synthesized by a complex Atom Transfer Radical Polymerization (ATRP).9,10 However, there were no studies reported on the field of PCE with innovative branched or star-shaped structure synthesized by a simple synthetic technique for improving cement paste fluidity. In this study, a PCE with a star-shaped structure was successfully synthesized by the simple synthetic reactions without having to resort to the ATRP method. Its detailed reaction conditions were determined by investigating catalyst, inhibitor and water-carrying agent dosages and esterification time. Additionally the structure and performances of PCEs with either star-shape or conventional structure were respectively characterized and compared to confirm their unique structure and excellent fluidity performances. Finally, the effects of adsorption behaviors of the star-shaped PCE in cement paste system were evaluated. RESEARCH SIGNIFICANCE To improve the workability, efficiency and other performances of PCE is an important subject for both PCE researchers and to improve concrete performance. The structure innovation based on the theory of molecular design creates a polymer with novel structure, and depending on this, the synthesized PCE displays excellent performances to achieve the improvement of workability efficiency. Furthermore, this research pays researchers’ attentions to how to design PCE structures to solve application problems. This novel-structural type of PCE can enrich the relevant theories and widen its application to further provide the theoretical basis of synthesis of the high-workability PCE. Consequently, this novel PCE is suitable for the chemical admixture of high-performance concrete. EXPERIMENTAL INVESTIGATION Materials Reaction raw materials include: Isobutenyl polyethylene glycol (IPEG, Mn=2348g/ mol, Mw=2484g/mol), acrylic acid (AA), pentaerythritol, ammonium persulfate (APS), phenothiazine, toluene, concentrated sulfuric acid, P-toluene sulfonic acid, hydroquinone, sulfamic acid and sodium hydroxide (NaOH). Reference cement P.I.42.5 was supplied by China Building Materials Research Institute (Beijing, China). The chemical and mineral compositions of reference cement are illustrated in Table 1.

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 185 Table 1–Chemical and mineral compositions of reference cement Composition SiO2, % Al2O3, % Fe2O3, % CaO, % MgO, % SO3, % Na2O, % Loss, % f-CaO, % C3S, % C2S, % C3A, % C4AF, %

Reference cement P. I. 42.5 22.93 4.29 2.89 66.23 1.92 0.35 0.70 1.48 0.64 58.78 21.38 6.49 8.77

Fig. 1–Schematic diagram of the synthesis of star-shaped PCE. Synthesis and calculation The pentaerythritol was mixed with AA at a molar ratio of 1:5 in a four-neck roundbottom flask with a stirrer, and then the catalyst, inhibitor and water-carrying agent were added in sequence, followed by heating to the set temperature with stirring. At the end of reaction which lasted for several hours, the esterification products can be obtained by vacuum-distillation to remove the water-carrying agent. The esterification product and IPEG with a molar ratio of 1:20 were diluted by distilled water in a four-neck round-bottom flask which was placed in a constant temperature bath at 65°C with stirring. When stirred homogeneously, AA and initiator in aqueous solution was dropwise added to the flask. After reacting at a constant temperature for 5 hours, the NaOH aqueous solution was used for adjusting the pH value to 6-7 followed by cooling to room temperature, thus yielding the final product. The schematic diagram of the synthesis of the star-shaped PCE is shown in Fig. 1. The specimens used for the characterization were further purified by repeated precipitation into the excess diethyl ether. The calculation formula of esterification rate (E%) is listed as the following equation:

E% = [(10.8 − C NaOH × VNaOH × 40 − m) / 10.8] × 100%

(1)

186 SP-302-14

Where CNaOH = molar concentration of sodium hydroxide solution, 1 mol/L; VNaOH = consumed volume of sodium hydroxide solution for neutralization, L; m = mass of catalyst containing crystal water, g. Items of investigation The 1H Nuclear Magnetic Resonance (1H NMR) spectra of PCEs were obtained at room temperature (25°C) with an ARX-400 spectrometer (Bruker Co., Germany) operating at a frequency of 400MHz, and the chemical shift values were expressed in δ values (ppm) relative to tetramethylsilane (TMS) as an internal standard. Samples for 1H NMR were prepared by dissolving samples in the solvent, i.e., deuterated water (D2O) as internal reference. The Fourier Transform infrared (FTIR) spectra of the PCE was prepared by mixing a fixed mass of solid PCE with KBr, which was then pressed into a disk, and analyzed by a VERTEX 70 Fourier transform infrared spectrometer (Bruker Co., Germany) at room temperature (25°C). The solid PCE samples were prepared via dissolution and precipitation at least three times, and then dried in vacuum at 60°C for 24 hours to constant weight. The fluidity of cement pastes treated with PCE were measured according to the standard method (GB 8077-2000 Uniformity Test Method for Concrete Admixture)11 described in the National Standards of the People’s Republic of China, at a water-cement ratio (W/C) of 0.29, and the dosage was based on the weight ratio of solid PCE to cement. To investigate the effect of PCE dosage on the paste fluidity, the fluidities of cement pastes mixing with PCE at different dosages were examined. Moreover, to study the maintenance of fluidity, cement pastes mixed with PCE were examined every 60 min within a total period of 120 min. For each test of cement paste fluidity, the cement paste mixture was poured into a truncated cone (2.36 in. [60 mm] height, 1.42 in. [36 mm] top diameter, 2.36 in. [60 mm] bottom diameter) on a glass plate, and then the cone was vertically removed. After 30 seconds, the diameter of paste was recorded as the fluidity of paste, and the resulting spread of the paste was measured twice. The second measurement was perpendicular to the first measurement, and the average was calculated to yield the spread value. The adsorption amounts of PCEs on the surfaces of cement particles were evaluated according to a reported method.12 The non-adsorbed portion of polymer remaining in solution at equilibrium condition was determined by analyzing the total organic carbon (TOC) content of the solution. In a typical experiment, 30 g of cement, 120 g of deionized water and the amount of PCE (remaining the same liquid concentration as the fluidity testing of cement paste, namely 0.15:29) were mixed and then the homogeneous slurry was filled into a 50 mL centrifuge tube to be centrifuged for 5 min at 6000 rpm in a TGL-16C High speed bench centrifuge produced by ShangHai Anting Scientific Instrument Factory (Shanghai, China). The supernatant was diluted 100 times with deionized water, and then the total organic carbon of the solution was determined on a Vario TOC cube instrument (Elementar Analysensysteme GmbH, Germany). The adsorption amounts of PCE were calculated from the difference between the TOC content of the polymer reference sample in the initial solution and that of the supernatant of cement paste. The adsorption amount changes with time were obtained by testing the PCE adsorption after 5 min, 30 min, 60 min, 90 min and 120 min, respectively. Measurements were generally repeated three times and the average was reported as the adsorbed amount.

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 187 Table 2–Equilibrium reaction temperature and esterification rate at different dosages of water-carrying agent Water carrying agent, g 50 60 70 75 80 85

Equilibrium reaction temperature, °C 128 122 119 118 117.8 117

Esterification rate, % 58 57.9 66.5 62 63 62.4

EXPERIMENTAL RESULTS AND DISCUSSION Effect of water-carrying agent dosages on the equilibrium reaction temperature and esterification rate The effect of water-carrying agent dosages on the equilibrium reaction temperature and esterification rate is shown in Table 2. From Table 2, the esterification rate is the highest when the dosage of water-carrying agent is 70 g and the corresponding equilibrium reaction temperature is 119°C. In the range of water-carrying agent dosage from 50 to 70g, the esterification rate increases with the increase of water-carrying agent dosage, which results from the higher water-carrying efficiency (corresponding to higher dosage of water-carrying agent), leading to higher esterification rate. However, with continual increase of water-carrying agent dosage, the esterification rate decreased. This is attributed to the excessive amounts of water-carrying agent, leading to a low reactant concentration and a low esterification rate. Correspondingly, the equilibrium reaction temperature decreased with the increase of water-carrying agent dosage, which is also caused by the reflux of a great deal of cooled water-carrying agent. Effects of catalysts and their dosages on the esterification rate The concentrated sulfuric acid and p-toluene sulfonic acid were selected as catalysts. The ratio of acid to pentaerythritol is 1.25 and the dosage of water-carrying agent is 70 g in this experiment. The effects of catalysts and their dosages on the esterification rate are shown in Fig. 2. From Fig. 2, the esterification rate is the highest when the molar ratio of p-toluene sulfonic acid to alcohol is 0.07 and the molar ratio of sulfuric acid to alcohol is 0.11. However, the real value of esterification rate is likely to be affected owing to the total amount of water containing a part of dehydrated water for the dehydrating property of concentrated sulfuric acid. Insufficient catalyst dosages can lead to incomplete esterification reactions; similarly, excessive catalyst dosages can also lead to low esterification rate, which is possibly because the excessive dosage of catalyst leads to the occurrence of side reactions. Overall, the p-toluene sulfonic acid is chosen as the esterification catalyst, and its molar ratio to alcohol is 0.07. Effects of inhibitors and their dosages on the esterification rate The hydroquinone and phenothiazine were selected as inhibitors in this experiment; besides, the ratio of acid to pentaerythritol, the water-carrying agent dosage, and the molar

188 SP-302-14

Fig. 2–Esterification rates at catalysts’ different molar ratio to alcohol. Table 3–Effects of inhibitors and their dosage on the esterification rate Catalyst

Hydroquinone

Phenothiazine

Dosage, % 2 3 4 4.5 5 0.2 0.5 0.8 1.1 1.4 1.7

Esterification rate, % 94.9 95.0 95.3 93.3 93.0 88.7 93.1 92.4 95.6 91.7 94.2

ratio of catalyst to alcohol are 1.25, 70 g and 0.07, respectively. The effects of inhibitors and their dosages on the esterification rate are shown in Table 3. The dosage is the molar percentage of inhibitor to AA. From Table 3, phenothiazine plays an inhibiting effect at the lower dosages; whereas the hydroquinone can exhibit an effective inhibiting effect only at the relatively higher dosages. It also can be seen from Table 3 that the esterification rates are the highest when the molar ratios of hydroquinone and phenothiazine to AA are 4% and 1.1%, respectively. With the continual increase of inhibitor dosage, the esterification rate decreases. This is probably because the acrylic acid is apt to homopolymerize with each other rather than esterification reaction when the dosages of inhibitors are too low, leading to the decrease of esterification rate. With the continual increase of inhibitor dosage, the free radicals produced from reaction system will be excessively captured, which can interfere with or even eliminate the

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 189

Fig. 3–Esterification rates for different reaction times. role of catalyst and leads to the decrease of esterification rate. Therefore, the proper molar ratios of hydroquinone and phenothiazine to AA are 4% and 1.1%, respectively. Effect of reaction times on the esterification rate The esterification rate is also affected by the reaction time, and thus the investigated reaction times were selected from 4 hours to 8 hours. In this experiment, the ratio of acid to pentaerythritol, the water-carrying agent dosage, and the molar ratio of catalyst to alcohol are 1.25, 70 g and 0.07, respectively. The effect of reaction times on the esterification rate is shown in Fig. 3. The results show that the esterification rate increases with the increase of reaction time in the range of 4-6 hours. The esterification rate directly affects the quality of esterification product, and further correlates to the performance of the final synthesized PCE. Besides, the esterification reaction time should be theoretically prolonged at a proper esterification temperature to ensure the esterification rate as high as possible. However, in Fig. 3, the esterification reaction rate gradually increases after reacting for 6 hours, and it changes slightly with the extension of reaction time. Based on economical considerations and energy conservation, the proper esterification time therefore is determined as 7 hours. Characterization of esterification product The esterification product in this study was characterized by 1H NMR (Fig. 4). The esterification reaction occurred between the hydroxyl groups of pentaerythritol and the carboxyl groups of AA. During the process of the esterification reaction, its methylenes initially linked to the hydroxyl groups change to link to ester groups. As a result, the electron withdrawing effect of ester structure leads to its peak shifting to the lower field. From the peak positions in Fig. 4, the primary peaks of esterification product in 1H NMR spectrum correspond to the three H atoms on the -CH=CH2 bonds (6.154ppm, 6.328ppm,

190 SP-302-14

Fig. 4–1H NMR spectrum of esterification product. 5.963ppm). The structure diagram displayed in Fig.4 clearly indicates the peaks and their corresponding H atoms in the molecular structure. All of the displayed characteristic peaks confirm the occurrence of esterification reaction and the indication of a relatively ideal molecular structure for the esterification product. Effects of AA/IPEG ratios on fluidity performances of paste mixing with synthesized PCE After the confirmation of structure of esterification product which is used for the subsequent polymerization, the relevant polymerization conditions also should be determined. The effects of AA/IPEG ratios on the fluidity performances of paste mixing with synthesized PCE at a dosage of 0.15% are shown in Fig. 5. The ratio of AA to IPEG affects PCE’s structure and performance in cement paste. From Fig. 5, the fluidity of cement paste increased slightly with the increase of AA/IPEG ratio in the range of 2.6-3.3. This is possibly because the effective adsorption sites increase caused by the increased proportion of carboxyl groups with the increase of AA/IPEG ratio. Thereafter, the fluidity of cement paste reached maximum at the AA/IPEG ratio of 3.3 and then decreased with a continual increase of AA/IPEG ratio. This is possibly because the continual increase of AA/IPEG ratio is equivalent to decrease the density of side chain, which is a key factor to determine the workability and efficiency of PCE. Too low density of side chain will lower the steric hindrance effect of side chain (polyoxyethylene), and thus leads to the decrease of cement paste fluidity. Effects of initiator/IPEG ratios on fluidity performances of paste mixing with synthesized PCE There were better fluidity performances of cement pastes at the initiator/IPEG ratio of around 0.3 by experiments, and thus the initiator/IPEG ratio was investigated in the range of 0.26-0.38. The effects of initiator (APS)/IPEG ratios on the fluidity performances of paste mixing with synthesized PCE at a dosage of 0.15% are shown in Fig. 6.

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 191

Fig. 5–Fluidities of cement paste mixing with PCE at different AA dosages.

Fig. 6–Fluidities of cement paste mixing with PCE at different initiator/IPEG ratios. From Fig. 6, the fluidities of cement paste increased and then decreased with the increase of APS/IPEG ratio; besides, the cement paste exhibited the best fluidity at the APS/IPEG ratio of 0.28. The dosage of initiator affects the polymerization rate and molecular weight in a free radical polymerization. Too low dosage of initiator will produce fewer primary radicals, and thus cause incomplete polymerization. With the increase of initiator dosage, the polymerization rate is accelerated and the reaction becomes sufficient; furthermore, the suitable initiator dosage is helpful to obtain the product with suitable molecular

192 SP-302-14

Fig. 7–1H NMR spectrum of the synthesized polymerization product. weight, which optimizes its adsorption rate on the surfaces of cement particles and ensures good fluidity retention. Too high dosage of initiator will reduce the molecular weight of PCE, which can weaken the steric hindrance effect and further lead to the low fluidity performance. Characterizations of polymerization product The synthesized polymerization product was characterized by 1H NMR to confirm its star-shaped structure. The 1H NMR spectrum of the synthesized polymerization product is shown in Fig. 7. From Fig. 7, the peaks at 4.5-5.0 ppm and 3.3-3.8 ppm were distinctly higher than other peaks, because they correspond to the H atoms of the polyethylene glycol connected with the arm structure. Also from Fig. 7, the peaks at 2.314 and 1.645 ppm correspond to the H atoms of methylenes, whose schematic structure diagram is also displayed in Fig. 7. It is well-known that the peaks corresponding to H atoms of methylenes will not be displayed at these positions if the esterification product does not polymerize with other monomers or forms other products. The total integral area for these two peaks also indicates many starshaped structures have formed. All of these characteristic peaks confirm the polymerization reaction and the expected polymerization product with a star-shaped structure. The FTIR spectra of the conventional comb-shaped PCE and synthesized star-shaped PCE are shown in Fig. 8, and these spectra were analyzed by means of other reported references.13,14 It can be seen from Fig. 8 that the two PCEs both exhibit characteristic peaks at around 1108cm-1 and 1554cm-1. The peak at 1108 cm-1 belongs to the vibration of polyethylene glycol chain, and the characteristic peak at 2500-3300 cm-1 corresponds to its stretching vibration. The peak at 1554 cm-1 belongs to the symmetric vibration of C=O bond in carboxylic acid. By contrast, there is a large peak at around 1772cm-1 in the spectrum of the star-shaped PCE but no peak at this position in the spectrum of the conventional

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 193

Fig. 8–FTIR spectra of the synthesized polymerization product. comb-shaped PCE. The peak in this position corresponds to the ester groups, and thus, this result proves the star-shaped structure of this PCE due to the esterification reaction. All of the analysis indicates that the star-shaped PCE has not only the conventional groups such as polyethylene glycol and carbonyl groups, but also the characteristic groups, i.e., ester groups, which confirm the completion of reactions and the achievement of the targeted structure. Fluidity performances of the pastes mixing with PCEs The star-shaped PCE was synthesized according to the above determined conditions; besides, the conventional PCE as reference was synthesized with the same conditions as the polymerization-step of star-shaped PCE. The fluidity performances of the cement pastes mixing with the two PCEs respectively at a dosage of 0.15% are shown in Fig. 9. From Fig. 9, the saturation dosages of the synthesized star-shaped and conventional comb-shaped PCEs were about 0.3% and 0.4%, respectively. These results indicate that excellent fluidity performance can be achieved for lower dosages of star-shaped PCE, demonstrating its high workability efficiency. Furthermore, the fluidity values of the synthesized star-shaped PCE were higher than those of conventional comb-shaped PCE over the range of experimental dosages. Adsorption mechanism of star-shaped PCE in cement paste As is well-known, the carboxyl groups of PCE adsorbs on the surfaces of cement particles when added to the cement paste; besides, the hydrophilic part, i.e., polyethylene glycols of PCE can retain the fluidity of cement paste. Thus, the PCE’s adsorption behavior on the surfaces of cement particles is an important factor to investigate the mechanism of star-shaped PCE in cement paste. The adsorption behaviors and the relationship between

194 SP-302-14

Fig. 9–Fluidities of cement pastes mixing with PCEs at different dosages (W/C=0.29).

Fig. 10–Adsorption behaviors of star-shaped and conventional PCEs in cement paste (PCE dosage=0.15:29). adsorption amount and fluidity for star-shaped and conventional PCEs in cement pastes are shown in Fig. 10 and Fig. 11 respectively. It can be seen from Fig. 10 and Fig. 11, the adsorption amount and rate of star-shaped PCE are higher than those of conventional comb-shaped PCE, but also the similar results were displayed in terms of relationship between adsorption amount and fluidity. This is caused by their different molecular structures. For the star-shaped PCE compared with conventional PCE, its multi-arm structure leads to a stronger steric-hindrance effect; moreover, it

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 195

Fig. 11–Relationship between adsorption amount and fluidity for star-shaped and conventional PCEs in cement paste. has smaller hydrodynamic volume15 which leads to a higher contents of polar groups per unit volume and has more adsorption sites which leads to a stronger affinity and a higher probability of surface adsorption. This is in accordance with the results reported by Kazuo Yamada.16 Also, with the hydration process, a part of conventional PCE molecules gradually lose the workability for the coverage of hydrate layer by its lower adsorption amount and more surfaces of cement particles. This well agrees with other researchers’ study17 investigating the effects of charge density of PCE on the dispersant behavior and showing that the dispersion effect of PCE is well correlated to the amount of adsorbed polymer and can be interpreted in terms of surface coverage. Then, we suppose that the star-shaped PCE still has other “arms” dispersed in the paste pore solution, and the hydrate layer is still covered by these free “arms”, leading to good dispersion retaining of cement particles in the cement paste. This speculative mechanism is diagramed in Fig. 12. Based on the above results and analysis, the good fluidity performances of star-shaped PCE are attributed to its special multi-arm structure, which has higher energy efficiency to achieve a better workability in cement pastes. This star-shaped PCE is one member of a large PCE family which has very broad performances. It is meaningful for researchers to investigate the PCE with novel structure, which can diversify the PCE family. CONCLUSIONS Based on the results of this experimental investigation, the following conclusions are drawn: 1. A star-shaped PCE was successfully synthesized through two-step reactions: the first esterification between pentaerythritol and AA, and the second free radical polymerization among the esterification product, IPEG and AA.

196 SP-302-14

Fig. 12–Adsorption mechanism of star-shaped PCE in cement paste. 2. The esterification rate can reach above 95% at a water-carrying agent dosage of 70g, a catalyst/alcohol molar ratio of 0.07:1, an inhibitor (hydroquinone) /AA molar ratio of 0.04:1 (or phenothiazine/AA molar ratio of 0.011:1), and a reaction time of 7 hours; furthermore, in the second polymerization step, the best fluidity of cement paste was achieved at the initiator/AA/IPEG ratio of 0.28: 3.3: 1. 3. The structure of esterification product was characterized by its 1H NMR spectrum, which confirms the occurrence of esterification; moreover, the structure of final synthesized star-shaped PCE was characterized by its 1H NMR and FTIR spectra, which confirm the completion of subsequent polymerization and the achievement of an ideal star-shaped structure. 4. The saturation dosages of the synthesized star-shaped PCE and a conventional combshaped PCE are 0.3% and 0.4% respectively, indicating star-shaped PCE’s dosage improved efficiency; besides, the synthesized star-shaped PCE exhibits better fluidity performance and adsorption behaviors. 5. The improved performances of star-shaped PCE are attributed to its novel branched multi-arm structure with higher energy efficiency, and thus it can be considered as a newtype PCE to provide the theoretical basis and technological application in cement and concrete research. AUTHOR BIOS Xiao Liu is an associate professor at the College of Materials Science and Engineering in Beijing University of Technology. He majored in Materials Science and Engineering and received his PhD from Beijing University of Chemical Technology. His research interests include functionalized polycarboxylate superplasticizer with high performance by molecular structure design, as well as the interface adsorption of polycarboxylate superplasticizer in cement paste system.

Preparation and Characterization of Star-Shaped Polycarboxylate Superplasticizer 197 Ziming Wang is a professor at the College of Materials Science and Engineering in Beijing University of Technology. He majored in Materials Science and Engineering and received his PhD from Beijing University of Technology. He is a vice executive secretary of Association of Concrete Admixture. His research interests include high-performance cement-based materials and rheology of cement paste. Jie Zhu is a Researcher Engineer at Beijing BBMG Cement Energy Technology Co., Ltd. Ming Zhao is a master student in Beijing University of Technology. Wei Liu is a master student in Beijing University of Technology. Dongjie Yin is a master student in Beijing University of Technology. ACKNOWLEDGMENTS The authors wish to express their gratitude and sincere appreciation to the National Natural Science Foundation of China (Grant number: 51208012), Research Fund of New Teachers for the Doctoral Program of Higher Education of China (Grant number: 3c009011201301), Project of Central Research Institute of Building and Construction Co., Ltd (Contract number: CBM2014Ky01-01) and Project of China Railway Engineering Materials Technology (Anhui) Ltd (Project number: 40009011201409) for financing this research work. REFERENCES 1. Büyükyağcı, A.; Tuzcu, G.; and Aras, L., “Synthesis of copolymers of methoxy polyethylene glycol acrylate and 2-acrylamido-2-methyl-1-propanesulfonic acid: Its characterization and application as superplasticizer in concrete,” Cement and Concrete Research, V. 39, No. 7, 2009, pp. 629-635. doi: 10.1016/j.cemconres.2009.03.010 2. Sakai, E.; Ishida, A.; and Ohta, A., “New trends in the development of chemical admixtures in Japan,” Journal of Advanced Concrete Technology, V. 4, No. 2, 2006, pp. 1-13. doi: 10.3151/crt1990.17.2_1 3. Cerulli, T.; Clemente, P.; Decio, M.; Ferrari, G.; Gamba, M.; Salvioni, D.; and Surico, F., “A new superplasticizer for early high-strength development in cold climates,” Seventh CANMET/ACI International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, edited by V, M, Malhotra., American Concrete Institute, Farmington Hills, 2003, 113 pp. 4. Lu, S., “H., Liu, G., Ma, Y, F., and Li, F., “Synthesis and application of a new vinyl copolymer superplasticizer,” Journal of Applied Polymer Science, V. 117, No. 1, 2010, pp. 273-280. 5. Plank, J., and Dai, Z., “M., Keller, H., Hossle, H., and Seidl, W., “Fundamental mechanisms for polycarboxylate intercalation into C3A hydrate phases and the role of sulfate present in cement,” Cement and Concrete Research, V. 40, No. 1, 2010, pp. 45-57. doi: 10.1016/j.cemconres.2009.08.013

198 SP-302-14

6. Voit, B., “New developments in hyperbranched polymers,” Journal of Polymer Science. Part A, Polymer Chemistry, V. 38, No. 14, 2000, pp. 2505-2525. doi: 10.1002/1099-0518(20000715)38:143.0.CO;2-8 7. Jikei, M., and Kakimoto, M., “Hyperbranched polymers: a promising new class of materials,” Progress in Polymer Science, V. 26, No. 8, 2001, pp. 1233-1285. doi: 10.1016/ S0079-6700(01)00018-1 8. Liu, X., and Wang, Z., M., and Cui, S, P., “Preparation method of high-performance star-shaped polycarboxylate superplasticizer,” US patent, 13/910, 785, 2013. 9. Dai, F.; Sun, P.; Liu, Y.; and Liu, W., “Redox-cleavable star cationic PDMAEMA by arm-first approach of ATRP as a nonviral vector for gene delivery,” Biomaterials, V. 31, No. 3, 2010, pp. 559-569. doi: 10.1016/j.biomaterials.2009.09.055 10. Francis, R.; Lepoittevin, B.; Taton, D.; and Gnanou, Y., “Toward an easy access to asymmetric stars and miktoarm stars by atom transfer radical polymerization,” Macromolecules, V. 35, No. 24, 2002, pp. 9001-9008. doi: 10.1021/ma020872g 11. GB 8077-2000, “Methods for Testing Uniformity of Concrete Admixture,” State Bureau of Quality and Technical Supervision of the People’s Republic of China, 2000. (in Chinese). 12. Plank, J.; Sachsenhauser, B.; and Reese, J., “D., “Experimental determination of the thermodynamic parameters affecting the adsorption behaviour and dispersion effectiveness of PCE superplasticizers,” Cement and Concrete Research, V. 40, No. 5, 2010, pp. 699-709. doi: 10.1016/j.cemconres.2009.12.002 13. Verdonck, B.; Gohy, J.; Khousakoun, E.; Jérôme, R.; and Prez, F., “D., “Association behavior of thermo-responsive block copolymers based on poly(vinyl ethers),” Polymer, V. 46, No. 23, 2005, pp. 9899-9907. doi: 10.1016/j.polymer.2005.07.079 14. Plank, J., and Yu, B., “Preparation of hydrocalumite-based nanocomposites using polycarboxylate comb polymers possessing high grafting density as interlayer spacers,” Applied Clay Science, V. 47, No. 3-4, 2010, pp. 378-383. doi: 10.1016/j.clay.2009.11.057 15. Aggarwal, S., “L., “Structure and properties of block polymers and multiphase polymer systems: an overview of present status and future potential,” Polymer, V. 17, No. 11, 1976, pp. 938-956. doi: 10.1016/0032-3861(76)90170-1 16. Yamada, K.; Takahashi, T.; Hanehara, S.; and Matsuhisa, M., “Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer,” Cement and Concrete Research, V. 30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6 17. Pourchet, S.; Liautaud, S.; Rinaldi, D.; and Pochard, I., “Effect of the repartition of the PEG side chains on the adsorption and dispersion behaviors of PCP in presence of sulfate,” Cement and Concrete Research, V. 42, No. 2, 2012, pp. 431-439. doi: 10.1016/j. cemconres.2011.11.011

SP-302-15

Influence of Diester Content in Macromonomers on Performance of MPEG-Based PCEs by Johannes Paas, Maike W. Müller, and Johann Plank Macromonomers for MPEG type of PCEs are produced through esterification of methacrylic acid (MAA) with methoxypoly(ethylene glycol) (MPEG) yielding the MPEG-MAA ester. However, PEG impurities present in MPEG may lead to MAA diester (PEG-di-MAA) formation. Such diester can cause crosslinking of the PCE polymer which might reduce its dispersing power. To investigate this effect, MPEG-MAA macromonomers containing 0 – 20 wt. % of PEG-di-MAA diester were used in PCE synthesis. It was found that when the PEG-di-MAA content in the macromonomer exceeds 2 wt. %, then dispersing effectiveness starts to decrease and the solution viscosity of the PCE increases. Surprisingly, incorporation of the diester into the PCE polymer does not occur randomly. Instead, two distinct species of crosslinked PCE molecules (Mw ~ 300.000 and ~ 3 mio g/mol) are formed within the first minutes of copolymerization. Apparently, the crosslinked PCE species counteract the dispersing effect of the main product. Keywords: cement; crosslinking; diester; dispersing performance; macromonomer; polycarboxylate ether; superplasticizer. INTRODUCTION The invention of polycarboxylate-based superplasticizers (PCEs) in the 1980s greatly broadened the application of superplasticizers in concrete.1,2 These comb-type copolymers contain polyethylene oxide side chains and show a very high water reducing capability combined with low dosage of polymer.3,4 In Europe and North America, methacrylate ester (MPEG)-based polycarboxylates are widely used.5,6 The mechanism behind the excellent dispersing effect of the PCEs is based on their interaction with cement hydrates.7-9 MPEG-PCEs can be synthesized by three different methods: (1) free radical copolymerization of methacrylic acid with MPEG-MAA macromonomer; (2) by grafting of MPEG onto a poly(methacrylic acid) backbone; and (3) by transesterification of the methylester of methacrylic acid with MPEG.

199

200 SP-302-15

Fig. 1 – Preparation of MPEG-MAA macromonomer through esterification of methacrylic acid In industrial manufacturing of MPEG-type PCEs free radical copolymerization involving an MPEG-MAA macromonomer presents the predominant route for preparation. Typically, MPEG-MAA macromonomers are produced through the esterification of methacrylic acid with ω-methoxy polyethylene glycol (MPEG) using e.g. an acid catalyst.10 Furthermore, the precursor compound MPEG is obtained via reaction of poly(ethylene glycol) (PEG) with methanol.11 However, depending on process conditions this reaction might be incomplete and residual PEG is then contained in the MPEG precursor. Thus, in the following synthesis step where methacrylic acid is reacted with such contaminated MPEG, not only the desired MPEG-MAA macromonomer (Fig. 1), but also a second macromonomer, PEGdi-MAA diester is obtained as by-product (Fig. 2). However, such PEG-di-MAA diester can crosslink the PCE molecules formed during the subsequent copolymerization step. Fig. 3 schematically illustrates how crosslinked PCEs are formed when the MPEG-MAA macromonomer is contaminated with PEG-di-MAA. In this work, through free radical copolymerization MPEG-PCEs were prepared from methacrylic acid and MPEG-MAA macromonomer containing 0 – 20 wt. % PEG-di-MAA as impurity at a molar ratio of 3.2: 1. The side chain of the PCEs consisted of 23 EO units. The synthesized copolymers were characterized with respect to their molecular properties using gel permeation chromatography. Furthermore, their dispersing performance was tested using “mini slump” testing. Finally, to understand the crosslinking mechanism, representative samples were pulled during the copolymerization reaction, and time-dependent formation of the crosslinked polymer fractions was tracked. From these experiments it was hoped to gain insight into the potentially negative effect of PEG-di-MAA impurities on the dispersing performance of PCE. Furthermore, as a

Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs  201

Fig. 2 – Formation of PEG-di-MAA diester as by-product in the macromonomer synthesis from PEG impurities contained in the MPEG precursor

Fig. 3 – Formation of crosslinked MPEG-PCE from MPEG-MAA macromonomer which is contaminated with minor amounts of PEG-di-MAA diester

202 SP-302-15

Table 1 – Physical properties and chemical composition of cement CEM I 52.5 N sample Properties / Oxides Specific gravity, kg/m3 (Ib/ft3) Blaine fineness, m2/kg (ft2/Ib) Loss of ignition, % SiO2 Al2O3 CaO MgO SO3 Na2O K2O TiO2 MnO Fe2O3

wt. % 3150 (197) 368.3 (67,822) 1.90 20.53 5.13 61.47 1.74 3.11 0.05 0.83 0.32 0.05 2.82

guideline for the industry, it was aimed to establish threshold values for the diester content which should not be exceeded. RESEARCH SIGNIFICANCE It has often been speculated that crosslinking of MPEG-PCEs might occur when the MPEG-MAA macromonomer is contaminated with PEG-di-MAA diester. However, no literature exists which provides scientific evidence for this effect. In our study, the impact of 0 – 20 wt. % diester contained in the macromonomer on molecular properties and dispersing performance of the resulting PCE was systematically investigated. The goal was to understand the consequences of such impurities for PCE performance and ultimately to develop guidelines for the industry which can help to ensure optimal quality for MPEG type PCEs. EXPERIMENTAL INVESTIGATION Materials Ordinary portland cement (OPC) type CEM I 52.5 N was used. The physical properties and oxide analysis of the cement sample are listed in Table 1. For PCE synthesis, as MPEG-MAA macromonomer (nEO = 23) Polyglykol MA 1000 70%, as PEG-di-MAA diester (nEO = 23) NK Ester 23G and methacrylic acid 99% were used. Na2S2O8 was uses as initiator and 3-mercaptopropionic acid ≥ 99% as chain transfer agent. For neutralization of the synthesized polymers, 30 wt. -% aqueous NaOH was used. Procedures Generally, the MPEG PCE samples were synthesized via aqueous free radical copolymerization from methacrylic acid (MAA) and ω-methoxy polyethylene glycol methacrylate ester (MPEG-MAA) holding 23 ethylene oxide (EO) units. Additionally, PCE samples incorporating 0 – 20 wt. % of PEG-di-MAA diester (nEO = 23) were synthesized using the same procedure by replacing part of the MPEG-MAA macromonomer with PEG-di-

Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs  203

Fig. 4 – Gel permeation chromatogram of the PCE copolymer prepared from pure macromonomer (no diester present) MAA diester. Details of the synthesis process are described in a previous publication.5 Molecular properties (Mn, Mw) of the synthesized PCE samples were obtained via gel permeation chromatography (GPC) using separation modul 2596 equipped with refractive index detector 2414 and three-angle static light scattering detector mini Dawn and QELS. Three UltrahydrogelTM columns (120, 250, and 500) were used for separation of the polymer fractions. As eluent, a mixture of 0.1 M NaNO3 and 0.1 g/L NaN3 dissolved in DI water adjusted to pH = 12 was used. Additionally, time-dependent formation of crosslinked PCEs was tracked by pulling representative 2 mL samples at defined intervals from the ongoing polymerization batch holding a total volume of ≥ 80 mL. The samples were immediately analyzed by GPC. Dispersing effectiveness of the synthesized PCE samples in cement paste (w/c = 0.3) was assessed via “mini slump” test adapted and modified after DIN EN 1015-3.12 The tests were carried out as follows: firstly, over one minute 300 g of cement were filled into a porcelain cup containing 90 mL of water, then left to soak for one minute and stirred for two minutes with a spoon. Thereafter, the cement paste was poured into Vicat cone (height 4.0 cm (1.57 in.), bottom diameter 8.0 cm (3.15 in.), and top diameter 7.0 cm (2.76 in.)), placed on a glass plate and filled to the brim. The cone was removed vertically and the diameter of the resulting cement cake was taken as slump flow (or spread) value. All tests were performed twice and the avarage value was reported as cement spread. Kinematic viscosities of the aqueous PCE solutions were obtained on an Ubbelohde 015T viscometer. Furthermore, measurements of shear-dependent dynamic viscosity were performed on a HAT Synchro-lectric viscometer. EXPERIMENTAL RESULTS AND DISCUSSION Synthesis and properties of non-crosslinked PCEs At first, a non-crosslinked PCE was synthesized from pure MPEG-MAA macromonomer free of PEG-di-MAA diester. Its GPC diagram is shown in Fig. 4. There, the refractive index detector (dRI) indicates a narrow peak of high intensity between 16.8 and 23.5 mL

204 SP-302-15

Fig. 5 – Dispersing performance of PCE polymers synthesized from macromonomers holding 0 – 10 wt. % diester elution volume which corresponds to the PCE polymer. Its molar masses were found at ~ 34,000 g/mol (Mw) and ~ 17,000 g/mol (Mn) respectively, thus signifying a polydispersity index of 2.0. The peaks at elution volumes of 23.8 to 25 mL are attributable to unreacted MPEG macromonomer and methacrylic acid. Based on the RI signal, a conversion of 89.1% for the monomers was achieved. The peak at elution volumes from 28.5 to 30.5 mL can be assigned to salt from the eluent (NaNO3). Furthermore, at elution volumes between 14.5 and 16.8 mL the light scattering signal (LS) detects a very intense peak which can be attributed to the homopolymer of methacrylic acid (PMAA). Its concentration is extremely low (< 0.1 wt. %), as indicated by the RI signal which is concentration-dependent while its molar mass is extremely high, as evidenced by the LS detector which is mass-dependent. In the next step, additional PCE polymers were synthesized whereby in the MPEG-MAA macromonomer up to 10 wt. % were replaced with PEG-di-MAA diester. The resulting polymers were characterized via “mini slump” test in cement paste. Dispersing performance of PCE samples The results obtained from dosage-dependent “mini slump” tests are presented in Figure 5. It was found that considering the typical error of this test, diester contents of 1 wt. % did not noticeably affect the dispersing performance of the PCE samples. Whereas, beginning at 2% addition of the diester and more clearly at 3%, higher PCE dosages are required to achieve the target spread of 26 ± 0.5 cm (10.26 inch). For example, at 3% diester content, the increase in PCE dosage required is ~ 14%, compared to the non-modified reference polymer. This trend continues quite linearly for polymers holding up to 5% diester. There, the increase in dosage vs. the reference polymer is 21%. However, at diester contents of > 5%, a considerable jump in dosage occurs whereby the PCE synthesized from 5.8% diester already requires 63% more dosage than the reference polymer. Interestingly, at even higher diester contents (8.1% and 10%) the dosages increase only slightly.

Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs  205

Fig. 6 – Gel permeation chromatogram of PCE copolymers prepared with 8.1 wt. % diester present in the macromonomer The results of the “mini slump” tests allow to conclude that beginning at 2 wt. % addition of diester, a minor decrease in PCE performance can be detected. This effect becomes more pronounced for 3% diester content and is particularly strong at diester additions of > 5%. Hence, to ensure high quality of PCE, the diester content present in the macromonomer should be kept below 2 wt. %, and preferably even at less than 1 wt. %. Such values seem to be feasible in large-scale industries manufacturing of PCE macromonomers.11 To understand the reason behind the negative effect of the diester content on dispersing performance, the composition of the synthesized PCE samples was analysed via GPC. Analysis of crosslinked PCEs As an example for the procedure in the analysis of the GPC spectra, the diagram for the PCE polymer prepared in the presence of 8.1 wt. % of diester is shown in Fig. 6. The chromatogram was divided into five parts: (1) elution volumes 14 -16 mL representing a highly crosslinked PCE; (2) elution volumes 16 - 18 mL referred to crosslinked polymer; (3) elution volumes 18.5 - 23.5 mL for non-crosslinked polymer; (4) elution volumes 23.5 - 28.3 mL representing unreacted monomers including macromonomer; and (5) elution volumes 28.3 - 30.5 mL for salt (NaNO3). Only data for the segments (1) – (3) in the diagrams will be discussed, because only they represent the PCE polymer. A comparison of the GPC diagrams of the PCE samples prepared with 0 – 8.1 wt. % diester content in the macromonomer is shown in Fig. 7. From this figure it can be recognized that, already at 1% diester addition, a shoulder develops, which increases in intensity by increasing diester contents. It represents a crosslinked PCE with an Mw of ~ 300,000 g/mol. At even higher diester contents, a third peak representing another, highly crosslinked PCE (Mw ~ 3,000,000 g/mol), was observed. From the GPC spectra, the individual contents of non-crosslinked, crosslinked and highly crosslinked PCE polymer fractions were determined. The results are summarized in Table 2. It was found that already at 1% diester content, the synthesized PCE polymer contains ~ 3.8% of crosslinked and 0.15% of a highly crosslinked fraction. Fortunately, such low contents of crosslinked PCE do not seem to severely affect the dispersing performance of

206 SP-302-15

Fig. 7 – Gel permeation chromatograms of PCE copolymers prepared with 0 – 8.1 wt. % diester present in the macromonomer the PCE (see Fig. 5). Whereas at 2% diester content, the total amount of crosslinked and highly crosslinked molecules reaches ~ 7% of total PCE mass, and this now starts to negatively impact the dispersing capacity (see Fig. 5). Even more, at > 5% diester content the amount of highly crosslinked PCE (Mw several mio. g/mol) suddenly surges considerably

Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs  207 Table 2 – Analysis of polymer fractions contained in PCE samples prepared with different diester contents in the macromonomer Content of diester in macromonomer [%] 0.0 1.0 2.0 3.0 4.0 5.0 5.8 8.1

Non-crosslinked Mw wt.-% [g/mol] [%]

Polymer fractions Crosslinked Mw wt.-% [g/mol] [%]

31,600 27,960 28,670 28,590 30,520 31,860 34,030 38,310

298,400 353,300 357,700 387,100 366,000 506,900 480,000

100.00 96.07 93.18 90.35 89.02 86.19 87.25 83.08

3.78 6.59 9.18 10.47 11.51 9.24 12.18

Highly crosslinked Mw wt.-% [g/mol] [%] 1,121,000 3,496,000 2,287,000 3,376,000 2,607,000 8,017,000 9,033,000

0.15 0.23 0.46 0.52 2.30 3.51 4.73

which explains the very significant increase in the dosage required for those PCEs (see Fig. 5). A graphical display of the individual diester content-dependent polymer fractions reveals that above ~ 3% diester addition, the amount of crosslinked PCE remains fairly constant whereas further addition of diester only leads to highly crosslinked PCE species (Fig. 8). The results signify that incorporation of the diester macromonomer into the PCE molecule does not occur statistically (randomly), but, the presence of diester leads to the formation of two distinctly different polymer species with characteristic molecular masses (Mw ~ 300,000 – 500,000 g/mol and ~ 3 mio g/mol) which are ~ 15 or 200 times higher than that of the non-crosslinked reference polymer. Consequently, depending on the amount of diester present, either a bimodal or trimodal polymer composition results. The amount of the PCE molecules linked together in the crosslinked polymers can be calculated by dividing Mw of the crosslinked polymer by Mw of the non-crosslinked reference PCE. Using this method it was found that in the crosslinked PCEs, 10 – 15 individual PCE molecules are linked together. For the highly crosslinked fraction, this number is ~ 30 – 300 molecules. This result implies that the portion of slightly crosslinked PCE is much less harmful to PCE performance than the highly crosslinked fraction. Its formation represents the main cause for deterioration of PCE performance. Time-dependent evolution of PCEs crosslinking To investigate the mechanism of crosslinking, time-dependent sampling was performed during the synthesis of the PCEs with 4.0% diester content in the macromonomer. Aliquot samples were pulled at 5, 15, 30, 60, 120, 180, 240 and 360 minutes of reaction time. The samples were then characterized via GPC measurement. Table 3 shows the results of the analysis. Surprisingly and most remarkably, the crosslinked PCE forms in relatively high amount (~ 25%) already at the beginning of the copolymerization. Subsequently, its content drops to ~ 3% within the next 4 hours of reaction time and then remains constant until the end of the copolymerization. Here, no highly crosslinked PCE fraction was detected.

208 SP-302-15

Fig. 8 – Mass fractions of non-crosslinked, crosslinked and highly crosslinked polymer contained in PCEs prepared from macromonomer holding 0 – 8.1% diester Table 3 – Time-dependent evolution of crosslinked PCE occurring during synthesis of PCE using a macromonomer holding 4% of diester Reaction time τ, [min] 5 15 30 45 60 120 180 240 300 360 final neutral

Non-crosslinked Mw wt.-% [g/mol] [%] 72,950 74.80 63,340 71.73 50,680 75.30 49,940 78.55 46,880 80.10 37,690 82.88 33,630 87.46 31,910 90.43 31,910 90.90 30,900 90.11 30,610 90.41 29,960 90.21

PCE Polymer Crosslinked Mw wt.-% [g/mol] [%] 244,800 25.20 303,000 28.27 338,000 24.70 362,000 21.45 356,500 19.90 331,400 17.12 323,600 12.54 331,400 9.57 341,300 9.10 321,700 9.90 326,100 9.59 317,200 8.72

Highly crosslinked Mw wt.-% [g/mol] [%] -

Viscosifying effect of crosslinked PCEs In order to quantify the viscosifying effect of the individual PCE polymers, measurements of the kinematic and dynamic viscosity of their aqueous solutions (solid content 25 wt. %) were performed. The results are displayed in Table 4. It was observed that starting at ~ 4% diester content, solution viscosity increases considerably. Even more, diester contents of ≥ 10% lead to excessively high solution viscosities. In fact, during the synthesis of these polymers formation of hydrogels occurred. This observation suggests that the crosslinked and highly crosslinked polymer fractions counteract the dispersing effect of the non-crosslinked PCE.

Influence of Diester Content in Macromonomers on Performance of MPEGBased PCEs  209 Table 4 – Kinematic and dynamic viscosity of synthesized PCE solutions (solid content 25 wt. %) Content of diester in macromonomer [%] 0.0 1.0 2.0 3.0 4.0 5.0 5.8 8.1 10.0 15.0 19.0 20.0

Solution viscosity kinematic ν [cSt] 19.6 21.7 22.9 25.7 25.0 31.7 31.3 43.0 666.7 -

dynamic η [cP] 39.0 39.7 40.7 41.7 42.7 49.0 54.7 56.0 448.0 1,742.7 2,422.0 1,538.7

CONCLUSIONS Several PCE polymers were synthesized to investigate a potential crosslinking caused by impurities of PEG-di-MAA diester contained in the macromonomer. It was found that starting from 2% diester content, dispersing effectiveness of the synthesized PCE polymers in cement paste (w/c ratio 0.3; CEM I 52.5 N) was progressively affected, less at lower diester content and very significantly at higher diester contents (≥ 5%). Excessive diester contents (≥ 10%) lead to the formation of hydrogels which can no longer disperse cement. Remarkably, incorporation of the diester into MPEG-PCEs does not occur randomly, but in the copolymerization process and depending on the diester content, one or two distinctly different species of crosslinked PCE molecules are formed, thus producing a bimodal or trimodal polymer distribution. Time-dependent analysis of the formation of crosslinked polymers during the synthesis revealed that the crosslinked species are formed within the very first minutes of copolymerization whereas by the progress of the reaction, non-crosslinked species become predominant. Our investigation suggests that MPEG macromonomers with diester contents of 1 wt. % or less are safe to obtain a PCE polymer of superior quality. When encountering difficulties with quality, a PCE producer may consider to determine the diester content present in the macromonomer via HPLC. The 1% threshold value represents what major macromonomer producers currently supply to the market. Thus, when using macromonomers of such quality, no negative effect from the diester content on PCE performance is to be expected. AUTHOR BIOS Johannes Paas studied chemistry at Technische Universität München, Germany. The present work was performed during his Master thesis at the Chair for Construction Chemistry.

210 SP-302-15

Maike Müller studied chemistry at Technische Universität München, Germany. Currently, she works on the the subject of structure – performance relationship for PCE superplasticizers within her PhD study at the Chair for Construction Chemistry. Johann Plank is full Professor at the Institute of Inorganic Chemistry of Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemicals there. Research interests include cement chemistry, chemical admixtures, organic-inorganic composite and nano materials, concrete, dry-mix mortar and oil well cementing. REFERENCES 1. Hirata, T.: Cement dispersant JP1984-18338 (1981), assigned to Nippon Shokubai. 2. Ebner, M.; Baumgartner, J.; Ohta, A.: Polycarboxylate based admixtures trend in Europe, Concr. J. 42 (2), 2004. 3. Ramachandran, V. S., and Malhotra, V. M., “Superplasticizers, Concrete Admixtures Handbook, second edition, V. S. Ramachandran, ed., Noyes Publications, Saddle River, NJ, 1996, pp. 410-517. 4. Plank, J., Current Developments on Concrete Admixtures in Europe, In: Proceedings of the Symposium “Chemical Admixtures in Concrete”, Dalian/China, 2004, pp. 13-27. 5.. Plank, J.; Pöllmann, K. Zouaoui, N.; Andres, P.R.; and Schaefer, C., “Synthesis and Performance of Methacrylic Ester-Based Polycarboxylate Superplasticizers Possessing Hydroxy Therminated Poly(ethylene glycol) Side Chain”: Cement and Concrete Research (38) 2008, pp. 1210-1216. 6. Plank, J. “PCE Superplasticizers – Cemistry, Applications and Perspectives”, 18. ibausil, 12-15 Sep. 2012, Weimar, pp. 91 – 102. 7. Zingg, A.; Winnefeld, F.; Holzer, L.; Pakusch, J.; Becker, S.; Figi, R.; and Gauckler, L., “Interaction of polycarboxylate-based superplasticizers with cements containing different C3A amounts,” Cement and Concrete Composites, V. 31, No. 3, 2009, pp. 153-162. doi: 10.1016/j.cemconcomp.2009.01.005 8. Ran, Q.; Somasundaran, P.; Miao, C.; Liu, J.; Wu, S.; and Shen, J., “Effect of the length of the side chains of comb-like copolymer dispersants on dispersion and rheological properties of concentrated cement suspensions,” Journal of Colloid and Interface Science, V. 336, No. 2, 2009, pp. 624-633. doi: 10.1016/j.jcis.2009.04.057 9. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., “Superplasticizers: how they work and why they occasionally don’t,” Concrete International, V. 16, No. 5, 1994, pp. 45-52. 10. Hirata, T.; Yuasa, T.; Shiote, K.; Nagare, K.; and Syogo, S., Cement dispersant, method for producing polycarboxylic acid for cement dispersant and cement composition, US Patent 6, 174, 980, 2001, assigned to Nippon Shokubai Co. 11. Crass, G. “Polyglycols as Macromonomers for PCEs from an Industrial Viewpoint”, Presentation at Opening Ceremony of TUM Center for Advanced PCE Studies, TU München, March 2014. 12. European Norm DIN EN 1015-3 “Prüfverfahren für Mörtel für Mauerwerk – Teil 3: Bestimmung der Konsistenz von Frischmörtel (mit Ausbreittisch)”, DIN Deutsches Institut für Normung e. V.

SP-302-16

Evidences about the Interactions between Grinding Aids and Cement Particles Surface by Valerio Antonio Paternò, Sara Ottoboni, Marco Goisis, and Paolo Gronchi The measure of wettability of cement particles and the evaluation of the type and the strength of active surface sites of clinker particles were carried out to score the efficacy of organic grinding aids. The first analysis, based on the Washburn method, allows measuring the contact angle of cement with different solvents and is directly related to the surface tension. The second analysis, based on the Hammett method, helps to classify the sites that may be responsible of the surface tension. Milled clinker (portland clinker CEM I 52, 5R; 0,025% w/w DEG and TEA) was investigated. Firstly the wetting rates of powder with 4 different solvents (ethanol, n-hexane, toluene, and formamide) were detected using a tensiometer. Then the powder dispersions in a solvent were titrated by acid solutions to get information on the acid/basic character of the surface sites. Techniques and results are shown. Keywords: additive; cement; clinker; contact angle; grinding aid; surface tension; wetting. INTRODUCTION Key engineering properties of concrete such as workability, reactivity, strength, and durability depend on the fineness of clinker.1 Unfortunately only a small amount of grinding energy is used to create new surfaces from the broken products.2-4 As a result cement grinding is a high energy intensive process consuming about 40% of all electrical energy spent in a plant.5 Organic grinding aids (GA), such as glycols, amine or poly(alcohols) are important tools to reduce energy consumption during mill operations. Indeed they obstacle the re-agglomeration of the fresh ground particles so improving the yield of the mill. However the knowledge of their action on the surface of cement particles is still approximate. The scenario is coarsely described and comprehends firstly the formation of some anisotropic energy distribution on the walls of the cleaved surfaces and then the interaction of the organic molecules with the active sites. It is generally supposed that they are rich of unsaturated valence forces which promote the interactions of the functional groups of the chemical with the mineral surface.6 The need of choosing an efficient organic 211

212 SP-302-16

molecule capable of maintaining separated the fresh cleaved surfaces after milling, push to examine in depth the interfacial situation. The analysis, however, is challenging due to the difficulty to monitor interfaces and to interpret available data. Many approaches, including molecular simulation, can be used with the aim to describe type and origin of the surface energy.7 Taking into account that similar characterization techniques are used in workings on heterogeneous catalysis, in this paper preliminary results of a study of wettability of cement powders by liquids with different molecular structure, of measurement of acid/basic strength and of density of active sites by acid/base titrations are reported. Plain clinker milled powder and clinker ground with diethylene glycol (DEG) and triethanolamine (TEA) as GA have been used. The comparison is focused at understanding the modification of the surface of cement particles w/wo grinding additives, and consequently the role of chemicals to maintain particles separated during the grinding process. RESEARCH SIGNIFANCE This research is focused at improving the comprehension of the effect of grinding aids on cement particles applying an original approach taken from the studies on heterogeneous catalysis to this investigation. MATERIALS AND METHODS Materials P-toluenesulfonic acid, (PTSA; Alfa Aesar, 98%), chlorophorm (CHCl3; Sigma Aldrich, 99-99.4%), n-hexane (HEX; Alfa Aesar, 95%+), methanol (MeOH; Sigma Aldrich, 99.9%), ethanol (EtOH; Fluka, 99.8%), xylene (XYL; Alfa Aesar, 98.5%), toluene (TOL; Alfa Aesar, 99%), N,N-dimethylformamide (DMF; Alfa Aesar, 99%), formamide (FA; Alfa Aesar, 99%) benzene (BEN; Alfa Aesar, 99%) were used as received. The plain milled clinker (CLKF) was prepared without gypsum. The chemical composition is reported in Table 1. The phase composition, according to the Rietveld method, is shown in Table 2. Methods Clinker grinding—Clinker samples were ground by a Ball mill, Bond type, BICO International, rotating at 70 rounds/min, equipped with 285 grinding balls (tot. weight 20.125g) of five sizes (1.5, 1.25, 1.0, 0.75, 0.625 inches) and charged with 1kg of clinker milled to 4000 Blaine fineness (blank sample name CLKF1) adding 0.025% (w/w) diethylene glycol (DEG; sample name CLKF2) or triethanolamine (TEA; sample name CLKF3). CLKF1CLKF2- CLKF3 sample’s particles size distributions are shown in Fig. 1. Size distribution analysis was carried out by Helos KFS equipped with dry dispersion system Rodos. Data elaboration by Fraunhofer model. Attention was paid to obtain a reproducible PSD (Particle Size Distribution) in order to avoid different porosity of the bed of the column for the Washburn tests. Experimental set-up—The Washburn tests8 and Hammett9 measures were performed using the usual laboratory apparatus and the instruments described below. Data were calculated from at the least 10 times repeated tests for each sample to assure a set of data for statistical evaluation. The reported values are the arithmetical medium value, while the standard deviation is maintained around the 10%.

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  213 Table 1—CLKF physical and chemical parameters (CLKF)* CLKF Oxide SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K 2O SrO Mn2O3 P 2O 5 TiO2 Total Loss on ignition Blaine fineness

% 21.31 5.03 3.70 64.55 1.57 1.29 0.33 1.28 0.07 0.01 0.20 0.13 99.88 0.44% 4000 cm2/g**

* Instrument: PANalytical Cubix XRF – 200W.

Table 2—Phase composition of CLKF according Rietveld method* CLKF Determination C3S C2S Cubic C3A C4AF MgO Free CaO K3Na(SO4)2 K2SO4 Total

% 58.1 19.6 3.02 14.04 0.03 2.00 0.09 1.03 97.91

* Instrument: Bruker D8 Advance.

The Washburn model The Washburn model (Eq. 1) was used to determine the powder contact angle6:

h2 =

r γ L cos θ t (1) 2µ

where h2 is the clinker height wetted by the liquid, r is the capillary radius, γL is the liquid surface tension, μ is the liquid viscosity, θ the contact angle.

214 SP-302-16

Fig. 1— Particle size distribution and cumulative distribution of milled Clinker (CLKF1-CLKF2-CLKF3). The symbols and the lines of the density distribution are overlapped. (1 kg = 2.204 lb) The model was modified and used in a suitable form (Eq. 2) for a powder bed. A linear relation between the mass of the liquid in the porous bed and the time for rising up is possible10:

m2 =

K ρ2 γ L cos θ t (2) µ

where the geometrical factor K (Eq. 3) is function of the porous bed radius Rc, and the bed porosity ε.

(

 r πR 2 c K=  2 

)

2

ε2   (3)  

It has to be underlined that the linearity of the Eq.2 depends on some critical parameters concerning the bed, as the packing factor, the void distribution and the capillaries radius, the porous bed uniformity, the PSD These are important features which can be considered as constant only after carefully and reiterated tests as done by us packing the samples by centrifuge. Liquid selection—The choice of the appropriate liquid is essential to the application of the thermodynamic model of van Oss-Chaudury and Good (OCG).10-13 Indeed it is fundamental to know some basic parameters such as the contact angle of a non-polar liquid for the determination of the γLV non-polar component and the contact angles of two polar liquids for the determination of the acid (γL+) and basic (γL–) components of the surface tensions. The chemical-physical characteristics of the selected liquids are shown in Table 3. The reference liquid

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  215 Table 3—Chemical-physical proprieties of the liquid used in the Washburn method Parameter µ (viscosity) ρ (density) γL (surface tension) γLLW (liquid-wall) γLA (A component) γLB (B component)

Units N*s/m2 x103 (psf s)

Toluene

Hexane

Ethanol

0.604 (12.61)

0.326 (6.81)

1.07 (22.35)

Kg/m3 (lb/ft3) 866.9 (54.11) 654.8 (40.87)

789.1 (54.08)

Methanol 0.544 (11.36) 791.8 (49.42)

Formamide 3.3 (68.92) 1130.0 (70.53)

mN/m (lb/ft)

28.5 (1.95)

184 (12.61)

22.4 (1.54)

22.5 (1.54)

58 (3.98)

mN/m (lb/ft)

28.5 (1.95)

184 (12.61)

18.8 (1.29)

18.2 (1.25)

39 (2.97)

mN/m (lb/ft)

0.0191 (0.001)

0.6 (0.04)

2.83 (0.19)

mN/m (lb/ft)

68 (4.66)

77 (5.28)

39.6 (2.71)

Fig. 2—Adsorption rate of different liquids on plain milled clinker (CLKF1). (1 kg = 2.204 lb) must completely wet the powder surface (so that θ=0 and cos θ=1) allowing to simplify the Eq. 2. Consequently the geometrical factor K (Eq. 3) can be determined from the curve slope Δm2/Δt provided that the other liquid properties (μ, γLV, ρ) are known. Figure 2 shows the absorption rate of several liquids and their maximum adsorbed amount. In order to fulfil the above statement, the reference liquid must not present any polar site. Moreover several liquids must be tested to calculate the polar acid (γL+) and basic (γL-) components of the surface tension (γLV). From the data reported in Table 3, toluene was assumed as reference liquid because the higher value of the product

216 SP-302-16



m2 µ (from Eq. (2)) t ρ2 γ

The choice agrees with literature.12 The experimental Washburn tests—A known amount of powder equals to 2 g (4 x 10-3 lbs) was placed in a in a glass tube of 5 cm (0.16 ft) height and 1 cm (0.032 ft) diameter, joined at the top to an arm of the tensiometer TE3 Lauda, and, at the other side, maintained into contact with a liquid through a porous membrane. Uniformity of the porous bed of the column was improved by packing with a centrifuge (see after). A continuous measure of the column weight progressively increasing due to the permeating liquid, was executed. The properties of the test liquids are reported in Table 3. Powder column packing—A reproducible, uniform porous bed of powder was obtained by centrifugation at 5000 rpm for 5 min.(Thermoscientific SL-16 centrifuge). The Hammett method The method enables to determine the basic strength expressed by an Hammett acidity function and, hence, the acid-basic strength distribution of a solid surface on a common scale. To determine the strength of basic sites of the powder suspended in toluene, the colour swing points of four indicators in the 1-8 pH range, possibly adsorbed on the surface previously treated with para-toluene sulfonic acid (PTSA) have been used. The acid was added as first component to the powder repeating the operation each time with different amount in such a manner to individuate the right addition for the clear colour swing. With the PTSA (Brönsted acid) amount exceeding the saturation of the basic sites (SBS) of the surface, the addition of the indicator gives a colour corresponding to acidity, while, with PTSA less than the SBS, the indicator colour corresponds to basicity. As the goal of the test is to quantify the basic sites related to the PTSA salt at neutral condition, the values of the Hammett parameter at different colour swing indicates the basic strength of the base conjugated to its salt. The amount of acid on a solid is usually expressed as the number or mmol of acid sites per unit weight or per unit surface area of the solid. Relatively to the Brönsted acid-base equilibrium (Eq. 4)

B + � H� + ↔ � BH + (4)

The Hammett and Deyrup’s acidity function Ho17 is:

H 0 = pKa + log

[B]  BH + 

(5)

Where [B] and [BH+] are the concentrations of the neutral base and its conjugate acid respectively, and

 f pKa = pK BH + = − log  aH + B f BH + 

  (6) 

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  217 Table 4—Geometrical coefficient and contact angle. (m =1 kg = 2.204 lb) Sample

Liquid Toluene Ethanol Hexane FA

CLKF1 CLKF2 CLKF3 K θ K θ K θ (geom. coeff.) (contact angle) (geom. coeff.) (contact angle) (geom. coeff.) (contact angle) (m5) x 10–16 (°) (m5) x 10–16 (°) (m5) x 10–16 (°) 1.45 Reference 1.42 Reference 1.50 Reference 1.36 20 1.25 28.2 1.32 28.3 1.45 84.3 0.85 84.4 1.71 83.5 1.28 27.5* 1.11 38.4 1.19 37.4

If Lewis equilibrium takes place by means of electron pair transfer from the adsorbate to the surface, Ho is expressed by

H 0 = pKa + log

[ B ] (7) [ AB ]

Where [AB] is the concentration of the neutral base which reacts with the acid or electron pair acceptor, A. During titration, stronger basic sites are neutralized earlier, while weaker ones later and the latter require stronger acids for neutralization. Therefore, it can be assumed that the weakest basic sites have been finally neutralized by an acid having an acid strength for which Ho = pKBH+. The experimental Hammett tests—0.2 g (4 x 10-4 lbs) of milled clinker was dispersed in 10 ml of toluene and a known amount of para-toluenesulfonic acid (PTSA) was added. In order to clearly evaluate the swing point, the addition of PTSA was repeated up to reaching the threshold value. Power sonication (40 KHz maximum) was carried out for 30 minutes to assure full dispersion of the powder; 2 ml (0.067 oz) of indicator were added later. The 1-8 pH range was covered by the following indicators: thymol blue (1.2-2.8 pH range), methyl yellow (3.3-4.0 pH range), methyl red (4.4-6.2 pH range) and neutral red (6.8-8.0 pH range). DISCUSSION Surface tension The results of the Washburn tests are reported in Table 4 and the curves representing mass of liquid (m2) vs. time (t) are presented in Figure 3 for the polar liquids, FA and EtOH. The same relationship is shown in Figure 4 for the non-polar liquid, TOL and HEX. All data have been collected at the same times and up to the equilibrium state. Contact angle— The contact angles of polar and non-polar liquids are very different: polar liquids, EtOH and FA, applied on samples treated with DEG (CLKF2) and TEA (CLKF3), give higher values respect to those obtained with the not treated powder, the increasing percentage being about 40% with both the liquids. On the contrary, the nonpolar liquid (HEX) gives the same value with non-treated and treated powder. The lower contact angles observed with the polar liquid respect to HEX can be explained in terms of remarkable interaction with the powder surface able to increase the spreading component of the surface tension (γSV).

218 SP-302-16

Fig. 3(a)(b)—Washburn plots for polar liquid with different molecular structure. (1 kg = 2.204 lb)

Fig. 4(c)(d)—Washburn plots for non-polar liquids, alkane and aromatic structured. (1 kg = 2.204 lb) Rate of liquid rising and amount of adsorbed liquid—Considering the apparent curve slope, that is proportional to the rate of liquid moving up (∆m2/∆t; Eq. 2), significant differences are present between the polar and the non-polar liquids for the samples w/wo GA (Figures 3 and 4). Table 5 shows the relevant slopes obtained by graphical interpolation of the ∆m2/∆t linear relationship. The gradient values decrease according to the sequence TOL>FA>HEX>EtOH while for the same liquid, we observe a CLKF3>CLKF1>CLKF2 sequence. From the liquid sequence we observe that the rising rate can be ascribed to the molecular structure, and consequently the more pronounced ability of EtOH to form hydrogen bond, could be responsible of the lowest rate. An increase of the interface tension γSL slows the spreading

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  219 Table 5—∆m2/∆t relationship (g2/s x 106; 1g = 2.204 x 10-3 lb) CLKF1 CLKF2

TOL 5.13 3.72

CLKF3

5.4

FA 5.03 3.42 4.5 initial 2.7 after 30s

HEX 3.55 3.05

EtOH 1.77 1.11

4.42

1.79

rate as appear from the balance condition (γSV> γLVcosϴ+γSV).The literature agrees with the lowest rising rate with EtOH as confirmed by studies of the adsorption on montmorillonite14 which claims that the observed strong adsorption is due to hydrogen bonding of the alcohol groups to the oxygen atoms of the surface.15,16 Furthermore EtOH has about the same rate with treated and non-treated powder (Tab. 5). Only the point CLKF3-FA does not clearly agree with a continuous decreasing sequence. However in this case the curve ∆m2/∆t has two linear segments with different gradients: initially the gradient is higher (4.5x10-6 g2/s) and after 30 s it becomes lower (2.7x10-6 g2/s). As far as concerns the amount of liquid captured by the powder at the equilibrium (from fig. 3 and 4, plateaus) and stored in the interparticle spaces, the amount of FA in plain CLKF1 bed is greater (not determined) than in CLKF3 and in the CLKF2 (Fig. 3b). With EtOH much smaller differences of the same parameter are observed (Fig. 3a). From Fig. 4 clearly appears that TOL is more adsorbed on the powders than HEX. The interpretation of the data needs a look at the clinker particle surface-liquid interaction. The conformations of TEA adsorbed on C3S were calculated by R.T. Mishra, et al.17 TEA bonds the silicate surface through hydrogen bonds (HB) and through the oxygen coordination of the hydroxyl groups with the Ca++ ions that are diffused all over the clinker surface. Due to the ratio of the hydroxyl groups, it is allowed to suppose that TEA forms more hydrogen bonds (HB) with silicate (probably in the present case the C3S phase) than DEG, that, on the contrary, easier coordinates the Ca++ ions in reason of the greater basicity of its electron pairs on the oxygen of the ether group. Moreover DEG is known to have only one OH link because the most probable molecular conformation has intramolecular HB of the second hydroxyl group with the ether central oxygen (Figure 5). Other bi-dentate (two HB bonds or one HB bond and one coordination bond) and tridentate (two HB bonds and one coordination bond) conformations of adsorbed DEG might be hypothesized but they appear less probable due to the great chain flexibility. The rising liquids can interact both with the adsorbed GA and the free surface. Clear indications may be acquired by the observed polar liquid behaviours considering that FA acts mainly by the carbonyl dipole creating coordinative bonds, while EtOH links the surface exclusively with the HB bonds. Similarly both the molecules link a Lewis base at the sites of the surface. Due to the great rate of FA liquid on CLKF1 (Tab.5) a prevalence of the coordinative bonds on strong HB bonds is then supposed on plain clinker. The amount of FA stored inside the spaces of the powder column seems to confirm this hypothesis. Now, comparing the FA rising rate and the amounts between TEA and DEG treated samples, TEA seems able to maintain more sites for coordinative bonds than DEG as both the values are higher. The observation agrees with the TOL adsorbing behaviour. TOL has polarizable electrons, the aromatic electron cloud, that is deformed by methyl group induc-

220 SP-302-16

Fig. 5—Monodentate HB and coordination of DEG. tive effect. The dipole promotes coordinative bonds causing rising rates greater than with FA. Similarly the behaviours with EtOH can be interpreted with the formation of strong HB bonds. The previously advanced hypotheses confirm that different types of active sites exist on the surface coordination bonds and second order HB bonds. The hypothesis agrees also with the chemical and physical data of the liquids: FA has the highest surface tension among those listed in Table 1 that derives from its greatest γLLW among the used liquids, the greatest acid γLA, and the moderate basic component of surface tension, γLB. In a separate experiment, the clinker CLKF1 was treated with vapour of DEG obtained from liquid DEG heated at 70°C for 92 h and transported to clinker by nitrogen carrier in order to completely saturate the powder surface. Examining the Figure 6, the comparison between the 0.025% DEG treated clinker (CLKF2) and the clinker that has been completely coated by DEG, shows that EtOH is adsorbed faster on the low DEG (EtOH-CLKF2) treated clinker. The plateau is not reached by the ETOH-CLKF1+DEG Ads sample, probably due to the different interaction at the interface (solid-vapour instead of solid-liquid) which may cause a different pore filling. Inversely, using FA, the absorption rate is high on the completely saturated sample and no remarkable difference of the amount of the adsorbed liquid is present. The observation agrees with the greater tendency of the DEG coated powder to form HB bonds rather

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  221

Fig. 6—Comparison between the completely coated surface and the surface obtained with 0,025% of GA (1 kg = 2.204 lb) than coordination bonds so decreasing the absorption rate of EtOH and, on the contrary, increasing that of FA. Investigation on surface tension origin—The results of the Washburn tests were evaluated with the Van Oss thermodynamic theory that, starting from the elaboration of the acidbasic interaction proposed by Fowkes10,15 elucidates the relevance of the electron-acceptor (Lewis acid) and the electron-donor (Lewis base) sites and their different behaviours. The free energy at the interface between the solid and the liquid phase is represented by the Eq. 8 where the total free energy is the sum of the non-polar interaction LW (Lifshitz-Van der Waals) and the polar interaction of acid and base Lewis sites (AB):

LW ∆GSL = � ∆GSL + ∆GSLAB (8)

Fowkes demonstrated that in case of non-polar interactions (Eq. 9):

LW LW ∆GSL = −2 γ LW (9) L γS

After the revision of vanOss, the free energy for polar interaction is described by the relation (Eq. 10)

∆GSLAB = −2 γ +L γ −S � −2 γ −L γ +S (10)

where γ+ is the acid component and γ- is the basic component of the interfacial polar tension and A and B indicate the Acid and Basic site. The relation of the total free energy (Eq. 8) becomes:

LW ∆GSL = −2 γ LW − 2 γ +L γ −S − 2 γ −L γ +S (11) L γS

222 SP-302-16

Table 6—Components of the surface tension ϒS (solid-liquid interface; mN/m) Sample CLKF1 CLKF2 CLKF3

Hex ϒS LW 5.56 5.54 5.71

FA ϒ S28.49 54.07 58.99

EtOH ϒ S+ 5.35 4.98 4.93

ϒtotal 30.25 38.37 39.80

That can be transformed into Eq. 8 (the Oss-Chaudary-Good equation) using the YoungDuprè equation (γL(1 + cosθ) = ∆GSL)12:

γ L (1 + cos θ) = 2

(

)

γ LLW γ SLW + γ +L γ −S + γ −L γ +S (12)

Using the Eq. 8 all the components of the surface tension of the solid, γLW, γ+ and γ-, can be measured. Three liquids with different surface tension were put in contact with the powder for the determinations of the solid surface tension. From the contact angles, applying the OCG theory, the powder surface tension can be calculated. In this study, HEX, FA and EtOH have been used for the γSLW, γS-, and acid component γS+ respectively. The results are listed in Table 6. The data indicate that the basic (polar ϒS-) component of the samples CLKF2 and CLKF3 milled with DEG or TEA respectively, increases respect to the not treated sample CLKF1 following the order CLKF3>CLKF2»CLKF1. The non-polar component ϒS LW does not change sensibly. Acid Base Strength Distribution The basicity (mmol/g) at the Ho value (basic site strength) is stated by the number of basic sites whose basic strength is equal to or greater than the Ho value.19 The basic sites (mmol/g) at Ho strengths are reported in Table 7. Firstly it appears an increase of the basic strength when DEG or TEA (GA) is used respect to the plain milled clinker. The values of the TEA treated samples are however lower than those of the DEG treated ones. Remembering that a high Ho=pKa (or pKBH+) value is related to a high basic population, the greater basicity value in presence of the GA additive may be interpreted as consequence of the direct reaction of PTSA with the additives. The proton-donating ability of the solid at the endpoint of titration is considered to be either due to the conjugate acids which were formed by the proton transfer from Brönsted acid solution to the solid surface or to the Brönsted acid which was physically adsorbed on the surface during the titration. Moreover, as previously reported, the conformations of TEA (and DEG similarly) adsorbed on the basic sites of the C3S surface,17 indicate the formation of HB between the OHs of –CH2-CH2-OH groups and the oxygen of the silicate together or alternatively between the coordination bond of the oxygen of the hydroxyl group and the Ca++ of the surface .

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  223 Table 7—Basic site strength HO

CLKF1 CLKF2 CLKF3

H0 ≥ 2.8 20 38 32

Basic sites (mmol/g) at Ho strengths H0 ≥ 3.3 H0 ≥ 4.8 40 0* 46 0* 34 0*

H0 ≥ 6.8 0* 0* 0*

* H0 ≤ 5x10e-5

However these types of bonds are weak and are broken easily by acids. The HB equilibrium and the coordination equilibrium

GA-O-H + B ↔ GA-OH---B HB bond

(13)



GA-HO + Ca++↔ GA-HO--- Ca++ coordination bond

(14)

are shifted to the left by the acid addition as a consequence of the acid character of the hydroxyl group present in the structure of the GA and of the basic strength of the oxygen in the silicate. A stronger HB means a greater amount of PTSA. The different behaviour between TEA and DEG might be attributed to their molecular structure that causes some greater inductive effect on the electron donor/acceptor character of the –OH group and to the steric hindrance. The structure is supposed to influence directly the shift of the equilibria towards left and to increase the number of the basic sites. FURTHER RESEARCH Our aim is to extend the examination to many grinding additives to obtain a large number of data on the interaction between the organic molecules and the clinker surface. The field of the study is of interest for many applications relating the behaviour of material surface at nano and micro dimension particularly in the cement sector. The molecular conformation analysis and the analysis of the kind and strength of surface active sites by chemical and instrumental methods, might give useful information to choice the right GA additive. CONCLUSION In this work several methods already present in literature have been employed to study the surface of milled clinker powder. Samples were obtained by grinding clinker with additives largely used in the industrial practice, such as DEG and TEA, at dosage equal to 0,025% w/w, and the results compared with those of the plain powder. The study is part of a larger investigation aimed at understanding the mechanism of action of grinding aids to design new molecules. The methodological approach performed in this study allows to calculate the total surface tension and its basic/acid components by the measure of the contact angles and the rate of wetting of the powder. Furthermore the approach allows to get information on the sites present on the surface. These indications are obtained by using polar (FA, EtOH) and non-polar (HEX, TOL) solvents, according to the Washburn method, as the liquids differently interact with the surface, i.e. by H atoms and/or by dipoles/WdW bonds and/or only by one of these. The results agree with the existence of two types of sites

224 SP-302-16

mainly, suitable for hydrogen bond and coordinative bond (dipole-induced dipole) respectively. The elaboration of the Washburn theory by a thermodynamic point of view permits to highlight the basicity of the sites according to the predominance of the basic component on the total surface tension. It increases in the order TEA>DEG>plain clinker. The Hammett method puts in evidence that the basic sites suitable for the HB bond (Lewis basic site) increase when the clinker is milled with DEG or TEA, respect to the plain clinker. In this framework DEG seems responsible of a greater population of basic sites than TEA. The explanation is not straight and relies on the hypothesis that grinding aids increase the amount of basic sites suitable for the HB bond. From these data it could be argued that the GAs cause an unforeseen and surprising increase of surface tension pushing to a deeper investigation in order to find possible different origins of their activity. It should be the case of adsorption energies,17 some shielding effect, or creation of an electric field around the surface of the particle. AUTHOR BIOS Valerio Antonio Paternò graduated in Materials Science Engineering at Politecnico di Milano, (Italy) in the field of hybrid materials and admixtures for cement and concrete. Sara Ottoboni graduated in Materials Science Engineering at Politecnico di Milano (Italy), where later occupied a post-degree position. She is focusing her study on Science and Engineering of surface of materials. Marco Goisis, graduated (MS) in Chemical Engineering at Politecnico di Milano, is specialist in additives for grinding cement, admixtures for concrete and cementitious products at Innovation Department, Italcementi. Since 1989 he has been working with Italcementi Group companies, in R&D and in the Innovation Departments. His current research interest includes carbon allotropes modified smart construction materials. Paolo Gronchi is associate professor at the Chemical, Materials and Engineering Chemistry Dept. of the Politecnico di Milano, Italy. His current research activity, beyond the industrial chemical research, concerns the structural investigation on polymer adsorbed on inorganic oxide and comprehends the synthesis of organic compounds. ACKNOWLEDGMENTS The authors acknowledge Italcementi Group for the financial and scientific support. REFERENCES 1. Bentz, D. P.; Garboczi, E. J.; Haecker, C. J.; and Jensen, O. M., “Effects of Cement Particle Size Distribution on Performance Properties of Portland Cement-Based Materials,” Cement and Concrete Research, V. 29, No. 10, 1999, pp. 1663-1671. doi: 10.1016/ S0008-8846(99)00163-5 2. Schlanz, J. W., “Grinding: An Overview of Operation and Design”, (1987) (http:// mrl.ies.ncsu.edu/reports/ 87-31-P_Grinding_ Operations_Design.pdf), 27 February 2015.

Evidences about the Interactions between Grinding Aids and Cement Particles Surface  225 3. Worrell, E.; Martin, N.; and Price, L., “potential for energy efficiency improvement in the US cement industry,” Energy, V. 25, No. 12, 2000, pp. 1189-1214. doi: 10.1016/ S0360-5442(00)00042-6 4. Madlool, N. A.; Saidur, R.; Hossain, M. S.; and Rahim, N. A., “A critical review on energy use and savings in the cement industries,” Renewable & Sustainable Energy Reviews, V. 15, No. 4, 2011, pp. 2042-2060. doi: 10.1016/j.rser.2011.01.005 5. Schneider, M.; Romer, M.; Tschudin, M.; and Bolio, H., “Sustainable, Cement Production - Present and Future,” Cement and Concrete Research, V. 41, No. 7, 2011, pp. 642-650. doi: 10.1016/j.cemconres.2011.03.019 6. Sohoni, S.; Sridhar, R.; and Mandal, G., ““The Effect of Grinding Aids, on the Fine Grinding of Limestone, Quartz and Portland Cement”, Clinker,” Powder Technology, V. 67, No. 3, 1991, pp. 277-286. doi: 10.1016/0032-5910(91)80109-V 7. Mishra, R. T.; Flatt, J.; and Heinz, H., “Force field for tricalcium silicate and insight into nanoscale properties: clevage, initial hydration, and adsorption of organic molecules,” The Journal of Physical Chemistry C, V. 117, No. 20, 2013, pp. 10417-10432. doi: 10.1021/ jp312815g 8. Washburn, E. W., “The dynamics of capillary”X, Journal of the American Physical Society, 2nd Ser. 17, 1921, pp.374-375. 9. Hammett, L. P., and Deyrup, A. J., “A series of simple basic indicators. I. The acidity functions of mixtures of sulfuric and perchloric acids with water,” Journal of the American Chemical Society, V. 54, No. 7, 1932, pp. 2721-2739. doi: 10.1021/ja01346a015 10. vanOss, C. J.; Chaudhury, M. K.; and Good, R. J., “Monopolar surfaces,” Advances in Colloid and Interface Science, V. 28, 1987, pp. 35-64. doi: 10.1016/0001-8686(87)80008-8 11. Ahadian, S.; Mohseni, M.; and Moradian, S., “Ranking proposed models for attaining surface free energy of powders using contact angle measurements,” International Journal of Adhesion and Adhesives, V. 29, No. 4, 2009, pp. 458-469. doi: 10.1016/j. ijadhadh.2008.09.004 12. Dang-Vu, T., and Hupka, J., “Characterization of porous materials by capillary rise method,” Physicochemical Problems of Mineral Processing, V. 39, 2005, pp. 47-65. 13. Van Oss, C. J.; Giese, R. F.; Li, Z.; Murphy, K.; Norris, J.; Chaudhury, M. K.; and Good, R. J., “Determination of contact angles and pore sizes of porous media by column and thin layer wicking,” Journal of Adhesion Science and Technology, V. 6, No. 4, 1992, pp. 413-428. doi: 10.1163/156856192X00755 14. Emerson, W. W., and Raupach, M., “The reaction of polyvinyl alchohol with montmorillonite,” Australian Journal of Soil Research, V. 2, No. 1, 1964, pp. 46-55. doi: 10.1071/SR9640046 15. Nguyen, T. T.; Raupach, I. M.; and Janik Csiro, L. J., “Fourier-transform infrared study of ethylene glycol monoethyl ether adsorbed on montmorillonite: implications for surface area measurements of clays,” Clays and Clay Minerals, V. 35, No. 1, 1987, pp. 60-67. doi: 10.1346/CCMN.1987.0350108 16. Parfitt, R. L., and Greenland, D. J., “The adsorption of poly(ethylene glycols) on clay minerals,” Clay Minerals, V. 8, No. 3, 1970, pp. 305-315. doi: 10.1180/ claymin.1970.008.3.08 17. Mishra, R. K., and Heinz, H., Zimmermann., Muller, T., Flatt, R., J., “Understanding the effectiveness of polycarboxylate as grindings aids”, Proceedings of Tenth CANMET/

226 SP-302-16

ACI International Symposium on Superplasticizer and Other Chemical Admixtures in Concrete, SP-288.16, V. M. Malhotra, ed. American Concrete Institute, Farmington Hills, MI, 2012, pp. 235-249. 18. Fowkes, F. M., “Additivity of intermolecular forces at interfaces. I. Determination of the contribution to surface and interfacial tensions of dispersion forces in various liquids,” Journal of Physical Chemistry, V. 67, No. 12, 1963, pp. 2538-2541. doi: 10.1021/ j100806a008 19. Wang, K.; Wang, X.; and Li, G., “Quantitatively study acid strength distribution on nanoscale ZSM-5, a,” Microporous and Mesoporous Materials, V. 94, No. 1-3, 2006, pp. 325-329. doi: 10.1016/j.micromeso.2006.03.049

SP-302-17

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs by Giorgio Ferrari, Vincenzo Russo, Massimo Dragoni, Gilberto Artioli, Maria Chiara Dalconi, Michele Secco, Leonardo Tamborrino, and Luca Valentini Portland cement is a multi-phase material, which can be simplified as a two-phase system with alite and C3A as the main constituents determining early properties. Alite is the most abundant phase and in a first approximation it is responsible for the development of mechanical strength during hydration, while C3A mainly affects the plastic behavior before set. In portland cement, superplasticizers are preferentially adsorbed onto C3A and its hydrates rather than alite, due to the different interaction with the mineral surfaces. Three different polycarboxylate superplasticizers (PCEs) were studied, based on copolymers of methacrylic acid and MPEG-methacrylate and characterized by different side chain length and different charge density. Their affinity to C3A and alite surfaces was determined through adsorption measurements on alite/gypsum and alite/gypsum/C3A mixtures. The results of the adsorption tests indicated that the charge density of PCEs, expressed as the ratio between carboxylic groups to ester groups, is the main parameter affecting the adsorption of the PCEs: the lower the charge density, the lower the adsorption on both the phases. The same parameter affects the induction period of alite phase, as demonstrated by in situ XRPD dissolution kinetics experiments, both in the presence and in the absence of C3A. These results can be put in relation with both the hindrance of adsorbed PCE molecules in the dissolution kinetic of alite and the concentration of PCE molecules in solution in conditions of saturation. KEYWORDS: alite; adsorption; C3A; comb copolymers; gel permeation chromatography; hydration; induction period; PCE superplasticizers; X-ray powder diffraction. INTRODUCTION Water reducing admixtures and superplasticizers (high-range water reducing agents) are nowadays extensively used to control rheology and to improve performances of normal and high performance portland cement-based concrete mixtures1,2. After naphthalene 227

228 SP-302-17

and melamine sulfonate based superplasticizers, poly-carboxylate ether (PCEs) became increasingly popular for their high efficiency in dispersing cement particles at low dosages. The molecular structure of such polymers basically consists in a polymer chain bearing anionic groups (backbone) with attached hydrophilic polyoxyethylene based (EO) side chains. This structure can be easily modified, by properly operating on the reaction conditions (type and ratio of monomers, length of side chains, etc.) and targeted to efficiently disperse the clinker and hydrating cement particles3-6 in order to reduce W/C, yield stress, viscosity and, ultimately, to improve final strength and durability of concrete mixtures. However, the addition of PCE superplasticizers invariably produces a reduction of the hydration rate of the alite phase, inducing a retardation of the hydration process. The reasons for the prolonged induction period induced by PCE has been variously attributed to (i) a reduced diffusion of water and calcium ions at the clinker particle surface caused by the adsorbed polymer, (ii) the calcium-polymer interaction in solution or (iii) the influence of the adsorbed polymers on the nucleation and growth of the hydration products.7-9 Although it is generally assumed that the Ca-PCE interactions are weak and therefore the calcium complexation should not largely affect the activity of the calcium ions in solution,8,9 it has been reported10 that, at dosages of PCEs above the saturation dosage, the nucleation and growth of C-S-H could be significantly affected by the PCEs molecules in solution, causing a retardation on C3S dissolution rate. Extensive research work has been performed in the attempt to characterize the interaction of the PCE dispersants with C3S, which is the main constituent of the clinker.11,12 All experimental evidence indicate that adsorption of the PCE polymers on the crystal surface has a critical inhibiting effect on the dissolution process of C3S (and alite as well). A number of investigations emphasize the correlation between the polymer molecular structure (charge density, ratio carboxylate/PEO groups, backbone and side-chains length) on the delay time, although the exact mechanisms still remain unclear.9,12,13 The observed drastic reduction in C3S dissolution induced by the polymer adsorption is increased at high concentrations of Ca(OH)2 in solution12 and the nucleation mechanism of C-S-H and the activation energy required for the nucleation and growth process are markedly changed in presence of superplasticizers.9 Some speculation has also been put forward concerning the possible interaction of the superplasticizer macromolecules and the hydration products, with the formation of intercalate phases, though the issue is still debated.14-16 The retardation in the dissolution of C3S is substantially modified in cement pastes because of the combined effect of C3A and calcium sulphates.17 In fact, it is known that superplasticizers are preferentially adsorbed on the surface of the aluminate phases compared to the silicate phases.18 In C3S-C3A-sulphate mixtures therefore a complex competition takes place between sulphate availability and adsorption of the polymers on the different clinker phases. The uptake of PCE by C3A and its hydration products, which potentially encompasses the formation of intercalate phases,19 sensibly reduces the retardation of the C3S hydration reaction and in extreme cases inhibits the dispersing effect of the plasticizers.20,21 RESEARCH SIGNIFICANCE Industrial research on new superplasticizers aims to develop new polymers capable to impart improved characteristics and performance to concrete mixtures, such as higher reduction of W/C, workability, workability retention and strength development. In order

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  229 Table 1. Structural characteristics of the different PCEs. Sample [P-(N-1)] PCE 17-2 PCE 17-3 PCE 113-9

Mw 36112 27892 75241

Mn 20148 14060 43335

Mz 62991 50341 135753

N; Rc 2.9 3.9 10.5

P 17 17 113

n 20.3 13 17.3

Degree of polymn. 58 50 77

where:

Mw = Weight average molecular weight



Mn = Number average molecular weight



Mz = Z-average molecular weight



P = Repeating EO units in side chain



n = Number of segment per macromolecule



N = Number of units per segment



N-1= Number of acidic groups per segment



Rc = Acidic groups/Ester groups

to attain these results, it is very important to study the interaction of superplasticizers with the main mineral phases of portland cement (alite, C3A) and to clarify their mechanism of action. As a further step in the understanding of the hydration process of cement in the presence of PCE superplasticizers and as a possible test of the proposed models, this study presents the results of in situ XRPD measurements of the hydration reactions and bulk absorption experiments using different alite/gypsum/C3A/ mixtures and PCE superplasticizers with different molecular structure. In this perspective, this article represents a contribution for a better comprehension of this complex and exciting matter. MATERIALS AND METHODS In this section, the characteristics of the different materials (both PCEs and mineral phases), the compositions of the model cement pastes and the analytical techniques and procedures for the different tests, are described. Polycarboxylate superplasticizers (PCEs) Three different PCEs based on random copolymers of methacrylic acid and polyoxyethylene methacrylate were synthesized in the laboratory, starting from methacrylic acid and methoxypolyethyleneglycol-methacrylate with different MPEG chain length. Molecular weight distribution analysis (MWD) was carried out with the Gel Permeation Chromatography (GPC) apparatus consisting of a Waters 515 pump (1ml/min, 0.06 in3/ min) equipped with a Waters 717 Plus auto-sampler. Separation was made by a set of four Waters Ultrahydrogel columns (7.8x3000mm, pore size 2x120Å, 250Å, 500 Å). Detection was performed by a Waters 410 Differential Refractometer detector (temperature 40°C, 104 °F). All data were evaluated using Waters Empower Pro software. All samples were dissolved in the mobile phase (NaNO3 0.04M 80% - Acetonitrile 20%) at 1% concentration. Molecular weight distribution of polymers was determined by using a calibration curve done with standards polysaccharides (Polymer Laboratories LTD). The charge density of the different PCEs, expressed as the ratio of carboxylic groups/ ester groups (Rc = CG/EG) in the backbone, was determined by potentiometric titration,

230 SP-302-17

Table 2. Composition of model cement pastes Sample alite/gyp alite/gyp/113-9 alite/gyp/17-3 alite/gyp/17-2 alite/5C3A/gyp alite/5C3A/gyp/113-9 alite/5C3A/gyp/17-3 alite/5C3A/gyp/17-2 alite/10C3A/gyp alite/10C3A/gyp/113-9 alite/10C3A/gyp/17-3 alite/10C3A/gyp/17-2

Alite (wt%)

Gypsum C3A (wt%) (wt%)

Water/solid W/S

95

0

5

0.5

90

5

5

0.5

85

10

5

0.5

Superplasticizer Dosage Type (wt% dry) None PCE 113-9 0.15 PCE 17-3 0.15 PCE 17-2 0.15 None PCE 113-9 0.15 PCE 17-3 0.15 PCE 17-2 0.15 None PCE 113-9 0.15 PCE 17-3 0.15 PCE 17-2 0.15

carried out with an acid/base titration with HCl 0.1N performed in alkali media with automatic titration system Mettler Toledo T70 with Rondo20 Auto-sampler. The different PCEs were classified according the general parameters defined by Gay and Raphaël22 for comb polymers in solution. According to this classification, a random comb copolymer can be represented as the repetition of n segments, each segment made by N repeating unit and containing one side chain of P repeating units. The characteristics of PCE comb copolymers used in the present work are shown in Table 1. Mineral phases and model cement paste composition Alite (C3S monoclinic polymorph), C3A (cubic polymorph), supplied by MR PRO, Meyzieu, France and reagent-grade gypsum were used to prepare the model cement pastes. Specific surface area of alite and C3A was measured by BET analysis with a Coulter instrument mod. 3100SA. Measurements were performed by volumetric He adsorption at low temperature (liquid nitrogen), after dehydration of the samples at 100 °C under vacuum. BET specific surface area was 1.292 m2/g (51.68 yr2/ounce) for alite and 1.812 m2/g (72.48 yr2/ounce) for C3A. The mean particle size, measured by means of laser diffraction, using a Malvern Mastersizer 2000, was 16 μm (6.3·10-4 in), 13 μm (5.1·10-4 in) and 70 μm (27.6·10-4 in), respectively, for C3S, C3A and gypsum. Model cements were prepared by mixing alite, C3A and gypsum in different proportions. The proportion of C3A was varied from 0 to 10 per cent on the whole composition mass and the amount of alite was correspondingly reduced. The amount of gypsum was fixed at 5 per cent of the whole mass, in order to maintain all the mixtures in properly sulphated conditions.23 The compositions of the model cement pastes are shown in Table 2. X-Ray Powder Diffraction measurements (XRPD) The alite/C3A/gypsum dry mixtures for XRPD diffraction were prepared by gently co-grounding the powders by hand in an agate mortar for 15 minutes. A saturated calcium hydroxide solution (lime water) containing the selected amount of the different PCEs, was added to the dry mixtures at a water to solid ratio W/S = 0.5 and then the pastes were

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  231 mixed by hand and an orbital mixer for 2 minutes. It has been reported that the use of a calcium hydroxide-saturated solution induces a decrease in the initial rate of dissolution of C3S.24 Such slower hydration kinetics allowed a better discrimination of the duration of the induction period for the different systems. Boron-glass capillaries with 0.5 mm internal diameter were used as sample holder for in situ XRPD measurements. The capillaries filled with paste were sealed with wax in order to prevent carbonation and water loss during the measurments. In situ XRPD data were collected using a PANalytical X’Pert Pro MPD diffractometer equipped with a PixCel detector and operating in transmission focusing geometry with a CuKα radiation. Incident and diffracted beam optics included an elliptical focusing mirror, 0.04 rad Soller slits, divergence and antiscatter slits of ½° aperture. The hydrating cement pastes were measured continuously for 24 hours collecting diffraction patterns (6°-66° 2θ interval) of 20 minutes duration. During measurements, the temperature inside the diffractometer case was maintained at 23.0°C ± 0.5°C (73.4 °F). Each diffraction pattern was analyzed using the Rietveld method as implemented in Topas v4.1. Adsorption measurements 30 grams (1.2 ounce) of the model cements, without preliminary co-grinding, were added to 15 grams (0.6 ounce) of saturated calcium hydroxide solution (lime water) (W/S = 0.5) containing the selected amount of the different PCEs and mixed for 5 minutes. The compositions of the mixtures were the same as those of Table 2 for XRPD measurements, but multiple dosages of the different PCEs were used: 0.05, 0.10, 0.15, 0.20 and 0.30 per cent of dry PCE by weight of powder. The pastes were then filtered and the filtrates were neutralized with diluted HCl and analysed by GPC. The amount of adsorbed polymer was calculated by comparing the areas of the chromatograms of the different PCE polymers before and after mixing, according to a previously described method.25 Different from the method based on TOC measurements, which measures the total amount of organic carbon in the solution after filtering or centrifuging, including unadsorbed polyglycol residues and unreacted monomers, by this method it is possible to distinguish all the different constituents and to determine only the effectively adsorbed polymer fraction. In this paper, adsorption data were referred only to the polymer fraction, not considering the other unadsorbed constituents. RESULTS AND DISCUSSION In this section, the results of XRPD and adsorption tests are presented and discussed. These results are further treated in order to investigate the mechanism of retardation of alite phase dissolution induced by PCEs at different dosages and in the presence of different amount of C3A. X-Ray Powder Diffraction tests (XRPD) The results of XRPD measurements on the dissolution of alite in the presence of different PCEs and different alite/gypsum/C3A compositions, are shown in Figures 1A, 1B and 1C. Each curve represents the relative intensity of the alite diffraction peaks (Rietveld scale factor) as a function of the hydration time. The sharp decrease of the intensity during hydration corresponds to the end of the induction period and the acceleration of the dissolution of alite to form C-S-H and CH (hydration). From Figure 1A (no C3A added), it is possible

232 SP-302-17

Figure 1. (A) Dissolution curves of alite in the presence of different PCEs (dos. 0.15% of dry polymer by weight) for mixtures with 0% C3A. (B) Dissolution curves of alite in the presence of different PCEs (dos. 0.15% of dry polymer by weight) for mixtures with 5% C3A. (C) Dissolution curves of alite in the presence of different PCEs (dos. 0.15% of dry polymer by weight) for mixtures with 10% C3A. Table 3. Retardation on alite dissolution induced by different PCEs (dos.0.15% by weight of dry polymer) in the presence of different amount of C3A. Type None 113-9 17-3 17-2

Type of PCE Dos. %wt 0.15 0.15 0.15

0% C3A 260 540 980 1280

End of induction period (minutes) 5% C3A 10% C3A 260 260 380 315 640 455 840 495

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  233

Figure 2. Curves of precipitation of portlandite in the presence of different PCEs (dos. 0.15% of dry polymer by weight) for mixtures with 0% C3A. to recognize that the different PCEs induced different periods of retardation in the alite dissolution. Particularly, at the same dosage of 0.15% of dry polymer by weight of powder, PCE 17-2 showed the highest retarding effect, followed by PCE 17-3 and PCE 113-9, the last showing the lowest retarding effect. The same order of retardation was observed in the presence of 5% and 10% of C3A (Figures 1B and 1C, respectively), but with a reduction of the induction period for all the mixtures. The end of the induction period for the different mixtures containing different PCEs and different amount of C3A, as determined by the XRPD curves, is reported in Table 3. These results confirmed that all the polymers induced a retardation in the dissolution of alite. Particularly, PCE 17-2, characterized by the lowest carboxylic acid to ester ratio and lowest anionic charge density on the backbone (Rc = 2.8) was the most retarding, followed by PCE 17-3 (Rc = 3.9) and PCE 113-9 (Rc = 10.5). It is noteworthy that PCE 113-9 has the longest EO side chain compared to PCE 17-3 and PCE 17-2. The addition of C3A did not exert any influence on the duration of the induction period of alite in the mixtures without PCEs and progressively reduced the induction period of mixtures containing PCEs. Furthermore, the sequence of retardation of the different PCEs was not altered by the presence of C3A, at both the dosages of C3A. These results could be in relation with the higher affinity of PCEs towards C3A, as previously reported by different authors.17,18 Similar behaviour for the different polymers was observed by monitoring the precipitation of portlandite, as shown in Figure 2 for the mixtures without C3A. Also in this case, all the polymers induced a retardation in the precipitation of portlandite, with the same sequence observed for the alite dissolution. Furthermore, for each polymer mixture, the occurrence of the precipitation of portlandite was concurrent with the increase of the dissolution of alite. Similar results were obtained for the mixtures with different dosage of C3A, not reported in this paper. The results of adsorption tests on the different model cements containing different PCEs are shown in Tables 4, 5 and 6. Table 4 reports the adsorption data for mixtures without C3A, Table 5 for mixtures with 5% C3A and Table 6 for mixtures with 10% C3A, respectively. Different from XRPD tests, which were conducted only at the dosage of 0.15% by weight of dry PCE, adsorption tests were performed over a range of dosages (0.05, 0.10. 0.15. 0.20 and 0.30% by weight of dry PCE by weight of powder). Results of adsorption

234 SP-302-17

Table 4. Adsorption data of different PCEs on mixtures alite/gypsum in the ratio 95/5 in the absence of C3A. Dosage %wt

%wt

0.05

100

0.10

100

0.15

100

0.20

64

0.30

41

PCE 113-9 PCE 17-3 PCE 17-2 Adsorption Adsorption Adsorption mg/m2 mmole/m2 mg/m2 mmole/m2 mg/m2 mmole/m2 %wt %wt (ounce/yr2) (mmole/yr2) (ounce/yr2) (mmole/yr2) (ounce/yr2) (mmole/yr2) 0.4 8.9E-06 0.4 2.8E-05 0.4 1.9E-05 100 100 (1.8E-06) (7.4E-06) (1.8E-06) (2.3E-05) (1.8E-06) (1.6E-05) 0.8 1.8E-05 0.8 5.5E-05 0.6 3.1E-05 100 81 (3.6E-06) (1.5E-05) (3.6E-06) (4.6E-06) (2.7E-06) (2.6E-06) 1.2 0.9 6.5E-05 0.7 3.7E-05 2.7E-05 79 64 (5.4E-06) (4.1E-06) (5.4E-05) (3.2E-06) (3.1E-05) 1.0 2.3E-05 0.8 5.8E-05 0.6 3.1E-05 53 40 (0.5E-06) (1.9E-05) (3.6E-06) (4.8E-05) (2.7E-06) (2.6E-05) 1.0 2.2E-05 0.7 4.6E-05 0.5 2.6E-05 28 23 (0.5E-06) (1.8E-05) (3.2E-06) (3.8E-05) (2.3E-06) (2.2E-05)

Table 5. Adsorption data of different PCEs on mixtures alite/gypsum/C3A in the ratio 90/5/5. Dosage %wt

%wt

0.05

100

0.10

100

0.15

100

0.20

69

0.30

45

PCE 113-9 PCE 17-3 PCE 17-2 Adsorption Adsorption Adsorption mg/m2 mmole/m2 mg/m2 mmole/m2 mg/m2 mmole/m2 %wt %wt 2 2 2 2 2 (ounce/yr ) (mmole/yr ) (ounce/yr ) (mmole/yr ) (ounce/yr ) (mmole/yr2) 0.4 0.9E-05 0.4 2.8E-05 0.4 2.0E-05 100 100 (1.8E-06) (0.8E-05) (1.8E-06) (2.3E-05) (1.8E-06) (1.7E-05) 0.8 1.8E-05 0.8 5.7E-05 0.6 3.1E-05 100 79 (3.6E-06) (1.5E-05) (3.6E-06) (4.8E-05) (2.7E-06) (2.6E-05) 1.2 2.8E-05 1.1 8.1E-05 1.0 4.8E-05 96 82 (5.4E-06) (2.3E-05) (5.0E-06) (6.8E-05) (4.5E-06) (4.0E-05) 1.1 2.5E-05 1.4 9.7E-05 0.9 4.7E-05 85 59 (5.0E-06) (2.1E-05) (6.3E-06) (8.1E-05) (4.1E-06) (3.9E-05) 1.1 2.5E-05 1.2 8.8E-05 1.0 4.8E-05 52 41 (5.0E-06) (2.1E-05) (5.4E-06) (7.3E-05) (4.5E-06) (4.0E-05)

tests were expressed in different ways: a) as percent by weight of PCE adsorbed with respect to the initial amount (%wt), b) as adsorption of milligrams of PCE by unit surface area (mg/m2, ounce/yr2) and c) as adsorption of millimole of PCE by unit surface area (mmole/m2, mmole/yr2). Adsorption in mg/m2 were calculated from the percent adsorption and the specific surface area of both alite (1.292 m2/g, 51.68 yr3/ounce) and C3A (1.812 m2/g, 72.48 yr3/ounce) phases, while adsorption in mmole/m2 were obtained by dividing adsorption in mg/m2 by the number average molecular weight (Mn) of the different PCEs. Results of Table 4 indicate that PCE 113-9 is completely adsorbed on alite up to the dosage of 0.15% by weight and PCE 17-3 is fully adsorbed up to the dosage of 0.10% by weight; finally, PCE 17-2 is completely adsorbed only up to the dosage of 0.05% by weight. In general, a strong interaction between PCE and the substrate is observed and no partition of the PCE between surfaces and solution is experimentally detected at low dosages. At higher dosages, the adsorption value, expressed as per cent by weight (%wt),

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  235 Table 6. Adsorption data of different PCEs on mixtures alite/gypsum/C3A in the ratio 85/5/10. Dosage %wt

%wt

0.05

100

0.10

100

0.15

100

0.20

100

0.30

73

PCE 113-9 Adsorption mg/m2 mmole/m2 (ounce/yr2) (mmole/yr2) 0.4 0.9E-05 (1.8E-06) (0.8E-05) 0.8 1.8E-05 (3.6E-06) (1.5E-05) 1.2 2.7E-05 (5.4E-06) (2.3E-05) 1.6 3.6E-05 (7.2E-06) (3.0E-05) 1.7 4.0E-05 (7.7E-06) (3.3E-05)

%wt 100 100 100 100 62

PCE 17-3 PCE 17-2 Adsorption Adsorption mg/m2 mmole/m2 mg/m2 mmole/m2 %wt (ounce/yr2) (mmole/yr2) (ounce/yr2) (mmole/yr2) 0.4 2.8E-05 0.4 1.9E-05 100 (1.8E-06) (2.3E-05) (1.8E-06) (1.6E-05) 0.8 5.6E-05 0.7 3.4E-05 88 (3.6E-06) (4.7E-05) (3.2E-06) (2.8E-05) 1.2 8.3E-05 1.1 5.4E-05 94 (5.4E-06) (6.9E-05) (5.0E-06) (4.5E-05) 1.6 11.1E-05 1.2 6.1E-05 79 (7.2E-06) (9.3E-05) (5.4E-06) (5.1E-05) 1.5 10.4E-05 1.5 7.2E-05 62 (6.8E-06) (8.7E-05) (6.8E-06) (6.0E-05)

Table 7. Adsorption of different PCEs on alite and C3A at saturation. PCE 113-9 17-3 17-2

Saturation concentration, mmole/m2 (mmole/yr2) Alite C 3A 2.2E-05 10.2E-05 (1.8E-05) (8.5E-05) 5.7E-05 47E-05 (4.8E-05) (39.1E-05) 3.1E-05 24.6E-05 (2.6E-05) (20.5E-05)

Selectivity Ratio at saturation SRC3A/alite 4.6 8.2 7.9

is always below 100%, indicating that unadsorbed polymer molecules remain dissolved in the mixing solution. This condition is considered representative of the saturation of PCE on alite phase and the corresponding concentration of adsorbed molecules is considered to be the saturation concentration of each PCE on alite phase. The saturation concentration of the different PCEs on alite phase was expressed as the average values of adsorption at saturation (in mmole/m2, mmole/yr2 of alite) and are reported in Table 7. Table 5 and Table 6, concerning mixtures containing 5% and 10% C3A, respectively, indicate that the saturation conditions were attained for higher dosages compared to the mixtures with only alite. In saturation conditions (adsorption less than 100% by weight), both alite and C3A particles resulted simultaneously saturated. By subtracting the saturation concentration of the different PCEs on alite phase, it is possible to calculate the saturation concentration for the different PCEs on C3A, at both the dosages of C3A. As for alite phase, the saturation concentration of the different PCEs on C3A was calculated from the adsorption concentrations at saturation at both the dosages of C3A (5% and 10%) and the average results (in mmole/m2 of C3A) are reported in Table 7. The assumption that the amount of PCE adsorbed on alite remains constant in the presence of C3A represents a limiting case, such that the figures displayed in Table 7 for the amount of PCE adsorbed onto C3A are minimum values. These results indicate that the saturation concentration of PCEs for C3A is much higher than for alite phase, confirming the higher selec-

236 SP-302-17

Figure 3. Increase of the time until end of induction of alite dissolution as a function of the concentration of unadsorbed PCE molecules in the solution. tivity of PCEs to C3A, as previously reported by other authors.17,18 When C3A is added to the mixture, it acts as a preferential sink for PCEs adsorption and leaves the surface of alite phase less hindered by PCE molecules. As a consequence, the induction period and the retardation of hydration are reduced, as shown from the dissolution kinetic curves (Figures 1A, 1B and 1C) and from the data of Table 3. Table 7 also indicates that the different PCEs have different affinity for C3A and alite phase. This difference can be expressed by the Selectivity Ratio SRC3A/alite, defined as the ratio of the saturation concentration on C3A and on alite phase for each PCE. According to Table 7, PCE 17-3 and PCE 17-2 are the more selective to C3A compared to PCE 113-9. Such results, derived from adsorption measurements, are in full agreement with the kinetic dissolution data measured by XRPD (Figures 1A, 2A, 3A and Table 3). In fact, the addition of C3A to the alite/gypsum mixtures reduced the retardation of PCEs 17-3 and 17-2 much more than for PCE 113-9. These results indicate that the different PCEs showed different affinity for C3A and alite phases and their hydrates. PCEs with shorter chain (PCE 17-2 and PCE 17-3) showed higher affinity for C3A and its hydrated phases than PCE 113-9. These findings are consistent with the results of other authors26,27 who demonstrated that PCEs intercalate in the hydrated aluminates phases and that PCEs with longer side chains showed lower intercalation compared with PCEs with shorter side chains. From the adsorption data and the Selectivity Ratios of the different PCEs, it is possible to evaluate the influence of the different PCEs on the retardation in alite dissolution both in saturated and unsaturated conditions. Retardation in saturation conditions The unadsorbed fraction of the different PCEs at the dosage of 0.15% by weight of dry polymer depends on the amount of C3A and the type of PCE. The concentration of the unadsorbed PCEs left in the mixing solution (W/S =0.5) can be easily calculated from the initial dosage of molecules of PCEs and the corresponding adsorption, expressed in per cent by weight (%wt) (Table 8). These results indicate that with PCE 17-2, a residual amount of polymer always remained dissolved in the solution, even with the highest

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  237 Table 8. Unadsorbed PCEs molecules in solution after mixing with alite containing different amount of C3A. Type of PCE

Dosage (% by weight of dry polymer)

113-9

0.15

17-3

0.15

17-2

0.15

Initial mmole/g of powder (mmole/ ounce) 2.7E-05 (9.0E-04) 10.0E-05 (33.3E-04) 6.8E-05 (22.7E-04)

Unadsorbed fraction, mmole/g of solution (mmole/g ounce of solution) 0% C3A 5% C3A 10% C3A 0

0

4.2E-05 (0.0011) 4.9E-05 (0.0012)

8.0E-06 (0.0002) 2.5E-05 (0.0006)

0 0 8.2E-06 (0.0002)

amount of C3A (10%). On the other hand, with PCE 17-3, unadsorbed polymer could be detected only in the mixture without C3A and with 5% of C3A, but not in the mixture with 10% C3A. Finally, no residual polymer in the solution were detected with PCE 113-9 at any dosage of C3A. The relationship between the end of induction period of the different mixtures (Table 3), measured by XRPD, and the unadsorbed fractions of Table 8 is shown in the following Figure 3. Figure 3 indicates that the induction period is prolonged as the concentration of PCE polymer molecules in solution increases. These results are in agreement with the previous work of Sowoidnich and Rössler,10 who reported an increase of the retardation on hydration by increasing the dosage of PCEs beyond the saturation concentration. These authors attributed these results to the effect of PCEs molecules in the pore solution on the nucleation and growth of C-S-H. This hypothesis is supported by Dragoni et al.,28 who found remarkable effects of PCEs on both the delay of the nucleation and growth of synthetic calcium carbonate and the morphology of the newly formed crystals. These authors hypothesize that similar effects could be produced by PCEs in cement based systems, with consequences on the delay of setting time. Furthermore, Figure 3 indicates that, at least in the range of investigated dosage, the retardation seems to be dependent on the number of the residual molecules in solution rather than the type of PCEs (PCE 17-2 and PCE 17-3). Retardation in undersaturation conditions In conditions of undersaturation, all the PCE molecules are adsorbed and no residual molecules are left in the solution, as in the case of PCE 113-9 for all the mixtures and of PCE 17-3 with 10% C3A (Table 8). The mmoles of adsorbed PCE molecules on alite, in conditions of undersaturation (100% of polymer adsorption, expressed as %wt), can be calculated from the adsorption data by assuming that the selectivity ratios of the different PCEs to alite and C3A are the same as those calculated in saturated conditions (Table 7). The relationship between the end of the induction period and the concentration of PCE molecules adsorbed in conditions of undersaturation is shown in Figure 4. Figure 4 indicates that the end of induction period is delayed by the increase of the coverage of alite surface by PCE molecules, confirming that the adsorbed PCE molecules act as an hindrance for alite dissolution at the interface solid/solution. As for the oversaturated condition at higher dosages, this effect seems to be dependent on the number of

238 SP-302-17

Figure 4. Increase of the time until end of induction of alite dissolution as a function of the coverage by PCE molecules in condition of undersaturation. molecules which cover the alite surface rather than the molecular structure of the examined PCEs. According to these results, it is possible to assume that the retardation on hydration induced by PCEs is due to a double contribution: 1) the primary effect of adsorbed molecules onto the alite surface which acts as an hindrance for alite dissolution and 2) the supplementary effect of unadsorbed molecules in the mixing solution which delays the nucleation of C-S-H, causing a further reduction on the alite dissolution. This supplementary effect might be even stronger in retarding the alite dissolution than the primary one. CONCLUSIONS The results on the simplified model cement alite/gypsum/C3A obtained by using the coupled complementary XRPD and GPC techniques, indicated the strong influence of C3A on the dissolution kinetic of alite in the presence of PCEs. The addition of C3A causes the reduction of the induction period of alite in the presence of PCEs, basically for the higher selectivity of these polymers to C3A and its hydrates compared to alite. Consequently, the alite surface is less hindered towards hydration and also the overall adsorption of PCEs is increased, with a corresponding reduction of unadsorbed PCE molecules in the solution. Both these effects influence the retardation on alite hydration, and two distinct mechanisms can be inferred: 1) the primary effect of adsorbed molecules onto the alite surface, which acts as an hindrance for alite dissolution and 2) the supplementary effect of unadsorbed molecules in the mixing solution which delays the nucleation of C-S-H and CH, causing a further reduction on the alite dissolution. In both the cases, for the examined PCEs the retardation seems to be mainly dependent on the number of polymer molecules and not on their molecular structure. The number of polymer molecules is related to the total number of charges, which represent both the anchoring sites for adsorption onto the surface and the active functions for chelation/complexation of calcium ions by the unadsorbed polymer in the solution. Nevertheless, the total number of charges introduced with the polymers cannot fully explain the retardation induced by PCEs. The charge density of PCEs, expressed as the ratio between carboxylic groups to ester groups, seems to be the main parameter influ-

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  239 encing the adsorption of the PCEs and, consequently the retardation: the lower the charge density, the lower the adsorption to both phases and the higher the supplementary effect due to unadsorbed PCEs molecules. Specific experiments will be addressed to clarify the role of the anionic charges of the adsorbed PCEs polymers on the dissolution rate of alite and on the potential chelation/ complexation of calcium ions in the mixing solution. Future work will be focused also on real cement systems in order to verify the results obtained on the model cements. AUTHOR BIOS Giorgio Ferrari graduated in Chemistry at Padua University in 1977 and presently is senior researcher in Mapei for concrete admixtures and development of sustainable concrete technologies. He is author of several scientific publications and international patents on concrete technology and of two books on environmental issues. Vincenzo Russo received his B.Sc. degree in chemistry at the University of Messina, Italy, in 2004 and M. Sc. degree in physical chemistry at the University of Pisa, Italy, in 2006. He received a Master of Arts degree in organic chemistry from Rice University, Houston, Texas, in 2009. He currently holds a research scientist position at Mapei S.p.A., Milan, Italy, where he works in the field of concrete admixtures. Massimo Dragoni graduated in Industrial Chemistry at the Milan University in 1992. Presently, he is senior researcher and responsible of the liquid chromatography section of the Analytical Department in Mapei. Gilberto Artioli holds a Laurea Degree from the University of Modena and a PhD in Geophysical Sciences from the University of Chicago. He is presently full professor at the University of Padua and Director of the CIRCe Centre for the investigation of cement materials. Maria Chiara Dalconi received a degree (B.S. and M.S.) in Geological Sciences in 1997 and a Ph.D. in Mineralogy, Petrology and Crystallography in 2001 from the University of Modena and Reggio Emilia (Italy). Presently she is assistant professor of Mineralogy in the Department of Geoscience at the University of Padua and her principal research activities are in mineralogy applied to cement materials. Michele Secco is postdoctoral Fellow at the Department of Civil, Environmental and Architectural Engineering of the University of Padua. He graduated in Science and Technology for the Archaeological and Artistic Heritage in 2008, and obtained his Ph.D. degree in Earth Sciences at the Department of Geosciences of the University of Padua in 2012. His research focuses on the mineral-petrographic, chemical, microstructural and physical-mechanical characterization of inorganic binders and structural materials (cement, lime, mortars and plasters, concrete).

240 SP-302-17

Leonardo Tamborrino received his master degree in Geological Sciences and Technologies from the University of Modena and Reggio Emilia in 2014, with exchange visits as guest scientist in the Marine Geology group of MARUM-University of Bremen. He is presently a post-graduate grant holder at the University of Padua for chemical-physical investigation on cement materials. Luca Valentini received a PhD in Earth sciences at the National University of Ireland Galway in 2009. He is currently a post-doctoral fellow at the Department of Geosciences, University of Padua. He is the author of several papers focused on computer modelling and 3D microstructural studies in the field of materials and Earth sciences. REFERENCES 1. Rixom, R., and Mailavaganam, N. P., “Chemical admixtures for concrete”, E&F Spon, London, 1999. 2. Flatt, R. J., and Schober, I., “Superplasticizers”, in “Understanding the rheology of concrete” (Ed. Roussel N.), Woodhead publishing, 1999, pp. 144-208. 3. Uchikawa, H.; Hanehara, S.; and Sawaki, D., “The role of steric repulsive force in the dispersion of cement particles in fresh paste prepared with organic admixture,” Cement and Concrete Research, V. 27, No. 1, 1997, pp. 37-50. doi: 10.1016/S0008-8846(96)00207-4 4. Ohta A., Sugiyama T. and Uomoto T., “Study of the dispersing effect of polycarboxylate-based dispersant on fine particles”, 6th CANMET, SP 195-14, 2000, pp. 211-227. 5. Flatt, R. J.; Schober, I.; Raphael, E.; Plassard, C.; and Lesniewska, E., “Conformation of adsorbed comb copolymer dispersants,” Langmuir, V. 25, No. 2, 2009, pp. 845-855. doi: 10.1021/la801410e 6. Ferrari, L.; Kaufmann, J.; Winnefeld, F.; and Plank, J., “Interaction of cement model systems with superplasticizers investigated by atomic force microscopy, zeta potential, and adsorption measurements,” J. Coll. Interf. Sci., V. 347, No. 1, 2010, pp. 15-24. doi: 10.1016/j.jcis.2010.03.005 7. Mollah, M. Y. A.; Adams, W. J.; Schennach, R.; and Cocke, D. L., “A review of cement-superplasticizer interactions and their models,” Advances in Cement Research, V. 12, No. 4, 2000, pp. 153-161. doi: 10.1680/adcr.2000.12.4.153 8. Winnefeld, F.; Becker, S.; Pakusch, J.; and Götz, T., “Effects of the molecular architecture of comb-shaped superplasticizers on their performance in cementitious systems,” Cement and Concrete Composites, V. 29, No. 4, 2007, pp. 251-262. doi: 10.1016/j. cemconcomp.2006.12.006 9. Cheung, J.; Jeknavorian, A.; Roberts, L.; and Silva, D., “Impact of admixtures on the hydration kinetics of Portland cement,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005 10. Sowoidnich, T., and Rössler, C., The influence of superplasticizers on the dissolution of C3S”, Superplasticizers and Other Chemical Admixtures in Concrete, SP-262, T.C. Holland, P.R. Gupta, V.M. Malhotra ed., American Concrete Institute, Farmington Hills, Mich., 2009, pp. 335-346 11. Ridi, F.; Dei, L.; Fratini, E.; Chen, S.-H.; and Baglioni, P., “Hydration kinetics of tricalcium silicate in the presence of superplasticizers,” The Journal of Physical Chemistry B, V. 107, No. 4, 2003, pp. 1056-1061. doi: 10.1021/jp027346b

The Influence of C3A on the Dissolution Kinetics of Alite/Gypsum Mixtures in the Presence of PCEs  241 12. Pourchet, S.; Comparet, C.; Nicoleau, L.; and Nonat, A., “Influence of PC superplasticizers on tricalcium silicate hydration”, 12th ICCC, Montreal, Canada, 2007. 13. Eusebio, L.; Goisis, M.; Manganelli, G.; Paternò, V. A.; and Gronchi, P., “Influence of comb-polymer structure on C3S phase hydration,” Mat. Sci. Appl., V. 4, 2013, pp. 35-44. 14. Matsuyama, H., and Young, J. F., “Intercalation of polymers in calcium silicate hydrate: a new synthetic approach to biocomposites,” Chemistry of Materials, V. 11, No. 1, 1999, pp. 16-19. doi: 10.1021/cm980549l 15. Popova, A.; Geoffroy, G.; Renou-Gonnord, M.-F.; Faucon, P.; and Gartner, E., “Interaction between polymeric dispersants and calcium silicate hydrates,” Journal of the American Ceramic Society, V. 83, No. 10, 2000, pp. 2556-2560. doi: 10.1111/j.1151-2916.2000. tb01590.x 16. Beaudoin, J. J.; Dramé, H.; Raki, L.; and Alizadeh, R., “Formation and properties of C-S-H-PEG nano-structures,” Materials and Structures, V. 42, No. 7, 2009, pp. 1003-1014. doi: 10.1617/s11527-008-9439-x 17. Zingg, A.; Winnefeld, F.; Holzer, L.; Pakusch, J.; Becker, S.; Figi, R.; and Gauckler, L., “Interaction of polycarboxylate-based superplasticizers with cements containing different C3A amounts,” Cement and Concrete Composites, V. 31, No. 3, 2009, pp. 153-162. doi: 10.1016/j.cemconcomp.2009.01.005 18. Nawa T. and Eguchi H., “Effect of cement characteristics on the fluidity of cement paste containing organic admixture”, 9th ICCC, 1992, pp. 597-603. 19. Plank, J.; Zhimin, D.; Keller, H.; Hossle, F.; and Seidl, W., “Fundamental mechanism for polycarboxylate intercalation into C3A hydrate phases and the role of sulfate present in cement,” Cement and Concrete Research, V. 40, No. 1, 2010, pp. 45-57. doi: 10.1016/j. cemconres.2009.08.013 20. Flatt, R. J., and Houst, Y. F., “A simplified view on chemical effects perturbing the action of superplasticizers,” Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8 21. Yamada, K.; Ogawa, S.; and Hanehara, S., “Controlling of the adsorption and dispersing force of polycarboxylate-type superplasticizer by sulfate ion concentration in aqueous phase,” Cement and Concrete Research, V. 31, No. 3, 2001, pp. 375-383. doi: 10.1016/S0008-8846(00)00503-2 22. Gay, C., and Raphaël, E., “Comb-like polymers inside nanoscale pores,” Advances in Colloid and Interface Science, V. 94, No. 1-3, 2001, pp. 229-236. doi: 10.1016/ S0001-8686(01)00062-8 23. Quennoz, A., and Scrivener, K. L., “Interactions between alite and C3A-gypsum hydrations in model cements,” Cement and Concrete Research, V. 44, 2013, pp. 46-54. doi: 10.1016/j.cemconres.2012.10.018 24. Juilland, P.; Gallucci, E.; Flatt, R. J.; and Scrivener, K., “Dissolution Theory applied to the Induction period in Alite Hydration,” Cement and Concrete Research, V. 40, No. 6, 2010, pp. 831-844. doi: 10.1016/j.cemconres.2010.01.012 25. Ferrari, G.; Cerulli, T.; Clemente, P.; Dragoni, M.; Gamba, M.; and Surico, F., “Influence of Carboxylic Acid-Carboxylic Ester Ratio of Carboxylic Acid Ester Superplasticizers on Characteristics of Cement Mixtures”, Superplasticizers and Other Chemical Admixtures in Concrete, SP-195, V.M. Malhotra ed., American Concrete Institute, Farmington Hills, Mich., 2000, pp. 505-519.

242 SP-302-17

26. Flatt, R. J., and Houst, Y. F., “A Simplified View on Chemical Effects Perturbing the Action of Superplasticizers,” Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8 27. Plank, J.; Dai, Z.; Zouaoui, N.; and Vlad, D., “Intercalation of Polycarboxylate Superplasticizers into C3A Hydrate Phases”, 8th CANMET/ACI International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, SP-239, V.M. Malhotra ed., American Concrete Institute, Farmington Hills, Mich., 2006, pp. 201-214 28. Dragoni, M.; Lo Presti, A.; Cerulli, T.; Biancardi, A.; Moretti, E.; and Salvioni, D., “Interaction of polycarboxylate-based superplasticizers with cement: synthetic calcium carbonate as model to investigate the structural evolution of C-S-H”, Proceeding of the 34th International Conference in Cement Microscopy, April 1-4, 2012 – Halle (Saale), Saxony-Auhalt, Germany.

SP-302-18

Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers by Jinzhi Liu, Jiaping Liu, Yong Yang, Dongliang Zhou, and Qianping Ran Slump loss of fresh concrete was a common issue in engineering construction, especially under high temperature and long distance transportation conditions. Therefore, slow-release polycarboxylate superplasticizers (PCEs) have been widely used to reduce the slump loss in various engineering projects. In this study, three kinds of PCEs with different proportions of hydroxyl ester groups (HEG) were synthesized and characterized by 1H-NMR and Gel Permeation Chromatography (GPC). The effects of the HEG content on dispersion retention, adsorption kinetics and zeta potential of fresh cement suspensions were systematically investigated to figure out the mechanism. For PCEs with the same molar ratio of carboxyl group and reactive polyether, the dispersion retention ability of PCEs is improved with the increasing of HEG ratio. HEG in PCEs can be slowly converted to carboxyl groups in the alkaline environment of cement suspension, which could enhance the adsorption of PCE molecules onto the surface of cement particles. Despite major of the initially adsorbed-PCE molecules might have been embedded in hydration products, free PCE molecules with released carboxyl groups in the solution can continuously adsorb onto the surface of cement particles and play a role in dispersion. This explains why slow-release PCEs have a dispersion retention effect on cement particles within a certain time. Keywords: polycarboxylate; slow-release; zeta potential; adsorption; dispersability retention. INTRODUCTION Polycarboxylate superplasticizers (PCEs) are widely used in different industrial fields to improve the performance of concrete. With the development of the construction industry and requirements of environmental protection, the proportion of ready-mixed concrete in building construction is growing. The slump loss of fresh concrete is a major problem, especially under high temperature and long distance transportation conditions. The traditional methods to solve the fluidity loss, such as changing the mixing concrete process, 243

244 SP-302-18

Table 1 Phase composition and physical properties of the cement Chemical composition (wt %) SiO2 19.5

Al2O3 6.16

CaO 64.3

MgO 1.04

Fe2O3 4.41

SO3 2.52

K2O 0.20

Na2O 0.48

Loss 1.39

Percent volume diameters/μm D50 VMD 16.38 20.76

Blain surface areas/cm2·g-1 2965

adding water to reshape concrete, adding water reducing agent repeatedly and set retarder, might not only increase the production cost, but also have a serious impact on the strength and durability of concrete.1-4 Therefore, slow-release PCEs have been widely used to reduce the slump loss in various engineering projects. Slow-release PCEs could disperse cement particles for a long time. The slump of fresh concrete mixed with slow-release PCEs could keep the slump for a long period while its loss is small. This characteristic can maintain good performance of fresh concrete under high temperature and long distance transportation conditions. Besides, the concrete obtained the good properties such as compressive strength, freeze-thaw resistance, and low-carbonation properties and so on. Robert Flatt5 considered there might be three forms of PCE molecules in early hydration stage: (1) One form is the adsorbed PCE molecule on the surface of unhydrated cement particles and cement hydration products. The initial dispersion effect comes from PCE of this form. (2) The second form is consumed by intercalation,coprecitation or micellization, which does not play any role in the dispersion process. (3) The last form is the remained free PCE molecule neither consumed nor adsorbed in the slurry, which might be in equilibrium with PCE molecule of the first form and thus closely associated with the long-time fluidity. In order to maintain the slump of concrete, the amount of PCE molecule of the last form should be increased. Herein hydroxyl ester groups (HEG) were introduced into PCE molecules. This type of PCE contained low content of carboxyl groups and thus low amount of PCE was adsorbed at initial stage. Based on the molecular design principles, three kinds of PCEs with different proportions of HEG were synthesized and characterized, and the mechanism of slow-released PCEs was explored. EXPERIMENTAL INVESTIGATION Materials An ordinary portland cement (52.5, Jiangnan Onoda Co., China), which meets the requirements of GB8076 standard was used throughout all the experiments. The cement composition was determined by X-ray fluorescence. The particle size analysis using a Helos-Sucell Laser particle size analyzer (SYMPATEC Instruments, Germany) showed the volume average mean particle size (VMD) to be about 20.76 μm(8.17×10-4in), and the percentage volume diameters d50 to be 16.38 μm(6.45×10-4in). The Blaine surface area is 2965 cm-2·g-1. The characteristics and compositions of the cement are given in Table 1. Synthesis of slow-release Superplasticizers Slow-release PCEs with different proportions of hydroxy ester groups (HEG) were prepared from methallyl ether of polyethylene glycol (MAPEG, Mw=2000) and hydroxyethyl acrylate (HEA) through a free-radical copolymerization in an aqueous medium at

Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers 245

Fig. 1 -Chemical structure of the slow-release PCEs comb polymer dispersants room temperature under a nitrogen gas atmosphere. Sodium persulfate (Na2S2O8) was used as the initiator and sodium hydrogen sulfite (NaHSO3) as a reductant and a chain transfer agent in order to control the molecular weight. The chemical structure of synthesized polymer samples was shown in Figure 1. Polymer samples with different MAPEG to HEA ratios (1:2, 1:4, 1:6) and a constant side chain length were synthesized and named respectively MAH2, MAH4 and MAH6. The composition and molecular weight of polymers were determined by 1H-NMR spectroscopy and a Shimadzu Refractive Index detector equipped with TSK-PWXL (TOSOH) columns, respectively. The architecture of prepared PCEs and their molecular characteristics are listed in Table 2. Dispersion effect of slow-release PCEs The fresh cement pastes were prepared at 20oC and a water to cement ratio (w/c) of 0.29 using a cement mortar mixing machine according to GB/T8077-2008. Water (87 g) containing a certain amount of PCEs was added to cement (300 g) and mixed for 2 min at low speed and further 2 min at high speed. The paste was filled in a cone with a height of 60 mm, an inner diameter at the bottom of 36 mm and at the top of 60 mm, respectively. The cone was pulled up and the spread diameter (F) of the cement cake was measured. The spread value (F) was the average of two perpendicularly crossing diameters. The cement paste was also used for the measuring its fluidity. From the spread (F), the dispersion ability (relative flow area ratio) was calculated by,6 Q was the area ratio of final spread and the bottom cone

Q= (F2-602)/602=F2/602-1

246 SP-302-18

Table 2 Structure information of slow-release copolymers Sample

Molar composition/% MAPEG HEA

Mna

Mw b

MAH2

33.3

66.7

24100

42900

MAH4

20

80

25900

43600

MAH6

16.7

83.3

24600

45100

a

Mn: number average molecular weight.

b

Mw: weight average molecular weight.

Adsorption of slow-release PCEs on cement particles Adsorption measurements were performed to detect the quantity of superplasticizer adsorbed on the cement particles by means of a total organic carbon analyzer (TOC), Multi N/C3100 (analytikjene AG, Germany). After mixing the cement with the solution containing the polymer, the amount of superplasticizer remaining in the solution can be measured from the separated solution phase from the suspension. The consumed polymer is estimated to be the difference in concentration before and after contact with the cement powder. 120g solution containing 0.2% of PCEs and 60g of cement were mixed. The sample solution was separated at 1, 5, 15, 30, 60 and 90 min by centrifuging at 10,000 rpm for 2 minutes. The supernatant was immediately decanted and diluted with deionized water for TOC. Zeta potential of cement particles containing slow-release PCEs The surface charge of colloidal systems is called zeta potential. The zeta potential data were collected with the electroacoustic-based ZetaProbe analyzer (Colloidal Dynamics Inc., American), zeta potential represents the potential separations between the dispersion medium and the stationary layer of water molecules and ions attached to the dispersed particle. The suspensions were prepared by dispersing 100 g of cement in 200 ml deionized water containing different dispersants under vigorous agitation, and then zeta potential of the cement paste was determined from 1 to 90 min. RESULTS AND DISCUSSION Slow-release PCEs with different proportions of hydroxyl ester groups (HEG) were synthesized and characterized by 1H-NMR. The effects of the HEG content on dispersion retention, the adsorption, and zeta potential of the concentrated cement suspensions were investigated in details. Structures of the prepared slow-released PCEs 1 H-NMR spectra of slow-released PCEs copolymer are shown in Fig. 2. The 1H-NMR spectra of the three copolymers are substantially the same. Peaks are the same position only different peak intensity. A case study in MAH4, the signals at δ = 1.0~1.8 ppm (b) and δ = 1.9~2.5 ppm (c) were attributed to CH2 protons and CHCOOR proton in the polymer backbone, respectively. The peak at δ=0.82 ppm (a) corresponds to the CH3 protons, and the peak at δ = 3.4~3.8 ppm (d) corresponds to the CH2 protons of the ethylene oxide repeating units in MAPEG. In addition, a small peak at δ = 4.21~4.28 ppm (f) is attributed to the CH2

Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers 247

Fig. 2- 1H-NMR spectra of slow-release PCEs copolymer of MAH4 (D2O). protons that are attached to the ester group, and the peak at δ = 3.86 ppm (e) corresponds to the CH2 protons attached to the hydroxyl group. Based on the 1H-NMR spectra, it was demonstrated that the preparation of slow-release PCEs was successful. Effect of slow-release PCEs dispersion of cement pastes The effect of PCEs with different proportions of HEG on dispersion retention was shown in Fig. 3. The spread of cement pastes containing different proportions of HEG shows the same behavior. As the proportion of HEG increased, the paste fluidity increased quickly. This result was in good agreement with the adsorption behavior. The dispersion ability of MAH6 was far greater than MAH2 after 30 min. So PCE with the higher HEG proportion had a better dispersion retention effect on cement particles within a certain time. Adsorption behavior of slow-release PCEs Adsorption of dispersant on solid–liquid interfaces is a crucial step in dispersing cement,7-9 and adsorption was the one of the most important measurements for illustrating the interaction between PCEs and cements. The adsorption amounts of PCE in the suspensions with the same dosage were shown in Fig. 4. For the three PCE polymers, the adsorption curves appeared the same trend. The adsorption amount increased slowly within 30 min. with this period, although the surfaces of cement particles contained empty sites, relatively small amount molecules can be adsorbed on the surface of cement particles owning to few adsorption groups (carboxyl group). At the same time, the remaining parts of PCEs in the water phase can be slowly converted carboxyl group and gradually adsorbed on the cement particles. After 30min, the adsorption of polymers on cement particles carried on continuing to quickly increase. MAH6 had the maximum amount of adsorption among

248 SP-302-18

Fig. 3- Influence of different HEG in PCEs on dispersion ability of cement paste three polymers, resulting from the larger proportion of adsorption groups (carboxyl group) which came from the hydrolysis of HEG. PCE molecules adsorbed more easily onto the surface of cement particles if the HEG ratio in PCE was larger, and with larger proportion of adsorption.10,11 Effect of slow-release PCEs on zeta potential of cement pastes The zeta potential of cement suspensions with slow-release PCEs with different HEG proportions and polyoxyethylene long side chain of the same length was measured against time (Fig. 5). The behavior of zeta potential development of the cement systems with MAH2, MAH4 and MAH6 are completely different. Zeta potential of cement with MAH2 underwent a gradual increase with time. Zeta potential of cement with MAH4 appeared with the same behavior before 72 min and slightly fell in later time. Zeta potential of the cement with MAH6 seemed similar with MAH4 and changed in larger range. When adding the PC, cement particles adsorbed carboxyl groups and formed electrical double layer on the cement particles. As time went on, HEG could be slowly converted to carboxyl groups (COO-) in the alkaline environment of cement suspension, which would have a direct influence on the zeta potential of the system, while the carboxyl groups in the polymer backbone might be wrapped by the long side chains.12 Therefore, different amounts of HEG per molecular caused different zeta potentials. Within 30min the zeta potential increase was caused by hydration: (1) More positive charge generated by the initial hydration of C3A; (2) The adsorbed PCE molecules might be embedded in hydration product. After 30min, the increase of zeta potential became slower (MAH2), or the zeta potential began to decrease (MAH4 and MAH6). This could be attributed to the hydrolysis and continuous adsorption of PCE molecules.

Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers 249

Fig. 4- Zeta potential of cement suspensions with different PCEs. The effect of slow-release PCEs on zeta potential and adsorption has been measured in order to elucidate the mechanisms of slow-release PCEs with different HEG proportions. When PCEs was added in the cement system, low amount of adsorbed PCE caused a small initial fluidity of cement paste. At the same time, HEG continuously hydrolyzed and converted to carboxyl group in the alkaline environment of cement. Thus, the proportion of carboxyl groups became larger and larger. As the hydration of cement went on, the PCE molecules adsorbed on the cement particles would be covered by hydration products gradually and a fluidity loss might be observed. The remaining parts of PCEs in the water phase can be adsorbed on cement particles and continue to play a role of dispersant. Therefore,the adsorption amount increased within a certain period. With the increase in the proportion of HEG, carboxyl group gradually increased which is providing adsorption on the cement particles. So more PCEs in cement slurry can keep the cement particles fluidity and adsorption equilibrium, and slump loss of cement fluidity became lower. PCEs with different carboxyl content have different adsorption capacity and adsorption rate on cement particles, which affect the zeta potential of cement particles. For MAH2, due to the low HEG proportion of MAH2, less carboxyl groups could be converted from HEG compared with MAH4 and MAH6. The amount of carboxyl groups of adsorbed PCE molecules were not enough to offset the positive charge increase from the released calcium ions during hydration process,8 the zeta potential increased gradually. But for PCEs with a high proportion of HEG (MAH4 and MAh6), higher proportion of carboxyl groups was released. At a certain time, the increasing positive charge of cement particles from the released Ca2+ might be neutralized by the adsorbed PCEs with higher carboxyl group proportions. Then the zeta potential began to decrease. In short, the slump retention mechanism of slow-release PCEs was caused by the hydrolysis of PCEs and continuous adsorption of PCE molecules on the surface of cement particles. The adsorption speed and adsorption equilibrium of PCEs to cement particles affects the dispersion retention of cement particles within a certain time.

250 SP-302-18

Fig. 5-Amount of slow-release PCEs adsorbed on cement particles versus time CONCLUSIONS In summary, this paper developed a slow-release type PCE which could effectively reduce the slump loss of cement paste. The mechanism of PCEs in dispersion retention was studied. This type of PCE might be significantly useful for high temperature and long distance transportation conditions of concrete. 1. Slow-release PCEs with different proportions of HEG were synthesized by radical polymerization and characterized by 1HNMR and GPC. The polymer structure was consistent with the experimental design. 2. With the increasing of HEG ratio while the same molar ratio of reactive polyether, the dispersion retention ability of PCEs was improved. 3. Researches on zeta potential and adsorption kinetics show that HEG in PCEs could be slowly converted to carboxyl groups in the alkaline environment of cement suspension, which could enhance the adsorption of PCE molecules onto the surface of cement particles. The adsorption speed and adsorption equilibrium of PCEs to cement particles affects the dispersion retention of cement particles within a certain time. AUTHOR BIOS Jinzhi Liu is a researcher of Jiangsu Research Institute of Building Science, Nanjing, China. She received her MS in Polymer Chemistry and Physics Zhejiang University. She has been working in the field of concrete admixtures, focusing on polycarboxylate superplasticizers. She is author or co-author of over 10 papers and patents on chemical admixtures. Jiaping Liu is a Vice-President of Jiangsu Research Institute of Building Science and General Manager of Jiangsu Bote New Material Ltd. He received his MS in Civil Engineering from Southeast University and PhD in materials science from Nanjing University of Technology. He has written and presented over 30 papers on his research activities

Preparation and Mechanism Study of Slow-Release Polycarboxylate Superplasticizers 251 related to high performance concrete, high performance admixture, deformation and cracking of concrete. Yong Yang is a researcher of Jiangsu Research Institute of Building Science, Nanjing, China. He received his MS in Polymer Chemistry and Physics from Lanzhou University. His research interests include polycarboxylate superplasticizers. He is author or co-author of over 20 papers and patents on chemical admixtures. Dongliang Zhou is a researcher of Jiangsu Research Institute of Building Science, Nanjing, China. He received his MS in Polymer Chemistry and Physics from Sichuan University. He has been working in the field of polycarboxylate superplasticizers. He is author or co-author of over 10 papers and patents on chemical admixtures. Qianping Ran is a Chief scientist at State Key Laboratory of High Performance Civil Engineering Materials and a Research Engineer (Prof.) in Admixture Product Technologies at Jiangsu Research Institute of Building Science, Nanjing, China. He received his MS in Material Science from Sichuan University and PhD in Polymer Chemistry and Physics from Nanjing University. He has been working in the field of concrete admixtures, focusing on the development of new polymeric superplasticizers. He is also author or co-author of over 60 papers and patents on chemical admixtures. REFERENCES 1. Schober, I., and Flatt, R. J., Optimizing polycarboxylate polymers. 8th CANMET/ACI International Conference Superplasticizers and Other Chemical Admixtures in Concrete Sorrento: ACI, 2006, 16 2. Yamada, K.; Takahashi, T.; Hanehara, S.; and Matsuhisa, M., “Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer,” Cement and Concrete Research, V. 30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6 3. Plank, J., and Winter, C., “Competitive adsorption between superplasticizer and retarder molecules on mineral binder surface,” Cement and Concrete Research, V. 38, No. 5, 2008, pp. 599-605. doi: [J]10.1016/j.cemconres.2007.12.003 4. Papayianni, I.; Tsohos, G.; Oikonomou, N.; and Mavria, P., “Influence of superplasticizer type and mix design parameters on the performance of them in concrete mixture,” Cement and Concrete Composites, V. 27, No. 2, 2005, pp. 217-222. doi: 10.1016/j. cemconcomp.2004.02.010 5. Flatt, R.-J., and Houst, Y.-F., “A simplified view on chemical effects perturbing the action of superplasticizers,” Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8 6. Yamada, K.; Hanehara, S.; and Honma, K., “Effects of the chemical structure on the properties of polycarboxylate-type superplasticizer,” Cement and Concrete Research, V. 30, No. 2, 2000, pp. 197-207. doi: 10.1016/S0008-8846(99)00230-6 7. Yoshioka, K.; Tazawa, E.; Kawai, K.; and Enohata, T., “Adsorption characteristics of superplasticizers on cement component minerals,” Cement and Concrete Research, V. 32, No. 10, 2002, pp. 1507-1513. doi: 10.1016/S0008-8846(02)00782-2

252 SP-302-18

8. Zingg, A.; Winnefeld, F.; Holzer, L.; Pakusch, J.; Becker, S.; and Gauckler, L., “Adsorption of polyelectrolytes and its influence on the rheology, zeta potential, and microstructure of various cement and hydrate phases,” Journal of Colloid and Interface Science, V. 323, No. 2, 2008, pp. 301-312. doi: 10.1016/j.jcis.2008.04.052 9. Yamada, K.; Ogama, S.; and Hanehara, S., “Controlling of the adsorption and dispersing force of polycarboxylate-type superplasticizer by sulfate ion concentration in aqueous phase,” Cement and Concrete Research, V. 31, No. 3, 2001, pp. 375-383. doi: 10.1016/S0008-8846(00)00503-2 10. Yoshioka, K.; Tazawa, E.; Kawai, K.; and Enohata, T., “Adsorption characteristics of superplasticizers on cement component minerals,” Cement and Concrete Research, V. 32, No. 10, 2002, pp. 1507-1513. doi: [J]10.1016/S0008-8846(02)00782-2 11. Plank, J., and Winter, C., “winter, Ch. Competitive adsorption between superplasticizer and retarder molecules on mineral binder surface,” Cement and Concrete Research, V. 38, No. 5, 2008, pp. 599-605. doi: 10.1016/j.cemconres.2007.12.003 12. Plank, J., and Hirsch, C., “Impact of zeta potential of early cement hydration phases on superplasticizer adsorption,” Cement and Concrete Research, V. 37, No. 4, 2007, pp. 537-542. doi: 10.1016/j.cemconres.2007.01.007

SP-302-19

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption by Dirk Lowke and Christoph Gehlen The effect of pore solution composition on zeta potential and superplasticizer adsorption has been investigated experimentally. The investigations were conducted on highly concentrated suspensions, containing quartz flour, limestone flour, cement and combinations of these materials. Furthermore cement-limestone suspensions with different types of cements and a varying ratio of cement to limestone were investigated. The results show that the zeta potential is significantly determined by pore solution. In a pore solution of highly concentrated cement suspensions the zeta potential can be characterized by the ratio of calcium to sulfate concentration. Furthermore it was shown that the superplasticizer adsorption is affected the zeta potential. At higher, more positive zeta potentials the superplasticizer molecules are more likely adsorbed onto the solid surfaces. Moreover, superplasticizer adsorption in limestone-cement suspensions is predominantly controlled by the composition of pore solution rather than the ratio of cement to limestone flour. If the ion concentration of the pore solution is artificially kept constant, the polymer adsorption is almost constant independent of the cement to limestone ratio in the suspension. Keywords: zeta potential; pore solution; superplasticizer adsorption. INTRODUCTION Rheological properties of high performance concrete like SCC and UHPC are affected by the surface properties of the particles, the properties of the liquid phase and the adsorbed polymers. It is necessary to consider the interactions of the colloidal particles in the cementbased suspension to understand the effect of these parameters on the rheological properties. The present contribution focuses on the effect of pore solution composition on zeta potential and superplasticizer adsorption in highly concentrated cement-based suspensions. The aim of the investigations was to provide a clear understanding of the mechanism of surface charge and zeta potential in cement suspensions with high solid fractions as well as the interactions between the composition of pore solution, zeta potential and superplasticizer adsorption onto surfaces of cement and mineral additions. 253

254 SP-302-19

For this reason, the experimental program was divided into to three test series: a) Zeta potential of quartz, limestone and cement in suspensions with high solid fractions - Effect of pore solution composition b) Superplasticizer adsorption onto quartz, limestone and cement - Effect of zeta potential c) Superplasticizer adsorption in limestone-cement suspensions - Effect of cement and limestone content ELECTRIC DOUBLE LAYER AND ZETA POTENTIAL Surface charge and Stern potential of oxides and carbonate minerals As soon as a cement or mineral addition particle is exposed to water, its surface becomes electrically charged. The potential ψ0 resulting from the surface charge is causing a predominant attraction of ions of opposite charge (counterions). As a result, an electrical double layer is formed between the charged particle surface and the liquid phase. From a historical point of view, the double layer model originates from the Helmholtz model (1879) and the Gouy-Chapman model (1910/13). Finally, both models were combined by Stern (1924).1 The Stern model describes the interface as a combination of a rigid layer of adsorbed ions on the surface, the so called Stern layer, and a diffuse layer of mobile counterions. The mechanisms causing a surface charge are complex. According to the Surface Complexation Theory2 surface charges are formed due to chemisorption and dissociation of water molecules onto unsaturated surface lattice ions. The oxygen atoms of chemisorbed water molecules fill the vacant cationic surface sites. Simultaneously, the anionic surface sites are stabilized by the transfer of dissociated protons from the chemisorbed water molecules. This hydration process leads to a surface composed of hydroxylated cationic sites (>KatOH0) and protonated anionic sites (>AnH0), where > represents the mineral lattice and Kat or An is the mineral cation or anion respectively.2 A surface potential ψ0 is finally formed by protonation (>KatOH2+) or deprotonation (>KatO–, >An–) of the hydration surface sites. For oxide minerals like SiO2, the anionic crystal lattice sites are formed by an oxygen atom of the mineral. Thus the primary hydration surface sites are described as hydroxylated surface cations (>SiOH0). By protonation or deprotonation of the hydroxyl groups surface charges are formed as follows: Positive surface sites: >SiOH0 + H3O+ → >SiOH2+ + H2O Negative surface sites: >SiOH0 + OH– → >SiO– + H2O Hence, the surface charge of oxides in water is controlled by the pH-value of the solution. At high pH-values, as usual in cement-based suspensions, a negative surface potential ψ0 is formed. For carbonate minerals like CaCO3, the surface consists of hydroxylated cationic sites (>CaOH0) as well as protonated anionic sites (>CO3H0). By protonation or deprotonation surface charges are formed, again depending on solution pH. Positive surface sites: >CaOH0 + H3O+ → >CaOH2+ + H2O Negative surface sites: >CaOH0 + OH– → >CaO– + H2O >CO3H0 + OH– → >CO3– + H2O In addition, the charge of a CaCO3-particle is determined by the adsorption of calcium ions (Ca2+), hydrogen carbonate ions (HCO −;3 ) and carbonate ions (CO 2 −;3 ). These ions

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption 255 originate from carbonic acid (H2CO3) as well as marginal dissolved amounts of calcium carbonate (14 mg/l at 20°C). Calcium ions: (CaCO3(s) + H2O ⇄ Ca2+ + HCO3– + OH–) >CO3-Ca+, >CaO-Ca+, >Ca-HCO30 Carbonate ions: (CO2 + 3H2O ⇄ H2CO3 + 2H2O ⇄ HCO3– + H3O+ + H2O ⇄ CO32– + 2H3O+) >Ca-HCO30, >Ca-CO3– The solubility of these ion species is controlled by the pH-value of the solution. Mainly hydrogen carbonate ions as well as calcium ion are existent in neutral pH-range. Providing a sufficient calcium concentration, CaCO3–particles dispersed in water in contact with atmospheric air therefore can have a positive Stern potential. With increasing pH-value, the number of dissolved carbonate ions increases. Due to adsorption of the divalent carbonate ions onto cationic surface sites, an increasingly number of negative charges are formed in the stern layer (>Ca-CO3–). Simultaneously, the concentration of dissolved calcium ions is decreasing, resulting in a reduction of positive sites like >CO3-Ca+ or >CaO-Ca+. Furthermore, negative surface charges are formed by deprotonation of the hydroxylated calcium sites or carbonate sites of the surface (>CaO–, >CO3–). Thus, the Stern potential of calcium carbonate in alkaline milieu becomes progressively negative. The role of ion adsorption in cement-based suspension The pore solution of cement-based suspensions is typically characterized by a high pH > 12 as well as a high ionic strength > 100 mmol/l. A considerable difference compared to the mechanism described before is the high concentration of dissolved ions in pore solution. Furthermore, besides dissolved ions from the crystal lattice of the particle surface, there are additional ion species, originated from readily soluble cement components. During the first minutes after water addition alkali sulfates (Na2SO4, K2SO4), free lime (CaO) and parts of the setting regulator (CaSO4•xH2O) are dissolved. Besides pH-controlling hydroxide anions, the pore solution contains further anions like sulfate and chloride as well as cations like calcium, potassium and sodium. Further Ca2+-ions and OH–-ions results from the hydrolysis of tricalcium aluminate and tricalcium silicate. These ions affect the charge of the dispersed particle significantly by adsorbing to the stern layer. In this context, the adsorption of divalent calcium cations and sulfate anions are of particular importance for the formation of the Stern potential. Surface charge equalization increases with increasing adsorption of ions to the Stern layer. Furthermore, in case of the adsorption of divalent ions a charge reversal is possible.1 For calcium carbonate suspensions, it could be experimentally confirmed that the effect of calcium ion concentration is of more importance than the effect of pH-value.3 Similarly, a positive charge reversal of an initially negative surface charge due to adsorption of cations was verified in experiment.4,5 Zeta Potential While the surface potential ψ0 in a cement-based suspension is - owing to ion adsorption - experimentally not accessible, the determination of the zeta potential ψZ allows the conclusions to the Stern potential ψS. The zeta potential ψZ of a particle surface is defined as the potential at a shear plane that arises from particle motion. The exact location of this shear plane is a matter of controversial discussions in the literature. While in various

256 SP-302-19

Table 1 – Characteristics of powder materials r [oz/in3] 1.79

As [ft2/in3]

3.3

r [lb/yd3] 5225

3.7

5225

1.79

653

3.0

5225

1.79

529

1.56

529

1.62

723

r [g/cm3]

As [m2/cm3]

Cement (C(a))

3.1

Cement (C(b))

3.1

Cement (C(c))

3.1

Quartz flour (Q)

2.7

3.0

4550

Limestone flour (L)

2.8

4.1

4720

582

Table 2 – Pore solution composition of suspension L.C(a) Ion species Concentration [mmol/l]

Ca2+ 16.9

Na+ 27.8

K+ 198.9

Cl– 33.6

SO42– 88.0

OH– 50.9

hypothesizes, the shear plane is within the diffuse layer, investigations done by Lyklema,6 Smith7 or Sprycha8 show that the shear plane is located directly at the interface between the Stern plane and the bulk electrolyte. Thus, the zeta potential can be used as a direct measure of the Stern potential depending on the measurement technique, compare.9 According to the manifold of influencing factors, the zeta potential of cement and mineral additions given in the literature vary over a wide range. A detailed overview is available in.10 In the subsequent investigations, the effect of pore solution on zeta potential and superplasticizer adsorption is discussed in detail. MATERIALS AND METHODS Mixture proportions Three Portland cements (C(a) - C(c)) as well as quartz flour (Q) and ground limestone (L) as mineral additions were used in the investigations. The density of the materials was determined by helium pycnometry and the surface area by nitrogen adsorption, Table 1. The mixes were prepared with a commercial polycarboxylate ether based superplasticizer (SP) with 35% solids in aqueous solution and a number average molecular weight of 69.000 g/mol. Three kinds of pastes were prepared for the investigations. L, Q, C(a) using 100 l/m3 (0.1 yd3/yd3) of water and 100 l/m3 (0.1 yd3/yd3) of ground limestone (L), quartz flour (Q) or cement (C) L*, Q* using 100 l/m3 (0.1 yd3/yd3) of artificial pore solution and 100 l/m3 (0.1 yd3/yd3) of ground limestone (L) or quartz flour (Q) L.C(a), Q.C(a) using 100 l/m3 (0.1 yd3/yd3) of water, 50 l/m3 (0.05 yd3/yd3) of ground limestone (L) or quartz flour (Q) and 50 l/m3 (0.05 yd3/yd3) of cement (C) The volumetric water-to-powder ratio Vw/Vp of all pastes was 1.0. The superplasticizer dosage was kept constant at 2.9 mg per cm3 of solids (1.7∙10-3 oz per in3 of solids). The composition of the artificial pore solution was equivalent to the pore solution of paste containing 100 l/m3 (0.1 yd3/yd3) of water, 50 l/m3 (0.05 yd3/yd3) of cement C(a) and 50 l/ m3 (0.05 yd3/yd3) of ground limestone (L), Table 2.

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption 257 Zeta Potential The zeta potential was determined using electro-acoustic spectroscopy (Quantachrome Dispersion Technology, Electroacoustic Spectrometer, DT 1200) 15 min after water addition. Electro-acoustic spectroscopy allows measurements of undiluted pastes with w/cratios in a range of 0.3 to 0.8. For the determination of the zeta potential a relative motion between the charged particles and the surrounding fluid is necessary. As soon as the particle or the surrounding fluid is moving, a part of the diffuse layer is slipping off. The higher the velocity, the larger is the part of the diffuse layer that is slipped off. As a result, the particle appears no longer electrically neutral. The electrical potential at the shear plane characterizes the zeta potential. For the electroacoustic zeta potential measurement, a high frequency acoustic wave (≈ 1 MHz) induces an oscillating motion of particles and ions. Due to the lower inertia, there is a larger movement of the ions in the double layer. This generates dipoles, creating a macroscopic detectable electrical field. Due to the high ion concentration of the pore solution of cement-based suspensions the measured electroacoustic signal TVI (total vibration current) contains a signal contribution from the colloidal particles CVI (colloid vibration current) and a signal contribution from the ionic background IVI (ion vibration current). Therefore, for the determination of the CVI, two measurements have to be performed: a) a separate measurement of the extracted pore solution for the determination of the IVI and b) a measurement of the suspension for the determination of the TVI. De-ionized water was used for the preparation of suspensions of cement and mineral additions. Water and solids were manually mixed for 3 min. 15 min after water addition pore solution was extracted from the pastes by means of a vacuum pump. Subsequently the IVI of the pore solution (15 ml / 0.5 fl oz) was determined. The determination of the TVI of a separately prepared suspension (500 ml / 16.7 fl oz) was again carried out 15 min after water addition. To prevent particle sedimentation the suspension was gently stirred during the measurement. Ion concentration of pore solution and Superplasticizer Adsorption In addition, 500 ml (16.7 fl oz) of suspensions were prepared to determine pH-value and ion concentration of pore solution as well as superplasticizer adsorption. Pore solution was extracted from the pastes 15 min after water addition by means of a cylindrical pressure device, see.1 The pH-value was determined by potentiostatic titration. Furthermore, the concentration of anions and cations was determined using ion chromatography and inductively coupled plasma optical emission spectrometry (ICP-OES) respectively. The total organic carbon (TOC) content of the pore solution and the superplasticizer solution were determined by high-temperature oxidation of the organic ingredients. The amount of adsorbed superplasticizer was calculated as the difference between the TOC of the added superplasticizer solution and the pore solution of the mortar. RESULTS AND DISCUSSION Zeta potential of quartz, limestone and cement - Effect of pore solution composition The zeta potential of mineral additions, cement and mixtures of mineral additions and cement dispersed in water or artificial pore solution is shown in Figure 1. For quartz in

258 SP-302-19

Figure 1 – Zeta potential of quartz, limestone and cement water (Q) the zeta potential is strongly negative at -52.8 mV. As there are almost no ions dissolved from the SiO2–particles, the zeta potential of quartz-water suspensions represents the charge conditions at the surface. The negative characteristic of the zeta potential is caused by the deprotonation of the hydroxyl groups >SiOH0 + OH– → >SiO– + H2O. In contrast, for limestone in water (L) the zeta potential is positive at + 17.3 mV. The positive zeta potential is primarily attributed to the adsorption of monovalent hydrogen carbonate anions as well as divalent calcium cations, which are the predominantly dissolved ion species in pore solution of limestone suspensions at neutral pH-range. While the adsorption of monovalent hydrogen carbonate ions induces a neutralization of the cationic calcium sites of the surface (>Ca-HCO30), the adsorption of divalent calcium ions causes a positive charge reversal of the anionic carbonate surface sites (>CO3-Ca+). In total, this results in a larger number of positive charges, i.e. a positive zeta potential. The zeta potential of the investigated cement (C(a)) and cement-addition mixtures (Q.C(a), L.C(a-c)) varies between -4.6 and +2.1. Thus, the absolute value of the zeta potential of the cement-based suspensions is significantly lower than for quartz or limestone dispersed in water, Figure 1. The main reason for the decrease is the high concentration of ions in the pore solution of cement-based suspensions. Immediately after water addition, ions are dissolved from the readily soluble compounds of the cement. The adsorption of counterions from the solution to the Stern plane causes a charge equalization, which is reflected in a significantly reduction of the absolute zeta potential value.

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption 259

Figure 2 – Effect of Ca/SO4-ratio on zeta potential of cement and mineral additions To illustrate the mechanism of counter ion adsorption, the mineral additions were dispersed in an artificial pore solution. The composition of the artificial pore solution is equivalent to the pore solution of the limestone-cement suspension L.C(a), Table 2. It was apparent that the absolute value of zeta potential of quartz and limestone dispersed in the artificial pore solution (Q* and L*) significantly decreases to -17.5 mV and -6.3 mV, respectively, Figure 1. Analogous to the cement-based suspensions, the ions dissolved in the artificial pore solution adsorb to the Stern layer of quartz or limestone, resulting in a charge equalization or even in a charge reversal, like in the case of limestone. The charge reversal of the limestone surface is attributed to two mechanisms: a) With increasing pH-value (increasing OH–-concentration) the number of negative surface sites increases (>CaO–, >CO3–) and b) the divalent sulfate anions (SO42–), which are dissolved in high concentration in the artificial pore solution, adsorb onto the remaining cationic surface sites. As a result, the zeta potential of the limestone particles in artificial pore solution is similar to the zeta potential of cement and cement-addition mixtures. In the case of quartz flour, again the high pH of the artificial pore solution causes an increase of negative surface sites (SiO–). Additionally, due to the adsorption of the divalent calcium ions, dissolved in the artificial pore solution as well, the number of negative charges decreases, resulting in a substantial decrease of the zeta potential absolute value. However, owing to the low solubility of calcium in water, the number of calcium ions dissolved in the artificial pore solution is limited. By contrast, in a real cement-based

260 SP-302-19

suspension, calcium ions can be dissolved constantly, causing a further reduction of the zeta potential, compare Q.C(a). According to the previous results and discussions, the ion concentration of the pore solution affects the zeta potential significantly. In principle all ion species in pore solution are able to adsorb to the Stern layer. However, the adsorption capacity is controlled by sign of charge, charge density (valence and ion size) and the size of the ionic hydration shell. Various investigations of ion adsorption onto mineral surfaces demonstrate that the adsorption capacity increases with increasing valence of the ions. Thus, with regard to ions dissolved in cement-based suspensions, divalent calcium cations as well as divalent sulfate anions can be classified as potential determining ions. For this reason the Ca/SO4-ratio was used to describe the effect of ion concentration on zeta potential quantitatively. The zeta potential of the investigated mineral additions, cements and addition-cement mixtures depending on the Ca/SO4-ratio of the pore solution is shown in Figure 2. It is apparent that the zeta potential increases with increasing Ca/ SO4-ratio. A Ca/SO4-ratio of about 0.4 to 0.6 characterizes the point of zero charge (pzc). Another important conclusion can be derived from the interrelation between Ca/ SO4-ratio and zeta potential ψz. Interestingly, the mineral additions dispersed in water or artificial pore solution (Q*, L*, L) as well as the cement-based suspensions (C(a), Q.C(a), L.C(a,b,c)) are within the same - almost material non-specific - functional correlation. For this reason, it seems reasonable to assume that all particles dispersed in cement-based suspensions exhibit comparable low Stern potentials. The sign and value of the Stern potential is primarily controlled by the adsorption of inorganic ions and can be described by the Ca/SO4-ratio. Superplasticizer adsorption onto quartz, limestone and cement – Effect of zeta potential From the point of view of thermodynamics, the adsorption of superplasticizer polymers onto a particle surface is associated with a heat release (enthalpy loss) and/or an entropy increase of the total system. An entropy increase results for example from a release of adsorbed ions or water molecules from the surface during polymer adsorption. On the other hand, enthalpic adsorption is caused by attractive electrostatic as well as van der Waals interactions. Adsorption due to electrostatic forces requires both a charge at the surface and a polymer bearing dissociated ionic groups. Polycarboxylate superplasticizers are polyanions containing carboxylate groups COO–. The adsorption of the negatively charged carboxylate groups can occur either directly onto cationic surface sites or indirectly onto cations, adsorbed to the Stern layer. In the latter case, owing the high charge density, mainly calcium ions are suitable as adsorption sites for the carboxylate groups. Thus, provided a sufficient high calcium concentration, superplasticizer adsorption is possible even in domains of initial negative surface charge. The electrostatically caused adsorption of superplasticizer polymers should therefore be directly depending on the charge conditions of the particle surface. For this reason, the effect of zeta potential on the superplasticizer adsorption was investigated. Therefore suspensions of water or artificial pore solution, superplasticizer and quartz flour, limestone flour, cement or mixtures of quartz, limestone and cement were prepared. The superplasti-

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption 261

Figure 3 – Effect of zeta potential on superplasticizer adsorption (Note: 1mg/m2 = 3.3∙10-6 oz/ft2) cizer content of all suspensions was kept constant at 2.9 mgpol/cm3P (in relation to the solid content = 1.7∙10-3 oz per in3 of solids). The fraction of adsorbed superplasticizer molecules was determined 15 min after addition of water and superplasticizer. After that time the cement hydration changes from the dissolution period to the induction period, characterized by a slowdown of the reaction kinetics.10,11 Thus almost equilibrium conditions of superplasticizer adsorption and desorption can be assumed. The effect of zeta potential on superplasticizer adsorption is shown in Figure 3. It is apparent that for strong negative zeta potentials < 15 mV only minor polymer adsorption of about 0.1 mgPol/m2P occurs (Note: 1mg/m2 = 3.3∙10-6 oz/ft2). At higher zeta potentials between -6.3 and +17.3 mV the superplasticizer adsorption increases by trend with increasing zeta potential from 0.39 to 0.56 mgpol/m2p. However, not every influence on superplasticizer adsorption is captured by this trend. In spite of a higher zeta potential at +2.1 mV for the limestone-cement mixture L.C(c) compared to the mixture L.C(b) at -2.3 mV, there is a lower superplasticizer adsorption for the mixture L.C(c). This is mainly attributed to lower hydration kinetics of cement C during the first 15 min after water addition. Thus, measurements using a heat flow calorimeter show a significantly lower maximum heat flow for the suspension L.C(c) at 1.3 mW/g than for the suspension L.C(b) at 8.1 mW/g, compare.10

262 SP-302-19

Figure 4 – Effect of cement and limestone content on superplasticizer adsorption (Note: 1mg/m2 = 3.3∙10-6 oz/ft2) Superplasticizer adsorption in limestone-cement suspensions – Effect of cement and limestone content The previous results show clearly that superplasticizer polymers adsorb onto cement as well as onto mineral addition particle surfaces. At this point, the question of selective polymer adsorption arises: In which amount does the polymer adsorb onto the different particle surfaces in suspensions containing both cement and mineral addition particles? In the case of zeta potential, it was shown that, due to the high concentration of divalent ions in pore solution and adsorption of these ions to the Stern layer, the charge conditions of different materials in a cement-based suspension are equalized. Furthermore it was discussed that superplasticizer polymers adsorb onto calcium ions of the Stern layer. Thus, it seems reasonable to assume that despite different material properties of cement and mineral additions, there is an equalization of the superplasticizer adsorption characteristics, as soon as these materials are present together in a cement-based suspension. To clarify this question, adsorption measurements of limestone-cement suspensions L.C(a) were carried out at different cement contents (0,0/12,5/25,0/37,5/50,0 Vol.%). Owing to the difference in specific surface of cement (Sv = 3,3 m2/cm3 = 582 ft2/in3) and limestone flour (Sv = 4,1 m2/cm3 = 723 ft2/in3) the total surface area of the solids is decreasing slightly with increasing cement content from 4.1 m2/cm3 to 3.7 m2/cm3 (723 ft2/in3 to 653 ft2/in3). The volumetric water/powder ratio Vw /Vp of the suspensions was kept constant at 1.0. Furthermore, the composition of the pore solution was kept constant at the same level as the pore solution at a cement content of 50 Vol.%. To compensate the decreasing ion concentration at decreasing cement contents, water was substituted by artificial pore solution according to Table 2 from 0% at a cement content of 50 Vol.% to 100% at

Effect of Pore Solution Composition on Zeta Potential and Superplasticizer Adsorption 263 a cement content of 0 Vol.%. In all cases, the dosage of superplasticizer was constant at 2.9 mgpol/cm3p (mass of polymer mpol in relation to total volume of cement and limestone Vp). Figure 4 shows the amount of adsorbed superplasticizer polymers in dependence of the cement content. The results indicate that the adsorbed amount of superplasticizer in relation to the specific surface of the solids is almost constant, independent of the cement content. Even for the limestone suspension without cement but artificial pore solution is the amount of adsorbed polymers at the same level as for the suspension containing 50 Vol.% of cement. According to these results, provided a constant composition of the pore solution, superplasticizer adsorbs in the same order onto cement and limestone. Thus, the adsorption characteristics of superplasticizer of the investigated limestone-cement suspensions, is preponderantly controlled by the composition of the pore solution rather than the kind of solids dispersed in the suspension. CONCLUSIONS The effect of pore solution composition on zeta potential and superplasticizer adsorption has been investigated experimentally. The investigations were conducted on highly concentrated suspensions, containing quartz flour, limestone flour, cement and combinations of these materials. Furthermore cement-limestone suspensions with different types of cements and a varying ratio of cement to limestone were investigated. The results show that the zeta potential of cement and mineral additions is significantly determined by the composition of pore solution. It seems reasonable to assume that all particles dispersed in cement-based suspensions exhibit comparable low Stern potentials. The sign and value of the potential is primarily controlled by the adsorption of inorganic ions. Moreover, the zeta potential cement-based suspensions can be characterized by the ratio of calcium to sulfate dissolved in the pore solution. Furthermore it was shown that the superplasticizer adsorption is affected by zeta potential. At higher more positive zeta potentials the superplasticizer molecules are more likely adsorbed onto the solid surfaces. Moreover, the superplasticizer adsorption in limestonecement suspensions is predominantly controlled by the composition of pore solution rather than the ratio of cement to limestone flour. If the ion concentration of the pore solution is artificially kept constant the polymer adsorption is almost constant independent of the cement to limestone ratio in the suspension. AUTHOR BIOS Dirk Lowke is Senior Researcher and Head of the Working Group Concrete Technology at the Centre for Building Materials (cbm), Technische Universität München in Munich (Germany). His major research fields are mixing, workability and rheology of fresh concrete as well as durability and time dependent deformation properties of hardened concrete. Christoph Gehlen’s work focusses on the description and prediction of the service life of mineral and metallic building materials in dependence of exposure conditions. He graduated at RWTH Aachen University. A doctorate followed in 2000. Together with two partners, he then founded an international engineering consultancy. After accepting an

264 SP-302-19

appointment at Stuttgart University, he took up his position at the TUM in autumn 2008. He’s involved in numerous national and international boards and committees. REFERENCES 1. Lowke, D., Segregation resistance and robustness of Self-Compacting Concrete. Optimization based on modelling interparticle interactions in cement based suspensions. (in German). PhD-Thesis, 2013 2. Foxall, T.; Peterson, G. C.; Rendall, H. M.; and Smith, A. L., “Charge determination at calcium salt/aqueous solution interface. Journal of the Chemical Society, Faraday Transactions 1,” Physical Chemistry in Condensed Phases, V. 75, 1979, pp. 1034-1039. 3. James, R.O.; Healy, T.W.: Adsorption of hydrolyzable metal ions at the oxide-water interface. II. Charge reversal of SiO2 and TiO2 colloids by adsorbed Co(II), La(III), and Th(IV) as model systems. Journal of Colloid and Interface Science 40(1972)1, pp.53-64 4. Koetz, J., and Kosmella, S., Polyelectrolytes and Nanoparticles. Berlin, Heidelberg: Springer, 2007 5. Kumar, A.; Bishnoi, S.; and Scrivener, K. L., “Modelling early age hydration kinetics of alite,” Cement and Concrete Research, V. 42, No. 7, 2012, pp. 903-918. doi: 10.1016/j. cemconres.2012.03.003 6. Lyklema, J.: Water at Interfaces: A Colloid-Chemical Approach. Journal of Colloid and Interface Science 58(1977)2, pp.242-250 7. Plank, J.; Sieber, R.; Schröfl, C.; Lesti, M.; and Gruber, M., Interactions between polycarboxylate superplasticizers, cement and microsilica in ultra-high strength concrete. In: 8th International Symposium on Utilization of High-Strength and High-Performance Concrete, Tokyo (Japan), 2008, pp.129-134. 8. Smith, A.L.: Electrokinetics of the Oxide-Solution Interface. Journal of Colloidal and Interface Science. 55(1976)3, pp.525-530 9. Sprycha, R., and Matijević, E., “Electrokinetics of Uniform Colloidal Dispersions of Chromium Hydroxide,” Langmuir, V. 5, No. 2, 1989, pp. 479-485. doi: 10.1021/ la00086a033 10. Stern, O., “Zur Theorie der elektrolytischen Doppelschicht. Zeitschrift für Elektrochemie und angewandte physikalische Chemie,” Z. Elektrochem., V. 30, 1924, p. 508 11. Van Capellen, P.; Charlet, L.; Stumm, W.; and Wersin, P., “A surface complexation model of the carbonate mineral-aqueous solution interface,” Geochimica et Cosmochimica Acta, V. 57, No. 15, 1993, pp. 3505-3518. doi: 10.1016/0016-7037(93)90135-J

SP-302-20

Optimization of the Structural Parameters and Properties of PCE Based on the Length of Grafted Side Chain by Zi-Ming Wang, Zi-Chen Lu, and Xiao Liu A series of polycarboxylic ether (PCE) superplasticizers were copolymerized through acrylic acid (AA) and isobutylene polyethylene glycol (IPEG) with different molecular weight (500, 1000, 1500, 2000, 2400 and 2700). The molecular weight, molecular weight distribution and reaction ratio were measured by gel permeation chromatography (GPC). The initial fluidity and flow retaining ability were evaluated through mini slump test at the same dosage (by mole and by weight). The results indicated that the dispersing capability of single PCE molecule firstly improved with the increase of backbone length, and then decreased after reaching a critical value. Both the appropriate backbone length and side chain density of PCE were affected significantly by the side chain length. In general, PCE with longer side chain, shorter backbone length and lower side chain density are appropriate to achieve better comprehensive properties. Finally, regression equations were proposed to predict the suitable backbone length and proper side chain density of PCE based on the length of grafted side chain. Keywords: PCE; IPEG; backbone; side chain; structural parameters. INTRODUCTION The PCE molecule has comb structure, which contains negative charged backbone and grafted side chains.1 One outstanding characteristic of comb like PCE is that its chemical structure could be designed and modified according to diverse requirements of concrete constructions. PCEs are characterized by different molecular weight and distribution, backbone length, grafted side chain length, type and density of functional groups. Change of any of the above mentioned structural parameters could alter the final molecular conformation and then affect the properties. Tremendous efforts were paid to study the relationship between structure and properties of PCE, but many of the conclusions are still incomplete and limited to the specific experiments. Yamada2 reported that PCE with longer side chain, lower backbone polymerization and higher contents of sulfonic groups showed higher dispersing power. Nawa3 studied the effect of chemical structure on steric stabilization of PCE and found that longer side chain could provide better dispersing ability with small amount of adsorption. Winnefeld4 concluded that the decreasing density of the side chains 265

266 SP-302-20

Table 1 Chemical and mineral compositions of standard cement (%, by weight) SiO2 22.93

Al2O3 4.29

Fe2O3 2.89

CaO 66.23

MgO 1.92

SO3 0.35

Na2Oeq 0.70

f-CaO 0.64

C3S 58.78

C2S 21.38

C3A 6.49

C4AF 8.77

enhanced workability while the length of side chain and its molecular weight had only a minor effect. Renkas5 found that maleic-based superplasticizers with longer backbones and side chains are more efficient. On the other hand, the acrylic acid derivatives, with shorter backbones and side chains and without carboxylic groups, appeared less efficient. RESEARCH SIGNIFICANCE All cited work discussed single structural parameter and the induced properties without considering the coordination and impact of these parameters to each other. It is commonly believed that all of these structural parameters were actually integrated together to decide the final property of the PCE. Hence, it is necessary to study how one structural parameter change with others to obtain the satisfied overall properties. EXPERIMENTAL INVESTIGATION Materials For the synthesis of series of PCE acrylic acid (AA) and isobutylene polyethylene glycol (IPEG) with different molecular weight (500, 1000, 1500, 2000, 2400 and 2700) were used as copolymerization monomers, mercaptoacetic acid as chain transfer agent and ammonium persulfate (APS) as initiator were used as received. Ordinary portland cement complying with the Chinese National Standard GB8076-1997 was used to prepare the cement pastes. Chemical and mineral compositions of this cement clinker are presented in Table1. Items of investigation Preparation and characterization of the synthesized PCEs—Series of PCEs with different side chain and backbone length were prepared by free radical polymerization in a three necked glass flask equipped with a mechanical stirrer, dosing units for both the monomers and the initiator. Firstly macromer and de-ionized water were charged into the flask and heated to 60°C[140°F],then the mixed solution of mercaptoacetic acid with AA and the APS solution were separately added dropwise into the flask at a constant rate in 180 minutes. After that the flask was kept at 60°C [140°F] for 120 minutes to complete the reaction. The obtained polymer solution (40 wt%) was cooled down to the room temperature and pH adjusted to 6~7 by 0.1 mol/L NaOH solution. Recipes for preparation of PCEs are shown in Table 2 and Table 3. The molar ratio of AA to macromer ([A/E]) was gradually increased when the other reactant amount were fixed. Through the mini-slump test, the [A/E] of synthesized PCE with the best property was ascertained. With [A/E] value fixed, the main chain length of PCE was changed through gradually increasing the molar ratio of mercaptoacetic acid to macromer ([T/E]) value. In this case, series of PCEs with different main chain length and side chain density were synthesized with different macromers. Gel permeation chromatography—Gel permeation chromatography (GPC) equipped with three different detectors, namely refractive index detector (RI), laser light scattering detector (LS) and viscosity detector (VIS) was used to measure the molecular weight (Mw), molecular

Optimization of the Structural Parameters and Properties of PCE Based on the Length of Grafted Side Chain  267 Table 2 Recipes for preparation of PCEs with gradual decreasing side chain density from different macromers Macromers IPEG-500 IPEG-1000 IPEG-1500 IPEG-2000 IPEG-2400 IPEG-2700

1.0 1.0 1.0 1.0 1.0 1.0

1.5 1.5 1.5 1.5 1.5 1.5

2.0 2.0 2.0 2.0 2.0 2.0

2.5 2.5 2.5 2.5 2.5 2.5

A/E (mole/mole) 3.0 3.5 4.2 3.0 3.5 4.2 3.0 3.5 4.2 3.0 3.5 4.2 3.0 3.5 4.2 3.0 3.5 4.2

4.5 4.5 4.5 4.5 4.5 4.5

5.0 5.0 5.0 5.0 5.0 5.0

----6.0 6.0 6.0 6.0

--------7.0 7.0

Table 3 Recipes for preparation of PCE with gradual decreasing main chain length from different macromers Macromers IPEG-500 IPEG-1000 IPEG-1500 IPEG-2000 IPEG-2400 IPEG-2700

0.05 0.05 0.05 0.05 0.10 0.05

0.08 0.09 0.08 0.09 0.17 0.10

T/E (mole/mole) 0.13 0.17 0.13 0.17 0.13 0.17 0.11 0.15 0.25 0.30 0.13 0.20

0.25 --0.25 0.22 0.40 0.25

0.30 --0.30 0.26 0.50 0.30

-------0.60 ---

A/E (mole/mole) 1.5 3.0 3.0 3.5 4.2 5.0

weight distribution (MWI) and the reaction ratio of the synthesized PCE.6 For all polymers, The eluent with 0.05 wt% sodium azide was prepared by dissolving 1g sodium azide in 2kg DI water. The PCE solutions with concentration 4mg/mL were prepared 1 day earlier to allow the PCE dissolve completely. After that, the solutions were filtered with 0.45μm filtering membrane before injection. The test time for every sample was 900s, and the flow velocity was 1mL/min. Mini-slump test—The mini-slump test was conducted according to the Chinese National Standard GB/T 8077-2000 to evaluate the fluidity of fresh cement pastes (FCP). The fluidity of the FCP was represented by the spread flow of the mini- slump test. The spread diameter was recorded as the average of two perpendicularly crossing diameters. To evaluate dispersing property of the prepared PCE precisely, the unreacted macromers were calculated based on the analysis of GPC chromatogram and then subtracted from the dosage. Two dosing methods were employed to better understand the structure-property relationship of PCE, namely the equivalent weight addition method (EWAM) and equimolar addition method (EMAM). The dosage of pure PCE was fixed at 0.15% by weight of cement in the case of EWAM, while the addition of PCE samples with the same side chain length were calculated by ensuring the same mole (molecule number) in the case of EMAM. The water to cement ratio of the cement pastes was fixed at 0.29 in both cases. For each cement paste, the spread diameter was respectively recorded at 5min, 60min and 120 min after mixing with water. The spread diameter at 5 min was defined as the initial fluidity and the values at 60 min and 120 min were used to evaluate the flow retaining ability of PCE. RESULTS AND DISCUSSION PCE structure analysis and macromer reaction ratio There are two problems affecting the precise evaluation of structure-property relationship. The first one is how to detect the “true” polymer content as the existence of unre-

268 SP-302-20

Table 4 Practical and theoretical molecular weight of synthesized PCEs [A/E] Value 1.0:1 2.0:1 2.5:1 3.0:1 3.5:1 4.2:1 4.5:1 5.0:1 6.0:1 7.0:1

Theoretical M 153900 105600 91800 81400 73400 64700 61700 57300 50400 45200

Practical Mw 67300 72700 105900 73500 74000 52300 61700 53700 58800 45400

Polydispersity 1.330 1.254 1.397 1.303 1.303 1.168 1.265 1.273 1.215 1.186

Effective polymer content/% 56.0 82.7 95.6 94.8 95.8 97.3 99.7 98.9 99.1 100.0

Fig.1 Cement paste fluidity by EWAM acted macromer will lead to the unequal adding amount, the other is the difference of real polymer structure and the supposed one. As we know, the free radical polymerization is hard to precisely control the polymer structure, and it is difficult to determine the precise structure of polymer presently. GPC provides us one helpful method to identify the desired structure of synthesized PCEs when only one structure parameter changed regularly while keeping others fixed. The real molecule weight was detected through GPC and then compared with the theoretical molecular weight to ascertain whether the right structure was obtained as designed. We here consider macromer IPEG-2400 as the example to discuss. Data are shown in Table 4. When the [A/E] value is over 2.5, the theoretical molecular weight and the practical molecular weight fitted well, which means the satisfied structure was obtained. However, the effective polymer content decreased sharply when the [A/E] was less than 2.5, which meant that lots of macromer were not co-polymerized into the PCE molecule, resulting in the difference of the theoretical Mw and the real Mw. Effect of side chain density The EWAM and EMAM were taken to evaluate the overall property of PCE and the single PCE molecule property when the side chain density decreased gradually, as shown in Fig.1

Optimization of the Structural Parameters and Properties of PCE Based on the Length of Grafted Side Chain  269

Fig.2 Cement paste fluidity by EMAM

Fig.3 Cement paste fluidity loss rate by EWAM to Fig.4. The fluidity loss rate is the ratio of fluidity loss value within t minutes to the initial fluidity indicating the flow retaining ability of PCE. The initial fluidity increased obviously with the decreasing side chain density while keeping the backbone length constant (Fig.1), which complies the former researcher work.2,7,8 At the same time the fluidity retaining ability weakened proportionally with the decreases of side chain density (Fig.3) in the case of EWAM. This phenomenon can be explained by the adsorption rate of PCE on the cement particles. The increase of anionic COO- group in the backbone improved the adsorption amount of PCE on the cement particles, which was beneficial to the initial fluidity growth. Nevertheless, as more PCE polymer was consumed at the beginning, less polymer were left to keep the fluidity. This effect was more evident in the case of EMAM (Fig.2 and Fig.4), which reflected the dispersing and fluidity retaining ability of single PCE molecule. Integration of considering the initial fluidity and retaining ability of PCE comprehensively, the proper [A/E] value is in the range of 3.0 to 4.2 when using IPEG 2400 as macromer

270 SP-302-20

Fig.4 Fluidity loss rate by EMAM Table 5 Molecular structural parameters of PCE with different backbone length Macromer

IPEG-500

IPEG-1500

IPEG-2400

[T/E] 0.05 0.10 0.17 0.25 0.30 0.05 0.10 0.17 0.25 0.30 0.05 0.10 0.17 0.25 0.30

Mw 36600 26500 15800 10700 > Q21). Their chemical compositions as well as their granulometrical and physical characteristics are shown in Table 1. The specific BET surface area was determined using nitrogen adsorption in accordance with ISO 9277. The particle size distribution was analyzed using laser granulometry. The fineness (x’) and the slope (n) of the particle size distribution were calculated according to Rosin, Rammler, Sperling and Bennet (RRSB). Their mineralogical compositions prior to and after the calcination are shown in Figure 1.

Interactions between Cements with Calcined Clay and Superplasticizers  301

Table 1—Chemical compositions, granulometrical, and physical characteristics of the calcined clays Parameter SiO2 Al2O3 Fe2O3 MgO CaO SO3 Na2O-Eq. react. SiO2, EN 196-2 1) density, EN 196-6 surface area, BET (N2) fineness x’, RRSB slope n, RRSB water demand, EN 196-3

Unit

mass % (XRF)

mass % mass % g/cm3 cm2/g µm mass %

Q11 (chloritic-illitic) 58.8 23.1 8.1 1.8 2.4 0.1 3.0 30.2 2.73 50,501 13.4 0.60 30.5

Q21 (illitic-kaolinitic) 71.8 21.8 1.6 0.6 0.0 0.0 2.1 33.4 2.63 23,043 12.5 0.71 27.0

1) requirement of EN 197-1: ≥ 25.0 mass %

Figure 1—X-ray diffraction pattern of the clays prior to and after the calcination. Cements—portland cement CEM I 42,5 R (CEM I) in accordance with EN 197-1 was used as a reference. Its chemical and mineralogical compositions as well as its granulometrical, mechanical, and physical characteristics are presented in Table 2. Laboratory-made cements were produced by intensive mixing of the CEM I with the particular calcined clay. The proportions of calcined clay in the cements were 20 mass % (portland-pozzolan cement CEM II/A-Q), 35 mass % (portland-pozzolan cement CEM II/B-Q) and 55 mass % (pozzolan cement CEM IV/B (Q)), respectively. The cements met the requirements of EN 197-1. Analyses data for the cements are reported in Table 3.

302 SP-302-23

Table 2—Chemical-mineralogical composition, granulometrical, mechanical, and physical characteristics of the portland cement CEM I 42,5 R Parameter loss on ignition SiO2 Al2O3 Fe2O3 MgO CaO SO3 Na2O-Eq. strength 1), EN 196-1 initial setting, EN 196-3 water demand, EN 196-3

Unit mass %

MPa min

CEM I 2.4 20.5 5.9 2.0 1.0 64.9 2.6 0.75 32/50/59 155

Parameter alite / belite C3Acubic/orthorhombic ferrite anhydrite hemihydrate gypsum free lime/calcite surface area, BET (N2) fineness x’, RRSB slope n, RRSB

mass %

27.5

fineness, EN 196-6

mass % (XRF)

Unit

cm2/g µm -

CEM I 59 / 12 11/3 3 1.1 2.9 1.5 0.5/4.2 11,438 19.7 0.85

cm2/g

3,650

mass % (XRD/ RV)

1) 2 / 7 / 28 d (1 MPa ≈ 0.145 ksi); RV: Rietveld analysis

Table 3—Granulometrical, mechanical, and physical characteristics of the laboratory-made cements cement type CEM I 42,5 R CEM II/A-Q CEM II/A-Q CEM II/B-Q CEM II/B-Q CEM IV/B (Q) CEM IV/B (Q)

abbrev.

CEM I Z20Q11 Z20Q21 Z35Q11 Z35Q21 Z55Q11 Z55Q21

calcined type of clay calcined proportion clay 0 mass % 20 mass % 20 mass % 35 mass % 35 mass % 55 mass % 55 mass %

Q11 Q21 Q11 Q21 Q11 Q21

fineness x’ µm

slope n -

19.7

0.85

17.6

0.77

15.5

0.74

15.8

0.74

15.2

0.74

15.5

0.66

13.2

0.71

BET surface cm2/g 11,438

WD mass%

strength 1) MPa

27,5

32/50/59

18,125

n. d.

n. d.

13,973

n. d.

n. d.

24,855

28.0

20/35/48

14,345

27.5

18/34/47

30,158

27.0

11/23/36

15,963

27.0

11/20/34

WD: water demand; 1): compressive strength 2, 7, 28 d (1 MPa ≈ 0.145 ksi); n. d.: not determined

Superplasticizers—two commercially available superplasticizers in accordance with EN 934-2 based on polycarboxylates, denoted as PCE11 and PCE22, were used. PCE11 was recommended by the manufacturer for the use in ready-mixed concrete, and PCE22 for precast concrete. Chemical and physical characteristics are given in Table 4. The effective charge of the superplasticizers in deionized water, in solution of potassium hydroxide (KOH) at pH 12.6, and in the pore solution of different cement pastes (w/c = 0.35) were determined using a particle charge detector titrating polydiallyldimethylammoniumchloride.

Interactions between Cements with Calcined Clay and Superplasticizers  303

Table 4—Chemical and physical characteristics of the commercial superplasticizers Parameter density, aerometer solid content, EN 480-8 TOC, EN 1484 molar mass Mw molar mass Mn SEC hydrodyn. radius gyration radius deion. Water KOH (pH 12.6) effective CEM I charge in Z35 pore solution Z55

Unit g/cm3 mass % g/dm3 Da nm

µeq/g

PCE11 1.066 30.3 153 81,000 22,000 6.7 27.2 -1,256 -4,763 -68 n. d. -50

PCE22 1.062 29.1 161 95,000 41,000 6.5 16.5 -1,124 -4,217 -274 -95 -245

SEC: Size Exclusion Chromatography with static and dynamic light scattering; n. d.: not determined

Mixing water—deionized water was used for the production of fresh cement paste and tap water for that of fresh concrete. Aggregates—Rhenish sand and gravel of the grading curve “B” in accordance with the German Standard DIN 1045-2, Annex L, with a maximum grain size of 16 mm (0.63 in.), were used. Proportion, mixing procedure, storage of fresh cement paste/fresh concrete, and separation of pore solution The constituent materials were stored at approximately 20°C (68°F). Fresh cement paste with a water to cement ratio of 0.35 (w/c = 0.35) as well as suspensions of deionized water and calcined clay (w/s = 0.35) were produced in a mortar mixer in accordance with EN 196-1 at approximately 20°C (68°F). Cement and water were mixed for 30 s at approximately 140 rpm. A break of approximately 30 s was taken to reincorporate the material that had adhered at the bottom and the walls of the mixing bowl into the fresh cement paste. Mixing was re-started for another 30 s at approximately 280 rpm. The superplasticizer was added approximately 90 s after the water addition and mixed for approximately 60 s at 280 rpm. Quantities of superplasticizer added always refer to the dry solid content of the admixture (Table 4) and with respect to the mass of cement. One cubic meter (1 yd3) of fresh concrete contained 355 kg (599 lbs) cement, 188 kg (317 lbs) water (w/c = 0.53), and approx. 1770 kg (2987 lbs) of aggregates. Concrete was mixed in accordance with EN 480-1 at approximately 20°C (68°F). For each dosage of superplasticizer, a fresh cement paste or concrete was produced to maintain uniform influences on the hydration and sorption processes. The quantity of water within the superplasticizer was always taken into account for the quantity of mixing water added. Samples were stored at approximately 20°C (68°F) in a plastic bowl (paste) or bucket (concrete), both covered with a damp cloth to avoid evaporation. Prior to testing, the samples were homogenized with a spoon or a scoop. Tests were conducted at approximately 20°C (68°F). Approximately 5, 10, 15, 30, 60, 90, and 120 min after the water

304 SP-302-23

addition, the particular fresh cement paste or suspension was vacuum-filtrated for approximately 60 s using a Buchner funnel with a blue-ribbon filter. Separated pore solution was filtered by a 0.45 µm PTFE syringe filter and stored flushed with argon in sealed tubes cold and dark until the analyses were conducted within the following week. Test methods Analyses of pore solution—the conductivity was measured with a conductometer on the original sample. The other analyses were done on subsamples. The concentration of OH- was determined by dynamic equivalence point titration with hydrochloric acid. By ion chromatography, the concentrations of Cl- and SO42- as well as Na+, K+, Ca2+, and Mg2+ were determined (cations: methane sulfonic acid, anions: mixture of sodium carbonate and sodium bicarbonate). Prior to the determination of the cations, the particular sub-sample was acidified with hydrochloric acid. The concentration of aluminum ions was determined in accordance with ISO 10566, that of silicon dioxide in accordance with the German Standard DIN 38405-21. The TOC content was determined in accordance with EN 1484. Zeta potential—the zeta potential of fresh cement paste as well as that of the suspensions with calcined clay and pore solution or deionized water was determined approximately 15 min after the addition to the respective liquid using an electro-acoustic method.27 The electro-acoustic background, resulting from dissolved ions in the particular solution (ion vibration current), was always taken into account for the calculation of the zeta potential. Sorption of superplasticizer—the sorbed quantity of superplasticizer (i.e. intercalated and/or adsorbed)28 was the difference between the TOC content determined in the pore solution and the quantity of TOC added with the addition of the superplasticizer. The TOC content of the particular pore solution without adding superplasticizer was always taken into account. Plastification and saturation dosage of fresh cement paste—the plasticizing effect of the polycarboxylates in fresh cement paste was investigated using a rotational rheometer for building material suspensions. Immediately after the end of the mixing procedure, the sample was filled into the rheometer’s vessel and a paddle was inserted into it. The vessel rotated at 60, 80, 100, 80, and 60 rpm for 5 min each and the torque was measured via the paddle. The torque at 80 rpm after approximately 20 min was the shear resistance. If the shear resistance could not be significantly reduced further by a larger dosage of superplasticizer, the rheological saturation of the particular combination of cement and superplasticizer at the testing conditions used was reached. This dosage was the saturation dosage (SD). Generally, dosages beyond the saturation dosage can cause segregation and retardation.11,29 Duration of plastification of fresh cement paste—this was determined with the mini slump-flow test on the basis of EN 12350-8 using a mini-slump cone in accordance with EN 1015-3 approximately 5, 10, 15, 30, 60, 90, and 120 min after adding water. The quantity of superplasticizer added was 90% of the particular saturation dosage. Consistency and saturation dosage of fresh concrete—the consistency of fresh concrete was determined by measuring the spread of the concrete after 15 shocks with the flow table test in accordance with EN 12350-5 at approximately 10 min after water addition. For the concrete composition and test conditions used, the saturation dosage of the particular cement and superplasticizer combination was reached if the concrete’s flow table spread

Interactions between Cements with Calcined Clay and Superplasticizers  305

Figure 2—Conductivity and ion concentration of pore solution in dependence of the proportion of Q21 in the cement (left) and normalized to the portland cement proportion, pcp (right). could not be significantly increased further by a larger admixture dosage and a paste ring started to form during the spread, indicating segregation. Retention of the concrete’s consistency was determined with an added superplasticizer quantity of 90% of the particular saturation dosage at the same testing ages as for the fresh cement paste. EXPERIMENTAL RESULTS AND DISCUSSION Influences of the type and proportion of calcined clay in the cement on the composition of the pore solution and the zeta potential With an increasing proportion of calcined clay Q21 in the cement the pore solution’s composition varied significantly (Figure 2, left side). The conductivity as well as the concentrations of Na+, K+, SO42-, and Cl- decreased in direct proportion to approximately zero (suspension of calcined clay and deionized water). Also, the OH- concentration decreased in almost direct proportion. These influences on the pore solution compositions are attributed to the substitution of the clinker and sulfate carriers, i.e. readily soluble alkalis and sulfates.20,21 With decreasing alkalinity of the pore solution the solubility of Ca2+ increases.30 Thus, the Ca2+ concentrations remained almost at the same level. The concentrations of Mg2+ were always in the range of the detection limit. The concentrations of Al3+ and Si4+ were in the range of µmol/L. During the testing period of approximately 120 min after adding water, the particular ion concentration remained almost unchanged. Increasing proportions of calcined clay Q11 in the cement influenced the composition of the pore solution to a comparable degree (not shown). Generally, changes in the pore solution’s conductivity, as a measure of its ionic strength, as well as in the ionic composition, especially OH-, SO42- and Ca2+, can influence the zeta potential as well as the constitution of polycarboxylates, depending on the particular polymer structure, and so their respective sorption behavior and performance potential.20,21,30,31 On the right side of Figure 2 the concentrations of the ions in the pore solution are normalized by the proportion of clinker and sulfate carriers in the particular cement. The concentrations of Na+, K+, SO42-, and Cl- remained unchanged emphasizing that these ions were released by the clinker and sulfate carriers. The increased concentration of Ca2+ was electrochemically neutralized by the increased concentration of OH-.

306 SP-302-23

Figure 3—Zeta potential of fresh cement paste in dependence of the type and proportion of calcined clay in the cement (left) and zeta potential of the constituents for their particular proportion in the respective cement (right); CEM I suspended in deionized water and calcined clays in the pore solution of the respective fresh cement paste. The zeta potential of the particular fresh cement paste in dependence of the type and proportion of calcined clay in the cement is shown in Figure 3 on the left side. The zeta potential of the paste with CEM I was slightly negative (approximately -6 mV). Increasing proportions of calcined clay in the cement shifted the zeta potential towards the iso-electrical point (IEP). This was slightly more pronounced using Q21. In general, the more the zeta potential of a suspension is near the IEP, the more this suspension is agglomerated and the demand for a dispersing admixture increases and vice versa. Here, agglomeration and the respective shift of the fresh cement pastes’ zeta potential towards the IEP was mainly induced by the changes in compositions of the pore solutions (cf. Figure 2) as well as due to the different zeta potentials of the particular cement constituents (Figure 3, right side). Portland cement was suspended in deionized water and the calcined clays in the particular pore solution (cf. Figure 2) to mimic the conditions for calcined clay within the cement paste. Sorption and plasticizing effect The sorption of the polycarboxylates and their particular plasticizing effect in dependence of the quantity of active substance added as well as of the cement type are shown in Figure 4. In combination with CEM I, the sorbed quantity of PCE11 increased slightly with the superplasticizer addition (Figure 4, upper left side). This is probably due to the low anionic charge of PCE11 (cf. Table 4) as well as the high concentration of SO42- in the pore solution of the paste with CEM I (cf. Figure 2, left side) and its negative zeta potential (cf. Figure 3, left side). Thus, the sorption was moderate and the majority of added PCE11 remained in the pore solution (depot effect). With an increasing proportion of Q21 in the cement, the sorption of PCE11 increased. This may be due to the decreasing SO42- concentration, the less negative zeta potential of the particular cement paste and Q21, respectively, and also to the large calcined clay surface. This also applies for increasing proportions of Q11, but was more pronounced due its specific surface area which was at least twice as large as that of Q21 (cf. Tables 1 and 3).

Interactions between Cements with Calcined Clay and Superplasticizers  307

Figure 4—Sorption and plasticizing effect of PCE11 (left) and PCE22 (right) in dependence of the quantity added as well as the type of cement. On the lower left side of Figure 4, the particular plasticizing effect of PCE11 is shown. Due to its moderate sorption in combination with CEM I, also its plasticizing effect was moderate. The saturation dosage was approximately 0.18 mass % of CEM I. Compared to the fresh cement paste with CEM I, the paste with cement with 35 mass % Q11 (Z35Q11) was plasticized to a greater extent with lower quantities of PCE11 due to the higher sorption. The saturation dosage decreased to 0.12 mass % of Z35Q11. The proportion of 55 mass % of Q11 in the cement (Z55Q11) enhanced its plasticizing effect further, but approximately the same dosage was necessary to reach saturation (0.13 mass % of Z55Q11). This is due to the significant increase in the specific surface area of the cement with increasing content of Q11. The particle size distribution and the particle packing also play a role and these change with the calcined clay content. The increasing proportions of Q21 led also to an increase in the plasticizing effect of PCE11 but to a significant decrease in the saturation dosage (0.07 mass% of Z55Q21). This was due to the increased sorption of PCE11 with decreasing clinker proportions and sulfate concentrations, as well as due to the lower specific surface area of Q21. On the right side of Figure 4, the results for PCE22 are shown. The sorption of PCE22 was higher in combination with CEM I compared to that of PCE11 (Figure 4, upper right side). This can be explained by its higher anionic charge (cf. Table 4) that enables a stronger sorption even at the high SO42- concentration in the pore solution of the fresh cement paste with CEM I. The sorbed quantities of PCE22 decreased with a decreasing proportion of clinker in the cement. This indicates a strong affinity of PCE22 to clinker. At 55 mass % of calcined clay in the cement, the sorption of PCE22 and PCE11 were comparable. As with

308 SP-302-23

Figure 5—Shear resistance of fresh cement paste in dependence of the sorbed quantity of active substance (left: PCE11, right: PCE22). The shear resistance is normalized to that of the respective paste without superplasticizer. PCE11, the sorbed quantity of PCE22 in combination with the cements containing Q11 was higher than in combination with cements containing Q21 due to their specific surface areas. On the lower right side of Figure 4, the particular plasticizing effect of PCE22 is shown. The strong sorption resulted in the strong plasticizing effect. The saturation dosage was approximately 0.12 mass % of CEM I as well as of the cements with Q11. With the increasing proportions of Q21 in the cement, the saturation dosage decreased slightly. Figure 5 shows the shear resistance of fresh cement paste in dependence of the sorbed quantity of active substance (left: PCE11, right: PCE22). The shear resistance is normalized to that of the respective paste without superplasticizer. On the left side of Figure 5 by the steeper slopes, it is clearly visible that the plasticizing effect of PCE11 increased with the proportion of calcined clay in the cement as well as with the specific surface area of the substitute. For PCE22, this was not such pronounced because of its stronger sorption even in combination with CEM I. Sorption and duration of plastification The sorption of the polycarboxylates and their particular duration of plastification in dependence of the cement type are shown in Figure 6. The dosage was always 90% of the particular saturation dosages (cf. Figure 4). The quantity of PCE11 sorbed in combination with CEM I was approximately 22 mass % to 30 mass % of the added quantity during the first 30 min after water addition (Figure 6, upper left side). After 120 min the quantity sorbed was approximately 45 mass %, only. Due to the depot effect, PCE11 remained in the pore solution to a large extent caused mainly by the high concentration of SO42- in the pore solution of the paste with CEM I (cf. Figure 2, left side) and its negative zeta potential (cf. Figure 3, left side) as well as benefiting from the low effective anionic charge of PCE11 (cf. Table 4). In combination with Z35Q21, the time-dependent sorption of PCE11 was comparable. In combination with Z55Q21, the quantity sorbed was approximately 40 mass % to 45 mass % during the first 30 min and approximately 55 mass % after 120 min due to the lowered concentration of SO42- and the increase in Ca2+ as well as the more positively charged surface (cf. left sides of Figure 2 and Figure 3). The substitution of the portland cement component with Q11 with larger specific surface area led to a further increase in the time-dependent sorption of PCE11. On the lower left side of Figure 6, the particular

Interactions between Cements with Calcined Clay and Superplasticizers  309

Figure 6—Sorption and duration of the plasticizing effect of PCE11 (left) and PCE22 (right) in dependence of the type of cement. duration of plastification of PCE11 is shown. In combination with all tested cements, the duration of the plasticizing effect was given within the test period of 120 min after water addition. On the right side of Figure 6, the results for PCE22 are shown. The TOC contents of the pore solutions increased in course of the testing period (not shown), indicating desorption of PCE22 (upper right side of Figure 6). Either PCE22 desorbed14 or more probably, as the SO42- concentration remained almost unchanged, PCE22 lost the grafted side chains by hydrolysis,32 both leading to an increased TOC content in the pore solution. On the lower right side of Figure 6, the particular duration of plastification of PCE22 is shown. As a result of the desorption process or rather the loss of side chains, the effect of the steric hindrance decreased considerably and the fresh cement pastes exhibited significant stiffening. Performance of fresh concrete The results obtained by the investigations on fresh cement pastes were verified by concrete tests. Therefore, CEM I (circle) and Z55Q21 (triangle) were used in combination with PCE11 (filled symbols) and PCE22 (open symbols). The results are presented in Figure 7. The results of concrete tests confirmed the results of the cement paste trials. Without the addition of superplasticizer, the concrete with Z55Q21 exhibited a softer consistency than the concrete with CEM I (Figure 7, left side, grey shaded symbols). This was largely due to the improved particle size distribution (cf. slope n, Table 2) and thus the improved packing density of the cement with calcined clay as well as the reduced binding of mixing water into initial hydration products. Hence, more “free water” was available to fluidize the fresh cement paste and the fresh concrete, respectively. The slightly larger

310 SP-302-23

Figure 7—Concrete’s consistency increase and consistency retention in dependence of the cement-PCE-combination used. cement paste volume caused by the slightly lower specific density of the cement with calcined clay (cf. Tables 1 and 3) also accounted for this. The expected performance of both polycarboxylates observed in combination with CEM I was scarcely present in combination Z55Q21. The fresh concrete with Z55Q21 could be plasticized with lower dosages of superplasticizer to greater extent (Figure 7, left side). Both polycarboxylates lost their characteristic performance and their plasticizing effect was almost identical. This was particularly the case for PCE11 (ready-mixed concrete). The reasons were explained previously. The consistency class F4 with a flow table spread up to 550 mm (≈ 21 in.) was already reached by approximately 0.09 mass % of PCE11 and PCE22, respectively, for the concrete with Z55Q21 compared to approximately 0.15 mass % of PCE22 or approximately 0.20 mass % of PCE11 needed for the fresh concrete with CEM I. On the right side of Figure 7, the particular consistency retention is shown. In combination with CEM I, both polycarboxylates exhibited their distinct performance. PCE11 retained concrete’s consistency and PCE22 showed a significant initial plasticizing effect combined with pronounced stiffening. In combination with Z55Q21, the specific performance of both polycarboxylates was diminished (cf. arrows on the right side of Figure 7). SUMMARY AND CONCLUSIONS Using pore solution analyses, zeta potential experiments, as well as rheological and consistency measurements on fresh cement paste and fresh concrete influences of the type and proportion of calcined clay in the cement on the sorption behavior and plasticizing effect of polycarboxylate-based superplasticizers were determined. Based on the results presented herein, the following conclusions can be drawn: 1. With an increasing substitution of the portland cement component (readily soluble alkalis and sulfates) with calcined clay, the fresh cement paste’s pore solution contained less dissolved alkali, hydroxide, and sulfate ions as well as more calcium ions. The ionic strength decreased. The zeta potential of the particular fresh cement paste was less negative. Changes in the ionic composition and ionic strength of the pore solution and the zeta potential of the fresh cement paste can influence the constitution, the sorption behavior and the performance potential of polycarboxylates depending on their particular polymer structure.

Interactions between Cements with Calcined Clay and Superplasticizers 311 2. Depending on the mineralogical composition of the clay and the calcining conditions, the specific surface area of the calcined clay and thus the quantity of superplasticizer necessary to reach the maximum flowability of fresh cement paste varied. 3. Decreasing concentrations of SO42- in the pore solution and less negative zeta potentials of fresh cement paste as well as an increased specific surface area depending on the type and proportion of calcined clay in the cement led to a stronger sorption and consequently to a stronger plasticizing effect of the less anionic superplasticizer recommended for the use in ready-mixed concrete (PCE11) and hence to a reduced consistency retention of fresh cement paste and fresh concrete, respectively. 4. The more anionic superplasticizer for the use in precast concrete (PCE22) sorbed to a greater extent even at high SO42- concentrations inducing always a strong initial plastification in combination with significant stiffening of fresh cement paste and fresh concrete, respectively. 5. The workability of fresh concrete containing cement with a clinker substitution of up to 55 mass % by calcined clay could be improved compared to that of the reference concrete with portland cement due to the higher volume of fresh cement paste as well as the rheological benefit of more “free water” caused by an increased cement’s packing density and less hydration products formed initially. Less superplasticizer was necessary for plastification and the specific performance of the polycarboxylates diminished increasing proportion of calcined clay in the cement. 6. For an ideal plastification of concrete, the type and dosage of the superplasticizer always have to be adjusted to concrete technological parameters such as the cement type and content, the w/c ratio, the quantity of fresh cement paste, the mixing process, the temperature, and the time until the concrete is placed. AUTHOR BIOS Civil engineer Jens Herrmann was awarded a scholarship from the Science Foundation of the German Cement Industry, Gerd Wischers Foundation, at VDZ’s Concrete Technology Department. Since 2012 he has been working as a Scientific Assistant at VDZ’s Cement Chemistry Department. He is a Ph.D student in material sciences at Clausthal University of Technology, Germany. His research interests include the workability of mortar and concrete, the performance of blended cements and their interactions with fluidizing and water-retaining admixtures. Jörg Rickert is a civil engineer and has been working at VDZ since 1996. He started his career working in the Concrete Technology Department and obtained a Ph.D in materials science in 2003. His topics of research have been cements and admixtures and their interactions in concretes. Jörg Rickert has been Head of the Cement Chemistry Department since 2009. He is a member of both national and international standardization committees. ACKNOWLEDGMENTS The IGF project 16726 N of VDZ gGmbH was sponsored by the Federal Ministry of Economics and Energy through the Federation of Industrial Cooperative Research Associations as part of the program to promote the Joint Industrial Research (IGF) based on a decision of the German Bundestag.

312 SP-302-23

REFERENCES 1. Schröter, N. and Fischer, P., “Entwicklungen und Trends bei Betonzusatzmitteln”, beton, V. 60, No. 6, 2010, pp. 226-231. 2. CEMBUREAU, ed., “CEMBUREAU statistics”, 2012. 3. Scrivener, K. L., “Options for the future of cement,” Indian Concrete Journal, V. 88, No. 7, 2014, pp. 11-21. 4. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., “Superplasticizers: How they work and why they occasionally don’t,” Concrete International, V. 16, No. 5, May 1994, pp. 45-52. 5. Andersen, P. J.; Roy, D. M.; and Gaidis, J. M.W.R. Grace & Co.. , “The effects of adsorption of superplasticizers on the surface of cement,” Cement and Concrete Research, V. 17, No. 5, 1987, pp. 805-813. doi: 10.1016/0008-8846(87)90043-3 6. Ferrari, L.; Kaufmann, J.; Winnefeld, F.; and Plank, J., “Multi-method approach to study influence of superplasticizers on cement suspensions,” Cement and Concrete Research, V. 41, No. 10, 2011, pp. 1058-1066. doi: 10.1016/j.cemconres.2011.06.010 7. Flatt, R. J., “Interparticle Forces and Superplasticizers in Cement Suspensions”, Ph.D theses, École Polytechnique Fédérale de Lausanne, EPFL, Départment de Matériaux, Lausanne, 1999, 301 pp. 8. Hirsch, C. M., “Untersuchungen zur Wechselwirkung zwischen polymeren Fließmitteln und Zementen bzw. Mineralphasen der frühen Zementhydratation”, Ph.D theses, TU München, München, 2005, pp. 261. 9. Jolicoeur, C., and Simard, M.-A., “Chemical admixture-cement interactions: Phenomenology and physio-chemical concepts,” Cement and Concrete Composites, V. 20, No. 2+3, 1998, pp. 87-101. doi: 10.1016/S0958-9465(97)00062-0 10. Regnaud, L.; Rossino, C.; Alfani, R.; and Vichot, A., “Effect of comb type superplasticizers on hydration kinetics of industrial Portland cements”, in: Palomo, A., Zaragoza, A., and López, A. (eds.), 13th International Congress on the Chemistry of Cement (Madrid, 04.-08.07.2011), Abstracts and Proceedings, 7 pp. 11. Spanka, G., Grube, H., and Thielen, G., “Wirkungsmechanismen verflüssigender Betonzusatzmittel”, beton, V. 45, No. 11/12, Nov./Dec. 1995, pp. 802-808/876-881. 12. Uchikawa, H.; Hanehara, S.; Shirasaka, T.; and Sawaki, D., “Effect of admixture on hydration of cement, adsorptive behaviour of admixture and fluidity and setting of fresh cement paste,” Cement and Concrete Research, V. 22, No. 6, 1992, pp. 1115-1129. doi: 10.1016/0008-8846(92)90041-S 13. Uchikawa, H.; Hanehara, S.; and Sawaki, D., “Effect of electrostatic and steric repulsive forces of organic admixtures on the dispersion of cement particles in fresh cement paste,” Cement and Concrete Research, V. 27, No. 1, 1997, pp. 37-50. doi: 10.1016/ S0008-8846(96)00207-4 14. Yamada, K., and Hanehara, S., “Interaction mechanism of cement and superplasticizers – The roles of polymer adsorption and ionic conditions of aqueous phase,” Concrete Science and Engineering, V. 3, No. 11, Sep. 2001, pp. 135-145. 15. Yamada, K., and Hanehara, S., “Working mechanism of polycarboxylate superplasticizer considering the chemical structure and cement characteristics”, in: Grieve, G. and Owens, G. (eds.), 11h International Congress on the Chemistry of Cement (Durban, 11.-16.05.2003), Durban: Halfway House, 2003.

Interactions between Cements with Calcined Clay and Superplasticizers 313 16. Flatt, R. J., and Schober, I., “Superplasticizers and the rheology of concrete”, in: Roussel, N. (ed.), Understanding the rheology of concrete, Woodhead publishing, 2012, pp. 144-208. 17. Sakai, et al., “Influence of molecular structure of comb-type and inorganic electrolytes on the dispersion mechanisms of limestone power”, in: Malhotra, V.M. (ed.), 7th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Berlin, 20.-23.10.2003), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2003, Special Publication SP 217, pp. 381-392. 18. Palacios, M. et al., “Compatibility of PC superplasticizers with slag-blended cements”, in: Holland, T.C., Gupta, P., and Malhotra, V.M. (eds.), 9th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Seville, 12.-15.10.2009), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2009, ACI Special Publication SP 262, pp. 97-112. 19. Regnaud, L. et al., “Interactions between comb-type superplasticizers and slag cement pastes”, in: Holland, T.C., Gupta, P., and Malhotra, V.M. (eds.), 9th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Seville, 12.-15.10.2009), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2009, Special Publication SP 262, pp. 139-151. 20. Herrmann, J., and Rickert, J., “Influences of Slag or Limestone on the Performance of Superplasticizers”, in: Malhotra, V.M. (ed.), 10th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Prague, 29.-31.10.2012), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2012, ACI Publication SP-288, pp. 317-328. 21. Herrmann, J., and Rickert, J., “Influences of slag or fly ash on cement-superplasticizer-interactions”, in: Gesellschaft Deutscher Chemiker, GDCh, (ed.), 1st International Conference on the Chemistry of Construction Materials (Berlin, 07.-09.10.2013), Frankfurt / M.: GDCh, GDCh-Monographie Bd. 46, pp. 457-460. 22. Ng, S., and Plank, J., “Interaction mechanisms between Na-montmorillonite clay and MPEG-based polycarboxylate superplasticizers,” Cement and Concrete Research, V. 42, No. 6, 2012, pp. 847-854. doi: 10.1016/j.cemconres.2012.03.005 23. Lei, L., and Plank, J., “Synthesis, properties and evaluation of a more clay tolerant polycarboxylate possessing hydroxy alkyl graft chains”, in: Malhotra, V.M. (ed.), 10th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete (Prague, 29.-31.10.2012), Farmington Hills, MI, USA: American Concrete Institute, ACI, 2012, Supplementary Papers, pp. 1-20. 24. Lei, L., and Plank, J., “A study on the impact of different clay minerals on the dispersing force of conventional and modified vinyl ether based polycarboxylate superplasticizers,” Cement and Concrete Research, V. 60, No. 6, 2014, pp. 1-10. doi: 10.1016/j. cemconres.2014.02.009 25. Nehdi, M. L., “Clay in cement-based materials: Critical overview of state-ofthe-art,” Construction & Building Materials, V. 51, 2014, pp. 372-382. doi: 10.1016/j. conbuildmat.2013.10.059 26. Schulze, S. E., and Rickert, J., “Pozzolanic Activity of calcined clays”, in: Malhotra, V.M. (ed.), 12th International Conference on Recent Advances in Concrete Technology and

314 SP-302-23

Sustainability Issues (Prague, 30.10.-02.11.2012). Farmington Hills, MI, USA: American Concrete Institute, ACI, 2012 (ACI Publication SP-289), pp. 277-287. 27. Dukhin, A. S., and Goetz, P. J., “Ultrasound for characterizing colloids - Particle sizing, zeta potential, rheology”, in: Möbius, D. and Miller, R. (eds.), Studies in Interface Sciences Vol. 15. Elsevier, Amsterdam, 2002, 372 pp. 28. Flatt, R. J., and Houst, Y. F., “A simplified view on chemical effects perturbing the action of superplasticizers,” Cement and Concrete Research, V. 31, No. 8, 2001, pp. 11691176. doi: 10.1016/S0008-8846(01)00534-8 29. Aitcin, P.-C., “Superplasticizers”, in: HBRC (ed.) 1st International Conference on New Cements and their Effects on Concrete Performance (Cairo, 16-18.12.2008), Cairo: Helwan University, 2008, 14 pp. 30. Diamond, S., “The Status of Calcium in Pore Solutions of Mature Hardened Portland Cement Paste”, Il cemento, V. 74, No. 4, 1977, pp. 149-156. 31. Sachsenhauser, B. K., “Kolloidchemische und thermodynamische Untersuchungen zur Wechselwirkung von Alpha-Allyl-Omega-methoxypolyethylenglykol-Maleinsäureanhydrid-Copolymeren mit CaCO3 und Portlandzement”, Dissertation, Technische Universität München, München, 2009, 272 pp. 32. Flatt, R. J., and Schober, I., „Superplasticizers and the rheology of concrete“, in: Roussel, N. (ed.), Understanding the rheology of concrete, Woodhead publishing, 2011, pp. 144-208.

SP-302-24

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives by Alexandre Govin, Marie-Claude Bartholin, and Philippe Grosseau Viscosity-modifying admixtures (VMA) are often introduced in the formulation of modern factory-made mortars in order to prevent segregation and to improve the homogeneity and workability of cement-based system. Among VMAs, organic admixtures, and more especially polysaccharides such as cellulose ethers (CE), are widely used, since they improve both rheological property and water retention capacity of the mortars. The present study examines the influence of chemical composition and structure of guar gum derivatives on water retention capacity (WR) and rheological behavior of fresh state Portland-based mortars. The investigation was also completed by adsorption isotherms. For this, original guar gum, HydroxyProplyl Guars (HPG) and hydrophobically modified HPGs were selected. The effect of the molar substitution (MSHP) and the degree of substitution (DSAC) was investigated. The results highlight that chemical composition of HPGs has a remarkable effect on fresh state properties of mortars. The original guar gum does not impact both WR and rheological behavior. Increasing MSHP leads to an improvement of the WR and the stability of mortars while the hydrophobic units further enhance WR and lead to a decrease in the yield stress and an increase in the resistance to the flow of admixed mortars. Keywords: cement; HydroxyPropyl Guar; mortar; rheology; water retention. INTRODUCTION Modern factory-made mortars are complex materials, in which several kinds of admixtures are added in order to obtain specific properties, from the fresh state to the hardened material. Indeed, since many years, concretes, mortars or cement grouts with high fluidity have been developed, since their use implies many economical and technical advantages. However, the use of highly flowable mixtures may lead to segregation or excessive bleeding and subsequently, durability issues. In order to overcome this problem by enhancing the 315

316 SP-302-24

sedimentation resistance while maintaining high fluidity, viscosity-enhancing admixtures (VEA) are frequently introduced within the formulation.1-4 Among these admixtures, natural polysaccharides or their derivatives (such as welan gum, starch derivatives or cellulose ethers) are the most widely used. Moreover, the incorporation of these VEAs in shotcrete or render mortar is useful to ensure sagging resistance for thick application on vertical support, and to allow sufficient fluidity for normal pumpability by supplying shear thinning rheological behavior.5 Indeed, these admixtures provide, generally, high yield stress and apparent viscosity at low shear rate but low resistance to flow at high shear rate.6 However, their mode of action is not fully understood, since results are sometimes contradictory. Water retention (WR) is another essential property of monolayer render at fresh state. Indeed, high water retention improves the cement hydration and limits the absorption of the mixing water by a substrate and thus provides good mechanical and adhesive properties to the mortar.7,8 Among admixtures enhancing water retention capacity of the freshly-mixed mortars, cellulose ethers (CE) are the most widely used. Nevertheless, hydroxypropyl guar (HPG) are now also well-established in the construction industry as water retention agent for mortars.9-12 Moreover, HPGs are already widely used in various industrial fields, such as textile printing, hydraulic fracturing process, oil production or paper manufacturing, due to their thickening effect.13,14 Consequently, since HPGs improve the two main properties of mortar, they appear as suitable admixtures to be used in render formulation. The aim of this study is to provide an understanding of the effect of chemical composition and structure of HPGs and its dosage on macroscopic properties of mortars. For this purpose, an original guar gum and five HPGs with specific chemical modifications, such as increase in MSHP or substitutions by hydrophobic units, were selected. The impact of admixtures on the water retention capacity and on the rheological behavior of mortars was investigated. RESEARCH SIGNIFICANCE Polysaccharides are commonly used in cement-based materials, but most studies focus on cellulose ethers or welan gum. Studies about hydroxypropyl guars are still scarce in the technical literature despite the fact that these molecules provide interesting properties comparable to those obtained with cellulose ethers. The aim of this study is to highlight the role of the chemical structure of hydroxypropyl guars on water retention properties and rheological properties. MATERIALS AND METHODS Mineral products Mineral products used in this study consist in blend Portland cement (Holcim), lime (Holcim), calcium carbonate (Calcitec V60, Mineraria Sacilese S.p.A.) and dolomite (Bombardieri and Leidi 0.1-0.4mm). The mineral compositions of the commercial Portland cement, CEM II/B-LL 32.5 R according to the European standard EN 197-115, used are given in Table 1. The phase composition was determined by Rietveld refinement method (Siroquant V2.5 software) after XRD analysis (D5000, Siemens) and the oxide composition was quantified by means of X-ray fluorescence spectroscopy. The median particle diameters by volume (d50%), determined by means of laser diffractometry with dry powder disperser, (Master-

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  317 Table 1 –Mineral composition (%, weight) of the investigated cement determined by XRF and XRD-Rietveld refinement Chemical composition (% wt)

Phase composition (% wt)

Oxides CaO

XRF 57.87

Oxides SO3

XRF 3.95

Phases C3S

XRD (Rietveld) 54.3

Phases Calcite

XRD (Rietveld) 28.9

SiO2

12.31

Na2O

0.99

C2S

3.5

Gypsum

3.0

Al2O3

5.25

K2O

1.66

C3A

4.7

Quartz

0.9

MgO

1.19

TiO2

0.16

C4AF

4.6

Free CaO

0.8

Fe2O3

4.05

LOI

13.7

Fig. 1 –Particle size distribution of raw materials constituting the mortar Table 2 –Median particle diameters by volume (d50) and specific surface area of the mineral phases d50 (µm) BET specific surface area (m2/g)

CEM II/B-LL 15

Lime 5

Dolomite Leidi 300

Dolomite Bombardieri 630

Calcite 20

2.40

5,67

0.43

0.34

1.44

(Note: 1 µm = 0.0000394 in, 1 m2/g = 0.00488 ft2/lbm)

sizer 2000 and Scirocco dispersing unit, Malvern), are 630 µm (248 10-4 in), 300 µm (118 10-4 in), 20 µm (7.9 10-4 in), 15 µm (5.9 10-4 in) and 5 µm (1.9 10-4 in) for the dolomite from Bombardieri, dolomite from Leidi, calcium carbonate, cement and lime, respectively. The particle size distribution and the specific surface area (determined by BET) are given in Fig. 1 and Table 2.

318 SP-302-24

Fig. 2 –Molecular structure of original guar gum (a) and HydroxyPropyl Guar (b) Table 3 –Qualitative description of the HPG used MS

DS

Additional substitution

HPG 1

Low

-

HPG 2

Medium

-

HPG 3

High

-

HPG 4

High

Short alkyl chain

HPG 5 HPG 6

High -

Higher DS than HPG 4 -

Short alkyl chain -

Organic admixtures Guar gum is a natural polysaccharide extracted from the seed endosperm of Cyamopsis tetragonolobus. This polymer consists in a β(1-4)-linked D-mannopyranose backbone with random branchpoints of galactose via an α(1–6) linkage (Fig.2(a)). Hydroxypropyl guars (HPGs) are obtained from the original guar gum via an irreversible nucleophilic substitution, using propylene oxide in the presence of an alkaline catalyst (Fig.2(b)). The manufacture of HPGs has the advantage of having a more reduced impact on the environment than cellulose derivatives. Indeed, guar gum is extracted by simple thermo-mechanical process, exhibits a higher chemical reactivity and is soluble in cold water thanks to its branched-chain structure with a lot of hydroxyl groups. Thus, the chemical modification of the original guar gum requires normal reaction conditions of temperature and pressure, does not generate large quantity of by-products, and requires minimal purification procedure.9 In this paper, five HPGs and an original guar gum provided by Lamberti S.p.A were studied. They exhibit roughly the same molecular weight, around 2.106g.mol-1 since they are all from the same original guar gum (noted HPG 6 in the paper).16 Table 3 provides a qualitative description of the polymers used. The qualitative substitution degrees are provided by the manufacturers. The molar substitution ratio (MSHP) represents the number of hydroxypropyl units per anhydroglucose unit and is less than 3 for the investigated HPGs. The degree of substitution (DSAC) represents the amount of alkyl chain per anhydroglucose unit. The only difference between HPGs 1, 2 and 3 is the molar substitution ratio, which increases, while HPGs 4 and 5 exhibit an additional substitution (short alkyl chains). The DSAC of HPG 5 is higher than that of HPG 4.

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  319 Methodology Mortars were prepared according to the following mixture proportions: 12% of cement, 3% of lime, 18% of calcium carbonate, 43% of dolomite Bombardieri and 24% of dolomite Leidi (by weight). The admixtures (0.05, 0.075, 0.1, 0.125 and 0.15%) were in addition to the total dry mixture (i.e. cement, lime, calcium carbonate and dolomite) and are expressed in weight percent by weight of binder (% bwob). Dry mixture was blended in a shaker (Wab, Turbula, Germany) for 10 min. Deionised water was added in order to obtained a liquid-to-solid ratio L/S = 0.22. The mixing procedure was in accordance with EN 196-1.17 The experimental methodology consisted in dividing the freshly mixed mortar into three parts in order to characterize several properties from the same mixing. A first part was used to characterize the rheological behavior of the mortar, the water retention study was performed on the second part and the third part of the freshly mixed mortar was centrifuged in order to determine the adsorption isotherms and the polymer concentration within the pore solution following a procedure described later. All tests were carried out, at least, in triplicate and at a controlled temperature because water retention, rheological behavior of the mortar and adsorption isotherm are temperature-dependent. A control test was also performed with a mortar without admixture. Water retention measurements The water retention capacity of freshly-mixed mortar can be assessed using different tests where the removed water after suction or depression is measured.18 In this study, the standard method used to estimate the water retention capacity of a mortar, was the test described in ASTM C1506-09.19 It had to be performed 15 min after mixing to measure the water loss of a mortar under depression. The standardized apparatus was submitted to a vacuum of 50 mm of mercury (6.6 103 Pa) for 15 min. Then, the water retention capacity, WR, was calculated using the following equation:

WR(%) =

W0 − W1 × 100 (1) W0

where W0 represents the initial mass of mixing water; W1 is the loss of water mass after aspiration. All the experiments were carried out at 23 °C (73.4 °F). Three classes of water retention (measured by ASTM method) of a fresh mortar can be specified according to the DTU 26.1.20 The first class (low water retention category) contains mortars that exhibit a water retention lower than 86%. The second class (intermediate) corresponds to values ranging from 86% to 94%. The last one (strong) is defined by water retention higher than 94%, corresponding to the required values in the field of rendering application. Rheological behavior The rheological measurements were performed with Rheometer MCR 302 (Anton-Paar), thermostated at 20 °C (68 °F). The rheological properties of fresh mortars were investigated with vane-cylinder geometry since this system is suitable for granular pastes like mortars.21,22 The gap thickness, distance between the periphery of the vane tool and the outer cylinder, was set at 8.5 mm (0.33 in), in order to be less sensitive to the heterogeneity

320 SP-302-24

of the mortar. Using a Couette analogy, the shear stress and shear rate were calculated from the torque and the applied rotational velocity respectively, after calibration with glycerol.23 The mortar was introduced into the measurement system at the end of the mixing cycle. At 10 min, the mortar was pre-sheared for 30 s at 100 s-1 in order to re-homogenize the sample and to eliminate its shear history because of thixotropic character of cementitious materials.24,25 After a period of rest of 5 min, the rheological measurements were started (total time = 15 min). At this time, the hydration rate is low enough which allows overcoming the irreversible effect of cement hydration on rheological behavior, especially at low shear rate.24 The imposed shear rate was decreased by step from 300 to 0.06 s-1 (16 steps). At each shear rate, the measuring time was adjusted in order to obtain a steady state whatever the formulation. The samples were systematically submitted to high shear rate (100 s-1) for 30 s before each imposed shear rate in order to resuspend particles of mortar within the mortar mixtures. The results were expressed as shear stress τ according to shear rate γ and the Herschel-Bulkley (HB) model was applied to fit the experimental data and used to describe mortars rheological behavior26:

τ = τ 0 + K γ n (2)

where τ0 correspond to the yield stress, K the consistency coefficient and n the fluidity index which characterizes shear-thinning behavior of mortar. Adsorption curves of HPGs on binder The adsorption isotherms were determined using the depletion method. The nonadsorbed polymer remaining within the pore solution was quantified by means of Total Organic Carbon (TOC) measurements. Prior to analysis, the pore solution was extracted from admixed or non-admixed mortar. The extraction was performed by means of two centrifugation steps. The first step consisted in the centrifugation of around 150 g (0.30 lbm) of mortar at 5000 rpm for 5 min. The supernatant was, afterward, centrifuged again at 14500 rpm for 10 min in order to avoid the presence of mineral particles within the solution. The supernatant was diluted with hydrochloric acid solution at 0.1 mol.L-1 (0.378 mol. gal-1). The total organic carbon was determined by combustion at 850 °C (1562 °F) with a Vario-TOC Cube (Elementar). The adsorbed amount of polysaccharides was calculated from the difference of TOC content of the HPG reference solution and the TOC content of the supernatant. EXPERIMENTAL RESULTS Impact of HPGs on the water retention property of fresh mortars Fig. 3 represents the evolution of the water retention capacity of fresh admixed mortars, according to the polymer dosage. The non-admixed mortar exhibits a low water retention capacity of about 72% ± 0.3%. Then, as expected, the water retention increases with the use of HPGs and with increasing polymer dosage, until reaching a plateau with very high WR values (>97%). In the range of polymer dosage studied, the WR values reached for HPGs 2, 3, 4 and 5, are greater than 94% and therefore belong to the strong WR class. One

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  321

Fig. 3 –Impact of polymer dosage on water retention capacity of fresh admixed mortars can also clearly notice the very limited impact of the original guar gum and, to a lesser extent, the one of HPG 1, regardless of dosage. These results suggest that the substitution of hydroxyl units from original guar gum by hydroxypropyl units increases the WR of mortars. Furthermore, the increase in the MSHP (from HPG 1 to 3) improves the WR capacity of mortar, since HPG 3 provides the higher WR despite lower dosage, followed by HPG 2 and then by HPG 1. The results highlight moreover the positive impact of additional alkyl chain on WR. Indeed, the highest WR are obtained with HPGs 4 and 5 for the lowest polymer dosages. Moreover, concerning the shape of WR curves, an abrupt change in slope can be noticed for mortars admixed with HPGs 2 to 5. This occurs for a decreasing polymer dosage from HPG 2 to HPG 3 and from HPG 3 to HPG 5. Adsorption curves of HPGs on binder Fig. 4 shows the adsorption isotherms of the hydroxypropyl guar and the original guar gum on Portland based-mortars. The results confirm the adsorption of original guar gum and HPGs on cementitious materials. It has been shown that the adsorption mechanism of galactomannose polysaccharides at solid–liquid interfaces involves strong hydrogen bonding.27 In the range of polymer dosage used in the present study, no plateau is reached, and this, whatever the admixture. For the original guar gum (HPG 6), the adsorption is totally linear (Fig. 4(a)). Moreover, its adsorption is the highest of all the tested polymers and corresponds to a total adsorption higher than 98.5% of the introduced polymer. The presence of hydroxypropyl substitutions on the guar leads to a decrease in the affinity of the polymer with the binder since the amount of HPG 1 adsorbed is lower by 35% than the original guar gum. Moreover, the adsorption is further reduced by the increasing values of MSHP by 46% and 64% for HPG 2 and HPG 3, respectively, with respect to HPG 6 (Fig. 4(a)). This tendency is consistent with previous studies on HPGs and cellulose ethers (CE).10,28 Fig. 4(b) highlights the effect of the additional alkyl chain on the adsorption. It appears that the hydrophobic side chains slightly intensify the adsorption of the hydrophobically

322 SP-302-24

Fig. 4 –Adsorption isotherms of HPGs 1 to 3 and original guar gum (HPG 6) (a) and HPGs 3 to 5 (b) on binder modified HPGs on surface of grains with respect to HPG 3. However, the adsorption of HPGs 4 and 5 is lower than that of HPG 2. For polymer dosages up to 0.1% bwob, the effect of the DSAC is negligible since the experimental data superimpose. Nevertheless, when HPG dosage is higher than 0.1% bwob, the affinity of HPG 5 with the binder becomes higher than that of HPG 4. From the TOC measurements, the real polymer concentration within the extracted pore solution was determined. Fig. 5 shows the evolution of this concentration versus the introduced polymer dosage. Excepted HPG 6, the amount of non-adsorbed polymer increases with increasing polymer dosage. According to the HPG, the concentration rises following this order: HPG 6 < HPG 1 < HPG 2 < HPG 5 < HPG 4 < HPG 3. Impact of HPGs on the rheological properties of fresh mortars The rheological results (not shown here) suggest that the thixotropy of the mortars does not affect the rheological measurements. Indeed, in the range of tested shear rates and thanks to the experimental procedure, the shear stress obtained by the increasing or decreasing shear rate ramps are superimposed which justifies the choice to consider only the decreasing ramps for all the rheological study.

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  323

Fig. 5 –Concentration of HPGs in extracted mortar pore solution Fig. 6 shows the evolution of the yield stress, extracted from Herschel-Bulkley model, for all the studied mortars with and without admixture. The mortar without admixture exhibits a yield stress value of around 45 Pa. From the presented results, three different classes of HPG, inducing different evolution of the yield stress with the polymer dosage, can be highlighted for admixed mortars. The first category is only composed of the original guar gum (HPG 6), which induces a quasi linear decrease in the yield stress of mortar when HPG dosage increases. On the contrary, HPGs 1, 2 and 3 lead to a continuous rise of the yield stress of mortars from 50-60 Pa to around 120 Pa with the increase in the HPG dosage from 0.05% to 0.15%. Finally, HPG 4 and 5 constitute the third class of admixture. The use of these admixtures leads an improvement of the yield stress compared to the nonadmixed mortar, whatever the dosages tested in the study. However, the improvement is not proportional to the admixture dosage. Indeed, our first dosage (0.05% bwob) leads to an increase in the yield stress. Beyond this dosage, increasing the dosage provides a slow and low decrease, before reaching a plateau. The value of the yield stress reached on the plateau is still higher than that of the mortar without admixture. The evolution of the consistency coefficient (K from Herschel-Bulkley equation) during the increase of polymer dosage is presented in Fig. 7. As in the case of the yield stress, the results can be divided into three classes of polymer. The first class is only composed of the original guar gum which provides a very low or negligible modification of the consistency coefficient with increasing polymer dosage compared to non-admixed mortar. HPGs 1 to 3, constituting the second group, induce first an increase followed by a plateau in the consistency coefficient. Finally, HPGs 4 and 5 lead to a continuous increase in the consistency coefficient of admixed mortars. Fig. 8 shows the evolution of the fluidity index (n) versus the polymer dosage for all the studied mortars. It is worth to note that whatever the mortars (non-admixed and admixed), the values of the fluidity index are lower than 1, meaning that they are all shear thinning. Due to the high standard deviation, the value of the fluidity index of mortars admixed with HPGs 1 to 3 and HPG 6 seem to be unchanged as the dosage of HPGs increase. However, HPGs 4 and 5 leads to a low increase followed by a continuous decrease in the fluidity

324 SP-302-24

Fig. 6 –Impact of polymer dosage on yield stress of fresh admixed mortars (HPGs 1 to 3 and original guar gum (HPG 6) (a) and HPGs 3 to 5 (b)) index until reaching values around 0.5. It means that the shear thinning behavior of mortars becomes more and more pronounced. DISCUSSION The effect of the original guar gum (HPG 6) on the studied macroscopic properties is negligible. This result is coherent with adsorption of the polymer from the Water Retention point of view. Indeed, since the adsorption is higher than 98.5% of the initial amount of polymer, very few molecules are still in the pore solution. The composition of pore solution is thus very close to that of the non-admixed mortar, leading to similar WR. Concerning the hydroxypropyl guar, the results from WR experiments are consistent with those of previous studies performed with HPGs or CEs and with the proposed mechanism.11,29,30 Indeed, the WR of admixed mortars is mainly governed by the ability of polysaccharidic admixtures to form a hydrocolloidal associated polymer molecules network and to induce overlapping of polymer coils within the pore solution.11,29,30 When the concentration of polymer increases in solution, the isolated polymer coils, existing at low polymer concentration, begin to come into contact with one another. This concentra-

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  325

Fig. 7 –Impact of polymer dosage on consistency coefficient of fresh admixed mortars (HPGs 1 to 3 and original guar gum (HPG 6) (a) and HPGs 3 to 5 (b)) tion is defined as the coil-overlap concentration (noted C*). Above this critical concentration, the polysaccharide aggregates stop the water flow by plugging the porous network of a thin polysaccharide-enriched filter cake at the interface mortar-substrate resulting in a sudden and sharp rise in WR curves.11 As previously mentioned, the abrupt change in slope is reached for a decreasing polymer dosage from HPG 1 to HPG 3. The only difference between these HPGs is the increasing substitution degree. According to literature, the increase in MSHP does not lead to a change in the C*.31 However, the increasing substitution degree leads to a decrease in polymer adsorption on mortar components (Fig. 4) and hence an increase in polymer amount in pore solution (Fig. 5). Consequently, the coil overlapping occurs at lower dosage. The results highlight furthermore the positive impact of additional alkyl chain on WR. The presence of additional alkyl chains (HPG 4 and 5), despite slightly higher adsorption than HPG 3, leads to the formation of polymer associates at lower polymer dosage. Indeed, the interconnection between alkyl chains creates intramolecular and intermolecular interactions through specific hydrophobic interactions which cause a decrease in the coil-overlapping concentration.32-34 Consequently, the abrupt change in slope is reached for a lower polymer dosage of HPG 4 and 5 than HPG 3.

326 SP-302-24

Fig. 8 –Impact of polymer dosage on fluidity index of fresh admixed mortars (HPGs 1 to 3 and original guar gum (HPG 6) (a) and HPGs 3 to 5 (b)) However, an increase in the DSAC (from HPG 4 to 5) can lead to a conversion of some intermolecular associations to intramolecular associations and hence an increase in the polymer dosage necessary to reach coil overlap.31 The rheological results (Fig. 6, Fig. 7 and Fig. 8) highlight that HPGs 1 to 3, HPGs 4 to 5 and HPG 6 behave quite differently. Indeed, HPGs 1-2-3 lead to a continuous increase in the yield stress, while HPGs 4-5 modify mainly the consistency coefficient and the fluidity index. This means that HPGs 1-3 increase the stability of mortars while HPGs 4-5 increase the resistance to the flow of admixed mortars. HPGs 1 to 3 affect the rheological behavior of the admixed mortars in the same way, i.e. an increase in the yield stress, a low increase followed by a plateau in the consistency coefficient and a negligible modification of the fluidity index when the polymer dosage rises. Fig 4 shows that HPGs adsorb onto particles constituting the mortar. Prima facie, this adsorption could be responsible for the increase in the yield stress because of bridging flocculation.28 However, despite a strong drop of the adsorption (50%) with the increase in the MSHP (from HPG 1 to 3) the yield stress also increases. This suggests that the non-adsorbed polymer may be responsible for the yield stress increase. The potential loss of bridging can be compensated by an increase

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  327 in the pore solution viscosity induced by the rise in the polymer concentration and/or by the depletion flocculation induced by the non-adsorbed coils (Fig. 5).35 Moreover, the presence of HPG coils within the pore solution leads to an increase in the consistency coefficient (K) compared to non-admixed mortars. However, the expected increase in K due to the rise of pore solution viscosity with the polymer dosage can be compensated by steric hindrance, leading to a plateau for K. Since the adsorption of HPG 6 onto the surface of the binder is higher than 98.5%, one expects to detect a very strong increase in the yield stress compared to non-admixed mortar. However, the rheological behavior of the admixed mortar with the original guar gum is very close to that of the non-admixed mortar. This result suggests that the entire molecule of the original guar gum could be mainly adsorbed onto the surface of only one particle, limiting therefore the bridging flocculation. The very high concentration of free hydroxide groups on the backbone of the guar could be responsible of this mechanism. Moreover, the adsorption of the guar molecule onto a single particle leads to an increase in the steric hindrance and in the dispersion and lubrication effects, leading to a low but continuous decrease in the yield stress. For dosages higher than 0.1% bwob, the concentration in polymer coils (HPG 6) into the pore solution begins to increase slightly (Fig. 5), leading to the beginning of the increase in the consistency coefficient and of the decrease in the fluidity index. The additional alkyl chain also modifies the rheological properties of mortars. Contrary to HPG 3, HPGs 4 and 5 lead to a strong and continuous increase in the consistency coefficient and a decrease in the fluidity index. These results highlight that mortars become more and more shear-thinning since the fluidity index decreases from 0.8 to 0.5. This rheological behavior gets more pronounced as the HPG dosage increases. These results are consistent with the fact that the hydrophobically modified HPGs leads to the formation of coil overlapping at lower HPG dosage (0.05% in this study) since the presence of additional alkyl chains enhances the entanglement. Above this dosage, entanglement occurs between polymer coils, inducing a shear thinning behavior to the solution. At low shear rate, the entanglement of polymer coils leads to a higher pore solution viscosity and thus higher mortar viscosity. When the shear rate increases, the polysaccharide chains align in the direction of the flow resulting in less and less effect on mortar fluidity. The shear thinning behavior of the solution, and thus of the mortars, amplify with the increasing polymer dosage. The yield stress is also impacted by the additional alkyl chain. Indeed, τ0, of mortar admixed with HPGs 4 and 5, increases for a dosage equal to 0.05% bwob then slowly decreases for dosages ranging from 0.05% to 0.075% bwob, before reaching a plateau for higher dosages (τ0 reached is still higher than that of the non-admixed mortar). This result could be explained by a change in the HPG conformation due to the presence of additional alkyl chains. Indeed, as previously mentioned, alkyl chain creates intramolecular and intermolecular interactions through specific hydrophobic interactions. Intramolecular association of hydrophobic units tends to force the polymer chain into a more compact conformation.34,36 Moreover, the conformation of hydrophobically modified polymers in aqueous solution involves the presence of alkyl chains mainly inside the coils in order to limit contacts between hydrophobic chains and water.34 Consequently, the hydrophilic groups, such as hydroxyl and hydroxypropyl, are preferentially on the outskirts of the coils, promoting the adsorption onto the surface of particles (Fig. 4 (b)) and therefore steric

328 SP-302-24

hindrance which implies a prevention of direct contacts between particles. Moreover, due to a more compact conformation and avoidance between water and hydrophobic units, the bridging ability of hydrophobically modified HPGs should decrease. All these points should lead to a decrease in the yield stress. CONCLUSIONS In this paper, we studied the effect of several guar gum derivatives on water retention property and rheological behavior of mortars. Based upon the results, it was found that the original guar gum was totally adsorbed onto particle surface, leading to a negligible modification of WR and rheological behavior with respect to the non-admixed mortar. Depending of the chemical structure of HPGs, it is possible to promote the water retention according to two different ways. First, by increasing the MSHP of HPGs, the amount of adsorbed polymer drops, which leads to an increase in the HPG concentration within the pore solution and therefore to lower HPG dosage necessary to reach coil overlapping. Second, by enhancing overlapping, the hydrophobically modified HPGs improve the effectiveness of WR agent at low dosage. HPGs also modify the rheological behavior of the mortars. As in the case of WR, the hydrophobic characteristic of HPGs is the preponderant parameter. Indeed, it was shown that additional alkyl chain mainly leads to a more shear thinning behavior of the mortar and to a rise in the consistency coefficient, while classical HPGs strongly increases the yield stress. AUTHOR BIOS Alexandre Govin: Assistant Professor at the École Nationale Supérieure des Mines de Saint-Étienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS 62362, 42023 Saint-Étienne Cedex 2, France, E-mail: [email protected] Marie-Claude Bartholin: Laboratory technician at the École Nationale Supérieure des Mines de Saint-Étienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS 62362, 42023 Saint-Étienne Cedex 2, France, E-mail: [email protected] Philippe Grosseau: Professor at the École Nationale Supérieure des Mines de SaintÉtienne, SPIN-EMSE, CNRS: UMR 5307, LGF, 158 Cours Fauriel, CS 62362, 42023 Saint-Étienne Cedex 2, France, E-mail: [email protected] ACKNOWLEDGMENTS The authors would like to acknowledge Lamberti S.p.A. for the provided products and for their technical and financial support. REFERENCES 1. Rols, S.; Ambroise, J.; and Péra, J., “Effects of different viscosity agents on the properties of self-leveling concrete,” Cement and Concrete Research, V. 29, No. 2, 1999, pp. 261-266. doi: 10.1016/S0008-8846(98)00095-7 2. Khayat, K. H., and Yahia, A., “Effect of welan gum-high-range water reducer combinations on rheology of cement grout,” ACI Materials Journal, V. 94, No. 5, 1997, pp. 365-372.

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  329 3. Lachemi, M.; Hossain, K. M. A.; Lambros, V.; Nkinamubanzi, P.-C.; and Bouzoubaâ, N., “Self-consolidating concrete incorporating new viscosity modifying admixtures,” Cement and Concrete Research, V. 34, No. 6, 2004, pp. 917-926. doi: 10.1016/j. cemconres.2003.10.024 4. Sonebi, M., “Rheological properties of grouts with viscosity modifying agents as diutan gum and welan gum incorporating pulverised fly ash,” Cement and Concrete Research, V. 36, No. 9, 2006, pp. 1609-1618. doi: 10.1016/j.cemconres.2006.05.016 5. Paiva, H.; Silva, L. M.; Labrincha, J. A.; and Ferreira, V. M., “Effects of a water-retaining agent on the rheological behaviour of a single-coat render mortar,” Cement and Concrete Research, V. 36, No. 7, 2006, pp. 1257-1262. doi: 10.1016/j.cemconres.2006.02.018 6. Khayat, K. H., “Viscosity-enhancing admixtures for cement-based materials— an overview,” Cement and Concrete Composites, V. 20, No. 2–3, 1998, pp. 171-188. doi: 10.1016/S0958-9465(98)80006-1 7. Bertrand, L.; Maximilien, S.; and Guyonnet, R., “Wedge Splitting Test: a test to measure the polysaccharide influence on adhesion of mortar on its substrate”, International Congress on Polymers in Concrete, Berlin, 2004. 8. Jenni, A.; Holzer, L.; Zurbriggen, R.; and Herwegh, M., “Influence of polymers on microstructure and adhesive strength of cementitious tile adhesive mortars,” Cement and Concrete Research, V. 35, No. 1, 2005, pp. 35-50. doi: 10.1016/j.cemconres.2004.06.039 9. Biasotti, B.; Giudici, M.; Langella, V.; and Pfeiffer, U., “Highly substituted hydroxypropylguar: a strong contribution to construction chemistry”, International Dry mix Mortar Conference, Nürnberg, 2011. 10. Poinot, T.; Govin, A.; and Grosseau, P., “Impact of hydroxypropylguars on the early age hydration of Portland cement,” Cement and Concrete Research, V. 44, 2013, pp. 69-76. doi: 10.1016/j.cemconres.2012.10.010 11. Poinot, T.; Govin, A.; and Grosseau, P., ““Importance of coil-overlapping for the effectiveness of hydroxypropylguars as water retention agent in cement-based mortars,” Cement and Concrete Research, V. 56, 2014, pp. 61-68. doi: 10.1016/j.cemconres.2013.11.005 12. Cappellari, M.; Daubresse, A.; and Chaouche, M.; “Influence of organic thickening admixtures on the rheological properties of mortars: Relationship with waterretention,” Construction & Building Materials, V. 38, 2013, pp. 950-961. doi: 10.1016/j. conbuildmat.2012.09.055 13. Risica, D.; Barbetta, A.; Vischetti, L.; Cametti, C.; and Dentini, M.; “Rheological properties of guar and its methyl, hydroxypropyl and hydroxypropyl-methyl derivatives in semidilute and concentrated aqueous solutions,” Polymer, V. 51, No. 9, 2010, pp. 19721982. doi: 10.1016/j.polymer.2010.02.041 14. Lapasin, R.; De Lorenzi, L.; Pricl, S.; and Torriano, G.; “Flow properties of hydroxypropyl guar gum and its long-chain hydrophobic derivatives,” Carbohydrate Polymers, V. 28, No. 3, 1995, pp. 195-202. doi: 10.1016/0144-8617(95)00134-4 15. Standard EN 197-1, “Cement − Part 1: Composition, specifications and conformity criteria for common cements”, 2012. 16. Poinot, T.; Benyahia, K.; Govin, A.; Jeanmaire, T.; and Grosseau, P.; “Use of ultrasonic degradation to study the molecular weight influence of polymeric admixtures for mortars,” Construction & Building Materials, V. 47, 2013, pp. 1046-1052. doi: 10.1016/j. conbuildmat.2013.06.007

330 SP-302-24

17. Standard EN 196-1, “Methods of testing cement − Part 1: Determination of strength”, 2006. 18. Patural, L.; Marchal, P.; Govin, A.; Grosseau, P.; Ruot, B.; and Devès, O.; “Cellulose ethers influence on water retention and consistency in cement-based mortars,” Cement and Concrete Research, V. 41, No. 1, 2011, pp. 46-55. doi: 10.1016/j.cemconres.2010.09.004 19. Standard C1506-09, “Standard test Method for Water Retention of Hydraulic Cement-Based Mortars and Plasters”. American Society for Testing and Material, 2009. 20. NF DTU 26.1, “Travaux d’enduits de mortiers “, 2008. 21. Bouras, R.; Kaci, A.; and Chaouche, M.; “Influence of viscosity modifying admixtures on the rheological behavior of cement and mortar pastes,” Korea-Australia Rheology Journal, V. 24, No. 1, 2012, pp. 35-44. doi: 10.1007/s13367-012-0004-3 22. Barnes, H. A., and Nguyen, Q. D., “Rotating vane rheometry - a review,” Journal of Non-Newtonian Fluid Mechanics, V. 98, No. 1, 2001, pp. 1-14. doi: 10.1016/ S0377-0257(01)00095-7 23. Ait-Kadi, A.; Marchal, P.; Choplin, L.; Chrissemant, A. S.; and Bousmina, M., “Quantitative analysis of mixer-type rheometers using the Couette analogy,” Canadian Journal of Chemical Engineering, V. 80, No. 6, 2002, pp. 1166-1174. doi: 10.1002/cjce.5450800618 24. Phan, T. H.; Chaouche, M.; and Moranville, M., “Influence of organic admixtures on the rheological behaviour of cement pastes,” Cement and Concrete Research, V. 36, No. 10, 2006, pp. 1807-1813. doi: 10.1016/j.cemconres.2006.05.028 25. Roussel, N.; Ovarlez, G.; Garrault, S.; and Brumaud, C., “The origins of thixotropy of fresh cement pastes,” Cement and Concrete Research, V. 42, No. 1, 2012, pp. 148-157. doi: 10.1016/j.cemconres.2011.09.004 26. Herschel, W. M., and Bulkley, R., “Measurements of consistency as applied to rubber–benzene solutions,” Proceedings of the American Society for the Testing of Materials, V. 26, 1926, pp. 621-633. 27. Wang, J.; Somasundaran, P.; and Nagaraj, D. R., “Adsorption mechanism of guar gum at solid–liquid interfaces,” Minerals Engineering, V. 18, No. 1, 2005, pp. 77-81. doi: 10.1016/j.mineng.2004.05.013 28. Brumaud, C.; Baumann, R.; Schmitz, M.; Radler, M.; and Roussel, N., “Cellulose ethers and yield stress of cement pastes,” Cement and Concrete Research, V. 55, 2014, pp. 14-21. doi: 10.1016/j.cemconres.2013.06.013 29. Bülichen, D.; Kainz, J.; and Plank, J., “Working mechanism of methyl hydroxyethyl cellulose (MHEC) as water retention agent,” Cement and Concrete Research, V. 42, No. 7, 2012, pp. 953-959. doi: 10.1016/j.cemconres.2012.03.016 30. Marliere, C.; Mabrouk, E.; Lamblet, M.; and Coussot, P., “How water retention in porous media with cellulose ethers works,” Cement and Concrete Research, V. 42, No. 11, 2012, pp. 1501-1512. doi: 10.1016/j.cemconres.2012.08.010 31. Volpert, E.; Selb, J.; and Candau, F., “Influence of the Hydrophobe Structure on Composition, Microstructure, and Rheology in Associating Polyacrylamides Prepared by Micellar Copolymerization,” Macromolecules, V. 29, No. 5, 1996, pp. 1452-1463. doi: 10.1021/ma951178m 32. Cheng, Y.; Brown, K. M.; and Prud’homme, R. K., “Characterization and Intermolecular Interactions of Hydroxypropyl Guar Solutions,” Biomacromolecules, V. 3, No. 3, 2002, pp. 456-461. doi: 10.1021/bm0156227

Modification of Fresh State Properties of Portland Cement-Based Mortars by Guar Gum Derivatives  331 33. Semenov, A. N.; Joanny, J.-F.; and Khokhlov, A. R., “Associating polymers: equilibrium and linear viscoelasticity,” Macromolecules, V. 28, No. 4, 1995, pp. 1066-1075. doi: 10.1021/ma00108a038 34. Simon, S.; Dugast, J.; Le Cerf, D.; Picton, L.; and Muller, G., “Amphiphilic polysaccharides. “Evidence for a competition between intra and intermolecular associations in dilute system,” Polymer, V. 44, No. 26, 2003, pp. 7917-7924. doi: 10.1016/j. polymer.2003.10.054 35. Palacios, M.; Flatt, R. J.; Puertas, F.; and Sanchez-Herencia, A., “Compatibility between Polycarboxylate and Viscosity-Modifying Admixtures in Cement Pastes”, Proceedings of the 10th International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, 2012, pp. 29-42. 36. Aubry, T., and Moan, M., “Rheological behavior of a hydrophobically associating water soluble polymer,” Journal of Rheology, V. 38, No. 6, 1994, pp. 1681-1692. doi: 10.1122/1.550566

332 SP-302-24

SP-302-25

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar with Clay-bearing Aggregates by Weishan Wang, Zuiliang Deng, Zhongjun Feng, Lefeng Fu, and Baicun Zheng Interaction mechanisms of polycarboxylate-based superplasticizer (PCE) and poly(vinyl alcohol) (PVA) with bentonite were systematically investigated. The adsorption of PCE onto bentonite in aqueous solution was carried out, and changes in the surfaces and microstructures of the resultant PCE/bentonite complex were characterized by FTIR, XRD, TGA and HRTEM. The results indicated a large adsorption amount of PCE onto bentonite ranging from 157 mg/g to 230 mg/g. The interlayers of bentonite were intercalated by PCE molecules with some surface adsorption. PVA adsorbed onto bentonite competitively with PCE which decreased the adsorption amount of PCE drastically. Cement mortar experimental data showed ether-based PCE had better clay tolerance than ester-based PCE. PVA as sacrifice agent can enhance the dispersibility of PCE for cement with clay. Keywords: adsorption; bentonite; clay; intercalation; manufactured sand; mortar; polycarboxylate superplasticizer; poly(vinyl alcohol). INTRODUCTION During the last few decades, there has been a growing dependency on manufactured sands caused by increasing regulatory pressure coupled with diminishing availability of naturally local occurring sands as fine aggregates for concrete and mortar. Generally, the presence of fracture microfines from manufactured sands can have different effects on the performance of concrete or mortar. Non-clay powders such as limestone,1 slag2,3 and fly ash.3,4 have been used successfully in concrete. However, the presence of clay as impurities is always considered harmful in concrete or mortar, such as increasing the water demand to provide a concrete or mortar of given workability,5 compromising the rheological property of concrete or mortar after absorbing water and swelling,6 weakening the bond between the cement paste and aggregates,7 and especially diminishing the dosage efficiency of superplasticizers.8 These attributes have been ascribed to the presence of microfines such 333

334 SP-302-25

Table 1-Main chemical composition of bentonite SiO2 72.99

Al2O3 16.33

MgO 4.67

Chemical composition (wt%) CaO Fe2O3 TiO2 3.18 1.78 0.35

K2O 0.17

Na2O 0.04

P2O5 0.05

LOI (wt%) 11.7

LOI: Loss of ignition at 1273 K.

as limestone9,10 and clay powder11-13 in manufactured sand, which can obstruct superplasticizers from dispersing cement particles. As is known from prior research,11-13 the microfines in manufactured sands consume a large amount of superplasticizer which can decrease the dispersing efficiency of polycarboxylate-based superplasticizer. Bentonitebased clay influences concrete workability most seriously. The objective of the present investigation was to study the interaction mechanism of bentonite with polycarboxylate superplasticizer (PCE) by adsorption isotherms and characterization methods such as FTIR, XRD, TGA and TEM. PCEs with different carboxyl density and side chain length were synthesized to reduce the sensitivity to clay. In addition, competitive adsorption by PVA with PCE was carried out to elucidate the possibility to find a sacrificial agent to weaken the adverse effect caused by clay. MATERIALS AND METODS Materials Natural bentonite was crushed, ground, sieved through a 200-mesh sieve and dried at 105oC (221oF) in an oven for 2h prior to use. Chemical analysis of bentonite was performed by sequential X-Ray fluorescence spectrometer (XRF-1800, Shimadzu Corporation) and the results are shown in Table 1. The specific surface area of natural bentonite was found to be 38.6 m2/g by BET method using nitrogen as an adsorbent (ASAP 2010N, Macromeritics). A commercial polycarboxylate-based superplasticizer labeled PCE-1 was selected as the adsorbate without further purification, which are comb-like copolymers based from a sodium polymethacrylate backbone and grafted side chains of polyethylene oxides (see Fig.1).The PCE architectures and its molecular characteristics are listed in Table 2. All other chemicals were of analytical grade. PVA with a degree of polymerization of 300 and degree of hydrolysis value of 87-89% was obtained from Kuraray Co., Ltd., Japan. Methods Adsorption equilibrium studies-Adsorption experiments were carried out by using a batch technique on a horizontal thermostated shaker (SPH-103B, Shanghai Shiping Laboratory Equipment Co., Ltd.) operated at 300 rpm. A stock solution of 5 g/L was prepared by dissolving a weighed amount of PC in deionized water. The experimental solution was prepared by diluting the stock solution with deionized water when necessary. A series of PC solutions with initial concentration range of 0.1-2.5 g/L were prepared by diluting the stock solution above. The bentonite suspension, 1g/L, was also prepared before 1 h. Known volumes of solutions and suspensions (50mL (3.05 cu in.), respectively) were

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  335

Fig.1-Chemical structure of polycarboxylatebased superplasticizer. Table 2-Characterization of the polycarboxylate-based superplasticizer Sample PCE-1 a

Length of side chain n n45:n22=1:2.5

Density of side chains b:a 2:1

Mna (g/mol) 18643

M wb (g/mol) 34822

PDIc (Mw/Mn) 1.87

Solid content (wt%) 40

Mn: number-average molecular weight.

b

Mw: mass-average molecular weight.

c

PDI: Mw/Mn = polydispersity index.

transferred into 100 mL (6.10 cu in.) erlenmeyers with glass stoppers. Then the dispersions were shaken for 30 min, which was found to be sufficient to reach adsorption equilibrium. After the adsorption was over, the dispersions were rapidly centrifuged using a laboratory centrifuge (TG16-WS, Shanghai Lu Xiangyi Centrifuge Instrument Co., Ltd.) at 12000 rpm (14800×g) (32.63×lb) for 10 min. Then residual PC concentrations of the supernatant solutions were determined by the standard calibration curve. The adsorption capacity of PC was calculated by the eq. (1):

qe =

(C0 − Ce )V (1) m

where qe is the adsorption amount of PC (mg/g) at equilibrium, C0 is the initial concentration of PC in solution (g/L), Ce is the equilibrium concentration of PC in solution (g/L), m is the mass of adsorbent used (g) and V is the volume of the dispersion (L). The reproducibility during concentration measurements was ensured by repeating the experiments at least two times under identical conditions and the average values were reported.

336 SP-302-25

Competitive adsorption of PVA with PCE was carried out by testing the adsorption amount of PCE with different concentration of PVA solution (0.025g/L, 0.05g/L, 0.10g/L, 0.15g/L, 0.40g/L and 0.60g/L). Characterization methods-FTIR spectra of samples were acquired by a 6700 Fourier transform infrared spectrometer (Nicolet) using KBr pressed disk technique. Natural bentonite and KBr were weighed and then were ground in an agate mortar prior to pellet making. The spectra were obtained by accumulating 32 scans at a resolution of 2 cm-1 in the range of 4000-400 cm-1. TG and DTG (differential thermogravimetric) curves were obtained simultaneously by using a STD Q600 thermal analyzer (TA). The measurements were conducted in highpurity flowing nitrogen atmosphere (100 mL/min) and the test temperature was from 25 oC (77 oF) to 800 oC (1472 oF) with a heating rate of 20 oC/min. Approximately 20 mg sample was heated in an oven alumina crucible. The X-ray studies were performed using the powder diffraction technique. The analysis was conducted using a D/MAX 2550 VB/PC diffractometer (Rigaku). The source of X-ray radiation was a sealed tube with a copper anode and nickel filter supplied by the generator (40kV, 100mA). Diffraction measurements were conducted with the 2θ angle of 2-80o at the scanning rate of 0.02 o/min. The basal spacing was calculated by using Bragg’s equation (Eq.(2)).

nλ = 2dsinθ

(2)

where n is an integer (n=1), λ is the wavelength of incident wave (λ=0.15418 nm), d is the spacing between the layers in the bentonite lattice and θ is the angle between the incident ray and the scattering planes. JEM-2100 high-resolution TEM (JEOL) was used to investigate the microstructure of natural and PCE/bentonite complex at an accelerating voltage of 200 kV. Bentonite samples were dispersed in ethanol ultrasonically for 30 min and dropped on Cu mesh grids coated carbon, then dried in a hot air oven at 50 °C (122 °F) for 10 min. Mortar tests-Cement mortars were made by mixing cement, fine aggregates, bentonite (2.0% bwoc) and polymer solutions. The mixing procedure was as follows: (1) cement and fine aggregate were added into the bowl and mixed in a Hobart mixer at a low speed for 1 min, (2) the polymer solution was added into the mixture and mixed at low speed for 1.5 min, (3) stopping the mixer, quickly scraping down into the batch any mortar adhered on the side of the bowl, then starting the mixer and keep on mixing for another 1 min at medium speed. The fluidity was measured at 20 oC (68 °F) by pulling out spread of the mortar from a cone of top diameter of 50 mm (2.165 in.), bottom diameter of 100 mm (3.937 in.) and height of 150 mm (5.906 in.). The spread was the average of two perpendicularly crossing diameters. RESULTS AND DISCUSSIONS Adsorption isotherms of PCE onto bentonite The adsorption capacity of PCE onto bentonite at equilibrium was affected by temperature from 303 to 333K (85.73 to 139.73 oF) (shown in Fig.2). As depicted in Fig.2, the PCE uptake increases as the temperature increases, indicating favorable adsorption occurs

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  337

Fig.2-Adsorption isotherms of PCE onto bentonite. at higher temperatures. The phenomenon may be due to the acceleration of originally slow adsorption or the creation of some new active sites on the adsorbent surface.14 In general, before the equilibrium time, the increase in PCE adsorption resulting from the increase of temperature shows a kinetically controlling process. After the equilibrium was achieved, the uptake decreased with increasing temperature indicating that the adsorption of PC onto bentonite is controlled by an endothermic process. Competitive adsorption of PCE with PVA onto bentonite The competitive adsorption isotherms of PCE with different concentrations of PVA onto bentonite were shown in Fig.3. The adsorption amount of PCE decreased with the increment in PVA concentration, indicating competitive adsorption behavior between PVA and PCE molecules. The variation trend in adsorption amount of isotherms with certain PVA concentration were different. When PVA concentration was less than 0.15 g/L, PCE adsorption amount increased initially, and then slowed down gradually until it reached a platform. When PVA concentration was more than 0.40 g/L, PCE adsorption amount increased at the beginning and then decreased without reaching equilibrium, which indicates that competitive adsorption behavior appears to involve exchanging of PVA molecules with PCE molecules adsorbed on bentonite.15,16 FTIR spectra The FTIR spectra of bentonite and PCE/bentonite complex are shown in Fig.4, depicting the major changes of bentonite before and after the adsorption of PCE. After the adsorption of PCE, the PCE/bentonite complex not only has characteristic bentonite bands, but also exhibited some new characteristic bands. As can be seen from Fig.4, the adsorption peak at 3620 cm-1 was assigned to the -OH stretching vibration of water molecules within the bentonite interlayer and weakly bonded

338 SP-302-25

Fig.3-Adsorption isotherms of PCE with different concentrations of PVA onto bentonite.

Fig.4-FTIR spectra of natural bentonite and PCE/bentonite complex. to the Si-O surface, and the broad peak centered at 3470 cm-1 was due to the -OH stretching vibration of adsorbed water. The peak at around 1640 cm-1 corresponded to the -OH deformation of water in both pristine bentonite and PCE/bentonite, but the peak intensity of PCE/bentonite was lower than natural bentonite. This might suggest the increased hydrophobicity of the bentonite surface because of the adsorption. Compared to natural bentonite, the spectra of PCE/bentonite showed two additional peaks at 2923 and 2879 cm-1, which were attributed to the asymmetric and symmetric stretching vibrations of the methyl and

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  339

Fig.5-The XRD patterns of natural bentonite and PCE/ bentonite complex methylene groups. This observation also indicated the presence of PCE molecules on the surface of PCE/bentonite complex. XRD diffraction The most widely used method for the study of intercalation surfactants in the galleries of phyllosilicates is XRD, which provides information on the interlayer structure of surfactant.17 The XRD patterns of bentonite adsorbed different concentrations of PCE are shown in Fig.5. An intense reflection at 2θ=7.14o (corresponding interlayer distance was 1.24 nm) was observed for natural bentonite, which was attributed to the d001 plane of montmorillonite. In comparison with natural bentonite, the d001 peak of 0.1 g/L PCE/bentonite complex shifted towards lowder angle 6.18o, corresponding to a basal spacing of 1.43 nm, indicating the expansion of the interlayer space due to the intercalation of PCE molecules. When PCE concentration increased from 0.5 g/L to 2.5g/L, the diffraction peak increased from 5.00o to 5.08o, and the corresponding interlayer distance were 1.76nm and 1.74nm, respectively. That means PCE molecules intercalated into bentonite interlayers during adsorption process, enlarging the interlayer space. TG analysis The TG and DTG curves of natural bentonite (see Fig.6) showed two main stages of mass losses. In the first stage, the evolution of physically adsorbed water on the surface of natural bentonite in the region of 22-175 oC (71.6-347 oF) gave rise to a peak maximum of 71 oC (159.8 oF) on the DTG curve. The peak at 134 oC (273.2 oF) which was accompanied by a mass loss of 0.30% was ascribed to the elimination of the water species coordinated to the interlayer cations. In addition, the second mass loss occurred at temperature ranging from 450-750 oC (842-1382 oF), where the TG curve displayed a step weight loss (about 5.88%) related to the release of structural OH of natural bentonite. The dehydroxylation

340 SP-302-25

Fig.6-TG/DTG curves of natural bentonite.

Fig.7-TG/DTG curves of PCE/bentonite. temperature of about 668 oC (1234.4 oF) is in agreement with the classical range of dehydroxylation temperature (600-700 oC) (1112-1292 oF) observed by other authors for cisvacant montmorillonites.18 For the TG and DTG curves of PCE/bentonite complex (shown in Fig.7), there arises two extra stages, which were about 5.37% from 300-450 oC (572-842 oF) and about 2.30% from 450- 570 oC (842-1058 oF) besides the two stages of mass losses corresponding to natural bentonite. This phenomenon is interpreted to suggest that two different types of association take place between the bentonite and the polymer. The first mass loss can be from thermal decompostion of the polymer adsorbed on the surface of bentonite, and the second mass loss associated with the molecules intercalated in the interlayers of bentonite.19 More-

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  341

Fig.8-TEM micrographs of natural bentonite (a) and PCE/ bentonite complex(b). over, Tmax (the temperature when the rate of weight loss reaches a maximum) of natural bentonite was observed to be at approximately 668 oC (1234.4 oF), in contrast, Tmax of PCE/ bentonite is 657 oC (1214.6 oF) and the intensity of the peak also decreased, which means the dehydroxylation process was altered by changes in the bonding nature of the interlayer water, possibly caused by the intercalation of PCE molecules. This was supported by the results obtained from FTIR and XRD studies. TEM observation In order to find further evidence of the basal spacing enlarging, the microstructure of natural bentonite and PCE/bentonite complex were evaluated by using TEM, which permits the direct observation of microstructural features of clays. In comparison with natural bentonite, it proved to be difficult to obtain detailed TEM micrographs since the high vacuum of TEM and the high-energy beam can remove the water or surfactant molecules that makes the layer structures collapse and prohibits the structures from being readily observed. Therefore, one must take photographs as soon as possible to obtain clear, accurate images during the period of TEM viewing, especially at high magnifications. In the PCE/bentonite complex, layer spacing of 1.42 nm was in good agreement with the XRD results, but in some areas, the layer spacing remained about 1.24 nm similar to natural bentonite. This suggests that not all the interlayers were intercalated by copolymers, resulting from irregular intercalation or structure collapse mentioned above. Meanwhile, a characteristic swelling of bentonite containing termination as shown by the arrow in Fig.8b also was observed. This suggested that the swelling of silicate layers maybe augmented by defect in the clay structure.20 Adsorption amount of PCE with different molecular structure onto bentonite Based on the reported intercalation of PCE molecules into bentonite interlayers, PCEs with different carboxyl density and side chain length were synthesized to investigate if it is possible to decrease the sensitivity of PCE to bentonite by changing its molecular structure. The molecular structure of PCE was shown in Table 3. The variation in adsorption amount of PCE is shown in Fig.9. For ester-based PCE (Fig.9a), adsorption amount of PCE decreased with increasing carboxyl density, and when nMAA/nMPEG increased to 0.96 (M400-3), 2.40 (M1000-3), respectively, the decreasing trend

342 SP-302-25 Table 3-Molecular structure of PCE Sample/Item mMAA/ (mMAA+mMPEG) M400 nMAA/nMPEG M1000 Sample/Item mAA/ (mAA+mTPEG) T1200 nAA/nTPEG T2400 T4000

Mxxx-1 6.44% 0.32 0.80 Txxx-1 4.58% 0.80 1.60 2.67

Mxxx-2 12.10% 0.64 1.60 Txxx-2 6.72% 1.20 2.40 4.00

Mxxx-3 17.11% 0.96 2.40 Txxx-3 8.76% 1.60 3.20 5.33

Mxxx-4 21.58% 1.28 3.20 Txxx-4 10.71% 2.00 4.00 6.67

Mxxx-5 25.60% 1.6 4.00 Txxx-5 12.59% 2.40 4.80 8.00

Note: Mxxx-y is ester-based PCE, and Txxx-y is ether-based PCE.

Fig.9-Comparison of adsorption amount of polycarboxylate with different molecular structure onto bentonite (a:esterbased, b-ether-based)

Mxxx-6 29.22% 1.92 4.80 Txxx-6 14.38% 2.80 5.60 9.33

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  343 of adsorption amount slowed down. The affinity between anionic bentonite and anionic PCE was mainly realized by intercalation of EO side chains into bentonite interlayers. When the carboxyl density of PCE main chain increased, the repelling force between adsorbate molecules and adsorbent particles increased correspondingly. The intercalation of side chains into bentonite interlayer became more difficult and the adsorption amount decreased. When molecular weight of side chain increased from 400 to 1000 with a certain carboxyl density, the adsorption amount had a little reduction (shown in Fig.9a). That was because the enlargement of side chain size was unfavorable for the intercalation. For ether-based PCE (Fig.9b), the variation in adsorption amount of PCE with carboxyl density and side chain increasing was similar to ester-based PCE. However, adsorption amount of ether-based PCE onto bentonite was much less than that of ester-based PCE. The adsorption amount of ether-based PCE ranged from 15 to 30 mg/g with comparison to 75-120 mg/g for ester-based PCE in our experiment. The ester-based PCE was synthesized with methoxy polyethylene glycol methacrylate (MPEG) as macromonomer and methacrylic acid as comonomer, and the ether-based PCE was synthesized with isoamyl alcohol polyoxyethylene ethers (TPEG) as macromonomer and methacrylic acid as comonomer. The methyl groups in ester-based PCE has strong shielding effect on negatively charged carboxyl groups. Otherwise, the carboxyl groups in ether-based PCE will play effective electrostatic repulsion between the polymer and bentonite particles, resulting in the reduction in adsorption amount. Mortar test The variation in cement mortar with 2%bwoc bentonite with PCE above was shown in Fig.10 (water/cement ratio-0.49, sand/cement ratio-2.07, PCE dosage-0.40%bwoc). For ester-based PCE, the fluidity increased initially, and then decreased with increasing carboxyl density. However, the effect of side chain size with same carboxyl density on fluidity didn’t show regular trend. For chain Mw of 400, maximum flow observed with Mxxx-5, whereas for chain Mw of 1000, Mxxx-3 exhibited maximum flow. For etherbased PCE, mortar fluidity increased with increasing carboxyl density with same side chain length. That is because the increase of carboxyl density lowered the harmful loss caused by bentonite and cement particles were dispersed well. In addition, mortar fluidity increased with increasing side chain length with same carboxyl density. With side chain length increasing, adsorption amount of PCE onto bentonite decreased, meanwhile, the steric hindrance of PCE molecules on cement particle surface decreased the agglomeration trend and improved the fluidity of cement mortar. Effect of PVA dosage on the fluidity of cement mortar with 2%bwoc was shown in Fig.11 (water/cement ratio-0.36, sand/cement ratio-2.07, PCE dosage-0.50% bwoc). Without PVA, mortar almost had no fluidity, which is ascribed to the large amount of consumption of PCE by bentonite resulting in inadequate dispersion of cement particles. When PVA dosage increased from 0.25% bwoc to 1.00% bwoc, the mortar fluidity increased from 130 mm (5.118 in.) to 310 mm (12.205 in.). PVA adsorbed onto bentonite with PCE competitively, and that decreased the PCE adsorption amount, weakening the consumption of PCE caused by bentonite. As a result, the insensitivity of cement dispersion with PCE was enhanced.

344 SP-302-25

Fig.10-Effect of PCE on the fluidity of cement mortar with bentonite. CONCLUSION (1) A large amount of PCE was found to be adsorbed onto bentonite ranging from 157 mg/g to 230 mg/g. The interlayers of bentonite are intercalated by PCE molecules with surface adsorption to a certain extent indicated by characterizing PCE/bentonite complex. (2) PVA adsorbs onto bentonite competitively with PCE which decreases the adsorption amount of PCE drastically. (3) Cement mortar experimental data indicates ether-based PCE has better clay tolerance than ester-based PCE. Increasing side chain length and carboxyl density in main chain are favorable for weakening the sensibility of PCE to bentonite. PVA as sacrificial agent can enhance the dispersibility of PCE for cement with clay.

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  345

Fig.11-Effect of PVA dosage on the fluidity of cement mortar with bentonite. AUTHOR BIOS Weishan Wang is a R&D Engineer at the Technology Center, Shanghai Engineering Research Center of Construction Admixtures (CAERC). He received his PhD from East China University of Science and Technology (ECUST). His research interests include surface/interface chemistry, rheology and compatibility of PCE with concrete raw materials. Zuiliang Deng is a PhD Candidate at school of resources and environment engineering of ECUST, and also a R&D Engineer at the Technology Center, CAERC. She received her MS from ECUST. Her research interests include applications of PCE in all kinds of high performance concrete. Zhongjun Feng is a R&D Engineer at the Technology Center, CAERC. He received his MS from ECUST. His research interests include molecular design of PCE and interaction of PCE with cement. Lefeng Fu is a senior R&D Engineer at the Technology Center, CAERC. He received his PhD from ECUST. His research interests include R&D of key common issues in concrete and PCE industry. Baicun Zheng is a professor at the Technology Center, CAERC. He received his PhD from ECUST. His research interests include chemical engineering, surface/interface chemistry, and nano materials.

346 SP-302-25

ACKNOWLEDGEMENTS The financial support (14YF1414700) from Science and Technology Commission of Shanghai Municipality was gratefully acknowledged. REFERENCES 1. Zhu, W., and Gibbs, J. C., “Use of different limestone and chalk powders in selfcompacting concrete,” Cement and Concrete Research, V. 35, No. 8, 2005, pp. 1457-1462. doi: 10.1016/j.cemconres.2004.07.001 2. Al-Jabri, K. S.; Hisada, M.; Al-Oraimi, S. K.; and Al-Saidy, A. H., “Copper slag as sand replacement for high performance concrete,” Cement and Concrete Composites, V. 31, No. 7, 2009, pp. 483-488. doi: 10.1016/j.cemconcomp.2009.04.007 3. Yazıcı, H.; Yiğiter, H.; Karabulut, A. Ş.; and Baradan, B., “Utilization of fly ash and ground granulated blast furnace slag as an alternative silica source in reactive powder concrete,” Fuel, V. 87, No. 12, 2008, pp. 2401-2407. doi: 10.1016/j.fuel.2008.03.005 4. Jaturapitakkul, C.; Kiattikomol, K.; Sata, V.; and Leekeeratikul, T., “Use of ground coarse fly ash as a replacement of condensed silica fume in producting high-strength concrete,” Cement and Concrete Research, V. 34, No. 4, 2004, pp. 549-555. doi: 10.1016/ S0008-8846(03)00150-9 5. Yool, A. I. G.; Lees, T. P.; and Fried, A., “Improvements to the methylene blue dye test for harmful clay in aggregates for concrete and mortar,” Cement and Concrete Research, V. 28, No. 10, 1998, pp. 1417-1428. doi: 10.1016/S0008-8846(98)00114-8 6. Norvell, J. K.; Stewart, J. G.; Juenger, M. C. G.; and Fowler, D. W., “Influence of clays and clay-sized on concrete performance,” Journal of Materials in Civil Engineering, V. 19, No. 12, 2007, pp. 1053-1059. doi: 10.1061/(ASCE)0899-1561(2007)19:12(1053) 7. Topçu, İ. B., and Uğurlu, A., “Effect of the use of mineral filler on the properties of concrete,” Cement and Concrete Research, V. 33, No. 7, 2003, pp. 1071-1075. doi: 10.1016/S0008-8846(03)00015-2 8. Leslie A. Jardine, Koyata, Kevin J. Folliard, et al., Admixtures and method for optimizing addition of EO/PO superplasticizer to concrete containing smectite clay-containing aggregates, US 6352952 B1, 2002.3.5. 9. Li, B. X.; Wang, J. L.; and Zhou, M. K., “C60 high performance concrete prepared from manufactured sand with a high content of mirofines,” Key Engineering Materials, V. 405-406, 2009, pp. 204-211. doi: [J]10.4028/www.scientific.net/KEM.405-406.204 10. Li, B.; Wang, J.; and Zhou, M., “Effect of limestone fines content in manufactured sand on durability of low- and high-strength concretes,” Construction & Building Materials, V. 23, No. 8, 2009, pp. 2846-2850. doi: [J]10.1016/j.conbuildmat.2009.02.033 11. Sakai, E.; Atarashi, D.; and Damaon, M., Interaction between superplasticizers and clay minerals [C]. Proceedings of the 6th International Symposium on Cement & Concrete. Japan: Japan Cement Association. 2006, 2: 1560-1566 12. Plank, J.; Liu, C.; and Ng, S., Interaction between clays and polycarboxylate superplasticizers in cementitious systems [C]. 9th CANMET/ACI International Conference in Superplaticizers and Other Chemical Admixtures (supplementary papers). ACI Special Publication, 2009, 279-298 13. Jeknavorian, A. A.; Jardine, L.; and Ou, C. C., Interaction of superplasticizers with clay-bearing aggregates [C]. 7th CANMET/ACI International Conference on Superplasti-

Interaction of Polycarboxylate-based Superplasticizer/Poly(vinyl alcohol) with Bentonite and Its Application in Mortar w/ Clay-bearing Aggregates  347 cizers and Other Chemical Admixtures in Concrete. America: American Concrete Institute. 2003, 143-159 14. Daifullah, A. A. M.; Girgis, B. S.; and Gad, H. M. H., “A study of the factors affecting the removal of humic acid by activated carbon prepared from biomass material,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 235, No. 1-3, 2004, pp. 1-10. doi: 10.1016/j.colsurfa.2003.12.020 15. Bae, J. A. E.-H. Y. U. N.; Song, D. O. N. G.-I. K.; and Jeon, Y. O. U. N. G.-W. O. O. N. G.Bae Jae-Hyun. , “Song Dong-Ik, Jeon Young-Woong. Adsorption of anionic dye and surfactant from water onto organomontmorillonite,” Separation Science and Technology, V. 35, No. 3, 2000, pp. 353-365. doi: [J]10.1081/SS-100100161 16. Al-Degs, Y., and Khraisheh, M. A. M., “Effect of carbon surface chemistry on the removal of reactive dyes from textile effluent,” Water Research, V. 34, No. 3, 2000, pp. 927-935. doi: [J]10.1016/S0043-1354(99)00200-6 17. Li, Y., and Ishida, H., “Concentration-dependent conformation of alkyl tail in the nanoconfined space: Hexadecylamine in the silicate galleries,” Langmuir, V. 19, No. 6, 2003, pp. 2479-2484. doi: 10.1021/la026481c 18. Drits, V. A.; Besson, G.; and Muller, F., “An improved model for structural transformations of heat-treated aluminous dioctahedral 2:1 layer silicates,” Clays and Clay Minerals, V. 43, No. 6, 1995, pp. 718-731. doi: 10.1346/CCMN.1995.0430608 19. Zhou, Q.; Frost, R. L.; He, H.; and Xi, Y., “Changes in the surfaces of adsorbed paranitrophenol on HDTMA organoclay—The XRD and TG study,” Journal of Colloid and Interface Science, V. 307, No. 1, 2007, pp. 50-55. doi: 10.1016/j.jcis.2006.11.016 20. Lee, S. Y., and Kim, S. J., “Expansion characteristics of organoclay as a precursor to nanocomposites,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 211, No. 1, 2002, pp. 19-26. doi: 10.1016/S0927-7757(02)00215-7

348 SP-302-25

SP-302-26

Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases by Camille Nalet and André Nonat The difference in the retarding effects induced by sugar alcohols on the hydration of pure tricalcium silicate and white cement pastes is investigated. The polyols studied which are stereoisomers (D-glucitol, D-galactitol and D-mannitol) generate a lower retarding effect on the hydration of white cement than on the hydration of pure tricalcium silicate. The presence of aluminate and sulfate phases in white cement pastes is shown to reduce the retarding effect induced by the molecules. Moreover, these alditols strongly complex aluminate in solution and adsorb on hydrating cement. The interactions of polyols with the anhydrous and/or hydrated aluminate phases and their effects on the hydration kinetics of white cement are discussed. Keywords: adsorption; aluminate; cement; complexation; hydration; kinetics; polyols; reactivity; stereochemistry. INTRODUCTION Portland cement is composed of silicate (C3S, C2S), aluminate (C3A, C4AF) and sulfate (CaSO4, xH2O) phases which interact between each other during hydration and make its study complicated. The generalized use of chemical admixtures confers to concrete specific properties either related to its fresh state (rheology modifiers, set accelerators or retarders…) or longer term properties (strength enhancers, durability improvers...).1 Although used since decades those additives are often considered from a performance point of view2,3 but their interactions with mineral surfaces are hardly resolved and understood. Hence, this study is focused on the understanding of the interactions of hexitols with white cement paste and their effects on its kinetic of hydration. Here, we compare the effect of the conformation of D-glucitol, D-galactitol and D-mannitol on the hydration kinetics of pure tricalcium silicate (C3S) and of white cement. Then, the interactions of the hexitols with the anhydrous and/or hydrated aluminate phases present during cement hydration are assessed.

349

350 SP-302-26

Table 1-Physical and chemical compositions of the cement phases. White cement Tricalcium silicate Specific surface area (m2/g) Mono. Alite (%) Tric. Alite (%) Belite (%) Ferrite (%) Cub. Aluminate (%) Ortho. Aluminate (%) Lime (%) Gypsum (%) Hemihydrate (%) Anhydrite (%) Calcite (%) Portlandite (%) Quartz (%)

Tricalcium aluminate-Gypsum-Hemihydrate

0.40 a

0.49 b

0.34 b

66.10 0.00 24.20 0.40 2.00 0.70 0.30 0.20 0.70 1.00 2.40 1.80 0.10

100.00 -

80.00 5.00 15.00 -

– = not measured items. a

Blaine method,

Calculated from particle size distribution assuming that the density of particles is homogeneous with the size, that the particles are spherical and considering the different densities (C3S: 3210 kg/m2 and C3A= 3030 kg/m2). b

RESEARCH SIGNIFICANCE With the diversity of existing cements and the increasing levels of Supplementary Cementitious Materials of Portland cement in modern concrete, the current limited knowledge on the interactions between chemical admixtures and mineral phases and their influence on the hydration mechanisms of cement phases represents a real limitation in the development of new products with improved properties. This study intends to contribute to fill this knowledge gap by focusing on the effects and interactions of hexitols with hydrating cement phases. EXPERIMENTAL PROCEDURE Materials A batch of white cement, a C3A-gypsum-hemihydrate mixture and triclinic C3S were used, Table 1. The molecules studied were D-glucitol (≥ 98%), D-galactitol (≥ 99%) and D-mannitol (≥ 98%) which were in a powder form. The preparation of all pastes, suspensions and solutions was made with water which was both distilled and deionised. Calcium oxide used in different experiments was obtained after decarbonation of calcium carbonate (98.5-100%) at 1000 °C [1830 °F] for 24 h. All the saturated lime solutions were obtained by equilibrating water at 25 °C [77 °F] with an excess of calcium oxide. Methods Isothermal calorimetry - The kinetics of hydration of different cement phases in presence of sugar alcohols were monitored by isothermal calorimetry at 23 °C [73 °F]. The mixing

Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases  351 Table 2-Chemical structure of the hexitols studied. D-glucitol

D-galactitol

D-mannitol

protocol was as follows: 1 g of mineral phases was put in a plastic ampoule and mixed with 0.4 mL of aqueous sugar solutions (0-27 mmol/L) at 3200 rpm for 2 min with a stirrer. After mixing, the ampoules were capped and inserted in the calorimeter. Adsorption - The adsorption of polyols on hydrating white cement (L/S=5) was measured during the induction period. Samples of the cement suspension were collected over time. The different samples were centrifuged at 9000 rpm for 5 min. Finally, the supernatants were filtered with a 0.20 μm [7.87 µin] syringe filter (PTFE). The adsorption of the organic molecules was determined by using the depletion method. The non adsorbed portion of molecules remaining in solution was measured by analyzing the Total Organic Carbon (TOC) of this solution with a TOC analyzer. Different C3A-gypsum-hemihydrate mixture contents in white cement - The C3A percentage of white cement was increased from 2.7% to 15% by adding different amount of C3A-gypsum-hemihydrate to the initial white cement. Ionic concentrations of cement pore solutions - The hydration of the aluminate phase of white cement in presence of the hexitols in saturated lime solutions (L/S=100) was followed by continuously measuring the concentration of aluminium in solution over time by Inductively Coupled Plasma – Atomic Emission Spectroscopy, ICP-AES. RESULTS AND DISCUSSIONS Effect of the stereochemistry of polyols on the hydration kinetics of pure C3S and of white cement In order to identify the effect of the stereoisomers D-glucitol, D-galactitol and D-mannitol (structures in Table 2) on the kinetics of hydration of C3S and white cement, calorimetric measurements were carried out during the hydration of these phases in presence of the molecules at different dosages. In the range of concentration studied (0-27mmol/L), these polyols do not change the slope of the calorimetric curves during the acceleration period as can be seen in presence of D-glucitol for example, Fig 1. Hence, the retardation is calculated as the difference in time between the maximum peaks of heat flow of the polyol samples and the reference without additive. Even though D-glucitol, D-galactitol and D-mannitol have the same chemical formula C6H14O6, they induce different effects on the hydration kinetics of pure C3S pastes and of white cement pastes (Fig 2 and Fig 3). This effect was also shown by Zhang4 who studied the setting behavior of cement and C3S pastes in presence of D-glucitol and D-mannitol. It is noted that D-glucitol is the most retarding polyol for all the systems studied, see Fig 2 and Fig 3. It is followed by D-galactitol and then D-mannitol when hydrating pure C3S (Fig 2) whereas in cement both molecules induce approximately the same delay in the hydration of C3S (Fig 3). The results suggest that white cement is less discriminating than pure C3S.

352 SP-302-26

Fig 1-Calorimetric curves followed during the hydration of C3S and white cement pastes in presence of a sugar alcohol, L/S=0.4.

Fig 2-Retardation of C3S hydration depending on the concentration of different stereoisomers (D-glucitol, D-galactitol, D-mannitol), L/S =0.4. In order to highlight the role of the aluminate phase present in white cement on the retarding effect induced by the presence of the hexitols, the retarding effects of each molecule on the hydration of white cement and of pure C3S were compared. In a first approximation, the retardation of the hydration of both white cement and of pure C3S is forced to fit a linear function of the concentration of molecules with the equation of the form y(x)= ax. Rr is defined as the ratio of the slope concerning the retardation of white cement to the slope concerning the retardation of pure C3S, both depending on the concentration of molecules. The Rr ratios induced by the different alditols are below 1 (Fig 4): the retardation of the hydration of white cement is lower than the retardation of pure C3S. Moreover, the molecules generate different Rr ratios between 0 and 1. This means that the molecules interact differently with the other components present in white cement paste, i. e. the aluminate (and/or its hydrates) and/or the sulfate phases. Given that the pKa of the polyols are above 135 and that the pH of the pore solution is around 12.5 during the hydration of cement, the amount of polyols with deprotonated alcohol groups is assumed

Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases  353

Fig 3-Retardation of white cement pastes hydration depending on the concentration of different sugar alcohols, L/S =0.4.

Fig 4-Comparison of Rr (ratio of the slope concerning the retardation of white cement pastes to the slope concerning the retardation of pure C3S pastes) in presence of different molecules. to be low. Since the interaction of the polyols in the pastes is not of electrostatic nature, sugar alcohols are then not supposed to compete with sulfate. Hence, the difference in the Rr ratios identified above should come from different affinities of the molecules with the anhydrous or/and hydrated aluminate phases. D-glucitol is supposed to have the higher and D-mannitol the lower affinity with the aluminate phase and/or its hydrates. Effect of the C3A-gypsum-hemihydrate mixture content of white cement on its kinetics of hydration in presence of alditols In order to investigate possible interactions of the polyols with anhydrous and/or hydrated aluminates, we focused on the kinetics of hydration of white cement with different C3A-gypsum-hemihydrate mixture contents. It was observed that varying the content of the C3A-gypsum-hemihydrate mixture from 2.7% to 15% in cement at constant concentration of alditols (20.25 mmol/L), the slopes of the calorimetric curves during the acceleration

354 SP-302-26

Fig 5-Calorimetric curves followed during the hydration of white cement with different C3A-gypsum-hemihydrate mixture contents with and without the presence of a sugar alcohol, L/S=0.4.

Fig 6-Comparison of the Tei (time ending the induction period from the calorimetric curve) as a function of the percentage of C3A in white cement pastes in presence of three different sugar alcohols (20.25 mmol/L), L/S=0.4. period were different, Fig 5. In this case, instead of considering the maximum heat flow which ends the acceleration period, the time ending the different induction periods (Tei) was compared. This time is defined at the intersection between the slope of the calorimetric curve during the acceleration period and the x-axis. In presence of the three polyols, Tei is reduced when increasing the amount of C3A-gypsum-hemihydrate mixture in cement and gets closer to the Tei of the references, Fig 6. By extrapolation, it can be argued that increasing the C3A-gypsum-hemihydrate mixture content in cement would inhibit the retarding effect of the alditols on the hydration of white cement. Thus, there is an interaction of the sugar alcohols with the aluminate part and/or its hydrates during the hydration of white cement which reduces their retarding effect.

Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases  355

Fig 7-Pseudo adsorption isotherm of sugar alcohols on hydrating white cement in saturated lime solution during the induction period, L/S=5. Adsorption and retarding effect of the hexitols on hydrating white cement The stereoisomers affect differently the retardation of the hydration of white cement. Hence, in order to understand why sugar alcohols retard the hydration of white cement, it makes the investigation of a possible link between the adsorption of polyols and their retarding effects on cement hydration a priority. Fig 7 shows the pseudo adsorption isotherm of the different sugar alcohols in saturated lime solution on hydrating white cement pastes. The term pseudo adsorption isotherm is used to highlight the fact that equilibrium is not reached and that the value of the adsorption corresponds to the amount of molecules adsorbed during the induction period to the amount of dry cement instead of the surface at this time of measurement as a function of the concentration of the remaining molecules in solution. The pseudo adsorption isotherm shows that there are different adsorption behaviors depending on the conformation of the molecules, Fig 7. D-galactitol and D-mannitol reach a saturation plateau whereas D-glucitol does not reach a plateau in the range of concentrations studied (0-54 mmol/L). In fact, for D-glucitol, the pseudo adsorption isotherm even diverges suggesting a massive precipitation at some point. In the initial linear part of the adsorption isotherm, for a fix concentration in solution, D-glucitol has the higher adsorption followed by D-galactitol and then D-mannitol. Thus, D-glucitol has the most important affinity with the anhydrous and/or hydrated component(s) of white cement compared to the other polyols. The evolution of the retardation of the hydration of C3S in white cement pastes as a function of the adsorption of polyols is shown in Fig 8. The retardation of the hydration of white cement pastes was measured by using the calorimetric curves obtained with and without additive (L/S=0.4). The last retardation value for each polyol (initial concentration 54 mmol/L) was estimated by extrapolation. The adsorption of hexitols was measured in white cement suspensions as described in the paragraph just above. For all the polyols, the retardation of the hydration of white cement increases as a function of the adsorption but the trends are different, Fig 8. We observe a low retardation of the hydration of white cement in presence of D-glucitol despite a high adsorption whereas for D-galactitol and

356 SP-302-26

Fig 8-Retardation of the hydration of white cement pastes induced by hexitols depending on their adsorption on hydrating cement during the induction period.

Fig 9-Aluminium concentration during the hydration of white cement in saturated lime solution with sugar alcohols (20.25 mmol/L), L/S=100. D-mannitol, it is the contrary. These results allow us to suppose that D-glucitol particularly adsorbs on C3A and/or its hydrates. Effect of polyols on the concentration of aluminate present in the pore solution during white cement hydration The hydration of C3A of white cement suspensions in presence of the alditols in saturated lime solutions have been followed by measuring the aluminium concentration in the pore solution over time by using ICP-AES, Fig 9. Results show that the presence of polyols in solution increases more than 10 times the aluminium concentration of pore solution. This result indicates that the hexitols complex aluminium in the pore solution. In 1996, using NMR and Raman spectroscopy, Smith6 already detected the existence of complexes between hexitols and aluminium in alkaline solutions.

Effect of the Stereochemistry of Polyols on the Hydration of Cement: Influence of Aluminate and Sulfate Phases  357 FURTHER RESEARCH The role played by the surface adsorption of alditols and/or of their aluminate complex and the one played by the aluminate complexation in solution on the hydration of C3A should be further investigated. Moreover, the impact of these interactions between polyols and hydrating C3A on the hydration of C3S present in cement should be examined in depth. CONCLUSIONS The retarding effects generated by polyols (D-glucitol, D-galactitol and D-mannnitol) are lower on the hydration of white cement than on the hydration of pure C3S and differ depending on the stereochemistry of their hydroxyl groups. Moreover, the increasing content of the C3A-gypsum-hemihydrate mixture shortens the induction period induced by the sugar alcohols during the hydration of white cement. Hence, the molecules are assumed to interact with anhydrous and/or hydrated aluminate phases. The polyols are shown to complex with aluminate in solution and to adsorb on hydrating white cement. D-glucitol induces a low retarding effect on the hydration of cement compare to its important adsorption on hydrating cement. For D-mannitol, it is the contrary. Then, D-glucitol is supposed to have the highest and D-mannitol the lowest affinity with anhydrous and/or hydrated aluminate phases. This way, the high amount of adsorbed D-glucitol would not be free any more to disturb the hydration of C3S in white cement and would induce a lower retardation. As for D-galactitol and D-mannitol are concerned, the amount of non adsorbed molecules after saturation on the surface of anhydrous and/or hydrated aluminate phases would particularly delay the hydration of C3S in white cement. AUTHOR BIOS Camille Nalet is PhD student at the University of Burgundy in the “Physical Chemistry of Cementitious and Colloidal Media” laboratory group (Dijon, France). She has focused her work on the influence of structural and chemical parameters of small organic molecules on the reactivity and hydration kinetics of cement phases. André Nonat is senior researcher at the CNRS/University of Burgundy in the “Physical Chemistry of Cementitious and Colloidal Media” laboratory group (Dijon, France). His research is based on the reactivity of cement with special interest in studying the mechanism of hydration reactions, the thermodynamics and microstructure of hydrates and the mechanism of the setting. ACKNOWLEDGMENTS The financial support of Nanocem, the European Consortium of Academic and Industrial partners is gratefully acknowledged. We are also thankful to the working group for advice and fruitful discussions. NOTATION C S A F

= CaO, = SiO2, = Al2O3, = Fe2O3.

358 SP-302-26

REFERENCES 1. Kovler, K. and Roussel, N., “Properties of fresh and hardened concrete”, Cement and Concrete Research 41, V. 41, 2011, pp. 775–792. 2. Cheung, J.; Jeknavorian, A.; Roberts, L.; and Silva, D., “Impact of admixtures on the hydration kinetics of Portland cement,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005 3. Flatt, R. J., and Schober, I., “Superplasticizers and the rheology of concrete”, Understanding the rheology of concrete, Woodhead Publishing, 2012, pp. 144-208. 4. Zhang, L.; Catalan, L. J. J.; Balec, R. J.; Larsen, A. C.; Esmaeili, H. H.; and Kinrade, S. D., “Effect of saccharide set retarders on the Hydration of Ordinary Portland Cement and Pure Tricalcium Silicate,” Journal of the American Ceramic Society, V. 93, No. 1, 2010, pp. 279-287. doi: 10.1111/j.1551-2916.2009.03378.x 5. Gaidamauskas, E.; Norkus, E.; Vaiciuniene, J.; Crans, D. C.; Vuorinen, T.; Jaciauskiene, J.; and Baltrunas, G., “Evidence of two-step deprotonation of D-mannitol in aqueous solution,” Carbohydrate Research, V. 340, No. 8, 2005, pp. 1553-1556. doi: 10.1016/j. carres.2005.03.006 6. Smith, P.; Watling, H.; and Crew, P., “The effects of model organic compounds on gibbsite crystallization from alkaline aluminate solutions: polyols,” Colloids and Surfaces A: Physicochemical and Engineering Aspects, V. 111, No. 1-2, 1996, pp. 119-130. doi: 10.1016/0927-7757(95)03488-9

SP-302-27

A New Accelerator Approach for Improved Strength Development by Franz Wombacher, Christian Bürge, Emmanuel Gallucci, Patrick Juilland, and Gilbert Mäder One of the main problems associated with supplementary cementitious materials (SCMs), which are used as clinker or cement replacement, is the slow strength development compared to pure OPC. This is especially evident in the early stages of cement hydration and may cause significant problems for the customers. Therefore, the demand for new and powerful accelerators only having a marginal influence on the workability of the concrete, is rising. These types of accelerators may find their application in normal ready-mixed concrete but, much more evident, in precast applications. In this paper, a new accelerator is presented, which can significantly improve the early strength (up to 2 days) of concrete. In addition to this, the components of this new approach do not bear any potential risk of corrosion for steel, be it normal reinforcement or prestressed steel. Keywords: hardening accelerator; calcium oxide; admixture; booster; early compressive strength; fly ash; limestone filler. INTRODUCTION The demand for effective hardening accelerators which have only little influence on the workability of concrete and mortar is increasing. This is based on different facts: On one hand, the use of supplementary cementitious materials (SCM) is increasing due to clinker and cement replacement in order to reduce carbon dioxide emissions.1,2 These binders are either produced as blended cements by the cement producer or mixed out of OPC and SCM at the batching plant. They often lack in sufficient strength development, which is especially evident at early ages, up to 2 days. On the other hand, precast concrete as a fast and reliable production method would be very efficient, however, steam curing is often necessary. As a result, a large amount of energy is consumed and the turnover of forms is low which can lead to a reduction in productivity.3 An increase in early compressive strength of concrete, either based on pure OPC or on blended systems can be achieved either by addition of hardening accelerators, either classical types or new ones based on calcium silicate hydrate (CSH) suspensions.4,5 359

360 SP-302-27

Classical hardening accelerators are often based on amino alcohols and soluble inorganic salts like nitrates and thiocyanates6,7 or even chlorides,8 whereas CSH-suspensions are produced from soluble calcium-salts (like nitrate) and a silicate source.4,5 In both cases, a potential risk of corrosion cannot be excluded.9 The discussion about this potential risk of these soluble inorganic salts is ongoing since long time10 and is not yet concluded. Especially the use of higher dosages of these new types of accelerators and, therefore, the salts might intensify the discussion again. RESEARCH SIGNIFICANCE Classical hardening accelerators and CSH-suspensions which promote the seeding are today used in several applications to especially improve the early strength development of cementitious systems, either in pure OPC or in blended systems. This paper deals with the use of a combination of calcium oxide and a new type of booster component for the improvement of the early compressive strength (up to 2 days) of cementitious materials. EXPERIMENTAL PROCEDURE Mortar and concrete mixes were investigated in the laboratory. The variables include different types of cements, mixes with fly ash (FA) and variations in the content of the investigated accelerator composition at 20°C (68°F). Flow table spread (FTS) and compressive strength were determined after varying time periods. Isothermal conduction calorimetry measurements were performed at 23°C (73.4°F) both on pastes at a water-cement ratio of 0.35 using a Thermometrics TAM AIR calorimeter and on standard mortar according to DIN EN-196-1 at a water-cement ratio of 0.4 using a Calmetrix I-Cal device. Materials Superplasticizers: High range water reducers HRWR 1 (PCE for precast concrete) and HRWR 2 (PCE for ready-mix concrete) Booster: an organic ester of phosphoric acid Classical hardening accelerator: ACC 1 (based on amino alcohol and inorganic compounds such as sodium thiocyanate and sodium nitrate) Cements: CEM I 42.5 N and CEM I 52.5 R, both from Holcim AG, Siggenthal, Switzerland. Fly ash (FA) was obtained from SAFA Saarfilterasche-Vertriebs GmbH Co. GK, Baden-Baden/Germany. Calcium oxide (CaO) was obtained from Kalkfabrik Netstal AG, Switzerland. CEM I 42.5 N has a Blaine fineness of 3060 cm2/g, CEM I 52.5 R a Blaine fineness of 4120 cm2/g, FA a Blaine fineness of 3930 cm2/g and CaO a Blaine fineness of 4360 cm2/g. The physical properties and chemical analysis of the OPCs, FA, limestone filler, quartz flour and calcium oxide are shown in Table 1. Aggregates: river sands 0/1 (0-1 mm, 0-0.04 in), 1/4 (1-4 mm, 0.04-0.16 in), 4/8 (4-8 mm, 0.16-0.31 in), stones 8/16 (8-16 mm, 0.31-0.63 in) and 16/32 (16-32 mm, 0.63-1.26 in) were all obtained from Carlo Bernasconi AG, Zurich, Switzerland, limestone filler

A New Accelerator Approach for Improved Strength Development  361

Table 1 – Physical and chemical compositions of OPC, FA, limestone filler, Calcium oxide and quartz flour FA 2.30 3930

CaCO3 Limestone filler 2.97 4840

CaO 3.31 4360

a)

a)

a)

a)

9000

2.6

1.9

3.8

2.9

0.21

SiO2 [%] Al2O3 [%] CaO [%] MgO [%] SO3 [%] Na2O [%]

19.9 4.6 62.8 2.0 3.1 0.2

20.3 4.7 63.1 2.0 3.6 0.2

0.9 1.5 2.4

-

0.1 0.07 95.9 0.7 0.2 -

98.8 0.8 0.02 0.01 0.01

K2O [%]

1.0

1.0

-

-

0.01

TiO2 [%] MnO [%] Fe2O3 [%] CaCO3 [%] MgCO3 [%]

3.0 -

3.0 -

0.1 95.5 1.7

0.03 n.d. -

0.03 0.02 -

CEM I 42.5 N 3.13 3060

CEM I 52.5 R 3.13 4120

a)

Loss on ignition [%]

Density [g/cm3] Blaine fineness [cm2/g] BET [cm2/g]

a)

-

SiO2 Quartz flour 2.63 a)

not determined

Table 2 - Mortar and concrete mixture proportions

Binder Fines (limestone or quartz flour) Sand 0/1 Sand 1/4 Aggregate 4/8 Aggregate 8/16 Aggregate 16/32

Mortar Lab mix, g (lb) 750 (1.65) 141 (0.31) 738 (1.63) 1107 (2.44) 1154 (2.54) -

Concrete Lab mix, kg (lb) Kg/m3 (lb/yd3) 8 (17.64) 320 (539.4) 2.17 (4.78) 86.8 (146.3) 8.7 (19.18) 348 (586.8) 9.15 (20.17) 366 (616.9) 5.65 (12.46) 226 (380.9) 6.5 (14.33) 260 (438.2) 11.3 (24.91) 452 (761.9)

(CaCO3) from Kalkfabrik Netstal AG, Switzerland and quartz flour (SiO2) was obtained from Dorfner GmbH, Germany. Test specimens Mortars were prepared from sand and aggregates up to 8 mm (0.31 in) and concrete up to 32 mm (1.26 in). The exact mixture proportions are depicted in Table 2. The w/b was chosen corresponding to the added HRWR. All admixture dosages are based on the weight of binder. The experiments were carried out at 20°C (68°F). Mortar prisms of the size 160 x 40 x 40 mm (1.57 x 1.57 x 6.30 in) were casted at 20°C (68°F) according to EN 196-1 for the determination of compressive strength and for the measure-

362 SP-302-27

Fig.1–Compressive strength of mortars: Influence of CaO, Booster and ACC1 on CEM I 52.5 R with quartz flour and HRWR1 at 0.5% by weight of binder (w/b 0.42) ment of the expansion. For the expansion measurement the prisms were demoulded after 24 hours and cured under water at 20°C (68°F). Items of investigation The flow table spread (FTS) and air content of the mortar and concrete mixes were determined according to EN 196-1 and EN 206 (concrete), respectively. The compressive strength of mortar prism and concrete cube specimens was tested at various ages according to EN 196-1 for mortar and EN 206 for concrete. The shrinkage or expansion of the mortar specimens were measured according to EN 196-1 as well. EXPERIMENTAL RESULTS AND DISCUSSION General Classical hardening accelerators such as ACC 1 are very efficient strength improvers.6 They are optimized to increase the compressive strength from early times up to later ages. The results of compressive strength tests in a classical precast application are shown in Fig. 1. The strength increase with ACC1 is significant, starting from 6 hours on. Calcium oxide at a dosage of 3% leads to a remarkable increase in compressive strength as well. The combination of calcium oxide with a classical hardening accelerator further improves the early strength development. The booster under investigation shows no or only limited impact when used alone. This seems to be reasonable, as it is a phosphate compound, which are normally used as retarding components in admixture formulations.11 However, when this booster component is used in combination with calcium oxide the most significant strength increases are obtained in this series. This is especially interesting as the amount of booster used is very low compared to the addition of the classical hardening accelerator. It was also noted that the initial workability of the mortars was drastically reduced when the booster component was added to the mix. This can be either due to a very fast initial reaction or due to a competitive adsorption between HRWR and booster. The effect of the booster with and without calcium oxide was also investigated via calorimetric experiments as shown in Fig. 2 and 3. It can be seen, that both plasticizer and

A New Accelerator Approach for Improved Strength Development  363

Fig. 2- Calorimetry curves of CEM I 42.5 N with HRWR2 and booster without CaO

Fig.3 –Calorimetry curves of CEM I 42.5 N plus 3% CaO with HRWR2 and booster booster induce a delay in the hydration kinetics, when no calcium oxide is added. When calcium oxide is present, the induced retardation by the plasticizer can be completely recovered thanks to the booster. This probably accounts for a specific interaction between the booster and calcium oxide in the early stages of the reaction.

364 SP-302-27

Table 3 - Influence of CaO, Booster and ACC1 on CEM I 52.5 R with HRWR1 at 0.5% by weight of binder at w/b 0.42 with limestone as filler Admixture

Dosage [%]

Reference Booster CaO CaO Booster ACC1 ACC1 CaO

0.075 3 3.0 0.075 2.0 2.0 3.0

Flow table spread at 3 min, mm (in)

4h

Compressive strength, MPa (psi) 6h

8h

175 (6.9)

0

3.0 (435)

6.2 (899)

170 (6.7) 165 (6.5)

0.9 (131) 1.3 (189)

2.6 (377) 3.9 (566)

5.4 (783) 7.9 (1146)

132 (5.2)

1.4 (203)

4.2 (609)

8.2 (1189)

154 (6.1)

1.2 (174)

4.0 (580)

10.2 (1479)

154 (6.1)

2.5 (363)

8.5 (1233)

15.7 (2277)

Trials with limestone In the trials shown in Table 3 the filler component was exchanged and limestone was used instead of quartz flour. Apart from that the experiments carried out are the same as the ones shown in Fig. 1. The strength of the reference mortar is definitely higher (100%) compared to the reference with quartz flour. This might indicate an accelerating effect of the fine calcium carbonate which is in line with the data from earlier studies.12,13 ACC1 yields similar results as before, alone and in combination with calcium oxide, proving the robustness of this type of accelerator. The use of calcium oxide alone gives strength values on a similar level as with quartz flour as well. However, the results with the booster component are not similar anymore. The booster alone even results in slightly lower values than the reference mortar. When the booster component is used in combination with calcium oxide, which has resulted in the highest strength gain before, basically no beneficial effect can be noted. The strength levels in this combination are similar as with the pure calcium oxide. This indicates that the type of filler has an influence on the performance of this booster component. To further evaluate this influence of the type of fines on the performance of the different systems, mortars with different ratios quartz flour / limestone filler were tested. The results thereof are shown in Table 4. The resulting strength values confirm the trends and results as obtained in the experiments before. The mortars with high range water reducer alone again show an increasing strength performance with an increasing amount of limestone powder. The trials with a classical hardening accelerator ACC1 yield similar high strength values irrespective of the type of fines, as do the mortars with the addition of calcium oxide only. The results with the combination calcium oxide / booster component show a clear dependence on the content of limestone addition. The higher the limestone content the lower the accelerating effect of this combination. However, the strength values with 75% limestone filler are still on the level of the results obtained with the addition of a classical hardening accelerator ACC1. Regarding the flowability of the mortars two things seem to be evident as well: The obtained values are generally lower when quartz flour is used, which might be a result of the higher fineness of this filler material. The values are drastically reduced when the booster component is used together with calcium oxide, especially when there is a significant accelerating effect.

A New Accelerator Approach for Improved Strength Development  365

Table 4 – Influence of the fines on the performance in CEM I 52.5 R with HRWR1 at 0.5% by weight of binder at w/b 0.42 Admixture

Dosage [%]

Reference Reference Reference

Type of fines 75% limestone 20% quartz flour 50% limestone 50% quartz flour

Flow table spread at 3 min, mm (in)

100% quartz flour

CaO

3.0

CaO

3.0

CaO CaO Booster CaO Booster CaO Booster

3.0 3.0 0.075 3.0 0.075 3.0 0.075

ACC1

2.0

ACC1

2.0

75% limestone 25% quartz flour 50% limestone 50% quartz flour 100% quartz flour 75% limestone 25% quartz flour 50% limestone 50% quartz flour 100% quartz flour 50% limestone 50% quartz flour 100% quartz flour

Compressive strength, MPa (psi) 4h 6h 8h

170 (6.7)

0

1.6 (232)

4.9 (711)

160 (6.3)

0

1.9 (275)

4.5 (653)

168 (6.6)

0

1.5 (218)

3.9 (566)

165 (6.5)

1.0 (145)

2.4 (348)

6.4 (928)

155 (6.1)

1.1 (159)

3.1 (449)

6.8 (986)

152 (6.0)

1.1 (159)

2.9 (421)

7.7 (1117)

125 (4.9)

2.0 (290)

6.4 (928)

11.2 (1624)

120 (4.7)

3.1 (449)

8.5 (1233)

14.2 (2060)

118 (4.6)

3.2 (464)

9.8 (1421)

15.5 (2248)

158 (6.2)

1.1 (159)

4.4 (638)

9.3 (1349)

158 (6.2)

1.1 (159)

4.4 (638)

10.9 (1581)

Table 5 - Increase of Booster on CEM I 52.5 R with 3% CaO with quartz flour as filler with HRWR1 at 0.5% by weight of binder at w/b 0.42 Admixture

Dosage [%]

Reference Booster Booster Booster Booster

0.05 0.1 0.15 0.2

Compressive strength, MPa (psi) 6h 8h

Flow table spread at 3 min, mm (in)

4h

157 (6.2)

1.1 (159)

2.9 (420)

7.6 (1102)

125 (4.9) 120 (4.7) 112 (4.4) 110 (4.3)

0.9 (131) 1.6 (232) 2.4 (348) 2.1 (305)

3.4 (493) 6.0 (870) 8.9 (1291) 8.6 (1247)

8.3 (1204) 12.1 (1755) 14.8 (2146) 14.5 (2103)

The dependence of the booster effect on the type fines would of course be a limiting factor as in real life the type of fines cannot be controlled or exchanged. The effect of the booster component was therefore further investigated and a series of mortar strength tests was performed with variable dosage of the phosphoric acid ester, keeping the calcium oxide content constant at 3%. The results of these tests are shown in Table 5 and 6, in mortars with limestone and quartz flour as filler component. It can be seen that an increase of the dosage of the booster component leads to higher strength at all ages tested, irrespective of the filler component used. In the case of quartz flour as filler this acceleration effect is already remarkable at an addition of 0.05% of the booster and seems to reach an optimum at 0.15%, at least in the dosage range tested.

366 SP-302-27

Table 6 - Increase of Booster on CEM I 52.5 R with 3% CaO with limestone as filler with HRWR1 at 0.5% by weight of binder at w/b 0.42 Admixture

Dosage [%]

Reference Booster Booster Booster Booster

0.05 0.1 0.15 0.2

Flow table spread at 3 min, mm (in)

Compressive strength [MPa] psi c) 4h 6h 8h

150 (5.9)

1.1 (159)

3.4 (493)

7.8 (1131)

130 (5.1) 130 (5.1) 128 (5.0) 126 (4.9)

0.8 (116) 0.9 (131) 1.3 (189) 1.6 (232)

3.0 (435) 3.9 (566) 5.5 (798) 6.6 (957)

7.6 (1102) 9.1 (1320) 11.1 (1610) 12.7 (1842)

Fig.4 –Compressive strength of concrete tests with CEM I 52.5 R, w/b 0.45 When limestone is used as the filler component the strength values are lower than with quartz flour at the same dosage of the booster compound. This is as expected from the earlier experiments. An acceptable performance is obtained at 0.15% and 0.2%. However, compared to quartz flour, an increase in the dosage of the booster component leads to a continuous increase in strength over the whole dosage range tested. It might therefore be assumed that a further dosage increase would lead to an even more pronounced acceleration; however, this would have to be verified. It is still evident that the workability of the mortars is drastically reduced in all cases and such a performance would not be realistically acceptable. Therefore, the authors decided to check the performance of the system in concrete and adjusted the initial plasticity of the tests to a comparable level via an increased dosage of the high range water reducer. The results of this test series are shown in Fig 4. The tests with booster need significantly more plasticizer (0.7% and 0.9% vs. 0.5% for the reference) to reach the targeted initial workability. The results in concrete reflect the behavior in the mortar systems tested before. As the concrete system contains limestone as filler component, the dosage of the booster was chosen as 0.15% according to the evaluation in the mortar system. The strength

A New Accelerator Approach for Improved Strength Development  367

Table 7 - Performance in blended system CEM I 42.5 N/FA (3:1) at w/b 0.42 and use of HRWR1 at 0.5% Admixture

Dosage [%]

Reference CaO CaO Booster

3 3 0.15

Compressive strength, MPa (psi) 12 h 16 h 20 h

Flow table spread at 3 min, mm (in)

9h

232 (9.1)

1.2 (174)

5.2 (754)

238 (9.4)

3.4 (493)

7.5 (1088) 15.2 (2205) 21.8 (3162) 26.2 (3800)

170 (6.7)

4.2 (609)

9.3 (1349) 18.2 (2640) 24.5 (3553) 29.0 (4206)

7.2 (1044)

24 h

14.1 (2045) 18.1 (2625)

increase of the booster system with CaO is high at 6 h and 8 h (300% vs. reference) and still significant after 1 day (20% higher than the reference). Due to the increased HRWR dosage, the negative impact of the system on the initial plasticity could be overcome as well. It is clear that such a powerful accelerator system can have its application in normal precast systems based on ordinary portland cements as was shown in the experiments up to now, but it might also be a very interesting approach in blended systems. The experiments in Table 7 show such an application in a system with 25% of fly ash as cement replacement. The results show that the booster component can also lead to a significant strength increase in this system. The effect of the calcium oxide is significant in this fly ash system, most probably due to an “activation effect” of the puzzolanic reaction. The combination of the booster and the calcium oxide leads to an impressive strength increase compared to the reference mortar starting from 8 h on. These results are also supported by calorimetric data. In Fig. 5 the time to reach the main peak of hydration is shown. It can be seen that the increase of the dosage of the booster component in a mix CEM I 42.5 N/FA (1:1) leads to a significant shift of the main peak of hydration to earlier times, which would mean a higher compressive strength at early age. One of the drawbacks of CaO can be its expansive reaction. This is desirable in some cases and it is, therefore, also used in some expansive systems. However an expansion at later ages might lead to a disintegration of the structure. Therefore, the effect of calcium oxide in the normal dosage of 3%, but also at 6% and 9% was investigated and the length change upon storage under water was measured. The behavior over time (91 days) is illustrated in Fig. 6. The calcium oxide under investigation only shows a higher expansion at a very high dosage of 9%. However, even this level of expansion should not bear any negative consequences.14 At the normal dosage of 3% only a negligible expansion is measured. CONCLUSIONS With the booster component a totally new accelerator could be found. From its chemical nature, being a phosphoric acid ester, it is supposed to be a retarding admixture and the experimental data also show that it delays the hydration and the strength development when it is used on its own. In combination with calcium oxide, however, this booster component is able to accelerate the strength development of ordinary portland cement systems as well as blended cements as the results of combinations with fly ash are indicating. The early compressive strengths obtained are higher than the ones obtained with classical types of accelerators containing inorganic salts. These inorganic salts bear the risk of promoting

368 SP-302-27

Fig. 5 –Time to reach main peak of hydration for CEM I / FA (50: 50) with 3% CaO, 0. 16% HRWR1 and increasing amount of booster (Calorimetry measurements)

Fig.6 –Length change of mortars (CEM I 52.5 R, HRWR1 0.7%, 0.15% booster, w/b 0.45), stored under water, with different amounts of CaO steel corrosion which could be even more pronounced as these accelerators have to be used at rather high dosage. The booster component can be used at very low dosages and is, from its chemical nature, not corrosive for steel. On the contrary, phosphates are often used as components in corrosion protection or corrosion inhibitor systems. Calcium oxide as main part of the accelerator system might even be beneficial in corrosion protection, being able to increase the pH of blended cement systems. The risk of delayed expansion, as it might occur with calcium oxide, seems to be limited in this case. The question about the chemical mechanism of the booster effect could not be answered yet. It seems to be clear that the combination calcium oxide/booster is relevant for a very

A New Accelerator Approach for Improved Strength Development  369

fast reaction as the workability drops dramatically in this case and has to be compensated with additional HRWR. However, even though additional HRWR might delay the hydration, the early strength is still extremely promoted. It seems that the booster component is less effective in cases where a lot of calcium carbonate is present. However, this reduced effectiveness can be compensated by increasing the dosage of the booster. This might indicate that the booster is either absorbed by or adsorbed on the calcium carbonate. The real reason for this reduced effectiveness is not known yet and therefore possible assumptions are pure speculation. The conclusion that the system calcium oxide / booster can be a new and powerful accelerator system in pure and blended cement systems with only little side effects, however, seems to be valid. AUTHOR BIOS Franz Wombacher is Head of the Technology Center of Sika Technology AG, Construction Chemicals & Mortars in Zurich/Switzerland. He received his M.S. in Chemistry and the Ph.D. in Inorganic Chemistry from the Swiss Federal Institute of Technology (ETH), Zurich/Switzerland. He is author of several papers and patents in the field of concrete admixtures, shotcrete accelerators and corrosion protection of concrete structures. Christian Bürge is Department Manager Admixtures at Sika Technology AG, Construction Chemicals & Mortars in Zurich/Switzerland. He received his B.S. in Chemical Engineering from the University of Applied Sciences, Winterthur/Switzerland. He is author of several papers and patents in the field of concrete admixtures. Emmanuel Gallucci is Head of the Central Research department on Construction Materials of Sika Technology AG. He received a PhD in Physical Chemistry from the University Lyon/France. He is author of several papers in the field of cement chemistry and admixtures. Patrick Juilland is Research Scientist at Sika Technology AG. He received his M.S. and PhD in Material Science form the Swiss Federal Institute (EPF) Lausanne/Switzerland. He is author of several papers in the field of cement chemistry, admixtures and shotcrete accelerators. Gilbert Mäder is Laboratory Manager at Sika Technology AG, Construction Chemicals & Mortars in Zurich/Switzerland. REFERENCES 1. Scrivener, K. L., “Options for the future of cements,” Indian Concrete Journal, V. 88, No. 7, 2014, pp. 11-21. 2. Bellmann, F. et al., “Reduction of CO2 emissions during cement clinker burning, part 1”, Cement International, V. 10, (1/2013), pp. 62 – 71. 3. Min, T. B.; Cho, I. S.; Park, W. J.; Choi, H. K.; and Lee, H. S., “Experimental study on the development of compressive strength of early concrete using calcium-based hardening

370 SP-302-27

accelerator and high early strength cement,” Construction & Building Materials, V. 64, 2014, pp. 208-214. doi: 10.1016/j.conbuildmat.2014.04.053 4. Thomas, J. J.; Jennings, H. M.; and Chen, J. J., “Influence of Nucleation Seeding on the Hydration Mechanisms of Tricalcium Silicate and Cement,” The Journal of Physical Chemistry C, V. 113, No. 11, 2009, pp. 4327-4334. doi: 10.1021/jp809811w 5. Nicoleau, L., “Accelerated growth of calcium silicate hydrates: Experiments and simulation,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1339-1348. doi: 10.1016/j.cemconres.2011.04.012 6. Wombacher, F.; Bürge, C.; Kurz, C.; and Marazzani, B., “Hardening accelerators for blended cements – New approaches to improve strength development”, Supplementary papers Tenth International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Prague, Czech Republic, Oct. 28 -31, 2012, pp. 49 – 59. 7. Schaefer, S., “Sodium Thiocyanate: The alternative to non-chloride accelerators”, Supplementary papers Eighth International Conference on Superplasticizers and Other Chemical Admixtures in Concrete, Sorrento (Italy), Oct. 29 – Nov.1, 2006, pp. 381 – 387. 8. Riding, K.; Silva, D. A.; and Scrivener, K., “Early Age Strength Enhancement of Blended Cement systems by CaCl2 and Diethanol-Isopropanolamine,” Cement and Concrete Research, V. 40, No. 6, 2010, pp. 935-946. doi: 10.1016/j.cemconres.2010.01.008 9. Eichler, W. R., and Manns, W., “Zur korrosionsfördernden Wirkung von thiocyanathaltigen Betonzusatzmitteln,” Betonwerk + Fertigteil-Technik, V. 82, No. 3, 1982, pp. 154-162. 10. Corbo, J. M., and Nmai, C. K., “Sodium Thiocyanate and the Corrosion Potential of Steel in Concrete and Mortar,” Concrete International, V. 11, No. 11, 1989, pp. 59-67. 11. Oppliger, M., and Tsohos, G., “An Introduction to Concrete Admixtures”, 2007, University Studio Press, Thessaloniki, Greece. 12. Lothenbach, B.; Le Saout, G.; Gallucci, E.; and Scrivener, K., “Influence of limestone on the hydration of Portland cements,” Cement and Concrete Research, V. 38, No. 6, 2008, pp. 848-860. doi: 10.1016/j.cemconres.2008.01.002 13. Ramachandran, V. S., “Thermal analyses of cement components hydrated in the presence of calcium carbonate,” Thermochimica Acta, V. 106, September, 1988, pp. 273-282. 14. Neville, A. M., “Properties of Concrete”, 4th edition, 1995, Pearson Educ. Ltd, Edinburg Gate, Harlow, England.

SP-302-28

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography by Christof Schroefl and Viktor Mechtcherine Water desorption from superabsorbent polymers (SAP) into cement-based pastes was characterized by neutron radiography imaging to promote the understanding of the mechanisms behind internal curing of concrete. Two anionic SAP samples were used which differed in their inherent sorption kinetics in cement pore solution (SAP 1: self-releasing; SAP 2: retentive). Portland cement pastes with W/C of 0.25 and 0.50 and a paste additionally containing silica fume (W/C = 0.42, SF/C = 1/10) were investigated. Desorption from SAP 1 initiated immediately. SAP 2 released water into all the matrices as well, even in the cement paste with the high W/C of 0.50. In the other two pastes, which require internal curing by principle, SAP 2 retained its stored liquid for as long as the dormant period of cement hydration. Intense desorption then set in and continued throughout the acceleration period and even beyond. These findings explain the pronouncedly higher efficiency of SAP 2 as an internal curing admixture when compared to SAP 1. Keywords: cement hydration; cement paste; internal curing; neutron radiography imaging; portland cement; silica fume; superabsorbent polymer. INTRODUCTION Superabsorbent polymers (SAP) have evolved as multifunctional admixtures in highperformance concretes due to their ability to quickly absorb water and release it slowly into the hydrating, cement-based matrix. They were found to be especially efficient as admixtures for internal curing to mitigate autogenous shrinkage in portland cement-based construction materials with low water-cement ratios (W/C, water by weight of cement). They significantly reduce the potential of crack formation caused by self-desiccation. By now, their functioning related to this effect of “internal curing” (IC) has well been described in principle.1-4 The majority of studies have been performance-oriented. Only few has been published on mechanisms and kinetics of water migration within the cementbased material in detail, including, for example migrating amounts of water from SAP into the surrounding matrix or the onset and duration of water release.4-6 Wyrzykowski et al.4 371

372 SP-302-28

have described a so-called “demand-supply mechanism” based on the findings with their individual SAP sample. They found that liquid is extracted from swollen SAP at that point of time which equals a first significant drop of internal relative humidity. Utilizing neutron tomography, Trtik et al.6 visualized and quantified the water release from one SAP particle into a cement paste. They linked these kinetics to the progress of cement hydration and concluded that the onset of water release from the SAP particle coincides with the transition from the dormant to the acceleration period. None of these manuscripts disclosed chemical details of the SAP particles. Comparative studies regarding differently composed SAPs and focusing on their molecular sorption mechanisms in cement-based materials have been very rare. One approach showed how different SAPs perform in a cement mortar depending on their cross-linking density and their relative density of anionic functional groups.7-10 Subsequently, it will be of essential importance to deepen the mechanistic knowledge based on experimental data within the cement-based pastes and link the characteristic behaviors of SAPs both to their chemical composition and the progress of cement hydration. Neutron radiography imaging has evolved as a powerful method in observing water migration inside hardening and hardened cement-based specimens, for example.6,11-14 The work at hand used neutron radiography imaging to quantify water migration from two individual, chemically different acrylate-based SAP samples into pastes of normal portland cement (W/C = 0.25 and W/C = 0.50) and one made of normal portland cement and silica fume (W/C = 0.42, silica fume 10 wt-% by weight of cement [bwoc]). RESEARCH SIGNIFICANCE Macroscopic, performance-based results on the effects of two superabsorbent polymer (SAP) samples are explained on behalf the sorption kinetics of these SAPs directly within cement mortars. Free sorption tests in extracted cement pore solution had up to date been the method of choice. It is now demonstrated that neutron radiography imaging is the most meaningful technique to quantitatively characterize and distinguish the sorption kinetics of differently composed SAP samples. Both rheological aspects of mortars and efficiencies in mitigating autogenous shrinkage could be explained on the basis of the presented results. EXPERIMENTAL INVESTIGATION Materials and specimen preparation Pastes of cement, SAP, water, and – where applicable – high-range water-reducing admixture or silica fume were prepared right before beginning neutron radiography imaging and calorimetry, respectively. The cement was a rapid hardening, normal portland cement type I with no extra defined properties, CEM I 42.5 R, according to DIN EN 197-1.15 Its composition is provided in Table 1. The silica fume was a dry powder with an average particle size of 85 nm (3.3 10-6 in.) and a purity of 97 wt-% amorphous SiO2. As high-range waterreducing admixture a polycarboxylate-based commercial product was used as obtained as aqueous solution. Two superabsorbent polymers were used, both of which were synthesized by means of block polymerization and crushed into small pieces after synthesis (R&D samples provided by an external partner). SAP 1 had the maximum relative anionicity; it was produced of acrylic acid as the only main monomer. Its cross-linking density is qualitatively rather high,

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  373 Table 1 – Chemical and mineralogical composition of the portland cement Chemical composition [wt-%] SiO2 19.4 Al2O3 4.9 Fe2O3 3.1 CaO 62.8 MgO 3.3 SO3 3.2 K2O 0.9 Na2O 0.2 CO2 1.7 H2O 0.6 Loss on ignition 2.4

Mineralogical composition (p-XRD, Rietveld refinement) [wt-%] C3S 54.2 C2S 17.0 C3A cubic 7.1 C3A orthorhombic 1.1 C4AF 7.1 Periclase 2.6 Quartz 0.1 Arcanite 0.9 Calcite 3.0 Anhydrite 3.9 Bassanite 1.4 Gypsum 0.1 Lime

0.7

Portlandite

0.8

Table 2 – Particle size distributions of SAP 1 and SAP 2 in the dry state d10 [µm] (in.) d50 [µm] (in.) d90 [µm] (in.)

SAP 1 224 (0.0088) 586 (0.0231) 1029 (0.0405)

SAP 2 386 (0.0152) 903 (0.0356) 1446 (0.0569)

and especially it is higher than that of SAP 2. The relative anionicity of SAP 2 is below that of SAP 1. SAP 2 is a cross-linked copolymer composed of the two main monomers acrylic acid and acryl amide. These SAP samples had been used in previous studies: in,7,8 present SAP 1 was denominated as “SAP B” and SAP 2 as “SAP D”. The present SAP 1 was called “SAP 2” in9 and SAP 2 is “SAP 4” in that study, respectively. “SAP-B” in10 was present SAP 1 and “SAP-DS, -DC, -DN” were three gradings of SAP 2. Their performance in cement-based mortar has been discussed in7-10 with respect to their correlation between structure/composition, impact on rheology and efficiency in mitigating autogenous shrinkage. Table 2 summarizes the characteristic values of their particle size distributions in the dry state (laser granulometry; dispersive liquid: propan-2-ol). Sorptivities of SAP 1 and SAP 2 in the filtrate of the cement used were quantified by the so-called “tea-bag method” as described in.7 Approximately 0.2 to 0.3 g (0.44 10-3 to 0.66 10-3 lbs.) of SAP particles with the exact mass being m1 were inserted in a tea-bag which had the tare weight m2 after pre-wetting. The tea-bag containing the polymer sample was completely immersed in a beaker filled with the excess of approximately 250 mL (0.33 10-3 yd3) of the cement pore solution. The beaker was sealed tightly with a plastic foil, which was only briefly released to conduct the weighing. In a time-resolved series, the tea-bag with the polymer was taken out and weighed, providing the soaked mass m3. It was carefully wiped with two dry cloths for a short time to remove surplus and weakly bound solution. Some minor amount of capillary water might remain between the individual particles, which was considered as to be irrelevant for the overall result. On the other hand, the

374 SP-302-28

Fig. 1 – Sorptivity of the SAP samples in filtrate of cement suspension for 24 hours as determined by the tea-bag method (CEM I 42.5 R, W/C = 4.3, cf.7) tea-bag was not to be squeezed so as to not interfere with the free sorption capability of the moistured SAP particles. The ratio of intaken liquid related to the initial dry mass of the polymer was calculated according to Eq. 1.

m absorbed = (m3 − m 2 − m1 ) / m1

(Eq. 1)

In the study at hand, the time was extended from three hours utilized in7 to 24 hours to cover the entire period of neutron radiography imaging (Fig. 1). Three kinds of pastes were regarded in the present study (Table 3). The SAP dosage was 0.3 wt-% bwoc in any case. This proportion accords to the common dosage for efficient internal curing of high-performance mortar.7-9 All mixtures were prepared in a small casserole by first thoroughly preblending all dry substances (cement, SAP and – where applicable – silica fume) with a spoon. The mixing water was added within 10 seconds. In the case of W/C = 0.25, the high-range waterreducing admixture had completely been pre-dissolved in this water. The pastes were manually stirred with the spoon for two minutes until a homogenous paste had formed. For neutron radiography imaging an appropriate portion of the material was filled into an aluminum container in such a way that no voids remained. All pastes were sufficiently flowable and this placement could be conducted without any necessity of further compaction. The paste volumes put into the containers were not controlled but an experimentally suitable portion was used only. This portion was extracted arbitrarily from the entire prepared amount. With the samples Cem-SF-042 the containers had the dimensions of width/height/ depth = 80/30/10 mm/mm/mm (3.14/1.18/0.39 in./in./in.), while smaller boxes were used for the other two types of paste. Their dimensions were width/height/depth = 30/30/5 mm/ mm/mm (1.18/1.18/0.20 in./in./in.). The top was closed to airtightness with a self-adhesive aluminum tape. Special attention was paid to a quick handling and transfer of the samples into the neutron beam and start of image recording. Naturally, the relative water contents are not detectable in the time between paste preparation and this earliest image recorded.

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  375 Table 3 – Paste compositions for neutron radiography imaging Material, property Cement Tap water W/C Silica fume (dry powder) SAP (dry powder) PCE high-range water reducing admixture (aqueous solution, solids content 35 wt-%)

Unit, dimension g g g/g g g

Cem-025 70.00 17.25 0.25 0 0.21

Cem-050 70.00 35.00 0.50 0 0.21

Cem-SF-042 70.00 29.40 0.42 7.00 0.21

g

0.70

0

0

Thus, the time between paste preparation and the start of image recording should be as short as experimentally feasible. With the pastes of W/C = 0.25 this time span was 22 minutes and for those of W/C = 0.50 it was 26 minutes. For the paste containing SAP 1, silica fume and featuring a W/C of 0.42 it was seven and for the respective one with SAP 2 it was eight minutes. Consequently, all representations and interpretation of sorption kinetics start at these respective times of delay after paste preparation. Neutron radiography imaging and image processing Neutron radiography imaging was conducted at the neutron imaging and activation group of Paul Scherrer Institut (PSI) in Villigen/Aargau, Switzerland, with their beam line NEUTRA (for details, cf.11,12). The neutrons stem from the spallation source SINQ, in which protons from the Swiss ring cyclotron impinge a lead target that releases the neutrons as a steady beam. Within the beam line, the samples were placed on position 2. After passing through the sample, the remaining neutron beam intensity is transformed into visible light by a 6LiF/ZnS scintillation screen, which is directed to a CCD camera via mirrors. Open beam and dark current images were acquired before each time series. The “open beam” (or “flat field”) image represents the spatial distribution of the neutron beam intensity, in which the full neutron flux directly impacts the scintillation screen without passing through any solid. The “dark current” image was recorded to measure the bias introduced by the background noise level of the CCD camera without any neutrons approaching (all shutters in the beam line closed). Pictures were processed and quantitatively evaluated in an established style of timeresolved differential image series for water migration as described earlier, for example in.13,14 The time series images were processed with the open source image-processing tool ImageJ using built-in functions and a plugin for image normalization developed by the Neutron Imaging and Activation Group at PSI.16 Quantification of relative water contents in time The target of quantitative image analysis was to elucidate the evolution of the water content in one spot of the sample with respect to the earliest state recorded. A relative water content with respect to the initial state was calculated via referencing each image of the time sequence to the earliest image recorded. The results are normalized pictures, in which the grey value of each pixel directly represents the relative water content with respect to the reference image.

376 SP-302-28

Fig. 2 – Relative water content inside the SAP particles and the cement matrix in the specimens with W/C = 0.25 Every spot of several pixels in size which gets continuously brighter in the course of time unambiguously shows one SAP particle. The SAP particles are the only components in the paste which are able to lower their water content in the course of cement hydration and become visible in the selected technology of measurement. As a consequence of the careful sample preparation procedure and the well-flowable consistency of all pastes, these spots are not to be ascribed to compaction pores. Each of these SAP spots was individually evaluated by measuring the relative grey value of a centered region of 25 to 150 pixels. In between these spots, plain matrix remained in a visually intermediate shading throughout the time of image recording. Quantitative results, however, disclosed a slight but still significant increase in the relative water content in the course of time. For self-control, the grey shading of the entire specimens was evaluated over time. This should identify whether exchange of moisture with the surrounding occurred to a detectable extent, which might indicate that the sealing was not hermetic. It was found that this overall water content did not change throughout the entire period of image recording. Hence, autogenous conditions were preserved all the time. EXPERIMENTAL RESULTS AND DISCUSSION Sorption kinetics of SAP 1 and SAP 2 in cement paste with W/C = 0.25 containing a polycarboxylate-based high-range water-reducing admixture The relative water contents inside the hydrogel particles and the SAP-free matrix in the specimens prepared with W/C = 0.25 shows Fig. 2. Any spot of significant continuous change in water content over time and at least 25 pixels in size was interpreted as one SAP particle. This way, 54 individuals of SAP 1 and 46 of SAP 2, respectively, were identified. The error bars indicate the standard deviations based on the average values in each time of image recording. The water content in the matrices increased slightly in both pastes. This indicates “internal curing”, that is water from the hydrogels relocates in the cement matrix.

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  377 Both SAPs released rather similar amounts of water as the overlapping error bars show. However, it is worth noting that the average values feature individual curve progressions for several hours. Right from the start of imaging, which is 22 minutes after paste preparation, SAP 1 released water. Contrarily, SAP 2 did not lower its water content for up to approximately 1.5 hours but kept it rather constant or potentially showed even some minor delayed absorption. However, SAP 2 initiated desorption after 1.5 hours. SAP 1 had disclosed itself as an “incontinent” hydrogel with respect to cement pore solution in the tea-bag test (Fig. 1). This fact reasons the desorption from SAP 1 right from the start in the cement paste as well. However, SAP 2 had been found to not give away absorbed extracted cement pore solution individually but to retain it for at least 24 hours (Fig. 1). Thus, desorption from SAP 2 has to be triggered by the hydrating cement paste. The progress of cement hydration at this very low W/C of 0.25 causes a rather quick lack of freely available water which finally results in “internal self-desiccation”2,3. Wyrzykowski et al.4 formulated a “demand-supply mechanism” to reason desorption from their individual SAP sample into the cement-based structure (W/C = 0.25). Their approach can be applied to present SAP 2 and explain its desorption. Sorption kinetics of SAP 1 and SAP 2 in cement paste with W/C = 0.50 Fig. 3 displays the relative amounts of water in the cement pastes prepared with a W/C of 0.50. Similar to the pastes with W/C = 0.25, the matrix is enriched in water in the course of time, while the SAPs are releasing intensely. Only 18 spots could undoubtedly be assigned to hydrogel particles in the specimen containing SAP 1. This amount seems to be rather low when compared to any other of the pastes. However, due to the overall rather high water content and even higher ones in the centers of the SAPs, sample-scattering and detector-backscattering of neutrons may impede quantification.17,18 In the present mode of evaluation, the 18 established spots were considered only and no further mathematical treatment was applied. The behavior of SAP 1 well corresponds to its inherent sorption kinetics as obtained by the tea-bag method (Fig. 1). However, the considerable error bars indicate a pronounced non-uniform behavior of the 18 individual SAP 1 particles, especially from four hours onwards. To account for this, each spot was individually re-considered. It turned out that a group of seven (A) had a similar and very high extent of release while eleven of them had similarly uniform extent of desorption (B) with a pronouncedly lower amount of released water (Fig. 4). As a trivial approach of interpretation, it may be claimed that at least two hydrogel particles were aligned in the direction of projection. However, due to the geometry of the sample container with a depth of 5 mm (0.2 in.) only, such an alignment may be considered as rather improbable. Additionally, if this were the case, the release may be expected to be double or three times right from the start and the splitting of the intensities not initiate at a delayed point of time. Positively speaking, one can conclude that the less active portion of particles (SAP 1 B) ran out of desorbable water as early as two to three hours after paste preparation whereas the other particles (SAP 1 A) continued. Local in homogeneities in the surrounding matrix or slightly differing individual polymer characteristics may be further reasons.

378 SP-302-28

Fig. 3 – Relative water content inside the SAP particles and the cement matrix in the specimens with W/C = 0.50; all particles of SAP 1 in average

Fig. 4 – Relative water content inside the SAP particles and the cement matrix in the specimens with W/C = 0.50; two portions of SAP 1 respected The pronouncedly enhanced desorbed amount of water in comparison to that in the paste featuring W/C = 0.25 (Fig. 2) may be explained as follows. In cement paste with a higher W/C a clearly bigger amount of Ca2+ ions dissolves into the pore solution.19,20 This enhanced Ca2+ concentration intensifies a screening of the water retention ability of the hydrogel and results in a significantly enhanced desorbed amount of water on the long term.21 Interestingly, there was no correlation between the size of a SAP spot and its released portion of water. While desorption from SAP 1 in the paste with the rather high W/C of 0.50 is reasonable in conjunction to its individual sorption kinetics as derived from the tea-bag test, the desorption from SAP 2 in this paste is not quite plausible at first glance. In the cement paste

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  379

Fig. 5 – Relative water content inside the SAP particles and the cement matrix in the specimens with W/C = 0.42 and containing 10 wt-% silica fume bwoc featuring W/C = 0.50 no internal curing is required in principle2,3 and a “demand-supply mechanism” based on the drop of relative humidity and capillary pressure buildup4,22 is most unlikely to apply. Hence, these approaches should be withdrawn in this case. As well, during the relevant times of shorter than seven hours after paste preparation, the ionic conditions in the liquid phase remain almost constant.19,20 The concentration of dissolved Ca2+ does not pronouncedly change during that time and should, hence, not come into question to trigger the release from SAP 2. It is worth noting that Figs. 3 and 4 clearly indicate water loss from SAP 2 right from the begin of image evaluation. Most likely, the concentration of Ca2+ by itself can be as high as to screen the inherent retentiveness of SAP 2 and have it release rather individually.21 However, the question of why SAP 2 released stored liquid in this kind of cement paste will have to be elucidated in future studies. Sorption kinetics of SAP 1 and SAP 2 in paste of cement and silica fume with W/C = 0.42 In contrast to the pastes discussed above, the third pair of specimens contained silica fume as a further fine and pozzolanic compound. The water contents in the paste portions as well as in the hydrogels SAP 1 and SAP 2, respectively, shows Fig. 5. 87 individual particles of SAP 1 and 63 of SAP 2 were found and evaluated in the corresponding pastes. These particle numbers are much higher than those in the other pastes but the sample volume has been significantly bigger as well. In the specimen containing SAP 1 there were two individual spots which released fairly well the double amount of all the residual ones throughout the entire time of neutron radiography imaging. This finding indicates that two particles were aligned along the direction of projection at these positions. As these were two instances only, they remained unconsidered in all further data evaluation. Similar to the plain cement-based pastes the water content in the SAP-free portions of the present specimens increased visibly. Water given away from the hydrogels redistributed into the matrices, which is the fundamental mode of “internal curing”1. SAP 1 released water right from the start of imaging, which was seven minutes after paste preparation.

380 SP-302-28

Fig. 6 – Isothermal calorimetry of the specimens with W/C = 0.42, containing 10 wt-% silica fume bwoc, heat flux The reason is its inherent sorption characteristic with respect to extracted cement pore solution (Fig. 1). Contrarily, SAP 2 kept its water content for as long as 2.5 to three hours and it may even be assumed a slight delayed absorption. After that time, intense release initiated. With respect to the sorption kinetics as measured by the tea-bag method, release from SAP 2 has to be directly driven by priming forces arising in the progress of cement hydration. Compared to the paste featuring W/C = 0.25 (Fig. 2), SAP 2 kept its initial level of humidity for a clearly longer time in the present paste. This finding indicates that the sucking forces arise much earlier in the paste with the lower W/C. On the long term, SAP 2 continued its desorption as long as the end of neutron radiography imaging while SAP 1 apparently levelled off after approximately 15 hours. Obviously, no further extractable water was left behind in SAP 1 at this instance whereas SAP 2 further on provided water to the hydrating matrix at a significant pace. Link of the findings from neutron radiography imaging to cement hydration by the example of the pastes made of cement and silica fume (W/C = 0.42) The progress of cement hydration was characterized by isothermal calorimetry as well (Figs. 6 and 7). The dormant period finalized by approximately 2.5 to three hours after paste preparation, no matter whether SAP were present or not. Interestingly, no pronounced effect on the main peak of hydration was observed. These findings are in contrast to a delay in time and a decrease in intensity as published by Justs et al.23 However, their system consisted of white portland cement and the W/C was varied between 0.20 and 0.30 whereas the present pastes featured W/C = 0.38 and 0.42, normal portland cement and silica fume. Because no chemical information about the polymer sample was provided in23 and both W/C and the mineralogy of the solid components differed, the apparent discrepancy of the experimental results cannot be elucidated mechanistically in this place. Conclusively, SAP 1 released its stored water starting amidst the dormant period and hence provided water at a stage when it cannot be used up by the paste in a chemically reactive way. On the other hand, SAP 2 did not release water during the first three hours after mixing to a significant extent (Fig. 5). In other words, SAP 2 did not “spill” its uptaken

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  381

Fig. 7 – Isothermal calorimetry of the specimens with W/C = 0.42, containing 10 wt-% silica fume bwoc, released heat water during the dormant period. By the start of the acceleration period, SAP 2 began providing water to the matrix. During the consecutive hours, SAP 2 accomplished a most intense supply of water that continued at the slowing down of the acceleration period and even beyond the turning point to the deceleration period. Hence, SAP 2 delivered its water coevally with the most intense requirement whereas SAP 1 released it prematurely and came to rest quite too early. The equivalent mechanism of liquid extraction from SAP 2 applies in the cement paste with the W/C of 0.25. SUMMARY AND CONCLUSIONS Desorption of water from two superabsorbent polymers (SAP) into cement-based pastes was characterized by neutron radiography imaging to promote the understanding of the mechanisms behind internal curing of concrete. The hydrogels differed in their chemical composition and structure and hence, their inherent sorption kinetics with respect to extracted cement pore solution. SAP 1 was a free releasing polymer according to the so-called “tea-bag test”, which desorbed quasi all of its initially uptaken liquid by three to five hours, whereas SAP 2 was retentive for as long as the period of measurement of 24 hours. Quantification of the referenced images recorded over time revealed that desorption from SAP 1 initiated very early after paste preparation in any of the pastes. This way, the results from the tea-bag test were confirmed. The release process from this kind of hydrogel is governed by its individual non-retentiveness with respect to the saline cement pore solution. Contrarily, SAP 2 was found to release water into all the matrices in spite of its retentiveness with respect to the extracted, filtered cement pore solution. In the pastes which require internal curing in principle (W/C = 0.25; W/C = 0.42 with silica fume) SAP 2 was expected to release its intaken liquid in the course of time because macroscopic measurements to assess internal curing had previously shown a high efficiency in mitigation of autogenous shrinkage. In fact, SAP 2 kept its internal water content constant for approximately three

382 SP-302-28

hours but afterwards it desorbed intensely. As was characterized by isothermal calorimetry the onset of desorption from SAP 2 was coeval to the start of the acceleration period of cement hydration. Hence, release from SAP 2 matched the period of most intense requirement of water. It may be claimed that the water is actively sucked out of the hydrogel particles by chemical and physical forces, which arise as a consequence of the structural build-up of a rigid skeleton. Interestingly, as well in the cement paste prepared with the high W/C of 0.50 water was released from the SAP 2 particles. This W/C is clearly too high for the demand of classical internal curing and hence, reasoning of this experimental result has to be postponed at the time being. It is worth noting that in any specimen the relative content of water in the SAP-free districts of matrix increased. This indicates that the water released from the hydrogel particles redistributes into the matrix and will there be available for internal curing. Regarding the application of SAPs as a chemical admixture in concrete technology the findings of the present paper explain their impact on the rheological properties in fresh cement-based pastes10 as well as the distinct efficiencies in mitigating autogenous shrinkage.7-9 The rheological characteristics of cement-based mortars were elucidated for up to 90 minutes after mixing.10 Plastic viscosity as well as yield stress developments in time had indicated that SAP 1 handed over water, which was absorbed during mixing, back to the mortar during this period. Neutron radiography imaging now disclosed that SAP 1 indeed desorbed a considerable amount of intaken water during this time, even in the resting paste without agitation by a mixer or a rheometric test equipment. Contrarily, three size fractions of SAP 2 used in10 were interpreted as to keep their stored water for that time or to potentially even absorb some further water after the end of mixing. The present study provided the proof of these statements as derived from rheometry. SAP 2 kept the amount of intaken water constant after paste preparation until the onset of the acceleration period of cement hydration. With respect to mitigation of autogenous shrinkage of a high-strength mortar,7-9 had demonstrated that the “retentive” sample of SAP 2 was pronouncedly more efficient than the “incontinent” SAP 1 as they had been classified by so-called “tea-bag” test. These macroscopic results can now be interpreted based on neutron radiography imaging and calorimetry. SAP 1 releases its water as early as during the dormant period, during which no chemical or physical requirement with respect to internal curing arises. Later on, SAP 1 tends to slow down its desorption intensity during the acceleration period. Obviously, this desorption behavior in time is not very appropriate for optimum mitigation of autogenous shrinkage. On the other hand, SAP 2 dispensed water into the matrix in a focused way during the acceleration period and even continued desorption after the turning point to the deceleration period. In chemical as well as physical terms, internal curing water from SAP 2 is most beneficially used by the cement-based matrix during these stages. Hence, the proper performance of SAP 2 in mitigating autogenous shrinkage of a high-strength mortar is a direct result of the well-timed release of water. AUTHOR BIOS Christof Schroefl is head of the research group for morphological characterization and analytics at the Institute of Construction Materials at the TU Dresden (Germany). He earned his Dr. rer.nat. degree in chemistry at TU München (Germany) on PCE-based

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  383 superplasticizers for UHPC in 2010. His research topics include the working mechanisms of chemical admixtures in and durability issues of high-performance cement-based materials. Viktor Mechtcherine is Full Professor and head of the Institute of Construction Materials at the TU Dresden since 2006. His scientific interests include the development, characterization and modeling of high-performance cement-based construction materials. Since 2007, he has chaired the RILEM TC 225-SAP “Application of Superabsorbent Polymers in Concrete Construction” and has initiated the RILEM TC RSC “Recommendations for Use of Superabsorbent Polymers in Concrete Construction” in 2014. ACKNOWLEDGMENTS Part of this research project (several of the experiments performed at the Swiss spallation neutron source SINQ, Paul Scherrer Institut, Villigen/Switzerland) has been supported by the European Commission under the 7th Framework Programme through the “Research Infrastructures” action of the “Capacities” Programme, Contract No: CP-CSA_INFRA2008-1.1.1 Number 226507-NMI3”. The support of the present study by this funding is gratefully acknowledged. The authors are obliged to Dr. Eberhard Lehmann, Dr. Anders Kaestner, Jan Hovind and Peter Vontobel (PSI) for experimental support and for teaching the fundamentals of image analyses and differential quantification. The authors gratefully thank SNF Floerger (Andrézieux Cedex/France) for the generous supply with superabsorbent polymers and for the permission to publish the data in the paper at hand. The research group of Prof. Dr. Johann Plank (TU München) is thanked for performing calorimetry. REFERENCES 1. Mechtcherine, V., and Reinhardt, H. W., eds., 2012, Application of superabsorbent polymers in concrete construction. State-of-the-art report of the RILEM TC 225-SAP, Springer, Heidelberg (Germany), 164 pp. 2. Jensen, O. M., and Hansen, P. F., “Water-entrained cement-based materials: I. Principles and theoretical background,” Cement and Concrete Research, V. 31, No. 4, 2001, pp. 647-654. doi: 10.1016/S0008-8846(01)00463-X 3. Jensen, O. M., and Hansen, P. F., “Water-entrained cement-based materials: II. Experimental observations,” Cement and Concrete Research, V. 32, No. 6, 2002, pp. 973-978. doi: 10.1016/S0008-8846(02)00737-8 4. Wyrzykowski, M.; Lura, P.; Pesavento, F.; and Gawin, D., “Modeling of internal curing in maturing mortar,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 13491356. doi: 10.1016/j.cemconres.2011.04.013 5. Nestle, N.; Kühn, A.; Friedemann, K.; Horch, C.; Stallmach, F.; and Herth, G., “Water balance and pore structure development in cementitious materials in internal curing with modified superabsorbent polymers studied by NMR,” Microporous and Mesoporous Materials, V. 125, No. 1-2, 2009, pp. 51-57. doi: 10.1016/j.micromeso.2009.02.024 6. Trtik, P.; Münch, B.; Weiss, W. J.; Herth, G.; Kästner, A.; Lehmann, E.; and Lura, P., 2010, Neutron tomography measurements of water release from superabsorbent polymers in cement paste, in: W. Brameshuber (ed.), International RILEM conference on material

384 SP-302-28

science, volume III: Additions improving properties of concrete, RILEM Proceedings PRO 77, RILEM Publications S.A.R.L., Bagneux (France), pp. 175-185 7. Schröfl, C.; Mechtcherine, V.; and Gorges, M., “Relation between the molecular structure and the efficiency of superabsorbent polymers (SAP) as concrete admixtures to mitigate autogenous shrinkage,” Cement and Concrete Research, V. 42, No. 6, 2012, pp. 865-873. doi: 10.1016/j.cemconres.2012.03.011 8. Schröfl, C., and Mechtcherine, V., 2012, Superabsorbent polymers to mitigate autogenous shrinkage of cement mortar, in: V. M. Malhotra (ed.), 10th International conference on superplasticizers and other chemical admixtures in concrete (Prague, Czech Republic), SP-288-18, ACI, Farmington Hills (MI/USA), pp. 265-278 9. Mechtcherine, V.; Schroefl, C.; and Gorges, M., 2013, Effectiveness of various superabsorbent polymers (SAP) in mitigating autogenous shrinkage of cement-based materials, in: F.-J. Ulm, H. M. Jennings, and R. Pellenq (eds.), Mechanics and physics of creep, shrinkage, and durability of concrete, ASCE, Reston (VA/USA), pp. 324-331 10. Mechtcherine, V.; Secrieru, E.; and Schröfl, C., “Effect of superabsorbent polymers (SAPs) on rheological properties of fresh cement-based mortars – Development of yield stress and plastic viscosity over time,” Cement and Concrete Research, V. 67, 2015, pp. 52-65. doi: 10.1016/j.cemconres.2014.07.003 11. Lehmann, E.; Pleinert, H.; and Wiezel, L., “Design of a neutron radiography facility at the spallation source SINQ,” Nuclear Instruments & Methods in Physics Research. Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 377, No. 1, 1996, pp. 11-15. doi: 10.1016/0168-9002(96)00106-4 12. Lehmann, E. H.; Vontobel, P.; and Wiezel, L., “Properties of the radiography facility NEUTRA at SINQ and its potential for use as European reference facility,” Nondestructive Testing and Evaluation, V. 16, No. 2-6, 2001, pp. 191-202. doi: 10.1080/10589750108953075 13. Zhang, P.; Wittmann, F. H.; Zhao, T.; and Lehmann, E. H., “Neutron imaging of water penetration into cracked steel reinforced concrete,” Physica B, Condensed Matter, V. 405, No. 7, 2010, pp. 1866-1871. doi: 10.1016/j.physb.2010.01.065 14. Lieboldt, M., and Mechtcherine, V., “Capillary transport of water through textile-reinforced concrete applied in repairing and/or strengthening cracked RC structures,” Cement and Concrete Research, V. 52, 2013, pp. 53-62. doi: 10.1016/j.cemconres.2013.05.012 15. DIN EN 197-1, 11/2011, Cement – Part 1: Composition, specifications and conformity criteria for common cements; German version 16. Rasband, W. S., 1997-2012, ImageJ, http://imagej.nih.gov/ij downloaded 2013, incl. plugins for image referencing developed and provided by members of Paul Scherrer Institut, Villigen/Aargau, Switzerland 17. Kardjilov, N.; de Beer, F.; Hassanein, R.; Lehmann, E.; and Vontobel, P., “Scattering corrections in neutron radiography using point scattered functions,” Nuclear Instruments & Methods in Physics Research. Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 542, No. 1-3, 2005, pp. 336-341. doi: 10.1016/j.nima.2005.01.159 18. Hassanein, R.; Lehmann, E.; and Vontobel, P., “Methods of scattering corrections for quantitative neutron radiography,” Nuclear Instruments & Methods in Physics Research. Section A, Accelerators, Spectrometers, Detectors and Associated Equipment, V. 542, No. 1-3, 2005, pp. 353-360. doi: 10.1016/j.nima.2005.01.161

Sorption Kinetics of Superabsorbent Polymers in Cement Pastes Quantified by Neutron Radiography  385 19. Lothenbach, B., and Winnefeld, F., “Thermodynamic modelling of the hydration of Portland cement,” Cement and Concrete Research, V. 36, No. 2, 2006, pp. 209-226. doi: 10.1016/j.cemconres.2005.03.001 20. Lothenbach, B.; Winnefeld, F.; Alder, C.; Wieland, E.; and Lunk, P., “Effect of temperature on the pore solution, microstructure and hydration products of Portland cement paste,” Cement and Concrete Research, V. 37, No. 4, 2007, pp. 483-491. doi: 10.1016/j. cemconres.2006.11.016 21. Richter, A.; Paschew, G.; Klatt, S.; Lienig, J.; Arndt, K.-F.; and Adler, H.-J. P., “Review on hydrogel-based pH sensors and microsensors,” Sensors (Basel, Switzerland), V. 8, No. 1, 2008, pp. 561-581. doi: 10.3390/s8010561 22. Slowik, V.; Schmidt, M.; and Fritzsch, R., “Capillary pressure in fresh cement-based materials and identification of the air entry value,” Cement and Concrete Composites, V. 30, No. 7, 2008, pp. 557-565. doi: 10.1016/j.cemconcomp.2008.03.002 23. Justs, J.; Wyrzykowski, M.; Winnefeld, F.; Bajare, D.; and Lura, P., “Influence of superabsorbent polymers on hydration of cement pastes with low water-to-binder ratios,” Journal of Thermal Analysis and Calorimetry, V. 115, No. 1, 2014, pp. 425-432. doi: 10.1007/s10973-013-3359-x

386 SP-302-28

SP-302-29

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers by Serina Ng and Harald Justnes The dispersing effectiveness of three polycarboxylate-based superplasticizers (PCE) was investigated in two blended cement systems containing entirely different SCMs; fly ash (FA) and calcined marl (CM) at replacement percentages of 20% and 60%. The methods of investigation employed include rheological studies, hydration profiling up to 24h, and packing density analysis. Generally, replacing clinker phases by FA decreased the dynamic yield stress and delayed hydration of the pastes due to increased PCE to clinker ratios, regardless of PCE type. Little variation except for cement with 60% FA replacement (FA60) was observed on the Bingham viscosity. On the other hand, CM competed with clinkers not only for water, but also for PCEs even in CM20, reducing the fluidity of the paste but maintaining a similar initial rate of hydration of the pastes. PCE possessing intermediate side chain lengths proved to be more effective for CM systems than PCEs possessing long side chains. Keywords: plasticizers; PCE; fly ash; calcined marl; blended cements. INTRODUCTION Green building and construction is the current trend in the building sector. Concrete, largely composed of cement (OPC), is one of the major materials employed. Cement had an annual production of about 3.5 billion tons in 2011.1,2 During its production, one ton of OPC clinker emits an equivalence of about one ton of CO2 when pulverized coal is used as a fuel, contributing to ~6% CO2 emission annually worldwide. This makes the cement industry the third largest CO2 emission source after housing and transport.3,4 Therefore, there is still a strong desire to reduce this CO2 emission from cement production. A most direct method to aid in this process is to find suitable substitutions for cement clinker. Greener and more environmentally friendly binders are thus sought after, and employing supplementary cementitious materials (SCMs) as replacement for clinker phases in blended cements is one of the main approaches. Incorporation of SCMs into cements often can result in a change in the properties of the blended cements, and thus altering the compatibility with other additions. One such chal387

388 SP-302-29

lenge is in the application of superplasticizers. Superplasticizers are commonly employed to modify the rheology of cement pastes to improve handling on site and/or increase and enable enhanced final properties and aesthetic of the composite or buildings. These polymers are commonly designed to work with pure OPC systems, where rapid adsorption onto the clinker, particularly C3A surfaces aids in the dispersion of the cementitious systems.5 In presence of SCMs, which can possess different chemical compositions and thus surface chemistry, etc, the dispersing ability of these polymers may be altered leading to different rheological behaviour arising from deviations in the interactions between these admixtures and binder particles. The fundamentals of the colloidal interface of the polymer-inorganic materials have been assessed in both isolated SCM systems6,7 and selected blended cement systems.8 Despite these extensive studies, an actual understanding of the applied rheological effectiveness of these superplasticizers is required to ensure optimal usage of the admixtures in such system. Additionally, the influence of these admixtures on blended cements, possessing higher replacement amounts of SCMs, is expected to create a greener construction industry. For this purpose, three different commercial polycarboxylate ether plasticizers (PCEs) were selected to test for their dispersing effectiveness with two different blended cements based on fly ash and calcined marl respectively. Fly ash was selected due to its prevalent utilization in commercial products, whereas calcined marl based on its abundance. The three different PCEs contain polymers with different PEO side chain lengths. Two cement replacement levels were selected for this investigation: 20% and 60% replacement by weight. Rheological tests were performed employing a parallel plate rheometer, while the packing density was measured by the centrifugal consolidation method.9 Finally, the heat of hydration of the cement slurries was investigated by isothermal calorimetry; and a correlation between the heat evolution profile and the rheological properties of the cement pastes was attempted. RESEARCH SIGNIFICANCE With the need for cheaper, greener and better alternatives to OPCs, high replacement levels by SCMs are predicted to be an important approach for the cement industry. This can lead to compatibility issues with other materials, such as superplasticizers, particularly PCEs, which have been commonly employed in the concrete industry construction. The authors believe that this study dealing with workability of superplasticizers with blended cements having high loading of SCMs, will prove to be highly useful. EXPERIMENTS AND METHODS Materials An ordinary Portland cement (OPC), a class F fly ash (FA) and a calcined marl (CM) calcined at 850 °C were employed in this investigation. The OPC and FA were supplied by Norcem A.S. Brevik, Norway, while the CM was from Saint-Gobain Weber, Oslo, Norway. Table 1 displays the chemical compositions of the OPC, FA and CM respectively. The calcined marl contains mainly smectite (>50%) and calcite (~25%). Minor amounts of kaolinite (~8%), quartz (~4%), siderite (~3%) and pyrite (~1%) are present. The specific Blaine surface areas of the OPC and FA were 382 and 357 m2/kg respectively. Whereas for CM, the specific surface area as measured by BET was 15.1 m2/g. Blaine and BET

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  389 Table 1–Chemical compositions of OPC, FA and CM SiO2 Al2O3 Fe2O3 CaO MgO P 2O 5 K 2O Na2O SO3 Alkali Total

OPC 20.8 4.6 3.5 61.6 2.4 0.2 1.0 0.5 3.5 1.1 99.2

FA 50.0 23.9 6.0 6.3 2.1 1.1 1.4 0.6 0.4 1.6 93.4

CM 49.6 18.1 10.6 14.1 2.9 0.2 2.4 0.7 98.6

values cannot be compared directly as Blaine is related to the “outer” surfaces of particles (i.e. voids between them), whereas BET measures both outer and inner surface of nonconnected pores and cracks. Three commercial polycarboxylate based superplasticizers (PCEs) were investigated. These PCEs were supplied by Mapei AS, Sagstua, Norway and were denoted as PC1, PC2 and PC3 in this investigation. PC1 and PC2 are both based on polymethacrylate type polycarboxylates. Commercially, PC1 is commonly employed in ready-mix systems and is comprised of a PCE with long side chains with low charge density. PC2 on the other hand, possesses intermediate workability and plasticizing effect, and is comprised of a polymethacrylate-based PCE with short side chains and high charge density. PC3 is made up of a combination of one polymethacrylate based PCE and a polyacrylate based PCE in the ratio of 1:9. The polymethacrylate based PCE here possesses long side chains and low charge density and is similar to that in PC1. The polyacrylate based PCE possesses very long side chains and very low charge density. The main differences between the polyacrylate based PCE and polymethacrylate based PCE are in their backbone structure, which can be explained by the fact that for the grafting of longer side chains onto PCE, a polyacrylate backbone is normally required. Therefore, the actual main difference between these two polymers was in the length of the side chains. PC3 is commonly used in the precast/ prestress industry and generally generates pastes with the shortest workability. All materials were utilised as received. For preparation of the blended cements, the OPC and FA/CM were manually mixed in the ratio of 4:1 and 2:3 to produce blended cements with CM contents of 20 and 60%, respectively. These blended cements were denoted as FA20, FA60, CM20 and CM60, respectively. Two different dry polymer dosages were employed; 0.2 and 0.4% dry polymer bwob (by weight of binder), respectively. Experimental procedures All cement pastes were prepared at a w/b of 0.36 to amplify the performance of the PCEs. When PCEs were added, the required dosages were homogenized in the water before adding to the dry powder mix over 30s. The mixture was blended under high shear for 1min utilizing a high shear mixer (Philips, 600W, capacity = 200mL), let stand for 5min and a final high shear mixing of 1min to avoid false setting. The high shear mixing was to

390 SP-302-29

mimic the high shear energy in a concrete partly imposed by coarse aggregate.10 In each mix, the amount of pastes prepared was ~205g to minimise weighing errors, and they were employed for the rheological, calorimetric and packing density analyses. Rheological measurements – About 2ml of the prepared pastes were smeasured within the first 12 min after first contact of dry powder with water. A Physica MCR 300 rheometer (Anton Paar, Graz/Austria) equipped with parallel plate geometry was utilised. The upper (rotor) and lower (stator) plates were serrated to a depth of 150µm, the gap distance was 1mm and measurements were conducted at 20°C. Before starting the measurements, the paste was sheared at 100s-1 for 1min. The up and down flow curves was measured over the range of shear rates from 2 to 150s-1 over a period of 6min. For analysis, the down flow curve was fitted with a linear regression at the high shear rates (threshold limits of >50s-1). The Bingham viscosity (μ2) can be obtained from the gradient of the linear regression, while the dynamic yield point (τd) of the cement paste was derived from the ordinate intercept of the linear regression line. All measurements were set to end and jump to next test segment when dQ/dt>10. The dynamic yield points present an indication of the amount of force needed to start shearing the cement slurry. Calorimetric investigations – About 8 g of the prepared pastes were weighed accurately into a glass vial, sealed with a lid and placed in an isothermal TAM Air calorimeter (TA Instrument, New Castle/USA). Measurements were performed up to 24 h from the point of first contact between dry powder and water against a calibrated reference of inert alumina powder of similar mass. The time of placement was recorded and all subsequent hydration profiles were calculated and tabulated after 1 h to exclude any excessive heat transfer arising from initial preparation. Packing densities – The packing densities of the cement pastes were determined by a modification of the centrifugal consolidation method, proposed by Miller.11 About 80 g of the paste was added into a 50 mL falcon tube and subjected to centrifugation at 4,000 rpm for 5 min. The supernatant was carefully removed with a pipette and the weights of the cement samples (before and after compaction/removal of liquid) were determined. The packing density of the cement slurry was calculated at this applied compaction energy as follows:

Packing density = Volbinder / Vol(binder+water) after centrifugation

(1)

where the volume of binder and water were calculated from the mass remaining in the residue after centrifugation utilizing the densities of cement (3.15kg/m3), fly ash (2.35kg/ m3), calcined marl (2.65kg/m3) and water (1.0kg/m3) respectively. For this calculation, it was assumed that only water was extracted during the centrifugal process. RESULTS AND DISCUSSION Rheological values quantify the fluidity of the cement pastes, which are very significant when the performance of PCE superplasticizers is needed. In particular, the ability of these polymers to deflocculate and disperse particles in solution or to reduce the water uptake for hydration, etc can be given a value. Coupled with calorimetry and packing density measurements, the driving force behind the performance of the PCE superplasticizers can be derived. The results here will be divided into four sections. First, the properties of neat

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  391

Fig. 1–(a) Viscosity versus dynamic yield stress of OPC, FA20, FA60, FA100 and CM20; (b) Cumulative heat evolved per binder mass of pastes containing 0 to 60% replacement of FA/CM at 3h, 6h, 12h and 24h (w/b = 0.36). cement pastes, both OPC and blended will be discussed. Thereafter, the interaction of OPC with PCE will be highlighted, presenting the normal effective behaviour of PCEs. In the third section, the properties of pastes with low replacement levels of SCMs in presence of PCEs will be clarified. Finally, the properties of pastes possessing high replacement level of SCMs (60%) with added PCEs will be discussed. Properties of neat OPC, FA and CM blended cements The properties of the neat OPC, FA20, FA60, CM20 and CM60 were first investigated (Fig. 1). At a w/b 0.36, the OPC paste exhibited a Bingham viscosity of 0.28 Pa·s and dynamic yield stress of ~230 Pa. With increasing FA replacement, a proportional decrease (R2 = 0.9317) in the rheological values of FA blended cements was observed. Replacement of clinker by FA appeared to decrease the dynamic yield stress greater than viscosity. This signifies that increasing FA content improves the flow of neat pastes, driven greatly by the reduction in dynamic yield stress of the pastes, possibly due to the spherical nature of the FA particles. On the other hand, when neat CM-cement pastes were tested, only CM20 was sufficiently fluid for initial rheological measurements (μ2 = 0.49 Pa·s, τd = 370 Pa) reflecting the high consumption (i.e. by adsorption and/or absorption) of water by CM, which can impair the early age hydration of OPC. The cumulative heat evolved from the hydration of OPC increased gradually up to 6h, which accelerated after that to attain a total heat of 12.1 J/g by 24h of hydration (Fig. 1b). When SCMs were introduced, a decrease in the cumulative heat of hydration with increasing SCM replacement was observed, regardless of the type of SCM employed. This observation indicated that the amount of OPC present dominated the hydration heat of the pastes. Two additional observations could be made. Firstly, regardless of replacement contents, the decrease in heat evolved relative to the neat OPC paste was larger for FA blended cements than the CM blended ones. Secondly, especially when analysing the heat evolved within the first 6h, it can be observed that FA caused a greater delay in early age hydration than CM right from the early age (% heat evolved relative to OPC by 6h: FA60 = 11%,

392 SP-302-29

Fig. 2–Packing densities of blended cements as a function of SCM loading. CM60 = 58%), indicating that these two SCMs contributed very differently to the hydration profiles of the pastes. When coupled with the cumulative heat evolved by 24h, it can be observed that FA acted mainly through filler effect, whereby hydration of FA20 and FA60 gave rise to 84% and 40% of the total heat evolved by a neat OPC (12.1 J/g). In the case of CM, hydration of CM20 and CM60 displayed relative heat of 90% and 49% respectively, indicating that CM contributed to the overall hydration of the paste. The decrease in proportional amount of heat evolved from CM20 to CM60 can be attributed to the decrease in OPC content and also the availability of water for hydration as a result of rapid initial consumption by CM. Then, the packing densities of the cement pastes were investigated. The packing density of the neat OPC was 0.496 at the compaction energy employed. When OPC was replaced by the SCMs, the packing densities increased as a function of SCM replacement (Fig. 2). It could be deduced at this point that similar mechanisms were at play for both SCMs. However, it is important to take into account that the densities of both FA and CM were lower than that of OPC (2.35kg/m3, 2.65kg/m3 and 3.15kg/m3 respectively), thus the variation in their packing densities were based on a sum of different factors. For FA systems, FA100 displayed a high packing density of 0.580 due to its low hydrating ability. Therefore, much of the water was released in the supernatant during centrifugation. The packing densities of FA20 and FA60 thus layed between that of FA100 and OPC. In the case of CM, the increase in packing density can be accounted for by the decrease in density of the binder when OPC was replaced with CM. In terms of affinity for water, CM consumed water rapidly and the saturation point was not reached at a w/CM of 0.36. Here, it is of interest to note that FA and CM are two very different SCMs, whereby FA improved the fluidity of the pastes while CM reduced the flow. In the next section, the impact of increasing SCM contents, comparing behavior of FA/CM on the PCEs, will be presented accordingly. Properties of OPC (FA0/CM0) with added PCEs First, the pastes without any SCMs will be discussed. In presence of 0.2%bwob of PCEs, the viscosity of the pastes remained in the same order as that of neat OPC, whereas

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  393

Fig. 3–(a) Viscosity versus dynamic yield stress of OPC, FA20 and FA60 with added 0.2%bwob PCEs; (b) Viscosity versus dynamic yield stress of OPC, CM20 and CM60 with added 0.2%bwob PC, w/b = 0.36. For CM60, 0.4%bwob of PCEs were investigated as only 0.2%bwob of PC1 was effective. Rheology of pastes is represented by unshaded markers corresponding to that of the neat pastes, i.e. neat OPC is shaded circle, whereas OPC + PCE are displayed as unshaded circles.

Fig. 4–(a) Cumulative heat evolved from the hydration of OPC pastes in absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36); (b) Packing densities of OPC pastes in the absence of presence of 0.2%bwob of varying PCEs. dynamic yield stress of all pastes decreased by at least a factor of 2.3 (PC2, Figure 3). With PC1 and PC3, the decrease was much greater with factors of >10, indicating that these PCEs dispersed the OPC effectively by reducing the dynamic yield stress (i.e. preventing flocculation or gel strength build-up). The variation in rheological impact can be attributed to the length of side chains (SC) present in the PCE, whereby PC3 with the longest side chain dispersed the paste best, whereas PC2 was the least effective as a result of the short side pendants present in this PCE. The impact of PCEs on the cumulative heat of hydration of OPC pastes can be observed from Fig. 4. During early age of hydration (up to 3h), a delay in hydration was observed for all pastes containing PCEs, potentially due to decreased dissolution and hydration of the clinker phases as PCEs adsorb topochemically onto the reactive C3A and alite phases.

394 SP-302-29

Table 2–Packing densities of OPC, FA pastes with PCEs after consolidating centrifugation OPC FA20 CM20 FA60 CM60*

None

PC1

PC2

NRGPC3

0.496 0.514 0.502 0.548 0.510

0.480 0.497 0.492 0.632 0.507

0.485 0.501 0.493 0.561 0.509

0.481 0.504 0.493 0.578 0.508

* 0.4%bwob of PCEs were investigated too. Results showed similar packing density as when 0.2%bwob of superplasticizers were employed.

PC3 retarded the least as it possessed the lowest coverage on the clinkers phases as a result of its low effective molecular coverage per clinker surface due to its high Mw (long side chains). Further hydration of the pastes showed deviations in the retarding effects of PCEs on OPC hydrations. By 12h of hydration, PC3 displayed a synergetic effect on the hydration of OPC paste, showing an 18% increment in cumulative heat evolved as compared to the neat OPC paste. This indicates that when PC3 is employed, a higher 1 day strength than that of neat OPC paste is to be expected. By 24h, all PCEs filled pastes possessed higher cumulative hydration heat relative to the neat OPC paste despite the prolongation of the induction period due to dispersion of cement agglomerates which resulted in more “free surfaces” for subsequent hydration. The packing density of the neat OPC was 0.496 at the compaction energy employed (Fig. 2). In the presence of PCEs, a decrease in packing densities was observed (Table 2). This resulted from an increase in the retention of water by the polymers which in effect contributed to the improved rheological properties of the cement pastes. In general, the packing density was the lowest when PC2 was added, followed by PC3 and PC1 respectively. No direct correlation to the side chain lengths could be drawn here. Properties of FA20/CM20 pastes with added 0.2% bwob of PCEs The impacts of PCEs on the rheological, heat evolution and packing densities of pastes with 20% SCM replacements were next investigated. 20% SCM replacement was selected as it represents the average normal amount of SCMs present in commercial cements currently available on the market.12 As observed in Fig. 3, addition of PCEs to the 20% SCM replaced pastes resulted in the decrease of both Bingham viscosities and dynamic yield stresses for all pastes. Similar to the OPC system, the dynamic yield stress of the pastes was more affected than the Bingham viscosity. More specifically, for the FA20 pastes, the trend in effectiveness of the PCEs was PC2 > PC1 > PC3, similar to that observed for OPC pastes. Additionally, the relative improvements in rheologies of the pastes were greater for FA20 than OPC, comparing the neat pastes to pastes with added PCEs. These findings confirmed that addition of FA to OPC not only diluted the system (higher effective OPC/PCE ratio), but also FA appeared to have no significant impact on the performance of PCEs at this level of replacement. In other words, independent of the nature of the PCEs, these polymers interacted and dispersed mainly with the OPC present in the system.

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  395

Fig. 5–Cumulative heat evolved from the hydration of (a) FA20 and (b) CM20 pastes in absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36). In the CM20 pastes, despite the improvement in rheological behaviours, the trend in performance of PCEs differed from that of OPC, whereby here, PC1 performed the best, followed by PC3 and finally PC2. This indicates that other than mere water consumption, CM may interfere with the performance of PCEs, particularly in the case of PC3. For the cumulative heat of hydration, a similar trend was observed here for FA20 pastes as compared to that in rheological measurements. Figure 5 presents the cumulative heat evolved from the hydration of FA20 and CM20 pastes in absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36). The retarding effect of PCEs on the hydration of FA20 was as follow: PC1 > PC2 > PC3, similar to OPC pastes. The variation in the degree of hydration for FA20 with and without PCEs was similar to that of OPC, confirming that PCEs mainly affect the hydration of OPC. In the case of CM replacement, PCEs addition displayed less influence on the overall heat of hydration of the paste, particularly after 24h. The apparent little impact of PCEs on the hydration of CM20, coupled with the significant influence on the rheological behaviour of the paste, indicates that PCEs may be prevented from excessive surface interaction with OPC for dispersion by other mechanisms than surface adsorption by the CM. This could also be accounted for by the unavailability of effective plasticizers to adsorb onto the silicate phases of the OPC, rendering little or no retardation in the hydration of this phase. Additional reaction between OPC clinker and CM may prevent the inhibition of hydration by the PCEs. Further elaborations will be discussed in high SCM loading system (CM60). The trend in the packing densities of the FA20 pastes was similar to that for OPC pastes as a result of the apparent inertness of the FA particles. The packing densities of CM20 with added PCE, on the other hand, showed clear deviation from that in OPC systems. Instead of measuring varying packing density due to different PCEs addition, all three CM20 with different PCEs registered a consistent packing density. This indicated that the PCEs were not available to interact with water molecules, thus confirming that they are not surface bound on the CM particles.

396 SP-302-29

Properties of 60% FA/CM blended cements with added PCEs Next, the impact of high SCM replacement of up to 60% was investigated (Fig. 3). As observed before, FA60 was highly fluid (τd = 80 Pa versus 230 Pa for neat OPC pastes), whereas no rheological values were measured for CM60 as it was no longer flowing. When 0.2%bwob of PCEs were added to these blended cements, drastic reduction in Bingham viscosity and dynamic yield stress were observed as expected for FA60, whereby the latter falls to almost 0Pa regardless of PCE employed. In the presence of 0.2%bwob of PC3, sedimentation of FA60 occurred, significantly over dispersing the binder system since the effective PCE/OPC ratio was increased due to insignificant FA interaction. In the case of CM60 pastes, addition of 0.2%bwob of PCE showed little flow, except when PC1 was added. This implied two things. Firstly, CM possessed a strong affinity for water and PCEs. Secondly, PC1 which possess intermediate side chain appeared to be the best dispersing PCEs for CM systems as CM appear to have a lower affintiy (thus increased dispersing power). No possible explanation could be given at this moment in time for the low Bingham viscosity of the CM60 paste with only 0.2%bwob of PC1. For closer analysis, 0.4%bwob of PCEs were dosed to the CM60 paste. Immediately, all CM60 pastes displayed fluidity, whereby Bingham viscosity was the greatest for PC1 > PC2 > PC3 (0.52, 0.47 and 0.43 Pa·s) respectively. The dynamic yield stress was highest for PC3, followed by PC2 and finally PC1 (300, 200 and 150 Pa), confirming that PCEs behave very differently in CM blended cements than in OPC. A likely explanation for this behavior could be attributed to the method PCEs being consumed by CM, as discussed earlier on in this paper. For the CM employed in this investigation, some uncalcined portions may remain (or partly calcined retaining layered structures) and these can be susceptible to absorb PCEs via their hydrophilic PEO side chains.13,14 In this way, CM has a higher capacity for PC3 which possessed very long side chains, thus excess PC3 (0.4%bwob) was needed to exert their dispersing effectiveness. On the other hand, PC1 could function better than PC3 at 0.2%bwob, but were also consumed by CM at higher CM replacement, normalising the performance of PC1 and PC3 to be similar to that in the OPC system. In case of PC2, the performance of this PCE was less affected as it possessed shorter side chains that were less vulnerable to intercalation in CM layers. However, due to the inherent dispersing power of PC2, it remained to be inferior to the other two PCEs in initial dispersion regardless of dosage. Further analysis on the cumulative heat of hydration revealed that at high replacement level of FA (60%), the FA60 pastes followed a similar trend as the OPC or FA20 pastes up to 6h hydration. Therefore, due to the high PCE/OPC ratio, the excessive amount of PCE blocked any active sites on the OPC particles, particularly alite phases, thus prolongating the induction period, and probably greatly reducing the early strength of concretes during the first 24 h. PC2 appeared to affect the hydration least, potentially due to the amount of ‘free surfaces’ it can generate relative to the other 2 PCEs as discussed before. In the case of CM (60%), the cumulative heat of hydration of CM60 was comparable with and without PCE addition, contrary to that observed in OPC or FA blended cement systems (Fig. 6). This can be attributed to the decreased effectiveness of PCE to delay hydration as they are consumed within the CM layers, while at the same time, the hydration of CM blended cement underwent a different mechanism route, which was dominated by the hydration of CM (Fig. 7). It appeared that a threshold amount of CM must be present in

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  397

Fig. 6–Cumulative heat evolved from the hydration of (a) FA60 and (b) CM60 pastes in absence and presence of 0.2%bwob of varying PCEs as a function of time (w/b = 0.36).

Fig. 7–Rate of heat of hydration per mass of binder of CM20 and CM60 with 0.2 and 0.4%bwob of PC2, w/b = 0.36 (Solid lines represent 0.4%bwob, whereas dotted line represent 0.2%bwob of PC2 respectively). the system before alternation of the mechanism occur. Nonetheless, the early heat of hydration (and thereby strength of concretes prepared with CM in presence of PCE) was thus not affected by the presence of these polymers. This finding is in line with that observed in the rheological measurements, where consumption of PCEs by CM caused a decrease in their dispersing ability. When PCEs were consumed by the CM through possible intercalation of the PEO side chains, the PCEs were no longer available to block the silicate surfaces on OPC and thus hydration was not retarded. The packing densities of blended cements with 60% SCM replacement can be observed in Table 2. In present of PCEs, the packing density of FA60 tended towards the characteristics of a pure FA paste. The high packing density of the FA60 and 0.2%bwob PC1 (0.632) can be accounted for by the high quantity of suspended FA particles in the supernatant, giving a ‘false’ high packing density. Therefore, a threshold balance between the amount

398 SP-302-29

of FA and plasticizer must be present to ensure a homogeneous and non-bleeding system. Packing densities of CM60, on the other hand, remained similar to that of CM20, whereby all pastes possessed similar packing densities regardless of PCE type added. This remained relatively constant even when the dosages of PCE were doubled to 0.4%bwob, showing the high affinity of PCEs by CM. CONCLUSIONS The effect of PCE superplasticizers on the rheological and hydration properties of FA and CM blended cements at 20% and 60% replacement was investigated and quantified. FA and CM are two very different materials which affected the rheological and early age hydration properties of the blended cements and the effectiveness of PCEs in totally different manner. In general, FA improved the rheology of cements, with or without PCEs while CM, on the contrary, decreased the flow of the cement pastes, and was particularly detrimental to the performance of PCEs possessing long side chains. FA exerted its presence in the blended cement through dilution effect. It was generally inert to hydration and its affinity to PCE is strongly outweighed by that from OPC. Therefore, the main influencing factor from FA in blended cement, regardless of replacement level was an increase in PCE/OPC ratio, allowing a reduction of PCE dosage needed in the dispersion of FA blended cements. On the other hand, CM was found to affect the rheological properties of the cement by first possessing a high demand of water, and also through an active role in hydration reactions with OPC. Upon addition of PCEs, PC3 failed to perform due to its rapid consumption by CM through the intercalation of PEO side chains of these PCEs between the incompletely calcined layers of the CM, rendering them ineffective as plasticizers. Thus, PC1 which possess only PEO side chains with intermediate length performed better in this investigation. A close look of these two SCM materials revealed that a new ternary binder could be possible, where OPC, FA and CM are added together to form a complementary system, where the different components act to compensate for the performance deficiency of one another, even in the presence of superplasticizers such as PCEs. AUTHOR BIOS Serina Ng is a Research Scientist at SINTEF Building and Infrastructure, Department of Materials and structures, Concrete group, Trondheim, Norway, since January 2013. She graduated with a PhD in construction chemistry at the Technical University of Munich, Germany. Her field of research includes construction materials, blended cements, admixtures and additives, insulation; ranging from the understanding of their fundamental properties to their applications. Harald Justnes is Chief Scientist at SINTEF Building and Infrastructure, Department of Materials and structures, Concrete group, Trondheim, Norway, where he has worked since 1985. His research covers the chemistry of cement, concrete, admixtures and additives from production, through reactivity, to durability. He graduated with PhD in inorganic chemistry at the Institute of Materials Technology, Inorganic Chemistry,

Study on the Rheology of Fly Ash Versus Calcined Marl Blended Cements with Polycarboxylate-Based Superplasticizers  399 NTNU, Trondheim, Norway, and has been Adjunct Professor in “Cement and Concrete Chemistry” at the same institute since 2000. ACKNOWLEDGMENTS The authors wish to express their gratitude to COIN consortium for the funding of this project. REFERENCES 1. World Cement Production, (2011). European Cement Association (2012). 2. Oss, H. G. v., Cement – Annual Publication 2012, In: Mineral Commodity Summaries, Cement Statistics and Information, USGS — U.S. Geological Survey (2012). 3. Cement Industry Energy and CO2 Performance – “Getting the numbers right”, World Business Council for Sustainable Development, The cement sustainability initiative (2011). 4. Roadmap, C. T., 2009 – Carbon emissions reduction up to 2050, World Business Council for Sustainable Development (2009). 5. Cheung, J.; Jeknavorian, A.; Roberts, L.; and Silva, D., “Impact of admixtures on the hydration kinetics of Portland cement,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1289-1309. doi: 10.1016/j.cemconres.2011.03.005 6. Lesti, M.; Ng, S.; and Plank, J., “Ca2+ Ion – mediated interaction between microsilica and polycarboxylate comb polymers in model cement pore solution,” Journal of the American Ceramic Society, V. 93, No. 10, 2010, pp. 3493-3498. doi: 10.1111/j.1551-2916.2010.03901.x 7. Habbaba, A., and Plank, J., “Surface Chemistry of Ground Granulated Blast Furnace Slag in Cement Pore solution and Its impact on the effectiveness of polycarboxylate Superplasticizers,” Journal of the American Ceramic Society, V. 95, No. 2, 2012, pp. 768-775. doi: 10.1111/j.1551-2916.2011.04968.x 8. Alonso, M. M.; Palacios, M.; and Puertas, F., “Compatibility between polycarboxylate-based admixtures and blended-cement pastes,” Cement and Concrete Composites, V. 35, No. 1, 2013, pp. 151-162. doi: 10.1016/j.cemconcomp.2012.08.020 9. Miller, K. T.; Melant, R. M.; and Zukoski, C. F., “Comparison of the compressive yield response of aggregate suspensions: pressure filtration, centrifugation, and osmotic consolidation,” Journal of the American Ceramic Society, V. 79, No. 10, 1996, pp. 25452556. doi: 10.1111/j.1151-2916.1996.tb09014.x 10. Williams, D. A.; Saak, A. W.; and Jennings, H. M., “The influence of mixing on the rheology of fresh cement paste,” Cement and Concrete Research, V. 29, No. 9, 1999, pp. 1491-1496. doi: 10.1016/S0008-8846(99)00124-6 11. Miller, K. T.; Melant, R. M.; and Zukoski, C. F., “Comparison of the compressive yield response of aggregate suspensions: pressure filtration, centrifugation, and osmotic consolidation,” Journal of the American Ceramic Society, V. 79, No. 10, 1996, pp. 25452556. doi: 10.1111/j.1151-2916.1996.tb09014.x 12. Thomas, M.; Optimizing the Use of Fly Ash in Concrete, Concrete, Portland Cement Association, (2007) pp. –24 IS548. http://www.cement.org/docs/default-source/ fc_concrete_technology/is548-optimizing-the-use-of-fly-ash-concrete.pdf?sfvrsn=4.

400 SP-302-29

13. Ng, S., and Plank, J., “Interaction mechanisms between Na Montmorillonite clay and MPEG-based polycarboxylate superplasticizers,” Cement and Concrete Research, V. 42, No. 6, 2012, pp. 847-854. doi: 10.1016/j.cemconres.2012.03.005 14. Svensson, P. D., and Hansen, S., “Intercalation of smectite with liquid ethylene glycol — resolved in time and space by synchrotron X-ray diffraction,” Applied Clay Science, V. 48, No. 3, 2010, pp. 358-367. doi: 10.1016/j.clay.2010.01.006

SP-302-30

A Study on the Cement Compatibility of PCE Superplasticizers by A. Lange and J. Plank It is well established among concrete producers that specific cements seem to be incompatible with most PCE products, thus causing excessive PCE dosages or even a total failure of the PCE. This effect is commonly referred to as “cement incompatibility” of PCE. The study here investigates the reasons for such incompatibility. First, it was found that only cements which upon contact with water instantaneously form large amounts of ettringite exhibit such incompatibility phenomenon. Their characteristics are elevated C3A content (> 7 wt.-%) and high initial heat of hydration. Second, it was observed that PCEs strongly influence early ettringite crystallization by acting as morphology modifying agent. Most PCEs transform common micro meter-sized ettringite into nano-sized crystals which bring about a huge surface area and thus require abnormal dosages of PCE to achieve dispersion. Such nano-sized particles can be separated from the cement paste by centrifugation where it appears as a viscous, gel-like top layer. From five chemically different PCE polymers tested, one (a modified APEG type) was identified as extremely compatible with all cement samples, whereas three other ones (two conventional MPEG and one APEG type) exhibited pronounced incompatibility with C3A rich cements. An IPEG PCE showed moderate cement compatibility. The phenomenon of cement incompatibility occurs only when the PCE is present in the mixing water, and disappears when PCE is added in delayed mode. Finally, a simple and quick test to identify cement–PCE incompatibility is proposed. Keywords: admixture; adsorption; cement dispersion; ettringite; high range water reducer; morphology; nano-size; polycarboxylate. INTRODUCTION Polycarboxylate based superplasticizers (PCEs) are widely used in concrete manufacturing, especially for high performance concrete such as ultra-high strength concrete or self-compacting concrete.1-3 PCEs were invented more than 30 years ago in Japan4 and since then spread all over the world with highly increasing production rates, especially in China. The first PCE products represented copolymers of sodium methacrylate and methoxypolyethylene glycol methacrylate macromonomers (the so-called MPEG type), but nowadays many other PCE products are on the market including allyl ether (APEG), methallyl ether (HPEG) and isoprenyl ether (IPEG or TPEG) based PCEs.5 401

402 SP-302-30

PCEs possess an anionic polymer trunk with polyglycol pendants. Their dispersing force originates from the ability to adsorb on sites of the cement particles exhibiting positive surface charge, i.e. mainly on the aluminate and ferrite phases (C3A and C4AF) and their initial hydration products such as e.g. ettringite or monosulfo aluminate.6 Cement clinker is composed of several mineral phases with different surface charges which causes strong agglomeration.7 The dispersing effectiveness of PCEs first is owed to charge neutralization of the positively charged domains and second to the steric hindrance provided by the polyglycol side chains.8 Without superplasticizer, a significant part of the mixing water is entrapped between the cement agglomerates, thus causing a stiff consistency of the mortar or concrete. PCE superplasticizers efficiently break down these agglomerates and free the entrapped water, thus inducing high flowability of the mixture. Compared to other superplasticizers including polycondensates, the key feature of PCEs is their outstanding dispersing performance at low water-to-cement ratios. However, applicators sometimes have experienced an incompatibility phenomenon of PCE with certain cements, whereby the admixture dosage became unreasonably high. In some cases it was totally impossible to disperse such cements with PCEs at all.9 It is well known that clay impurities can impede the performance of PCE superplasticizers,10 but even in clay-free concretes, high dosages were observed when cements containing a high amount of C3A were used. Early investigations indicated that such negative effect was owed to interaction of the PCE with early cement hydration products.11,12 Especially in ultra-high strength concrete (UHSC) where the PCE commonly is added to the mixing water to reduce mixing time, incompatibility between PCEs and cement was frequently observed13. In this study, compatibility of several structurally different PCE superplasticizers with nine specifically selected ordinary Portland cements (OPCs) was analyzed with respect to the dispersing performance of the PCEs in cement paste, initial heat of hydration within the very first minutes of cement hydration and via extraction of the initial hydration products from the cement pastes. Furthermore, the effect of point of addition time of PCE was tested to clarify how fast these initial hydration products are formed in the cement paste. The initial hydration products were identified using X-ray diffraction, elemental analysis and thermogravimetry. From this data, interaction between the PCE superplasticizers and the initial hydration products was elucidated and an explanation for the incompatibility phenomenon was sought. Finally, the impact of PCE superplasticizers on the morphology of ettringite synthesized from solution was studied to confirm the results obtained in cement. RESEARCH SIGNIFICANCE The incompatibility phenomenon of PCEs with certain cements is addressed in this study. For optimum cost-effectiveness, PCE dosages should be as low as possible, but it is applicators experience that sometimes unusually high dosages are needed in some systems. This phenomenon is analyzed with respect to the influence of PCEs on initial cement hydration products such as ettringite. It was found that almost all PCEs act as morphology modifying agent for ettringite, resulting in nano-sized ettringite particles which possess a huge surface area and high PCE uptake capacity.

A Study on the Cement Compatibility of PCE Superplasticizers  403

EXPERIMENTAL PROCEDURES PCE Samples— A total of five PCE samples was studied. Two of them (samples MPEG-7 and MPEG-25) were methacrylic acid-co-ω-methoxy polyethylene glycol (MPEG) methacrylate ester polymers with side chains holding 7 and 25 EO units respectively. The molar ratio of methacrylic acid and MPEG-MA ester was adjusted in such a way that optimal dispersing performance (i.e. lowest dosage) was achieved. For MPEG-7, the MA: MPEG-MA ratio was 1.2:1 while for MPEG-25 it was 3:1. Furthermore, two PCE samples (denominated as APEG-34 and APEG-34AM) were allyl ether (APEG) based polycarboxylates with a side chain holding 34 EO units. APEG-34AM contained allyl maleate as additional monomer. The molar ratios between allyl ether, maleic anhydride and allyl maleate were 1:1:0 (for APEG-34) or 1:1:1 (for APEG-34AM) respectively. The fifth PCE sample (IPEG-25) was a copolymer of acrylic acid and ω-hydroxy polyethylene glycol isoprenyl ether (IPEG) with 25 EO units in the side chain. Again, the molar ratio was adjusted for maximum dispersing performance and was 2.7:1 (acrylic acid: IPEG ether). All polymers were self-synthesized. For preparation of the MPEG PCEs, two solutions were prepared, one containing the MPEG-MA ester, methacrylic acid and mercaptopropionic acid as chain transfer agent (solution I, 50 wt.-% in DI water). The other solution contained the radical initiator sodium persulfate (3 mol.-% relative to the monomers, 1 wt.-% in DI water, solution II). The reactor was equipped with stirrer, nitrogen inlet and thermometer and charged with a small amount (~30 mL, 1 fl. oz.) of DI water. The vessel was then heated to 80 °C (176 °F). Both solutions were fed continuously into the reactor, solution I within 3 hours and solution II within 4 hours. After completion of the addition, the mixture was stirred for one more hour, cooled and neutralized with 30 wt.-% NaOH solution. In preparation of the APEG-PCEs, all monomers were placed into a round bottom flask equipped with stirrer, heated to 90 °C (194 °F) and polymerized in bulk. Benzoyl peroxide was used as radical initiator. Initiator dosage was 4 mol.-% with respect to the monomers. The radical initiator was added uniformly as powder over 1.5 hours. When the reaction was finished, DI water was added to yield an aqueous solution with a solid content of ~50 wt.-%. After cooling to room temperature, the samples were neutralized with 30 wt.-% NaOH solution. The IPEG PCE was synthesized by placing the IPEG ether macromonomer as 50 wt.-% solution into the reactor as above and heated to 80 °C (176 °F). Again, acrylic acid was fed into the reactor within 3 hours and the initiator solution (same amount as for MPEG PCEs) within 4 hours. The final procedure was identical to that for the MPEG PCEs. The chemical structures of all synthesized PCE polymers are shown in Figure 1. All polymer samples were used without further purification. The polymers were characterized by gel permeation chromatography using Waters 2695 separation module equipped with 2414 RI detector (Waters) and a Dawn EOS 3 angle light scattering detector (Wyatt Technology). A dn/dc of 0.135 mL/g was used to calculate molar masses relative to polyethylene oxide.14 The analytical results are shown in Table 1. Cement— Nine different cements were selected for this study. All of them were ordinary Portland cements (CEM I) ranging from 32.5 N to 52.5 R. Additionally, an API Class G oil

404 SP-302-30

Figure 1–Chemical structures of the polycarboxylate samples synthesized for this study Table 1–Analytical properties of the synthesized PCE samples Polymer MPEG-7 MPEG-25 APEG-34 APEG-34AM IPEG-25

Mw [g/mol] 44,300 67,600 63,100 78,400 93,800

Mn [g/mol] 23,300 28,200 22,500 24,500 36,100

PDI [Mw/Mn] 1.9 2.4 2.8 3.2 2.6

nEO [mol] 7 25 34 34 25

A Study on the Cement Compatibility of PCE Superplasticizers  405

Table 2–Oxide analysis of cement samples as determined via X-ray fluorescence analysis Cement sample CEM I 32.5 N CEM I 32.5 R CEM I 42.5 N CEM I 42.5 R CEM I 42.5 R HS CEM I 52.5 N CEM I 52.5 R CEM I 52.5 R HS API Glass G

CaO wt.-%

SiO2 wt.-%

Al2O3 wt.-%

Fe2O3 wt.-%

MgO wt.-%

Na2O wt.-%

K2O wt.-%

SO3 wt.-%

TiO2 +MnO +P2O5 wt.-%

67.1 62.8 64.7 65.7 65.8 67.1 66.4 66.3 62.8

20.3 19.1 19.7 24.2 22.4 23.5 20.5 21.0 21.8

3.77 5.10 3.81 3.95 3.51 3.98 5.26 3.36 4.27

4.18 3.18 2.37 1.55 4.08 1.25 2.80 5.43 5.30

0.64 2.94 1.38 0.59 0.76 0.58 1.19 0.74 0.69

0.14 0.27 0.15 0.09 0.15 0.01 0.04 0.17 0.14

0.64 0.90 0.76 0.66 0.63 0.77 0.54 0.35 0.56

2.76 3.26 2.73 2.47 2.33 2.66 2.98 2.08 2.34

0.47 2.45 4.40 0.79 0.34 0.15 0.29 0.57 2.10

Table 3–C3A and CaO contents of cements, initial hydration energies and amount of ettringite gel formed in the presence of 1.0% of PCE sample MPEG-25 (1 J/g = 0.43 Btu/lb) Cement sample CEM I 32.5 N CEM I 32.5 R CEM I 42.5 N CEM I 42.5 R CEM I 42.5 R HS CEM I 52.5 N CEM I 52.5 R CEM I 52.5 R HS API Glass G

C3A wt.-% 1.90 9.85 6.00 7.57 1.77 8.04 8.90 1.56 1.20

CaO (free) wt.-% 0.11 0.97 0.15 0.04 0.06 0.03 0.10 0.27 0.10

Spec. surf. area (Blaine) [cm2/g] 3,319 3,589 3,889 4,972 3,280 3,299 4,803 4,332 2,998

5 min hydration energy [J/g] 1.71 14.83 11.66 22.18 0.88 5.49 25.74 0.22 1.99

100 min hydration Amount of ettringite energy gel [J/g] 2.86 no gel 23.88 very large 13.65 large 23.95 very large 3.11 no gel 9.67 medium 31.28 very large 5.65 no gel 2.75 no gel

well cement was tested. The X-Ray fluorescence (Oxide) analysis of the cements is exhibited in Table 2. Their phase compositions were analyzed by quantitative XRD (Rietveld) and their free lime content was assessed using the Franke method.15 The cement samples were selected such as to cover a broad range of C3A contents from 1.2 to 9.85 wt.-% (see Table 3). PCE performance test— The dispersing force of the PCE samples was determined using a “mini slump test” carried out as follows: First, a constant water-to-cement ratio (w/c) of 0.3 was chosen. At this w/c ratio, the dosages of the polymers required to reach a spread of 26 ± 0.5 cm (10.2 inch) were determined. Generally, the polymers were dissolved in the required amount of mixing water placed in a porcelain cup. The amount of water contained in the PCE solution was subtracted from the amount of mixing water. Next, within 5 seconds, 350 g (12.35 oz.) of cement were added to the mixing water and thoroughly agitated manually for a total period of 4 minutes. The cement paste was then poured

406 SP-302-30

into a Vicat cone (height 40 mm (1.57 inch), top diameter 70 mm (2.76 inch), bottom diameter 80 mm (3.15 inch)) placed on a glass plate and the cone was lifted vertically. The resulting spread of the paste was measured twice, the second measurement being in a 90° angle to the first and averaged to give the spread value. When PCE was added in a delayed mode, then the cement was first mixed with water only, and the PCE solution was added after 0.5, 1 or 2 minutes respectively. Total mixing time was kept constant at 4 minutes. Initial hydration energy—The initial hydration energy was determined utilizing an isothermal calorimeter (TAM Air Thermometric, Järfälla, Sweden). For this purpose, 4 g (0.141 oz) of cement and 2 g (0.07 oz.) of mixing water were placed into the calorimeter in two separate compartments and were left to equilibrate until the calorimeter showed a constant base line. Using the titration cell, the mixing water was added to the cement placed in the vial and the resulting cement paste was carefully mixed for 1 minute. The measurement proceeded over 100 minutes and the hydration energy released during this period was captured. All measurements were then repeated under adiabatic conditions using a Dewar flask (inner height: 9 cm (3.54 inch); inner diameter: 7 cm (2.76 inch)) in which the temperature increase caused by the hydrating cement paste was determined. In this case the measurement was stopped after 5 minutes. Both methods yielded very similar trends. Separation of initial ettringite gel— It was found that the initial hydration products (mainly ettringite) can be separated from the cement via centrifugation (20 minutes at 10.000 g). There, they form an aqueous top layer above the cement residue. For the separation, the same cement paste like in the “mini slump test” was prepared, except that here the w/c ratio was increased to 0.5 and PCE dosages were set constant at 1% by weight of cement. At this high w/c ratio and PCE dosage, the cement pastes exhibit strong bleeding. After mixing, the cement paste rested for 5 minutes. Then the bleeding water (which is where a significant portion the nano-sized ettringite accumulates) was taken up using a pipette and centrifuged as described above. A gel-like top layer with ~ 85 wt.-% solids content was obtained. The gel layer was carefully removed and dried at 30 °C under atmospheric pressure. It is highly important to mildly dry the product, as ettringite easily can dehydrate to metaettringite which is X-ray amorphous.16 The resulting colorless powder was analyzed using X-ray diffraction (Bruker AXS D8 Advance, Karlsruhe, Germany), elemental analysis (Elementar vario EL, Hanau, Germany) and thermogravimetry (Netzsch STA 409 TG-MS, Selb, Germany). Synthetic ettringite—Ettringite was precipitated from solution by dissolving 0.255 g (0.009 oz.) of Al2(SO4)3 ∙ 18 H2O in 10 mL (0.338 fl. oz.) of water. To this solution, 100 mL (3.38 fl. oz.) of a saturated Ca(OH)2 solution (~1.6 g/L) were rapidly added under vigorous stirring. Instantaneous precipitation of ettringite was evident as the solution turned turbid.17 Influence of the PCE samples on ettringite crystallization was analyzed by dissolving the PCE in the aluminum sulfate solution. The amount of PCE was kept constant at 0.35 g (0.0123 oz.) (~ 10 wt.-% of the amount of ettringite formed). One half of the resulting dispersion was centrifuged for 20 minutes at 10.000 g and carefully dried under atmospheric pressure. The resulting colorless powder was analyzed using X-ray diffraction. The other part of the dispersion was immediately analyzed for particle size distribution using dynamic light scattering (ZetaSizer Nano ZS, Malvern Instruments, Worcestershire, United Kingdom).

A Study on the Cement Compatibility of PCE Superplasticizers  407

Table 4–PCE dosages required in cement paste to achieve a 26 cm (10.2 inch) slump flow (w/c ratio = 0.30) Cement sample CEM I 32.5 N CEM I 32.5 R CEM I 42.5 N CEM I 42.5 R CEM I 42.5 R HS CEM I 52.5 N CEM I 52.5 R CEM I 52.5 R HS API Glass G

MPEG-7 0.25 > 1.00 0.35 > 1.00 0.15 0.22 > 1.00 0.31 0.10

MPEG-25 APEG-34 APEG-34AM [PCE dosages in % by weight of cement] 0.11 0.21 0.10 > 1.00 > 1.00 0.25 0.16 0.29 0.14 > 1.00 > 1.00 0.22 0.09 0.14 0.08 0.12 0.20 0.12 0.90 0.45 0.31 0.12 0.30 0.12 0.06 0.11 0.05

IPEG-25 0.10 0.51 0.13 0.35 0.08 0.10 0.26 0.11 0.05

EXPERIMENTAL RESULTS AND DISCUSSION Cement dispersion effectiveness of PCEs—The dispersing force of the PCE samples was determined in a “mini slump test” using cement paste. In this test, a paste was prepared at a w/c ratio of 0.3 and the PCE dosage was adjusted to yield a slump flow of 26 ± 0.5 cm (10.2 inch). The PCE was dissolved in the mixing water prior to cement addition. This test was carried out for the nine different cement samples and for all five PCE polymers. Table 4 shows significant variations in PCE dosages required for the different cements. Dosages of > 1% suggest that this PCE is not able to effectively disperse the cement. It is also apparent that PCE sample APEG-34AM exhibits the most consistent performance (i.e. best cement compatibility), followed by IPEG-25. For example, for APEG-34AM the dosages vary between 0.05 and 0.31%, and for IPEG-25 between 0.05 and 0.51% by weight of cement (bwoc), thus indicating relatively stable performance with different cement samples. Opposite to this, the three other PCE samples exhibit much higher variations in their dosages, and they cannot fluidize at least two cement samples at all. The data from Table 4 also clearly indicate that cement samples CEM I 32.5 R and CEM I 42.5 R seem to be most difficult to disperse. Both cements exhibit relatively high C3A contents (9.85% and 7.51% respectively). Accordingly, for the PCE compatibility of these cements the order as follows was established (C3A content in wt.-%): API Class G (1.20%) > CEM I 42.5 R HS (1.77%) > CEM I 52.5 N (8.04%) > CEM I 32.5 N (1.40%) > CEM I 52.5 R HS (1.56%) > CEM I 42.5 N (6.00%) > CEM I 52.5 R (8.90%) > CEM I 42.5 R (7.57%) > CEM I 32.5 R (9.85%). Yet, the C3A content alone does not seem to represent the only criteria for incompatibility, because one cement sample (CEM I 52.5 N) possesses a high amount of C3A (8.04%), but still is easy to disperse. Furthermore, in additional tests the dispersing performance was assessed at delayed addition of the PCEs. As cement, the “difficult” sample CEM I 42.5 R (C3A content 7.57%) was used. In the following, results for PCE sample MPEG-25, a PCE with low cement tolerance, will be described. Its dosage was fixed at 0.20% bowc. At first, the PCE was dissolved in the mixing water, as described above. In the other 3 tests, the PCE was added in a delayed mode after 0.5, 1.0 and 2.0 minutes respectively. Figure 2 shows the development of slump flow with increasing delay of PCE addition. It became apparent that the phenomenon of cement incompatibility disappears completely when the PCE sample is added only 1 min

408 SP-302-30

Figure 2–Slump flow of CEM I 42.5 R (w/c = 0.3) depending on the point of addition of 0.20% bwoc of MPEG-25 PCE after the cement has been mixed with water. Similar results were obtained when the other “incompatible” PCE samples were tested with the “difficult” cements. These findings allow to conclude that the phenomenon of incompatibility between specific cements and PCE polymers can be avoided when PCE is added in a delayed mode. However, especially when formulating mortars or concretes possessing low w/c ratios, the preferred mode of addition includes placing the PCE in the mixing water to achieve faster wetting of the cement and aggregate particles, thus to reduce mixing time. Consequently, still a need exists for PCE polymers which work well with a broad diversity of cement samples, even when the PCE is placed in the mixing water. Another conclusion from this test is that the incompatibility phenomenon seems to be linked to processes occurring during the very first seconds of cement hydration. This gave reason to study the initial heat release from hydration of the cement samples. Initial heat of hydration of cements—The initial hydration energy released by the cements was determined in an adiabatic calorimeter for the first 5 minutes of hydration and in an isothermal calorimeter for the first 100 minutes of hydration. The results are presented in Table 3. The amounts of heat measured using the isothermal calorimeter are generally slightly higher than those obtained under adiabatic conditions as hydration time was longer in the isothermal test. For both instruments, a similar trend was found. The heat flow curves obtained from the isothermal test are presented in Figure 3. Note that there, the curves for cement samples CEM I 32.5 N, CEM I 42.5 R HS and API Class G oil well cement are almost identical and therefore cannot be distinguished. In the first 5 minutes of cement hydration, the initial hydration energies vary greatly between the cements tested. They range from almost 0 J/g (0 Btu/lb) for CEM I 52.5 R HS to more than 25 J/g (10.75 Btu/lb) for sample CEM I 52.5 R. Furthermore, it became evident that the amount of initial hydration energy correlates with PCE dosages: Cement samples which release substantial amounts of heat are more difficult to disperse and require higher PCE dosages. Generally, the initial hydration energies of cements possessing a very low C3A content (< 2% in CEM I 32.5 N, CEM I 52.5 R HS, CEM I 42.5 R HS and API Class G) are extremely low (0.22 – 1.99 J/g (0.09 – 0.86 Btu/lb)), suggesting that the amount of heat released is linked to

A Study on the Cement Compatibility of PCE Superplasticizers  409

Figure 3–Heat flow curves of neat cements (w/c = 0.5) hydrated over 100 minutes in an isothermal calorimeter. the C3A content. However, one cement exhibiting a high C3A content of 8.04% (sample CEM I 52.5 N) exhibits a rather low initial hydration energy of 5.48 J/g (2.34 Btu/lb). This indicates that the amount of immediately soluble sulfate or the modification of C3A present plays an additional role here. Additionally, the impact of the specific surface area a cement sample on the amount of ettringite gel was considered. However, the fact that cement sample CEM I 32.5 R possesses a relatively low specific surface area (see Table 3), but still produces copious amounts of ettringite, suggests that the surface area of a cement sample plays an insignificant role here, if any. Further investigations will be necessary to clarify this point. The measurements on the initial hydration energy were repeated in the presence of the PCE samples to determine whether any of them affects the amount of heat released. However, no significant differences could be observed. This finding led the authors to conclude that the incompatibility phenomenon is linked to hydration products which crystallize almost immediately after the cement comes into contact with water. Analysis of colloidal gel—By coincidence it was observed that pastes prepared from “difficult” cements and PCEs form a white, gelous top layer when centrifuged for 20 minutes at 10.000 g (Figure 4). Especially PCEs exhibiting poor cement compatibility such as MPEG-25 or APEG-34 produced a significant volume of this layer. Also, the amount of gel was especially high for the samples CEM I 32.5 R, CEM I 42.5 R and CEM I 52.5 R which before were identified as those which are difficult to disperse (Table 3). The amount of gel produced by PCE MPEG-25 and the cement samples are exhibited in Table 3. There, it is evident that cements which are difficult to disperse produce large amounts of this gel, and vice versa. For example, from 300 g (10.59 oz) of “difficult” cement sample CEM I 32.5 R ~ 6 g (0.21 oz) of a transparent, waxy gel with a solids content of ~ 85 wt.-% is obtained. Figure 4 displays how the amount of gel can vary with cement composition (i.e. the C3A content) when PCE polymer MPEG-25, a PCE with low cement tolerance, is used. Relative to the different PCE samples, the amount of gel was very similar among the MPEG-type PCEs and for APEG-34 (= the less cement compatible PCEs). Whereas, IPEG-25 produced slightly less gel and for APEG-34AM, hardly any gel formation was detectable. Thus, based on the amounts of gel produced, the order as follows with respect to cement compatibility was established: APEG-34AM > IPEG-25 >> APEG-34 > MPEG-25

410 SP-302-30

Figure 4–Comparison of the amounts of ettringite gel obtained as top layer after centrifugation of cement paste (w/c = 0.5) prepared in the presence of 1% bwoc of PCE sample MPEG-25 using (from left to right): CEM I 52.5 R HS (no gel), CEM I 52.5 N (medium amount) and CEM I 32.5 R (large amount).

Figure 5–XRD pattern of the gel layer formed in the presence of PCE sample MPEG-25 atop CEM I 32.5 R paste after centrifugation. > MPEG-7. This order suggests that even within the same chemical group of PCE, different cement compatibilities can occur. XRD analysis revealed that this gel contains pure ettringite (Figure 5), as was confirmed also by elemental analysis of the dried powder. However, in the presence of 1% bwoc of PCE sample MPEG-25, a carbon content of 7.73 wt.-% was detected in the gel, thus indicating significant adsorption of PCE on the surface of colloidal ettringite. Furthermore, TG-MS analysis of the ettringite gel containing PCE was performed (Figure 6). Dehydration of ettringite occurs in the temperature range between 100 °C and 180 °C (212 – 356

A Study on the Cement Compatibility of PCE Superplasticizers  411

Figure 6–TG-MS analysis of ettringite gel produced from CEM I 32.5 R hydrated in the presence of 1 wt.-% PCE sample MPEG-25. °F) while the PCE begins to decompose above 200 °C (392 °F) as evidenced by the mass signal of m/z = 44 corresponding to the release of CO2. The opaque appearance of the gel indicates that the ettringite crystals must be extremely small and thus can adsorb large amounts of PCE. Generally, an ettringite gel was not observed when no PCE was present in the cement paste. Apparently, the PCE polymers either hinder ettringite formation substantially or – more likely – act as morphology modifying agent which reduce the size of the early ettringite crystals to colloidal or even nano scale. Furthermore, formation of the ettringite gel was studied at delayed addition of PCE sample MPEG-25. However, no ettringite gel was obtained even when the PCE was added only 30 seconds after the cement paste had been mixed. This again instigates that the incompatibility phenomenon is derived from an instantaneous interaction between cement und PCE which occurs as soon as water is added. To summarize, the amounts of gel formed correlate well with the PCE dosages required in the performance test and also with the initial hydration energies. In other words, cements which are difficult to disperse are characterized by high initial hydration energy, large amounts of colloidal ettringite gel and abnormal PCE dosages. The experiments also suggest that specific PCE molecules can strongly impact ettringite crystallization. They seem to act as morphology modifying agent which greatly reduce the aspect ratio of the early ettringite crystals. To investigate this aspect, the crystallization of pure synthetic ettringite was studied in the presence and absence of the PCE polymers. Pure ettringite crystals were preferred over an ettringite gel obtained from cement to eliminate any effects derived from cement impurities. For applicators of PCEs who want to find out whether poor performance of a PCE product is owed to cement incompatibility, the simple test as follows is recommended: mix your cement sample at w/c = 0.5 with water containing 1 wt.-% of your PCE polymer, then centrifuge the bleeding water of the paste for 20 min at 10.000 g. If a significant

412 SP-302-30

amount of opaque gel appears at the top of the centrifugate, then incompatibility between this particular cement and the admixture is confirmed. This test allows assessing whether mal-performance of a PCE product is owed to incompatibility with cement, or derives from other factors such as e.g. the presence of clay impurities etc. Impact of PCE on ettringite crystallization— Following the method of Struble,17 synthetic ettringite was prepared via precipitation from aluminum sulfate solution combined with a saturated calcium hydroxide solution. Note that this synthesis route for ettringite is quite different compared to the formation mechanism in cement, still valuable results relating to the effect of PCE superplasticizers on ettringite morphology can be retrieved. Ettringite precipitates were collected from both without PCE as well as in the presence of PCE. XRD analysis confirmed that pure ettringite was formed in all cases. Visual inspection of the samples already revealed substantial differences in the crystal sizes of the ettringite. The product obtained in the absence of PCE settled quickly and appeared as milky white suspension, whereas samples produced in the presence of PCEs, especially of the MPEG PCEs and of APEG-34, were stable, did not settle, exhibited only slight turbidity or were almost transparent. Concentration of these samples via centrifugation yielded the same viscous gel as found before when using cement pastes. Next, the suspensions holding the precipitated ettringite were characterized by dynamic light scattering. For the sample where no PCE was present, using laser granulometry particle sizes of ~10 µm (3.9∙10-4 inch) (d50 value) were detected. Opposite to this, the sizes of the ettringite samples obtained in the presence of PCEs were much smaller: MPEG-7 718 ± 14 nm (2.8∙10-5 inch); MPEG-25 613 ± 11 nm (2.39∙10-5 inch); APEG-34 1,536 ± 38 nm (5.99∙10-5 inch); IPEG-25 1,720 ± 93 nm (6.71∙10-5 inch) and APEG-34AM 2,050 ± 65 nm (7.99∙10-5 inch). Remarkably, except for MPEG-7 the sizes of the ettringite crystals correlate very well with the PCE dosages required for difficult cements, the amounts of gel formed and the initial hydration energies. More specifically, PCEs which fail with “difficult” cements produce large amounts of a gel holding particularly small colloidal ettringite crystals. MPEG-7 does not fit well into this order because per se it presents a relatively poor dispersant, as a result of its short side chain. This polymer always requires substantially higher dosages than the other PCE samples tested here. The experiments on precipitation of pure ettringite from solution also revealed that ettringite formation and precipitation is instantaneous, i.e. turbidity immediately occurs when aluminum sulfate and calcium hydroxide come into contact. Even more, the precipitation was complete within ~ 10 seconds. This observation is consistent with the results on delayed PCE addition to cement paste (Figure 2). There, the negative effect on PCE performance was observed only within the first seconds. After 1 min of hydration, the negative impact from the “difficult” cement sample had vanished. CONCLUSIONS The study suggests that the incompatibility phenomenon occurring between specific PCE polymers and certain cements relies on modification of the ettringite crystal morphology. Based on these findings, “incompatibility” between a PCE and cement sample is likely to occur when the following conditions exist:

A Study on the Cement Compatibility of PCE Superplasticizers  413

a) The cement sample exhibits an elevated C3A content (> 7 wt.-%) and contains significant amounts of immediately soluble sulfates (alkali sulfates, CaSO4 – hemihydrate) to produce large amounts of ettringite. b) The PCE sample strongly impacts ettringite morphology and produces nano-sized instead of the common meso-sized, large crystals. The nano-sized ettringite possesses a multiple of the surface area of conventional ettringite formed in the absence of PCE, and therefore consumes large amounts of superplasticizer via adsorption. FURTHER RESEARCH The authors acknowledge that this study only represents a first step in understanding the complexity of this incompatibility phenomenon. For example, it will be most useful to investigate why specific PCE molecules interact differently with the different surfaces of ettringite crystals. Molecular simulations evidencing the interaction energies might be useful to establish a correlation between PCE molecular structure and ettringite morphology. Additionally, the nano-sized ettringite crystals need to be characterized using high-resolution electron microscopy. Attempts to achieve this in the present study failed because under the impact of the vacuum and electron beam, the ettringite crystals immediately dehydrated to amorphous meta ettringite. AUTHOR BIOS M. Sc. Alex Lange studied chemistry at Technische Universität München, Germany. Currently, he is a Ph.D. student at the Chair for Construction Chemicals at the same university. His research focuses on the synthesis and working mechanism of polycarboxylate-based superplasticizers and the effect of molecular architecture on PCE performance. Prof. Johann Plank is full Professor at the Institute of Inorganic Chemistry at Technische Universität München, Germany. Since 2001, he holds the Chair for Construction Chemicals there. His research interests include cement chemistry, chemical admixtures, organic inorganic composite and nano materials, concrete, dry mortar and oil well cementing. ACKNOWLEDGMENT The authors greatly thank Clariant and Nippon Oil and Fats Company for generously supplying monomers for the PCE synthesis. A. Lange also wishes to thank the “TUM Center for Advanced PCE Studies” for financing the final part of his Ph.D study. REFERENCES 1. , “. Okamura, H., Ouchi, M., “Self-Compacting Concrete,” Journal of Advanced Concrete Technology, V. 1, 2003, pp. 5-15. 2. Sakai, E.; Kakinuma, Y.; Yamamoto, K.; and Daimon, M., “Relation between the Shape of Silica Fume and the Fluidity of Cement Paste at Low Water to Powder Ratio,” Journal of Advanced Concrete Technology, V. 7, No. 1, 2009, pp. 13-20. doi: 10.3151/ jact.7.13 3. Plank, J.; Schröfl, C.; Gruber, M.; Lesti, M.; and Sieber, R., “Effectiveness of Polycarboxylate Superplasticizers in Ultra-High Strength Concrete: the Importance of PCE

414 SP-302-30

Compatibility with Microsilica,” Journal of Advanced Concrete Technology, V. 7, No. 1, 2009, pp. 5-12. doi: 10.3151/jact.7.5 4 . T. Hirata, “Cement dispersant”, JP1984-18338, 1981. 5. Plank, J., ““PCE Superplasticizers – Chemistry, Application and Perspectives” 18. ibausil,” Weimar, V. 1, Sep. 2012, pp. 91-102. 6. Bonen, D., and Shondeep, S., “The Superplasticizer Adsorption Capacity of Cement Pastes, Pore Solution Composition, and Parameters Affecting Flow Loss,” Cement and Concrete Research, V. 25, No. 7, 1995, pp. 1423-1434. doi: 10.1016/0008-8846(95)00137-2 7. Yoshioka, K.; Tazawa, E.; Kawai, K.; and Enohata, T., “Adsorption characteristics of superplasticizers on cement component minerals,” Cement and Concrete Research, V. 32, No. 10, 2002, pp. 1507-1513. doi: 10.1016/S0008-8846(02)00782-2 8. Yoshioka, K.; Sakai, E.; Daimon, M.; and Kitahara, A., “Role of Steric Hindrance in the Performance of Superplasticizers for Concrete,” Journal of the American Ceramic Society, V. 80, No. 10, 2005, pp. 2667-2671. doi: 10.1111/j.1151-2916.1997.tb03169.x 9. Agarwal, S. K.; Masood, I.; and Malhotra, S. K., “Compatibility of Superplasticizers with different cements,” Construction & Building Materials, V. 14, No. 5, 2000, pp. 253-259. doi: 10.1016/S0950-0618(00)00025-8 10. Lei, L., and Plank, J., “A Study on the Impact of Different Clay Minerals on the Dispersing Force of Conventional and Modified Vinyl Ether Based Polycarboxylate Superplasticizers,” Cement and Concrete Research, V. 60, 2014, pp. 1-10. doi: 10.1016/j. cemconres.2014.02.009 11. Prince, W.; Espagne, M.; and Aïtcin, P.-C., “Ettringite formation: A crucial step in cement superplasticizer compatibility,” Cement and Concrete Research, V. 33, No. 5, 2003, pp. 635-641. doi: 10.1016/S0008-8846(02)01042-6 12. Aitcin, P.-C.; Jolicoeur, C.; and MacGregor, J. G., “Superplasticizers: How they work and why they occasionally don’t,” Concrete International, V. 16, 1994, pp. 45-52. 13. .] Schröfl, C., Gruber, M., Plank, J., “Structure-performance relationship of polycarboxylate superplasticizers based on methacrylic acid esters in ultra high performance concrete”, in: Fehling, E.; Schmidt, M.; and Stürwald, S., Ultra High Performance Concrete (UHPC) – Second International Symposium on Ultra High Performance Concrete, Kassel, 2008, 383-390. 14. Teresa, M.; Laguna, R.; Medrano, R.; Plana, M. P.; and Tarazona, M. P., “Polymer characterization by size-exclusion chromatography with multiple detection,” Journal of Chromatography. A, V. 919, No. 1, 2001, pp. 13-19. doi: 10.1016/S0021-9673(01)00802-0 15. Franke, B., “Bestimmung von Calciumoxyd und Calciumhydroxyd neben wasserfreiem und wasserhaltigem Calciumsilikat,” Zeitschrift fur Anorganische und Allgemeine Chemie, V. 247, No. 1-2, 1941, pp. 180-184. doi: 10.1002/zaac.19412470115 16. Zhou, Q.; Lachowski, E. E.; and Glasser, F. P., “Metaettringite, a decomposition product of ettringite,” Cement and Concrete Research, V. 34, No. 4, 2004, pp. 703-710. doi: 10.1016/j.cemconres.2003.10.027 17. Struble, L. J., “Synthesis and Characterization of Ettringite and Related Phases”, Proceedings of the 8th International Congress on the Chemistry of Cement, Rio de Janeiro, Brazil, 6, 1986, 582-588.

SP-302-31

Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination by Hessam AzariJafari, Mohammad Shekarchi, Javad Berenjian, and Babak Ahmadi Use of pozzolanic materials such as natural zeolite as portland cement replacement helps to reduce amount of CO2 emission due to clinker production. Natural zeolite also improves mechanical and durability properties of concrete. It is common to use natural zeolite as a rheological modifying admixture in flowing concrete. However, many cases were reported that zeolite blended cements showed severe workability loss. The object of the analysis is to investigate compatibility of different chemical-based superplasticizers and effect of superplasticizers’ combination on workability retention of concrete made with zeolite blended cement. The results show that combination of lignosulfonate admixture with naphthalene and polycarboxylate based admixture not only reduces the superplasticizer’s demand to achieve certain workability retention, but also helps to reduce slump loss. Keywords: natural zeolite; superplasticizers’ combination; superplasticizer’s demand; workability loss. INTRODUCTION Since ancient times, natural pozzolans such as zeolite have been broadly used in limebased mixtures. Nowadays, zeolite is consumed as a supplementary cementitious material (SCM) in many countries that have enough resources of zeolitic tuffs. Natural Zeolite (NZ) is a three dimensional structured pozzolan containing alumino-silicate mineral. NZ can enhance microstructural properties of cement paste by decreasing the size and amount of C-S-H crystals.1,2 Earlier studies have shown that Portland cement replacement by NZ (5% to 55%, by mass) contributes to consumption of calcium hydroxide which is formed during hydration of portland cement.3-6 Pozzolanic activity and effects of zeolite powder on the hardened properties of concrete have been widely investigated by researchers1,2,7-11 On the fresh properties of zeolite blended cements, Ramezanianpour et al.8 reported that high NZ replacement level (30%) promotes segregation resistance of low portland cement SCC mixtures, which was measured by sieve segregation test. A study by Ranjbar et al.9 showed that use of NZ (up to 20%) contributes to satisfying the SF2 criteria for SCC 415

416 SP-302-31

Table 1– Physical Properties of cementitious materials. Portland cement SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O L.O.I Specific Gravity (kg/m3) Blaine (m2/kg)

Chemical composition (%) 21.25 3.38 3.56 63.1 1.96 1.71 0.216 0.56 1.87 3140 320

NZ 67.79 13.66 1.44 1.68 1.20 0.50 2.04 1.42 10.23 2200 320

mixtures (slump flow diameter of 660 to 750 mm [26 to 29.52 in.]). Criterion of SF2 is satisfactory for normal applications. Similar results were reported by Cioffi et al.12 at 40% NZ replacement level. Still, influence of NZ inclusion on the fresh state behavior of cementitious materials has been investigated only to a basic level. Some observations regarding the decreased workability, higher viscosity, and accelerated workability loss (in the case of NZ incorporation) are reported by some researchers.1,6,8,9,13 RESEARCH SIGNIFICANCE While the compatibility issue has been investigated for various pozzolan-admixture combinations, it has not been carried out for NZ pozzolan. In fact, some observations with respect to the higher workability loss of NZ included materials are often reported, but detailed investigation on the involved parameters has been neglected by the researchers. In addition, effect of lignosulfonate- based admixture (LS) as partial replacement of naphthalene sulfonate formaldehyde (NSF) and Polycarboxylate ether (PC) on workability loss and dosage on plain concrete and zeolite-incorporated concrete were studied. EXPERIMENTAL INVESTIGATION Based on the earlier research works on Iranian NZ, 10% was selected as the cement replacement level in zeolite-incorporated concrete (NZ concrete).9,10,13 Also, each of the chemical admixtures was employed within the concentration range recommended by the producer. Materials The cement used in this study was ASTM C150 Type II portland cement. The clinoptilolite type NZ was from the quarries in the north of Semnan, central region of Iran. The physical and chemical compositions of the cementitious materials are listed in Table 1. The concrete aggregates were composed of two fractions: 0-6 mm [0-0.23 in.] and 6-19.5 mm [0.23- 0.77 in.]. The lower fraction (0-6 mm [0-0.23 in.]) which was used in concrete mixtures had specific gravity of 2.67, water absorption of 2.8% and fineness modulus of

Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination  417

Fig. 1 - Sieve analysis of combined aggregates 3.06. The coarse aggregate is crushed type. Specific gravity and water absorption of the coarse aggregate (6-19.5 mm [0.23- 0.77 in.]) were 2.75 and 1.8%, respectively. The sieve analysis of total aggregates of the mixtures is presented in Fig. 1. In order to investigate influence of superplasticizers on zeolite blended concrete, Ligno sulfonate (LS), naphthalene sulfonate formaldehyde (NSF), and Polycarboxylate ether (PC) were selected. The superplasticizers were the most widespread chemical admixtures in 3 different generations. As they performance are more effective in different w/c range, it might be a suitable choice for studying the effect of NZ workability retention. The properties of these three commercially available admixtures are presented in Table 2. Each admixture was used single and combined. The combined admixtures included LS and NSF or PC at different ratios. The ratios of superplasticizers’ combination were 30-70, 50-50, and 70-30. Mixture proportions of concrete including different dosages and combinations of chemical admixtures are listed in Table 3. Mixtures including no zeolite (Plain concrete) as control mixtures were made for comparison purposes. These mixtures involve only portland cement as binder. Items of investigation The initial slump for the mixtures was kept constant at 175±20 mm [6.89±0.78 in.]. It should be noted that the slump was measured according to ASTM C14314 at the end of mixing procedure and also at 15, 30, 45, and 60 minutes (after initial water-cement contact). The fresh concrete was placed in slump cone. The cone was lifted and the height of fresh concrete cone was measured at two locations using a caliper. The average value was used as the slump. Depending on the water reduction ability of the superplasticizing

418 SP-302-31

Table 2– Properties of chemical admixtures Superplasticizer Type poly-carboxylate Ether (PC) Naphthalene Sulfonate Formaldehyde (NSF) Lignosulfonate (LS)

Specific Gravity 1.14

pH 6.3

Solids Content (%) 40

Chloride Content (%)

1.20

7.2

50

< 0.01

1.17

6.1

50

Table 3- Mixture proportions of concrete mixtures

Mixture type Non-zeolite mixtures(Plain concrete) Zeolite mixtures(NZ concrete)

Stone Aggregate 0-6 mm(Kg/ 6-19.5 mm(Kg/ m3) m3)

Portland cement(Kg/ m3)

Natural zeolite(Kg/ m3)

W/C Ratio

350

-

0.42

889

892

315

35

0.42

883

885

agents, one w/cm ratio was selected for each of the three admixtures, and adequate dosages were added to obtain the desired initial slump (175±20 mm [6.89±0.78 in.]). EXPERIMENTAL RESULTS AND DISCUSSION Required dosage of single and combined superplasticizers The effect of LS combination with NSF and PC with different ratios of 30-70, 50-50, and 70-30 on dosage was investigated. Fig. 2 and 3 demonstrate the effect of superplasticizers’ combination on dosage of the admixtures for plain and NZ concrete, respectively. It should be noted that the measured slump before addition of the chemical admixtures was 75±5 mm [2.95±0.2 in.] for plain concrete. At w/c 0.42, the measured slump for NZ mixtures was 20±5 mm [0.79±0.2 in.]. This reduction of slump is usually justified by the honeycomb like structure of NZ crystals which have extremely small pores and channels, varying in size from 3×104–4×104 μm. These characteristics enable NZ to absorb water by over 30% of its dry weight.1,15 Comparing SP dosage reveals that incorporation of NZ has increased superplasticizers demand. The most increase in SP demand belongs to PC which is nowadays the most common type of superplasticizer for use in special concretes such as SCC. The PC dosage has been increased by 0.5% at NZ concrete compared to plain concrete. Increasing the rate of superplasticizers demand might be related to high specific area, which is existed in NZ. Therefore, a high portion of superplasticizers might be absorbed by NZ. As a result, the superplasticizer demand increased. Dispersing mechanisms in PC superplasticizer are based on steric hindrance and electrostatic repulsion of copolymers included in PC.16 The electrical charge induced on cementitious materials particles with PCs is lower than for NSF and LS.3 As it was shown in Fig. 2 and 3, inclusion of LS as partial replacement of NSF and PC reduced the required dosage in NZ concrete. The combination of 70% PC and 30% LS could result in the lowest superplasticizer demand in NZ mixture. Compared to PC, combination of 70% PC and 30% LS have resulted 50% reduction in dosage. It should

Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination  419

Fig. 2 – Required dosage for obtaining Initial slump of 17.5±1 cm in plain concrete

Fig. 3 – Required dosage for obtaining Initial slump of 17.5±1 cm in NZ concrete be noted that PC and NSF have stronger ability of water reduction. It can be inferred the single use of PC and NSF lead to high amount of adsorption on the cementitious materials particles. The presence of LS could reduce the adsorption of the other superplasticizers by NZ. Therefore, combination of the admixture could result in less superplasticizer demand to achieve a certain workability level. Similar results were reported by Tingshu et al. and Zidong in plain concrete.17,18 Workability retention of single and combined superplasticizers Workability retention of plain concrete and NZ concrete in 1 hour are presented in Fig. 4 and 5, respectively. As it is demonstrated, presence of LS in plain concrete and NZ concrete leads to higher workability retention rather than single use of superplasticizers. Single use of NSF led to higher slump loss, while the workability loss was less severe for PC incorporated mixture. Therefore, single use of NSF in NZ concrete is not recommended in long time agitation and hot environments. These observations could be attributed to the fact

420 SP-302-31

Fig. 4 – Total Slump loss in plain concrete after one hour from batching

Fig. 5 – Total Slump loss in NZ concrete after one hour from batching that NZ decreases the concentration of alkaline ions (Na+, K+, Ca2+) in the pore solution through ion exchange and adsorption.19 So, the amount of alkalinity in pore solution will be decreased. Reports have shown that NSF show more severe slump loss in low-alkali aqueous solutions due to higher adsorption level.20 Therefore, it might be the reason for severe slump loss of NZ concrete incorporating NSF. Workability retention of combined superplasticizers was also investigated. Combination of LS with the other superplasticizers result a significant reduction in slump loss. According to Fig. 4, combination of 30% PC with 70% LS could reduce slump loss 10 cm in one hour in plain concrete (compared to single use of PC). Blending 30% NSF and 70% LS could enhance slump retention for 75 mm [2.95 in.] compared to single use of NSF. Regarding to Fig. 5, blending LS with the other superplasticizers could result in higher slump retention compared to single use of NSF and PC. The observed performance might be related to dual mechanism of LS and PC or NSF. Some authors reported that LS can delay hydration of cement particles significantly.21,22 Retarding effect of LS is a supporting hand for electro static repulsion and steric hindrance of NSF and PC. In NZ concrete, single use of LS reached to the lowest slump loss. High dosage of LS in the mixture (1.2% of cementitious materials weight) provide

Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination  421 higher amount of dispersing materials. In constant amount of cementitious materials, higher dosage of superplasticizers leads to higher amount of superplasticizers in pore solution. Therefore, the higher amount of the superplasticizer at longer time interval can replace with consumed adsorbed molecules. Consequently, the mixture can maintain its workability for longer time. However, there is limitation in use of LS in concrete. Researchers have shown that water reduction ability of LS is restricted.21 As a result, it might not be a suitable choice for NZ incorporated mixtures with lower w/cm ratios. Blending 30% PC and 70% LS can be the second choice for compensating slump loss in NZ concrete. CONCLUSIONS Based on the results of this experimental investigation of single and combined use of three different chemical- based superplasticizers (lignosulfonate, naphthalene sulfonate formaldehyde, and polycarboxylate ether) on workability retention of plain and natural zeolite blended concrete, the following conclusions are drawn: 1. Incorporation of natural zeolite in concrete mixtures led to a substantial reduction on initial slump of the concrete due high capacity of NZ water absorption. 2. The obtained results indicated that simultaneous inclusion NSF admixture was not efficient for NZ blended cement. This observation could be related to the high alkali adsorption capacity of natural zeolite particles. 3. Inclusion of LS was an efficient measure to mitigate the workability loss of NZ blended concrete made with NSF and PC admixture. (Especially at incorporation of 70% LS with 30% NSF or PC). 4. Lignosulfonate-based admixture shows the best performance in slump retention of natural zeolite incorporated concrete. It might be due to sufficient amount of free LS in the pore solution to compensate the consumed LS as well as its retarding effect. AUTHOR BIOS ACI member Hessam AzariJafari is a PhD candidate at Civil Engineering Department, University of Sherbrooke, Sherbrooke, Quebec, Canada. He is a member of ACI Committees 212 (Chemical Admixtures) and 237 (Self-Consolidating Concrete). His research interests include chemical admixtures, supplementary cementitious materials, and life cycle assessment of building materials. Mohammad Shekarchi is a professor at construction materials institute (CMI) at University of Tehran, Tehran, Iran. His research interests include durability properties of concrete, fresh properties of concrete and dimensional changes. Javad Berenjian is an assistant professor at Tabari Institute of Higher Education, Babol, Iran. His research interests include self-consolidating concrete and effect of chemical admixtures in concrete. Babak Ahmadi is a PhD candidate at Civil and Environmental Engineering Department, Amirkabir University of Technology. He is also a research assistant at construction materials institute (CMI) at University of Tehran. His research interests include supplementary cementitious materials and durability properties of concrete in harsh environment.

422 SP-302-31

REFERENCES 1. Ahmadi, B., and Shekarchi, M., “Use of natural zeolite as a supplementary cementitious material,” Cement and Concrete Composites, V. 32, No. 2, 2010, pp. 134-141. doi: 10.1016/j.cemconcomp.2009.10.006 2. Dousti, A.; Rashetnia, R.; Ahmadi, B.; and Shekarchi, M., “Influence of exposure temperature on chloride diffusion in concretes incorporating silica fume or natural zeolite,” Construction & Building Materials, V. 49, 2013, pp. 393-399. doi: 10.1016/j. conbuildmat.2013.08.086 3. Burgos-Montes, O.; Palacios, M.; Rivilla, P.; and Puertas, F., “Compatibility between superplasticizer admixtures and cements with mineral additions,” Construction & Building Materials, V. 31, 2012, pp. 300-309. doi: 10.1016/j.conbuildmat.2011.12.092 4. Kocak, Y.; Tascı, E.; and Kaya, U., “The effect of using natural zeolite on the properties and hydration characteristics of blended cements,” Construction & Building Materials, V. 47, 2013, pp. 720-727. doi: 10.1016/j.conbuildmat.2013.05.033 5. Snellings, R.; Mertens, G.; Cizer, Ö.; and Elsen, J., “Early age hydration and pozzolanic reaction in natural zeolite blended cements: Reaction kinetics and products by in situ synchrotron X-ray powder diffraction,” Cement and Concrete Research, V. 40, No. 12, 2010, pp. 1704-1713. doi: 10.1016/j.cemconres.2010.08.012 6. Uzal, B., and Turanlı, L., “Blended cements containing high volume of natural zeolites: Properties, hydration and paste microstructure,” Cement and Concrete Composites, V. 34, No. 1, 2012, pp. 101-109. doi: 10.1016/j.cemconcomp.2011.08.009 7. Narasimhulu, K.; Gettu, R.; and Babu, K., “Beneficiation of Natural Zeolite through Flash Calcination for Its Use as a Mineral Admixture in Concrete,” Journal of Materials in Civil Engineering, 2013 8. Ramezanianpour, A. A.; Kazemian, A.; Sarvari, M.; and Ahmadi, B., “Use of Natural Zeolite to Produce Self-Consolidating Concrete with Low Portland Cement Content and High Durability,” Journal of Materials in Civil Engineering, V. 25, No. 5, 2013, pp. 589-596. doi: 10.1061/(ASCE)MT.1943-5533.0000621 9. Ranjbar, M. M.; Madandoust, R.; Mousavi, S. Y.; and Yosefi, S., “Mousavi, Yasin, S., Youosefi, S., “Effects of natural zeolite on the fresh and hardened properties of selfcompacted concrete,” Construction & Building Materials, V. 47, 2013, pp. 806-813. doi: 10.1016/j.conbuildmat.2013.05.097 10. Valipour, M.; Pargar, F.; Shekarchi, M.; and Khani, S., “Comparing a natural pozzolan, zeolite, to metakaolin and silica fume in terms of their effect on the durability characteristics of concrete: A laboratory study,” Construction & Building Materials, V. 41, 2013, pp. 879-888. doi: 10.1016/j.conbuildmat.2012.11.054 11. Valipour, M.; Pargar, F.; Shekarchi, M.; Khani, S.; and Moradian, M., “In situ study of chloride ingress in concretes containing natural zeolite, metakaolin and silica fume exposed to various exposure conditions in a harsh marine environment,” Construction & Building Materials, V. 46, 2013, pp. 63-70. doi: 10.1016/j.conbuildmat.2013.03.026 12. Cioffi, R.; Colangelo, F.; Caputo, D.; and Liguorim, B., “Influence of High Volumes of Ultra-Fine Additions on Self-Compacting Concrete,” ACI Publication, vol. SP-239-9, 2006. 13. Sabet, F. A.; Libre, N. A.; and Shekarchi, M., “Mechanical and durability properties of self consolidating high performance concrete incorporating natural zeolite, silica fume

Enhancing Workability Retention of Concrete Containing Natural Zeolite by Superplasticizers’ Combination  423 and fly ash,” Construction & Building Materials, V. 44, 2013, pp. 175-184. doi: 10.1016/j. conbuildmat.2013.02.069 14. ASTM, “ASTM C143-12: Standard Test Method for Slump of Hydraulic-Cement Concrete ” 2012. 15. M. FA, Mineralogy and geology of natural zeolites. Newyork: Reprint of Mineralogical Society of America’s Reviews in Mineralogy, 1993. 16. H. Liu, Pang, H., Ou, J., Zhang, L., Dai, Y., Liao, B., “Effect of cross-linked polycarboxylate-type superplasticizers on the properties in cementitious system,” Journal of Applied Polymer Science, vol. 131, 2014. 17. S. F. HE Tingshu, WANG Fuchuan,WANG Huian, “INFLUENCES OF COMBINATION OF SUPERPLASTICIZERS AND RETARDERS ON PROCESS OF CEMENT HYDRATION,” Journal of the Chinese Ceramic Society, vol. 06, 2007. 18. Zidong, L., “Reasearch on polycarboxylates water-reducing admixture combined with modified lignosulfonate,” Ready-Mixed Concrete, vol. 01, 2012. 19. Feng, N., and Hao, T., “Mechanism of natural zeolite powder in preventing alkalisilica reaction in concrete,” Advances in Cement Research, V. 10, No. 3, 1998, pp. 101-108. doi: 10.1680/adcr.1998.10.3.101 20. Kim, B.-G.; Jiang, S.; Jolicoeur, C.; and A[UNKNOWN ENTITY &idie;]tcin, P.-C., “The adsorption behavior of PNS superplasticizer and its relation to fluidity of cement paste,” Cement and Concrete Research, V. 30, No. 6, 2000, pp. 887-893. doi: 10.1016/ S0008-8846(00)00256-8 21. V. S. Ramachandran, “Effect of Sugar-free Lignosulphonates on Cement Hydration,” 1979. 22. Zhen Ping Sun, L. Z., “Xiong Rong Huang, Hui Yang, Liang Liang Shui, “Polycarboxylate-Lignosulfonate Copolymerized High Performance Water Reducing Agent: Preparation and Application,” Advanced Materials Research, V. 687, 2012, pp. 435-442.

424 SP-302-31

SP-302-32

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF by Kazuki Matsuzawa, Daiki Atarashi, Masahiro Miyauchi, and Etsuo Sakai The calcination temperature in cement manufacturing can be reduced by the addition of sulphate and fluoride containing compounds and it is possible that sulphate and fluoride ions elute to the suspension after calcination. This paper describes the influence of sulphate and fluoride ions on the action of polycarboxylate based superplasticizer in cement paste. When the amount of K2SO4 or KF was increased, the viscosity of the cement paste with superplasticizer increased. The amount of adsorbed superplasticizer was decreased by K2SO4 addition but increased by KF addition. The fluidity with polycarboxylate based superplasticizer containing more functional groups was less affected by K2SO4 addition. In contrast to the case of K2SO4 addition, the increase in the degree of viscosity by KF addition was not dependent on the amount of functional groups. The specific surface area increased with K2SO4 or KF addition. Keywords: polycarboxylate based superplasticizer; cement paste; fluidity; sulphate ion; fluoride ion; adsorption of superplasticizer. INTRODUCTION High fluidity concrete is necessary to save labor and reduce the time spent in the construction of concrete. In addition, reduction of the water ratio is required to increase the strength and the durability of the concrete, but sufficient fluidity is required for construction. For consistent fluidity and strength, polycarboxylate based superplasticizers are used as the air-entraining and high range water-reducing admixture. Polycarboxylate based superplasticizers are comb-type polymers that adsorb to the surface of cement particles by their carboxyl functional groups.1 Polycarboxylate based superplasticizer stabilizes the dispersion of particles by the steric hindrance effect of the graft chains, and this type of superplasticizer can produce high fluidity with less dosage than other types of superplasticizer. In addition, the molecular structure of polycarboxylate based superplasticizers can be easily modified by changing the monomers. High temperatures (>1450°C: >1720 K) are necessary for alite production in cement manufacturing, and a large amount of energy is consumed in this calcination process. The amount of energy for the calcination is equal to 80% of all energy consumed in cement 425

426 SP-302-32

manufacturing.2 Furthermore, high temperature calcination results in deterioration of the furnace and NOx emission. Because of these problems, decrease of the calcination temperature and reduction of energy consumption by the addition of sulphate and fluoride compounds have been investigated.3,4 Compounds that can decrease the calcination temperature are separated into two groups based on the mechanism of the temperature decrease. Fluxes enable decrease of the calcination temperature by affecting the liquid phase diagram, amount of liquid, viscosity of the liquid and surface energy. Sulphate containing compounds are potential fluxes. On the other hand, mineralizers enable a decrease of the calcination temperature by affecting the solid phase diagram. Mineralizers widen the temperature range in which the alite can exist to include lower temperatures. Fluorides are potential mineralizers. There have been many studies of the use of fluxes and mineralizers. For example, Raina and Janakiraman added CaSO4 and CaF2 to the raw materials of cement, and evaluated the degree of temperature decrease by the amount of free-lime after calcination.5 In their research, it has been revealed that fluxes and mineralizers bring about a 220°C temperature decrease at most by the addition of 1–2 mass% CaSO4/CaF2 to the raw materials. The reduction of the energy consumption ratio in the manufacturing process is about 2.2% for every 100°C temperature decrease.6 Therefore, the energy reduction ratio is calculated to be about 4.5% for the addition of CaSO4/CaF2 at most. However, when sulphate and fluoride are added to the raw material, it is possible that the sulphate and fluoride ions become soluble by calcination and elute to the suspension. Authors have investigated limestone powder (CaCO3) without hydration reactions as a model system, and it was revealed that the fluidity of the CaCO3 paste with polycarboxylate based superplasticizer considerably decreased by the addition of sulphate and fluoride ions.7 It has been revealed that ions that can form an insoluble salt with the calcium ions, such as sulphate and fluoride ions, decrease calcium ions on the surface of CaCO3 particles. It is thought that polycarboxylate based superplasticizers adsorb by complexing the calcium ions on the surface of particles by their negatively charged functional groups. Therefore, the adsorption of these superplasticizers will be hindered by the decrease of calcium ions on the surface of the CaCO3 particles. This adsorption hindrance effect of inorganic ions decreases the fluidity of the CaCO3 paste with superplasticizers. Although there are many previous studies on CaCO3 paste, the influence of fluoride ions on the fluidity of the cement paste with polycarboxylate based superplasticizer has not been investigated. Moreover, regarding the influence of the sulphate ion, there are only studies of the suspension with a smaller amount of sulphate ions than is needed in low-temperature calcination technology. To perform the low-temperature calcination by mineral salt addition in industry, more effective superplasticizers are desirable and detailed analysis is required about the mechanism of the fluidity change of cement paste by inorganic ions. In this study, authors discuss the influence of K2SO4 and KF addition on the fluidity, amount of adsorbed superplasticizer, specific surface area, and hydration reaction of cement paste with polycarboxylate based superplasticizer. Additionally, authors investigate the mechanism for the fluidity decrease of cement paste with superplasticizer by the sulphate and fluoride ions.

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF  427

Fig. 1– Molecular structure of P-n Table 1– Polymerization ratio and mean molecular weight of P-n α

β

γ

Mean number of graft chains in a molecule

P-10

1

1.0

0.005

47.5

P-34

1

1.0

0.014

13.8

Mw [g/mol] ([lb/mol]) 29100 (64.2) 23100 (50.9)

RESEARCH SIGNIFICANCE This research evaluates the fluidity decrease of cement paste with superplasticizers by inorganic ions from a viewpoint of low-temperature calcination technology and this research can contribute to the performance of low-temperature calcination. There are no other studies of the influence of fluoride ions on the fluidity of the cement paste with polycarboxylate based superplasticizer except for the authors’ studies.8,9 This research discusses the fluidity decrease of cement paste by inorganic ions with the adsorption action mechanism of superplasticizer, and can provide some countermeasures against the fluidity decrease by low-temperature calcination of cement by the addition of inorganic salts. EXPERIMENTAL INVESTIGATION Materials The polycarboxylate based superplasticizers used in this study (P-10 and P-34) were α-allyl-ω-methoxypolyethylene–maleic anhydrite copolymers with graft chains of polyethylene oxide. The number in the superplasticizers’ name indicates the mean polymerization degree (n) of the graft chains. P-10 has shorter graft chains and contains more functional groups per unit mass than P-34. The comb-type polymers may also contain impurities, such as low molecular weight non-grafted polyethylene oxide oligomers. Since these impurities do not act as dispersing agents, their concentration was not included in the total superplasticizer concentration. Figure 1 shows the molecular structure of the polycarboxylate based superplasticizers (P-n). Table 1 shows the monomer polymerization ratio of P-n (α: β: γ), the mean number of the graft chains in one molecule (length of the main chain), and the mean molecular weight of P-n measured by gel permeation chromatography analysis. Table 2 shows the chemical composition ratio of the ordinary Portland cement (OPC) used in this study. Table 3 shows the mineral composition ratio of OPC calculated by Bogue’s equation. For sulphate and fluoride ions addition, K2SO4 (>99.0%, Kanto Chem-

428 SP-302-32

Table 2– Composition of OPC [mass%] CaO 64.96 Na2O 0.32

SiO2 20.94 K2O 0.48

Al2O3 5.45 TiO2 0.27

Fe2O3 2.83 P2O5 0.31

MgO 1.54 MnO 0.08

SO3 2.05

Table 3– Mineral composition of OPC [mass%] 3CaO·SiO2 59

2CaO·SiO2 16

3CaO·Al2O3 10

4CaO·Al2O3·Fe2O3 9

Annotation: These values are calculated by Bogue’s equation.

ical Co., Inc., Japan) and KF∙2H2O (>95.0%, Wako Pure Chemical Industries, Ltd., Japan) were selected because the influence of potassium on the hydration of cement is small. Specimens The cement paste for the measurements was made by adding superplasticizer solution, K2SO4 solution and KF solution to OPC. The mass ratio of water to OPC (W/C) was fixed at 0.32. After mixing by hands with a stainless steel spoon in a rubber cup for 5 min, the cement paste was used as the specimens for the experiments. Methods The fluidity of the ordinary Portland cement paste was measured by a rotational cylinder viscometer (Haake MARS III Z41-TI, Thermo Fisher Scientific K.K, Japan) at 20°C (293 K), and the shear stress was changed linearly from 0.54→200→0.54 Pa (0.000078→0.0290→0.000078 lbf/in2) in 240 s. The fluidity was evaluated by the apparent viscosity at a shear stress of 200 Pa (0.0290 lbf/in2). In the measurement of the amount of superplasticizer adsorbed to the solid, the adsorption time was fixed at 5 min and the temperature was 20°C (293 K). The liquid phase in the paste was separated by centrifuging at 8200 m/s2 (26900 ft/s2) for 10 min. After separation, particles of which diameter is larger than 0.20 µm was removed by filtering from the liquid and the unadsorbed superplasticizer concentration in the liquid was measured using a total organic carbon analyser (TOC-L CSH/CSN, Shimadzu Corporation, Japan). The amount of adsorbed superplasticizer was calculated from the superplasticizer concentration in the initial solution and the liquid phase of the paste. To determine the specific surface area of the solid in the cement paste, the hydration time was fixed at 5 min and the temperature was 20°C (293 K). The hydration was stopped by acetone, and the solid phase was separated from the liquid by centrifuging at 8200 m/s2 (26900 ft/s2) for 10 min. After separation, the solid phase was dried under low pressure (0.01 MPa = 1.5 lbf/in2) for 24 hours at 20°C (293 K), and the surface area of the solid was measured by the Brunauer–Emmett–Teller (BET) method with N2 adsorption using and a surface area analyser (Gemini V2380, Micromeritics, USA). The degassing of specimens before the specific surface area measurement was done in N2 gas flow for 3 hours at 40°C (313 K). Also the solid phase was analyzed by X-ray diffraction (XRD) using the wave length of CuKα1 = 0.15406 nm (6.0654 ninch). To determine the hydration reaction of the cement paste, the heat liberation rate of the cement paste was measured

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF  429

Fig. 2– Relationship between the amount of added a) K2SO4, b) KF and the apparent viscosity using a sandwich-type conduction calorimeter (SCM-12L, Tokyo Riko, Japan). The heat liberation rate was measured from 3 to 90 hours after hydration start at 20°C (293 K). The range of the amount of added K2SO4 was determined as 0-2.18 mass% of OPC and the amount of KF was determined as 0-0.745 mass% of OPC. The amount of sulphate and fluoride ions contained in 2.18 mass% K2SO4 and 0.745 mass% KF of OPC is equal to the amount of sulphate and fluoride ions in 1 mass% SO3 and 0.5 mass% CaF2 of OPC. At this addition ratio, the amount of K2SO4 molecules is 0.125 mol per 1 kg of OPC (0.0567 mol/ lb) and the amount of KF molecules is 0.128 mol per 1 kg of OPC (0.0582 mol/lb). EXPERIMENTAL RESULTS AND DISCUSSION Influence of Sulphate Addition Influence of K2SO4 Addition on Fluidity—Figure 2 a) shows the relationship between the amount of added K2SO4 and the apparent viscosity of the cement paste with polycarboxylate based superplasticizer (P-n). The dosage of P-n was fixed at 0.192 mass% of OPC. As previously reported,8,9 when the amount of added K2SO4 was increased, the apparent viscosity of the cement paste with P-n increases. Comparing P-10 with P-34, the increase in the degree of apparent viscosity with P-10 by K2SO4 addition was smaller than with P-34. Therefore, the fluidity of cement paste with the polycarboxylate based superplasticizer with more functional groups (P-10) was less susceptible to K2SO4 addition. This result is the same as the result of CaCO3, where it has been reported that the fluidity of CaCO3 paste containing polycarboxylate based superplasticizer with more functional groups is less susceptible to inorganic ion addition.1,10 Influence of K2SO4 Addition on the Amount of Adsorbed Superplasticizer—Table 4 shows the relationship between the amount of added K2SO4 and the amount of adsorbed P-n to the solid phase in cement paste. The dosage of P-n was fixed at 0.192 mass% of OPC. Compared with the case of no K2SO4 addition, when a small amount of K2SO4 (0.0626 mol/kg=0.0284 mol/lb) was added, the amount of adsorbed P-n per unit mass solid increased and the amount of adsorbed P-n per unit area decreased. At this concentration, it can be considered that the fluidity decrease was brought by the adsorption hindrance mechanism mainly. However, compared with the case of 1.09 mass% K2SO4 addition,

430 SP-302-32

Table 4– Relationship between the amount of added K2SO4 and the amount of adsorbed P-n per unit mass solid or unit area in cement paste P-10 0.192 mass% +K2SO4 +K2SO4 0.0626 mol/kg 0.125 mol/kg No K2SO4 (0.0284 mol/lb) (0.0567 mol/lb) 0.487 mg/g 0.566 mg/g 0.731 mg/g (0.487 mlb/lb) (0.566 mlb/lb) (0.731 mlb/lb) 0.533 mg/m2 0.468 mg/m2 0.490 mg/m2 2 2 (1.82 mlb/in ) (1.60 mlb/in ) (1.67 mlb/in2)

No K2SO4 0.303 mg/g (0.303 mlb/lb) 0.376 mg/m2 (1.29 mlb/in2)

P-34 0.192 mass% +K2SO4 0.0626 mol/kg (0.0284 mol/lb) 0.383 mg/g (0.383 mlb/lb) 0.306 mg/m2 (1.05 mlb/in2)

+K2SO4 0.125 mol/kg (0.0567 mol/lb) 0.515 mg/g (0.515 mlb/lb) 0.319 mg/m2 (1.09 mlb/in2)

Table 5– Relationship between the amount of added K2SO4 and the specific surface area of the solid in the paste P-10 0.192 mass% P-34 0.192 mass% +K2SO4 +K2SO4 +K2SO4 +K2SO4 0.0626 mol/kg 0.125 mol/kg 0.0626 mol/kg 0.125 mol/kg No K2SO4 (0.0284 mol/lb) (0.0567 mol/lb) No K2SO4 (0.0284 mol/lb) (0.0567 mol/lb) 0.913 m2/g 1.21 m2/g 1.49 m2/g 0.806 m2/g 1.25 m2/g 1.61 m2/g (0.645 Min2/lb) (0.855 Min2/lb) (1.05 Min2/lb) (0.570 Min2/lb) (0.885 Min2/lb) (1.14 Min2/lb)

when a large amount of K2SO4 (0.125 mol/kg=0.0567 mol/lb) was added, the amount of adsorbed P-n per unit area increased. Such a phenomenon when the amount of adsorbed superplasticizer increased with the addition of a large amount of sulphate ions has not been previously reported. Although the amount of adsorbed P-n increased with a large amount of K2SO4, the fluidity of the paste with P-n did not increase with K2SO4 addition, as shown in Figure 2 a). Therefore, the decrease in the fluidity with much K2SO4 addition in cement paste cannot be solely explained by the adsorption hindrance mechanism. Comparing P-10 and P-34, the amount of adsorbed P-10 was larger than the amount of adsorbed P-34. It is supposed that the difference of the adsorbed amount relates to the number of functional groups in the superplasticizer. It can be considered that P-10 has stronger adsorption ability than P-34 because P-10 has more functional groups than P-34. Influence of K2SO4 Addition on the BET Specific Surface Area—Table 5 shows the relationship between the amount of added K2SO4 and the BET specific surface area of the solid in the paste with P-n. Figure 3 shows the XRD pattern of the solid phase in cement paste with 0.125 mol/kg (0.0567 mol/lb) K2SO4 addition. The dosage of P-n was fixed at 0.192 mass% of OPC. When the amount of added K2SO4 was increased, the specific surface area increased. Since Ettringite was found in the paste with K2SO4 addition, it is supposed that the cause of the increase of specific surface area is hydrated products generation by the reaction between calcium aluminate and added sulphate ions. It can be considered that the increase in the amount of adsorbed superplasticizers per unit mass solid as shown in Table 4 is related to the cause of the increasing of specific surface area, resulting in the increasing of the number of adsorption site for superplasticizers. The decrease in fluidity and the increase in the amount of adsorbed superplasticizers on the solid with large amount of K2SO4 addition suggests that superplasticizers adsorbed to the hydrated products, resulting in the concentration of superplasticizers decrease and the amount of adsorbed superplasticizers to the

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF  431

Fig. 3– X-ray diffraction pattern of the solid in cement paste with K2SO4 addition unhydrated cement particles decrease. Therefore, it is supposed that the influence of the hydrated products generation on the fluidity should be considered in the case of much sulphate addition. Influence of Fluoride Addition Influence of KF Addition on Fluidity—Figure 2 b) shows the relationship between the amount of added KF and the apparent viscosity of the cement paste with superplasticizers. The dosage of P-n was fixed at 0.192 mass% of OPC. As previously reported,8,9 when the amount of added KF was increased, the apparent viscosity of the cement paste with P-n increased. In contrast to the case of K2SO4 addition, the increase in the degree of apparent viscosity with P-10 by KF addition was almost equal to the degree with P-34. This result is also different from the result of CaCO3, where it has been reported that the fluidity of CaCO3 paste containing polycarboxylate based superplasticizer with more functional groups is less susceptible to inorganic ion addition.1,10 Influence of KF Addition on the Amount of Adsorbed Superplasticizer—Table 6 shows the relationship between the amount of added KF and the amount of P-n adsorbed to the solid phase in cement paste. The dosage of P-n was fixed at 0.192 mass% of OPC. When the amount of added KF was increased, the amount of adsorbed P-n increased. In particular, when the amount of added KF was 0.745 mass% of OPC, the amount of adsorbed P-10 was 74% of the dosage of P-10. Although the amount of adsorbed P-n increased with KF addition, the fluidity of the paste with P-n decreased with KF addition, as shown in Figure 2 b). This result is also different from the results of CaCO3, where the amount of adsorbed superplasticizer decreased by the adsorption hindrance of inorganic ion addition.11 The adsorption hindrance mechanism cannot explain the decrease in fluidity with KF addition to cement paste with P-10. The results of the fluidity and the amount of adsorbed superplasticizer suggest other mechanisms influence the fluidity.

432 SP-302-32

Table 6– Relationship between the amount of added KF and the amount of adsorbed P-n per unit mass solid or unit area in cement paste P-10 0.192 mass% +KF 0.0641 mol/kg No KF (0.0291 mol/lb) 0.487 mg/g 0.858 mg/g (0.487 mlb/lb) (0.858 mlb/lb) 0.533 mg/m2 0.593 mg/m2 2 (1.82 mlb/in ) (2.03 mlb/in2)

+KF 0.128 mol/kg (0.0582 mol/lb) 1.43 mg/g (1.43 mlb/lb) 0.773 mg/m2 (2.64 mlb/in2)

No KF 0.303 mg/g (0.303 mlb/lb)

P-34 0.192 mass% +KF 0.0641 mol/kg (0.0291 mol/lb) 0.645 mg/g (0.645 mlb/lb)

+KF 0.128 mol/kg (0.0582 mol/lb) 0.928 mg/g (0.928 mlb/lb)

/

/

/

Table 7– Relationship between the amount of added KF and the specific surface area of the solid in the paste

No KF 0.913 m2/g (0.645 Min2/lb)

P-10 0.192 mass% +KF 0.0641 mol/kg (0.0291 mol/lb) 1.45 m2/g (1.02 Min2/lb)

+KF 0.128 mol/kg (0.0582 mol/lb) 1.84 m2/g (1.30 Min2/lb)

Influence of KF Addition on the BET Specific Surface Area—Table 7 shows the relationship between the amount of added KF and the BET specific surface area of the solid in the paste with P-10. The dosage of P-10 was fixed at 0.192 mass% of OPC. When the amount of added KF was increased, the specific surface area increased. It is supposed that the cause of the increase of specific surface area is some types of fine particles generation by the reaction regarding the added fluoride ions. It can be considered that the increase in the amount of adsorbed superplasticizers is related to the cause of the increasing of specific surface area. The decrease in fluidity and the increase in the amount of adsorbed superplasticizers on the solid with KF addition suggests that superplasticizers preferentially adsorbed to the fine particles, resulting in the amount of adsorbed superplasticizers on the cement particles decrease and the fluidity of cement paste decrease. Influence of P-10 Addition on the Hydration of Cement Paste Figure 4 shows the relationship between the dosage of P-10 and the time in which the heat liberation rate of cement paste was largest (T1). In the case of no KF addition, when the dosage of P-10 was increased, T1 increased and the hydration reaction was retarded. In contrast, in the case of 0.745 mass% KF addition to OPC, when the dosage of P-10 was increased, the hydration reaction was not retarded. These results also suggest that P-10 preferentially adsorbed to the fine particles generated by KF addition and the amount of P-10 adsorbed to cement particles decreased with KF addition. CONCLUSIONS Based on the results of this experimental investigation, the following conclusions are drawn:

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF  433

Fig. 4– Relation between the dosage of P-10 and T1 1. Compared with no K2SO4 addition, when a small amount of K2SO4 (0.0626 mol/kg = 0.0284 mol/lb) was added, the apparent viscosity of the cement paste with polycarboxylate based superplasticizer increased. The amount of adsorbed polycarboxylate based superplasticizer to the unit mass solid phase in the cement paste increased by K2SO4 addition but the amount of adsorbed polycarboxylate based superplasticizer per unit area decreased. In contrast, compared with a small amount of K2SO4 addition, when a large amount of K2SO4 (0.125 mol/kg=0.0567 mol/lb) was added, the amount of adsorbed superplasticizer increased and the viscosity of the cement paste increased. The fluidity of the cement paste with the polycarboxylate-based superplasticizer with more functional groups was less susceptive to K2SO4 addition than the polycarboxylate-based superplasticizer with less functional groups. 2. It is suggested that the causes of the fluidity decrease with K2SO4 addition are not only the adsorption hindrance mechanism but also the adsorption of superplasticizer to the hydrated compounds generated by sulphate ions and calcium aluminate. 3. On the other hand, when KF was added to the cement paste with polycarboxylate based superplasticizers, the apparent viscosity of the cement paste and the amount of adsorbed superplasticizer to the solid in the paste increased. This suggests that the adsorption hindrance mechanism by ions does not explain the viscosity increase with KF addition to cement paste. In contrast to the case of K2SO4 addition, the influence of the number of functional groups on fluidity change by KF addition was small. 4. It is suggested that the cause of the fluidity decrease with KF addition was the preferential adsorption of superplasticizer to some types of fine particles generated by fluoride ion addition. AUTHOR BIOS Kazuki Matsuzawa is a doctoral candidate of Metallurgy and Ceramics Science, Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his BS in 2013 from the Tokyo Institute of Technology. His research interests are the action mechanisms of chemical admixtures, the fluidity of cement paste, and chemical reactions in cement.

434 SP-302-32

Daiki Atarashi is an Assistant Professor of Metallurgy and Ceramics Science, Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his BS in 2001, MS in 2003 and Dr. Eng. in 2006 from the Tokyo Institute of Technology. His research interests are the action mechanisms of chemical admixtures, the fluidity of cement paste and the material design of high-recycled-content and reduced-CO2 emission cement. Masahiro Miyauchi is an Associate Professor in Metallurgy and Ceramics Science, Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his PhD. in 2002 from the University of Tokyo. He worked at TOTO Ltd. from 1995 to 2006 and at the National Institute of Advanced Industrial Science and Technology from 2006 to 2011. His research interests include photo-electrochemistry, photocatalysis, solar cells, wet chemical synthesis of nanoparticles, and quantum dots. Etsuo Sakai is a Professor in Metallurgy and Ceramics Science, Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his Dr. Eng. from the Tokyo Institute of Technology in 1979. His research interests include construction chemistry, material recycling and material design of high-recycled-content and reduced-CO2 emission cement. ACKNOWLEDGMENTS The authors would like to thank NOF Corporation for the synthesis of the polycarboxylate based superplasticizers used in this research. REFERENCES 1. Sakai, E.; Atarashi, D.; Kawakami, A.; and Daimon, M., “Influence of Molecular structure of Comb-Type Superplasticizers and Inorganic Electrolytes on the Dispersion Mechanisms of Limestone powder,” SP-217, 2003, pp. 381-392. 2. Qian, Z., “CO2 abatement in the cement industry”, IEA CLEAN COAL CENTRE, CCC/184 ISBN 978-92-9029-504-4, 78 (2011) 3. Yoshikawa, T., “Low-temperature Sintering Technology for Cement Clinker Using Mineralizers and Fluxes,” JOURNAL OF RESEARCH of the TAIHEIYO CEMENT CORPORATION, No., V. 161, 2011, pp. 66-73. (in Japanese) 4. Klemm, W. A.; Jawed, I.; and Holub, K. J., “Effects of calcium fluoride mineralization on silicates and melt formation in portland cement clinker,” Cement and Concrete Research, V. 9, No. 4, 1979, pp. 489-496. doi: 10.1016/0008-8846(79)90046-2 5. Raina, K., and Janakiraman, L. K., “Use of mineralizer in black meal process for improved clinkerization and conservation of energy,” Cement and Concrete Research, V. 28, No. 8, 1998, pp. 1093-1099. doi: 10.1016/S0008-8846(98)00082-9 6. Gardeik, H. O., “Effect of the clinkering temperature on the specific energy consumption in cement clinker burning,” Zement Kalk Gips, V. 34, 1981, pp. 169-174. 7. Sakai, E.; Kawakami, A.; Hamamoto, H.; Honda, S.; Itoh, A.; and Daimon, M., “Influence of Various Types of Inorganic Salts on Dispersion Mechanisms of Comb-Type Polymer Containing Graft Polyethylene Oxides Chains,” Journal of the Cement Society of Japan, V. 108, No. 10, 2000, pp. 904-908. doi: (in Japanese)10.2109/jcersj.108.1262_904

Fluidity Change of Cement Paste with Superplasticizer by K2SO4 and KF  435

8. Matsuzawa, K., Atarashi, D., Miyauchi, M., and Sakai, E., “Influence of Sulphate Ion and Fluoride Ion on the Fluidity of Cement Paste with Polycarboxylate Based Superplasticizer Having Different Molecular Structure”, Cement Science and Concrete Technology Vol. 66, JAPAN CEMENT ASSOCIATION pp.59-64 (2014) (in Japanease) 9. Matsuzawa, K.; Atarashi, D.; Miyauchi, M.; and Sakai, E., “Influence of Potassium Sulphate and Potassium Fluoride on the Fluidity of Cement Paste with Polycarboxylate Based Superplasticizers Having Different Molecular Structure”, The 8th International Symposium on Cement & Concrete Proceedings, Nanjing/China (2013) (CD-ROM) 10. Atarashi, D., Sakai, E., Honda, S., Itoh, A., and Daimon, M., “Adsorption and Dispersion Mechanisms of Comb-type Superplasticizer Containing Grafted Polyethylene Oxide Chains”, Journal of the Ceramic Society of Japan, Supplement 112-1, PacRim5 Special Issue, 112 [5], S1304-S1307 (2004) 11. Sakai, E.; Kawakami, A.; and Daimon, M., “Dispersion mechanisms of comb-type superplasticizers containing grafted poly(ethylene oxide) chains,” Macromolecular Symposia, V. 175, No. 1, 2001, pp. 367-376. doi: 10.1002/1521-3900(200110)175:13.0.CO;2-9

436 SP-302-32

SP-302-33

Cement Recycling System Using Sodium Gluconate by Daiki Atarashi, Yutaka Aikawa, Yuya Yoda, Masahiro Miyauchi, and Etsuo Sakai The objective of this study was to establish a recycling system for cement sludge by hydration control and an evaluation of the residual cement content of sludge. The adsorption behavior of sodium gluconate on cementitious materials was clarified in this study. The saturated absorption behavior between sodium gluconate and alite follows a Langmuir adsorption isotherm. The delayed hydration time depends on the concentration of residual sodium gluconate. When the residual concentration of sodium gluconate was over 0.018–0.020 mass%, cement hydration did not proceed. The relationship between the heat liberation after 1 day and the amount of non-hydrated alite as calculated by X-ray diffractometry is linear. The amount of non-hydrated alite can be estimated by using calorimetric data. Finally, the amount of non-hydrated alite in sludge water can be estimated by using calorimetric data, magnesium hydrate hexahydrate (Mg(NO3)2·6H2O). Keywords: sludge water; sodium gluconate; set retarder; rate of heat liberation; heat liberation. INTRODUCTION In today’s society there is a need to implement low carbon recycling systems and therefore, the reduction and reuse of waste sludge water from the production of ready-mixed concrete at factories is an important issue. Data from 2006 indicate that ~1 million tons of sludge are generated annually in Japan. Sludge is a mixture of cement and very fine aggregate that occurs as a waste at construction sites, or forms when residual recycled concrete is sieved to remove aggregates. Cleaning water that contains a sludge component, which is referred to as sludge water, is formed when water is used to clean equipment such as ready-mixed-concrete mixers or agitators. To reduce the quantity of sludge formation at ready-mixed-concrete factories, Japanese Industrial Standard A 5308 “Ready-mixed Concrete” prescribes a method of using mortar that adheres to the inside of truck agitator drums. This allows for a solid sludge component of up to 3% (the mass of the solid sludge component as a percentage of the unit cement 437

438 SP-302-33

quantity in the concrete mix) to exist in sludge water when it is used as concrete mixing water. However, if sludge water is added, concrete fluidity is reduced and the quantity of water required to produce the concrete increases. This method is not used widely because of its high cost and complicated management of sludge water. The number of ready-mixedconcrete factories that use sludge water as concrete mixing water is low and the amount of industrial waste that they generate after the dewatering process is large. Sludge water contains fine particles that originate from cement that has already been hydrated, and from non-hydrated cement and aggregates. By using a set retarder such as sodium gluconate (GLNa), the hydration reaction of the unhydrated cement in the sludge water can be controlled (Atarashi et al. 2012). The sludge water can then be reused effectively to replace a portion of the new cement when producing ready-mixed concrete. This could reduce the required cement production quantity, which would contribute significantly to reductions in CO2 emissions. A set retarder can delay the hydration of cement within the sludge water. However, few studies exist of the mechanism of action of the effect of set retarders on cement hydration. To achieve more widespread recycling of sludge using set retarders, the authors need to clarify the mechanism of cement hydration control and establish a method to evaluate the residual cement content of sludge. It is necessary to establish a method to determine the amount of non-hydrated cement in sludge water so that this non-hydrated cement can be used as cement. To date, the authors have found that there is a correlation between the percentage of reacted alite (C3S), which contributes to the generation of early strength in cement, and the liberated heat of hydration as obtained from a conduction calorimeter (Atarashi et al. 2012, 2013). A prototype conduction calorimeter that is capable of measuring the heat of hydration of cement has been produced jointly with Tokyo Riko, a calorimeter-manufacturing company (Sakai et al. 1997). It is suggested that the measurement of heat using a calorimeter can be used effectively in cement quality control and inspection (Sakai et al. 2010). In this paper, the authors discuss the adsorption of GLNa and its set-retarding mechanisms. The recycling of cement sludge is discussed, especially hydration control by GLNa and a method to evaluate the residual cement content of sludge. The objective of this research was to formulate a method for the rapid evaluation of the quantity of cement remaining in sludge water whose hydration reaction has been suppressed using GLNa. This could enable the effective reuse of residual cement in sludge water. In the method, the amount of non-reacted cement remaining in the sludge water is determined using a conduction calorimeter. Furthermore, a method for the rapid evaluation of the quantity of non-reacted cement in sludge water, for which the hydration has been suppressed, was developed using a metal salt. The objective was to integrate these research results to establish a new cement recycling system to use the cement contained in sludge water effectively. RESEARCH SIGNIFICANCE Our research could enable the effective reuse of residual cement in sludge water. The amount of unreacted cement remaining in the sludge water is determined using a conduction calorimeter. Furthermore, a method for the rapid evaluation of the quantity of unreacted

Cement Recycling System Using Sodium Gluconate  439

cement in sludge water, for which the hydration has been suppressed, has been developed using a metal salt. Integration of these research results would make it possible to establish a new cement recycling system to use the cement contained in the sludge water effectively. EXPERIMENTAL INVESTIGATION Sample preparation—The water to powder (synthesized C3S and ordinary Portland cement, OPC) ratio was 0.5 and the mixing time was 10 min. The GLNa dosage was 0, 0.1, 0.2, 0.4, 1.0, and 2.0%. Hydration reaction of cement—The hydration ratio of C3S or C3A was calculated using an X-ray diffractometry (XRD) internal standard method. Adsorption and residual concentration of GLNa in the liquid phase—The liquid phase in the paste was obtained by centrifugal suspension. The residual concentration of GLNa in the liquid phase was measured using a total organic carbon analyzer (TOC-5050A, Shimadzu). The adsorbed amounts of GLNa were calculated from the concentration of GLNa in the initial solution and the liquid phase after adsorption testing. A 10-min-adsorption time was used to reach equilibrium. Measurement of heat liberation—The cumulative heat liberation and rate of heat liberation of C3S and OPC with or without GLNa were measured using a multi-channel conduction (sandwich-type) calorimeter (Tokyo Riko Co. Ltd). The time required for the maximum heat liberation rate (T1) was estimated and used to characterize the cement hydration. Measurement of BET specific surface area—After 10 min, hydration was stopped by adding a large amount of acetone and the samples were dried on an aspirator (1.0×104 Pa). The specific surface area of the hydrated cement with various concentrations of GLNa was measured using N2 gas adsorption. Determining the residual quantity of cement in sludge water—Simulated sludge was produced with a water cement ratio (W/C) of 4.0 without the addition of GLNa and magnesium hexahydrate (Mg(NO3)2·6H2O). It was mixed manually for 3 min at 20°C, and then left for 4 h at 20°C. After allowing the hydration reaction to proceed for 1 to 24 h at 20°C, the hydration was stopped, and the quantity of residual cement (C3S) was determined by comparing the areas under the peaks obtained from XRD. Sample hydration was stopped for samples that had been hydrated between 0 and 24 h, and recommenced for 24 h at 20°C. The integrated quantity of the heat of hydration liberated from the cement paste was measured using a twin conduction calorimeter (Tokyo Riko, Tokyo, Japan). By comparing the XRD quantitative method and the calorimetric data, the validity of the method for determining the quantity of residual C3S in the residual cement was determined. Rapid evaluation of the quantity of residual cement in sludge water using metal salt and a conduction calorimeter—The cement paste (produced using OPC for research purposes) had a W/C of 4.0, and a GLNa content of 0–0.2 mass%. Mixing was carried out manually for 3 min at 20°C. GLNa was added after hydration for 1 h at 20°C to produce simulated sludge water in which the hydration was suppressed. Mg(NO3)2·6H2O was added at 0–8.0 mass% with respect to the cement, and the mixture was sampled. The heat of hydration properties of this simulated sludge water were determined using a conduction calorimeter.

440 SP-302-33

Table 1– Chemical and mineral compositions and properties of OPC

OPC

SiO2 21.6 C3S 64.4

*

Chemical compositions (mass %) Al2O3 Fe2O3 CaO MgO Na2O K2O 5.3 2.3 64.5 2.11 0.25 0.55 Mineral compositions* (mass %) Blaine (cm2/g) C2S C3A C4AF 11.6 9.5 8.0 3090

Cl 0.11

SO3 1.8 Density (g/cm3) 3.17

Calculated by Bogue’s equation

Materials C3S was synthesized in the laboratory from industrial raw materials and reagents using an electric furnace. The cement used in this study was OPC with properties as listed in Table 1. GLNa was used as the set retarder, and Mg(NO3)2·6H2O, which has an accelerating effect on the retarded cement, was used as the metal salt (Harada 1995, 1996). EXPERIMENTAL RESULTS AND DISCUSSION Hydration of cement minerals Fig. 1 shows the hydration reaction ratio of C3S and C3A in the initial hydrated cement. The hydration ratio of C3S is ~6–8% at 1 h and no significant change was observed from 1 to 4 h. This phenomenon is explained by the induced mechanism of C3S. C3A has a higher hydration activity and the reaction ratio of C3A was ~30% at 1 h. After 4 h of hydration, the hydration reaction ratios of C3A and C3S were ~30 and 10%, respectively. In other words, 70% of C3A and 90% of C3S still remain as hydration active cement. The authors could reuse this cement effectively to reduce the environmental loading. Adsorption of GLNa on C3S Fig. 2 shows the isotherm of GLNa on C3S. The amount of adsorbed GLNa on C3S (V) increased gradually and the adsorption became saturated at ~0.8 mass% residual concentration of GLNa. The relationship between P/V and residual GLNa concentration (P) is linear (figure omitted, (Atarashi et al. 2012)). The adsorption of GLNa on C3S is thought to be a Langmuir-type adsorption and the cross-sectional area of a GLNa molecule as calculated from the experimental results was 0.303 nm2 under saturation. Hydration control of C3S with GLNa Fig. 3 shows the relationship between the concentration of residual GLNa and the time of maximum heat liberation rate (T1). The retarded hydration time is also dependent on the residual GLNa concentration. This result indicates that the retardation time can be controlled by controlling the residual GLNa concentration. Hydration was retarded for 220 h with a 0.05% residual concentration of GLNa in the liquid phase. The hydration of C3S was retarded for 40 h with a 0.019% residual concentration of GLNa in the liquid phase.

Cement Recycling System Using Sodium Gluconate  441

Fig. 1– Reaction ratio of C3S and C3A during initial cement hydration

Fig. 2– Adsorption isotherm of GLNa on C3S Hydration control of OPC with GLNa The relationship between residual GLNa concentration and the specific surface area of the hydrated cement after 4 h hydration is shown in Fig. 4. The specific surface area of the

442 SP-302-33

Fig. 3– Comparison of T1 of GLNa on C3S OPC did not change when the residual GLNa in the liquid phase was increased. When the residual concentration of GLNa was greater than 0.018–0.020 mass%, the cement hydration did not proceed. Development of method to determine the quantity of residual cement in sludge water using calorimetric measurement An investigation was carried out to establish a method to determine quantitatively the amount of residual cement in sludge water using a conduction calorimeter. Fig. 5 shows the relationship between the quantity of residual (Non-reacted) cement (C3S) in the sludge water and the cumulative heat of hydration liberated during the 24-h-period after recommencing the hydration. The cumulative heat of hydration of the non-hydrated cement is expressed as a relative heat of hydration with the cumulative heat of hydration in 24 h being 100. As the quantity of residual cement in the sludge water increased, the relative heat of hydration in 24 h increased, and a correlation was found between the two. The conventional quantitative determination of the amount of non-hydrated cement remaining in cement must be measured using an XRD internal reference method, which requires skill in measurement and time. The measurement of the heat of hydration as proposed in this research enables the quantity of residual cement in sludge water to be estimated simply and in ~24 h.

Cement Recycling System Using Sodium Gluconate  443

Fig. 4– Relationship between concentration of residual GLNa and specific surface area of hydrated cement after 4 h

Fig. 5– Relationship between residual ratio of C3S and relative heat liberation in 24 h

444 SP-302-33

Fig. 6– Effect of GLNa and Mg(NO3)2·6H2O on the heat liberation of cement hydration Investigation of rapid method to determine the quantity of cement in sludge water using magnesium nitrate If it were possible to estimate the quantity of residual cement in sludge water rapidly, it may be possible to use the cement in the sludge water from ready-mixed-concrete factories effectively and thereby reduce the environmental load. Fig. 6 indicates the example data of effect of GLNa and Mg(NO3)2·6H2O on the heat liberation of cement paste. Hydration is retarded by adding of GLNa. And by using Mg(NO3)2·6H2O, retardation of cement hydration is disappeared. Fig. 7 shows the effect of Mg(NO3)2·6H2O on the heat of hydration properties of sludge water to which GLNa has been added. The time T1, which indicates the maximum rate of heat liberation of plain paste, is ~11 h as shown by the dashed line on the graph. The T1 for plain paste increases significantly with the addition of GLNa. As reported previously (Song et al. 2004, 2008, Atarashi et al. 2012), this occurs because GLNa is absorbed onto the cement surface thereby suppressing hydration and making it possible to control the reaction of the cement by controlling the GLNa dosage. For example, when this dosage is 0.05 and 0.1 mass%, T1 is 14 and 29 h, respectively, and at 0.15 and 0.20 mass%, T1 is more than 50 h. With Mg(NO3)2·6H2O addition, T1 decreases as the dosage increases. For example at a GLNa dosage of 0.1 mass%, when 1.0 and 2.0 mass% Mg(NO3)2·6H2O are added, the T1 is 15 and 11 h, respectively. The latter value is almost the same as the T1 value of the plain paste without the addition of GLNa and Mg(NO3)2·6H2O. It has therefore been shown that by sampling a portion of sludge water in which hydration has been prevented by GLNa and Mg(NO3)2·6H2O addition, it is possible to eliminate the suppression of cement hydration. This rapid method to estimate the amount of nonreacted C3S is required for practical use of sludge water.

Cement Recycling System Using Sodium Gluconate  445

Fig. 7– Influence of Mg(NO3)2·6H2O on T1 CONCLUSIONS In this paper, the authors discuss the adsorption of GLNa and its set-retarding mechanisms. The authors also discuss the recycling of cement sludge, especially hydration control by GLNa and a method for the evaluation of the residual cement content of sludge. A method has been established for the rapid evaluation of the quantity of cement remaining in sludge water whose hydration reaction has been suppressed using GLNa. The results are as follows. The adsorption behavior of GLNa on cementitious materials was clarified. The saturated absorption behavior between the GLNa and C3S follows a Langmuir adsorption isotherm. The delayed time of hydration depended on the concentration of residual GLNa. When the residual concentration of GLNa was over 0.018–0.020 mass%, cement hydration did not proceed. The relationship between the heat liberation after 1 day and the non-hydrated amount of C3S as calculated by the XRD method is linear. The authors can estimate the amount of non-hydrated C3S by using calorimetric data. Finally, the authors can estimate the amount of non-hydrated C3S rapidly, by using calorimetric data and Mg(NO3)2·6H2O. AUTHOR BIOS Daiki Atarashi is an Assistant Professor in Metallurgy and Ceramics Science at the Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his B.S. in 2001, M.S. in 2003, and Dr. Eng. in 2006 from the Tokyo Institute of Technology, Tokyo, Japan. His research interests are the action mechanisms of chemical

446 SP-302-33

admixtures, the fluidity of cement paste, and the material design of high recycled content cement and low CO2 emission cement. Yutaka Aikawa is a Researcher of Metallurgy and Ceramics Science at the Graduate School of Science and Engineering, Tokyo Institute of Technology, Japan. He received his Dr. Sci. in 1993 when he worked at Taiyo Yuden Co., Ltd. His research interests include the theory of void fractions in particle systems and theoretical cement hydration. Yuya Yoda is a Researcher at Shimizu Co., Ltd., Shimizu Institute of Technology, Center for Structural and Production Engineering. He received his B.S. in 2009 and M.S. from Tokyo Institute of Technology, Japan in 2011. His research interests are the relationship between the heat of hydration and the reaction rate of high-recycled-content cement and low-CO2-emission cement. Masahiro Miyauchi is an Associate Professor in Metallurgy and Ceramics Science at the Graduate School of Science and Engineering, Tokyo Institute of Technology. He received his B.S. in 1993 and M.S. in 1995 from Tokyo Institute of Technology, Tokyo, Japan. He received his Ph. D. from the University of Tokyo, Tokyo, Japan in 2002. His research interests include photo-electrochemistry based on semiconductor nanomaterials for the reduction of environmental load. Etsuo Sakai is a Professor in Metallurgy and Ceramics Science at the Graduate School of Science and Engineering, Tokyo Institute of Technology. He received his Dr. Eng. from the Tokyo Institute of Technology, Tokyo, Japan in 1979. His research interests include construction chemistry, material recycling and material design of low-carbon cement. REFERENCES Atarashi, D.; Kamio, T.; Aikawa, Y.; Miyauchi, M.; and Sakai, E.2014 , “Method for Estimating Quantity of Non-Hydrated Cement in a Cement Recycling System,” Journal of Advanced Concrete Technology, under review. Atarashi, D.; Song, Y.; Nishimura, T.; and Sakai, E., Control of Cement Hydration by Sodium Gluconate in Recycling System, The 10th CANMET/ACI International Conference Superplasticizers and other chemical admixtures in concrete Supplementary Papers, pp.197-208(2012) Japan Society of Civil Engineering: The report of Properties and performance estimation of concrete using admixtures (2007) in Japanese Japanese Industrial Standard (JIS) A5308 2009: Ready-mixed concrete J.Nakamoto et.al: Properties of Concrete with Sludge (in Japanese), Cement Sci. and Concrete Tech., No.53, pp.318-323 (1999) Sakai, E.; Atarashi, D.; Kawakami, A.; and Daimon, M.2003 , “Influence of Molecular structure of Comb-Type Superplasticizers and Inorganic Electrolytes on the Dispersion Mechanisms of Limestone powder,” ACI, V. SP-217, pp. 381-392. E. Sakai, E. Maruya, S. Hagiwara and M. Daimon: Material design of cement for increased waste usage and quality control systems of cement by using of various types of calorimeter.” Cement & Concrete, 756, 48-52. (2010) in Japanese

Cement Recycling System Using Sodium Gluconate  447

Sakai, E.; Tsutsumi, K.; and Daimon, M.1997 , “Measurement of cement hydration by means of sandwich type calorimeter,” Cement Science & Concrete Technology, V. 51, pp. 68-71. in Japanese Young, J. F.1972 , “A Review of the Mechanisms of Set-Retardation in Portland Cement Pastes Containing Organic Admixtures,” Cement and Concrete Research, V. 2, No. 4, pp. 415-434. doi: 10.1016/0008-8846(72)90057-9

448 SP-302-33

SP-302-34

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete by Lucia Ferrari and Pascal Boustingorry Self compacting concretes for precast applications were scaled-down to concrete-equivalent grouts where only the surrounding paste around bigger particles was studied in a rheometer. Low shear rate steady-state flow curves feature a non-monotonous variation of shear stress versus shear rate, with a minimum stress obtained for a critical rate. The thixotropy description initially developed by Roussel, Le Roy and Coussot1 was successfully applied to the data in order to model this behaviour. Extending the study over different superplasticizers showed that their influence depends on the molecular architecture. Specific concrete tests were developed in order to assess workability and formworkfilling retention while applying as little energy as possible to the material. The trends observed at the rheometer scale were confirmed showing that structure build-up kinetics has a major influence on concrete placing and that superplasticizers may help control it up to some extent. Keywords: thixotropy; self-compacting concrete; superplasticizer; formwork filling. INTRODUCTION In the last decade, self compacting concretes (SCCs) have spread widely in the precast market owing to the ability to cast without vibration. This aspect enabled to reduce human operations and exposition to noise and other health-related risks. Superplasticizers are essential to the production of SCCs for which very low yield stresses are desired2,3 but they should also provide more complex flow characteristics, such as speeding up the filling of intricate geometries or dense rebar networks. Such features require a thorough control of rheological properties, including yield stress and plastic viscosity,4 and beyond. Early on during this research it appeared that common laboratory tools such as Abrams slump cone, V-Funnel flow time, J-Ring flow or even L-Box tests failed to distinguish subtle differences between concretes. A rheological approach was then developed for a more precise insight on the flow properties and specific tools were developed to further differentiate concrete properties at lab scale, according to customer feedback. A close look at casting operations indicated that the behaviour at low shear rates was most significant. Low shear rate flow was shown to be largely influenced by suspension 449

450 SP-302-34

Table 1-Cement analysis Blaine specific surface 4445 cm2/g

Weight % Na2Oeq 0.18

Weight % Na2Oeq 0.87

Table 2–Grout composition Cement – g (oz) 196.1 (6.9)

Limestone Filler – g (oz) 89.6 (3.2)

Sand 0/0.160mm – g (oz) 97.8 (3.5)

Sand 0/0.315mm 20.7 (0.7)

Water – mL (US fl oz) 93.6 (3.2)

structure build-up kinetics, often summed-up under the term ‘thixotropy’5. This aspect is the key topic of the present paper, from the scale of a concrete-equivalent grout to full concrete scale evaluations, both in the laboratory and in the field. RESEARCH SIGNIFICANCE This paper focuses on the often overlooked influence of ageing on the rheological behaviour through the shape of flow curves and on the formwork-filling properties of SCCs. The influence of superplasticizer nature and dosage is also discussed, and some evidence is shown that polycarboxylate molecular design allows controlling such properties. EXPERIMENTAL INVESTIGATION AT THE GROUT SCALE Materials The materials used are Portland cement from Port La Nouvelle, Lafarge, the basic characteristics of which are shown in Table 1. Its high specific surface makes it suitable for precast applications where rapid setting is needed. Limestone filler was supplied from the FACO company, Vaiges quarry (France). The Millisil C4 0/0.16 mm sand was supplied by the Sifraco Company (France) and the 0/0.315 mm sand was supplied from Sablières Palvadeau (France). All admixtures were used as aqueous solution of roughly 20% by weight of polymer in tap water with the addition of a suitable defoamer. Equivalent grout mix proportioning The concrete mix proportioning was scaled down through the use of an approach inspired by multiscale studies previously published.6 It relies upon applying a cutoff to the concrete grading curve at an arbitrary particle size – in this study, 315 µm. After normalizing to 100% passing, a target grading curve is obtained which is then matched as closely as possible by a blend of the binders and fine sands. This methodology may be considered as a way to simulate the grout surrounding the largest aggregates in the concrete, while allowing working in a rheometer where the sample is sheared in a very small gap, of the order of several millimeters. The composition of the grout mixture is described in Table 2. Experimental procedure – grout rheology Water and admixtures were weighed in a Krups YY8506FD mixer bowl, the dry powders were added during the first 30 seconds of mixing at speed 1 with a leaf-shaped blade. The

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 451 Table 3-Prestressed SCC mix proportioning CEM I 52,5R Cement kg/m3 (lb/yd3) 350 (590)

Limestone filler kg/m3 (lb/yd3) 160 (270)

0-4 mm sand kg/m3 (lb/yd3) 642 (1082)

4-10mm crushed stone kg/m3 (lb/yd3) 859 (1448)

8-16 mm Riverbed stone kg/m3 (lb/yd3) 114 (192)

Total Water L/m3 (gal/yd3) 176 (35.5)

mixing speed was increased to speed 7 for 1 minute and then stopped for 30 seconds (to scrape the sides of the bowl) before applying a last mixing stage of 1 minute at speed 7. The sample was loaded on the lower plate of a Kinexus Pro rheometer (Malvern Instruments, U.K.) equipped with a serrated parallel plate geometry (1 mm gap). The procedure started five minutes after the beginning of mixing with a pre-shear at 200 s-1 during one minute, followed by logarithmic shear rate steps from 200 to 0.1 s-1. Each stress data point is sampled after the steady state is reached whenever possible in order to build the flow curve. At the same time minislump tests are performed (cone dimensions: upper diameter 18 mm – 0.71 in, lower diameter 36 mm-1.42 in, height 54 mm-2.13 in) using a pneumatic lifting fork for reproducible results. After the flow curve measurement, the structure is reset to zero with an oscillating shear period during one minute with a strain amplitude of 100% and a 1 Hz frequency. Then a constant stress of 4 Pa is applied to the material in order to observe the structure buildup close to rest through the increase of viscosity with time. This applied stress value was chosen to match the stress applied by the weight of the largest aggregate in the system, according to the following rough calculation7:

σ0 ≈

∆ρgd / 2 10.5

1.

Δρ: Specific weight difference between the falling object and the suspended fluid, kg/m3.g: gravity constant, 9.81 m/s2.d: particle diameter, m Eq. 1 yields approximately 4 Pa for a diameter of 10 mm, a paste density of 1800 kg/m3 (112.4 lb/ft3) and an aggregate density of 2600 kg/m3 (162.3 lb/ft3). This section of material testing is called ‘ageing’ in the rest of the paper. Experimental procedure – Concrete flow The concrete mix proportioning tested in this study represents a typical composition of a self compacting concrete for prestressed applications (see Table 3). The cement and limestone filler are the same as the ones used in the grout-scale study. The aggregates were supplied from relevant sources across the French territory. Using a laboratory SKAKO horizontal mixer the sand and the gravel were blended and pre-soaked with a fraction of the batching water for 5 minutes. The binders were then added in the dry state. The admixtures were dosed in a pail with the remaining water, and then the whole was added into the mixer under stirring for 3 minutes and 30 seconds. In order to keep the water-to-cement ratio constant, the amount of water brought by the admixtures was substracted from the total amount of water. General flow properties were assessed with an Abrams cone and a V-shaped funnel for the slump flow and flow time properties respectively.

452 SP-302-34

Given the need for differentiating thixotropy-induced flowability variations at low shear rate, a robust and simple tool was designed to evaluate such behaviour on concrete. A proprietary device named below ‘Double-box’ was built under the form of a rectangular container separated in two compartments of equal size by a removable gate. One compartment (side 1) is filled with concrete right after the end of mixing and the inner wall is then lifted 2, 6 or 10 minutes later. The time taken by the concrete to reach the opposite end (side 2) of the box is measured for each resting time and gives a practical evaluation of structure build-up at rest during the considered time interval. This device differs from the U-box or U-tube sometimes used for SCC in that its height is much smaller so that a lesser amount of concrete is used and a lower kinetic energy is imparted to the material. ANALYTICAL INVESTIGATION Roussel et al1,8 proposed a model for taking thixotropy into account in the flow curves of a suspension. It is based on the assumption that apparent viscosity is linked to a parameter λ which measures the ‘structure degree’ of the suspension as expressed in Eq. 2:

η = η∞(1 + λn)

2.

η: apparent viscosity (Pa.s)η∞: viscosity Newtonian plateau at very high shear rates (Pa.s) λ: structure coefficient (unitless)n: exponent which quantifies the influence of λ on the viscosity (unitless). The structure parameter is a result of the competition between a time-dependent increase and a shear rate-dependent decrease and thus obeys a relaxation law expressed by Eq. 3:

dλ 1 = − αλγ dt θ

3.

θ: time constant for the structure build-up rate (s).α: coefficient describing the efficiency of shear in breaking up the structure (unitless). Eq. 3 shows how structure level increases with an assumed constant rate through time if no shear is applied, whereas shear induces a slow down of increase, or even a decrease of structure if the applied shear rate is high enough. At steady state under a constant shear rate Eq. 2 may be rewritten under the form of a steady state flow curve by simply stating that

dλ 1 = 0 = − αλ eq γ which yields the steady dt θ

state value for the structure degree:

λ eq =

1 αθγ

4.

Eq. 2 then becomes:

(

)

−n η = η∞ 1 + (αθγ )

5.

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 453

Fig. 1-Example flow curves from an experiment. (a) Shear stress vs. shear rate, (b) Apparent viscosity vs. shear rate Or, as expressed in terms of stress:

(

)

−n σ = ηγ = η∞ γ + (αθ) γ 1− n



6.

Eq. 5 or Eq. 6 may be considered as flow curve equations since they describe a unique relationship between apparent viscosity (resp. stress) and shear rate. They may be used as models for fitting to experimental data, provided the considered data are obtained at a steady state. COMPARISON OF PREDICTIONS AND EXPERIMENTAL RESULTS Fig. 1a shows an example flow curve from a typical experiment of our study. It shows that the stress vs. shear rate curve features a non-monotonous shape which departs from the expected result of a monotonous decrease in stress with a decrease in shear rate. Quite interestingly, provided that n > 1, Eq. 6 predicts that flow curves plotted as stress vs. shear rate should feature a non-monotonous trend with a minimum occurring for a ‘critical shear rate’ γ c defined by:

dσ d γ

(

= 0 = η∞ 1 + (αθ) γ = γ c

−n

(1 − n) γ c− n )

7.

γ c: critical shear rate (s-1). After solving, Eq. 7 eventually yields:

γ c =

( n − 1)1/ n αθ

8.

454 SP-302-34

Fig. 2-Transients recorded by the rheometer for two applied shear rates (a) below and (b) above the critical shear rate The curve on Fig. 1 presents a critical shear rate around 3 s-1. Inputting the result of Eq. 8 into Eq. 6 gives a value for the corresponding ‘critical shear stress’: 1



n ( n − 1) n σ c = η∞ αθ −1

9.

Eq. 8 and Eq. 9 show that the existence of such a critical stress is only compatible with an exponent n strictly greater than one. If a stress below this critical value is applied to the material then no steady homogeneous flow may be achieved.1 In this sense the critical stress may be considered as a form of dynamic yield stress reached at a shear rate which in practice may be far from zero. Further insight about this question may be found in paper SP-017 ‘The influence of Superplasticizers on the Flocculation Degree of Cement Suspensions’ in the present Conference Proceedings. Correspondingly, trying to force the material into flowing at a shear rate below γ c may not result into a homogeneous and steady flow. As a matter of fact simply observing the transient signals recorded by the rheometer shows that with an applied shear rate below the critical value the regulation loop is unable to set a steady flow (Fig. 2a) whereas beyond the critical value a steady state is reached within 5 seconds (Fig. 2b). As shown in Fig. 1b, a shear-thickening behaviour often appears in our experiments beyond a shear rate of the order of 10 s-1. As a result the apparent viscosity vs shear rate curve features a minimum ηmin for a critical shear rate γ st .This led us to modify Eq. 5 by adding a dissipation term βγ , following the approach by Hot and Roussel9:

(

)

−n η = η∞ 1 + (αθγ ) + βγ

10.

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 455

Fig. 3 - Grout flow curves as a function of superplasticizer dosage. (a) Apparent viscosity as a function of shear rate, (b) Stress as a function of shear rate. Eq. 10 was then used for fitting to flow curve data and extracting the parameter values. Though Eq. 10 slightly departs from Eq. 6 the above discussion about critical stress and shear rate remains qualitatively valid. EXPERIMENTAL RESULTS AND DISCUSSION Grout scale - Material response to a change of superplasticizer dosage In order to check the model response and determine the most relevant parameters, Eq. 10 was fitted to experimental data obtained with an increasing dosage of a superplasticizer. Fig. 3 shows flow curves obtained for increasing dosages of a common polycarboxylate superplasticizer. As expected, apparent viscosity and yield stress decreases at all shear rates when dosage increases. Fig. 3a stresses out the need of at least an inertial term in the model in order to better simulate the increase in apparent viscosity beyond a critical shear rate. Fig. 3b shows the minimum shear stresses occurring at critical shear rates below which no steady-state may be achieved; consequently Eq. 10 was not fitted to these data according to the approach described in the paper by Roussel et al.1 Fig. 4 displays the parameter values obtained by the modelling of the flow curves according to Eq. 10. Increasing the dosage leads to an increase in the product αθ which means that either structure build-up is slower (θ increases) or shear break-up efficiency is higher (α increases). η∞ decreases when superplasticizer dosage increases which implies that the high shear rate state of the suspension is more deflocculated. The parameter n does not seem to have a meaningful dependence on the superplasticizer content. β increases with dosage meaning that the inertial dissipation increases according to the deflocculation degree. These conclusions allow interpreting the role of a superplasticizer as a means to slow down thixotropy, increase the efficiency of shear for breaking up the suspension structure, and reduce the overall aggregation degree.

456 SP-302-34

Fig. 4 - Model parameters as a function of superplasticizer dosage after a fit of Eq. 10 to the data By fitting the model to a well-known superplasticizer property, this preliminary study provided a better understanding of the physical meaning of the different parameters and showed that αθ and η∞ are most relevant for the description of low-shear rate flow. For a lesser influence of thixotropy on flow, a higher αθ and a lower η∞ are preferred. For a further insight into thixotropy, the ageing curves will now be discussed. Fig. 5 shows the time evolution of the apparent viscosity under a constant applied stress of 4 Pa. The grout response strongly depends on the dosage. At the lowest dosage the apparent viscosity sharply increases over several orders of magnitude in less than one minute, as a sign of a very fast stiffening at rest. When dosage increases, the rate of viscosity increase slows down with an apparent sudden transition between 0.6 and 0.7% beyond which the grout enters a creep regime. Quite interestingly, this regime change seems to occur when the ratio between the applied stress σ0 and the critical stress σc is higher than one. This is another possible interpretation of the critical stress: any applied stress above this value will induce some flow whereas the material will keep stiffening if the applied stress is smaller. To conclude, the increase in superplasticizer dosage seems to have further effects on the considered system than merely decreasing yield stress. The observations discussed above

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 457

Fig. 5 – Influence of superplasticizer dosage on the ageing of grout samples under a constant probe stress of σ0 = 4 Pa. show that ageing is slowed down by the addition of superplasticizers and that the influence of ageing onto the shape of flow curves is somewhat mitigated. Grout scale - Comparison of different superplasticizer formulations at equal mini-slump flow The method described above was used to select the most efficient superplasticizers before testing at the concrete scale. Dosages were adapted in order to obtain a target mini-slump flow of 105 ± 5 mm; this target value was determined from the slump flow of the corresponding concrete. Three examples from the study are displayed in Fig. 6, SP1 being a common precast superplasticizer and SP2 and SP3 being two products specifically designed to ‘reduce thixotropy’. Despite the grouts being prepared at equal mini-slump flow, slight critical stress differences may be observed in Fig. 6b. This was quite unexpected since the general idea is that flow spread is directly linked to yield stress.10,11 Here, it may show the influence of thixotropy on the slump test itself, which is possibly magnified by a high paste volume fraction compared to a concrete mixture. Model parameters fitted according to Eq. 10 are shown in Fig. 7. SP2 and SP3 seem to induce higher values of αθ, which corresponds to a slower structure build-up or a higher shear efficiency, both inducing better flow properties. They also have a similar influence on n while SP1 yields a slightly higher n value. η∞ shows the same trend as the critical stress and SP3 seems to be the most efficient superplasticizer to decrease this parameter, which confirms its deflocculating efficiency. The hydrodynamic dissipation term β seems to be equivalent for the three measurements. It seems then that SP2 and SP3 are more efficient to prevent thixotropy from interfering with flow.

458 SP-302-34

Fig. 6 - Grout flow curves for three different formulations at equal mini-slump flow. (a) Apparent viscosity as a function of shear rate, (b) Stress as a function of shear rate.

Fig. 7 - Model parameters for the three selected superplasticizers after a fit of Eq. 10 to the data

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 459

Fig. 8– Influence of three different formulations on the ageing of grout samples under a constant probe stress of σ0 = 4 Pa. Fig. 8 shows the ageing behaviour induced by superplasticizers SP1, SP2 and SP3 right after the pre-shear stage following the flow curve measurement. SP1 provides a very fast stiffening, with the apparent viscosity increasing by several orders of magnitude in a matter of seconds. SP2 and SP3 prevent this sharp increase by keeping the apparent viscosity below 10 Pa.s during almost 25 minutes before the eventual increase due to hydration. SP3 seems to be able to set a lower viscosity than SP2 despite a slightly lower dosage. Once again, the influence of the ratio between the applied stress and the critical stress is shown; With SP1, as σ0/σc < 1, the grout suddenly stiffens in less than one minute. SP2 and SP3 are able to lower the critical stress below 4 Pa hence σ0/σc > 1 and the grout enters a slow-viscosity-increase regime. This is another sign of their ability to slow down thixotropy. Concrete scale tests of SP1, SP2 and SP3 presented in the next section perfectly illustrate the relevance of the above rheological study. Concrete scale – Flow properties after different resting times SCCs were prepared with the three superplasticizers described above. The dosages were determined in order to achieve a 700-720mm slump flow with the Abrams cone. ‘Doublebox’ filling times are displayed in Fig. 9. For a 2-minute resting time, all three products seem to perform equally. For a 6-minute resting time, SP1 seems less efficient in slowing down the structure build-up at rest, with results in a slightly longer filling time, while SP2 and SP3 have a similar performance. For a 10-minute resting time SP3 allows achieving a significantly faster filling after gate lifting, which confirms it is the best admixture for limiting thixotropic effects. Compressive strengths were assessed on 15x15x15 cm3 cubes cured at 10°C (50°F) during 16h and at 20°C (68°F) for 24h. Fig. 10 shows that the flow improvement brought by SP3 does not induce a decrease in strength, even in the harsh conditions of 16h curing at 10°C. In the same conditions SP2 induces a small strength decrease due to its slightly higher dosage.

460 SP-302-34

Fig. 9- Double-box filling times after periods of rest of 2, 6 and 10 min

Fig. 10- Compressive strength at 16h for a curing cycle at 10°C and 24h for a curing cycle at 20°C The concrete-scale study thus confirms the observations at the grout scale, SP3 being the most efficient superplasticizer when it comes to improve the concrete flow properties, even after periods of rest. FURTHER RESEARCH Given the influence of hydration on the ageing thus the rheology of concrete it may be inferred that all cases where hydration is accelerated (either by temperature or the use of an admixture) will be concerned by the notions introduced in the present paper. Some further investigation beyond the scope of the present study already showed how civil engineering concretes, where water to cement ratios are quite lower than traditional ready mix concretes, also feature a certain amount of thixotropy somewhat limiting the placing of concrete in complex formworks or dense rebar conditions. The application of the present approach may then spread way beyond precast applications.

The Influence of Paste Thixotropy on the Formwork-Filling Properties of Concrete 461 It was shown that some superplasticizers are able to slow down the ageing due to cement hydration while no retardation was observed, at least on early strength in cold conditions. The fundamental mechanisms involved remain unclear and should be investigated in the future. CONCLUSIONS The present study shows how a scaled-down grout-like material may allow a successful simulation of the behaviour of full-scale concrete with the beneficial wealth of information brought by rheometric experiments. Cement hydration was observed as an ageing mechanism having a noticeable influence of the rheological behaviour, namely the shape steady state flow curves, which feature a minimum for a characteristic pair of values of critical stress and critical shear rate. A theoretical framework designed for the description of thixotropic materials was successfully applied to the data, allowing a better understanding of the role of superplasticizers in this interplay between ageing and flow. They may be considered not only as yield-stress-decreasing admixtures, but also as agents able to slow down the ageing due to hydration, increase the efficiency of shear thus leading to a steeper shear-thinning behaviour and decrease the overall structure level of the cement paste, thus improving the flow properties of concrete. The grout-scale observations were confirmed at the concrete scale in laboratory trials. It is worth noting that these results were extended an applied on numerous jobsites in France, Spain, Poland and India, leading to the launching of a whole product range. AUTHOR BIOS Lucia Ferrari is the Physical Chemistry manager in the main research and development laboratory of CHRYSO in France. She received her PhD from the Technische Universität München (Germany) after she completed her research work with the EMPA in Dübendorf (Switzerland) under the supervision of Dr Frank Winnefeld and Pr. Dr. Johann Plank. Pascal Boustingorry is the Head Manager of the Interface Physical Chemistry Team in the main research and development laboratory of CHRYSO in France. He received his PhD from the INP Grenoble and the School of Mines in Saint Etienne (France). Their main research interests are the interaction of organic molecules with cement suspensions and the links between superplasticizer chemical architecture and the flow properties of building materials. REFERENCES 1. Roussel, N.; Le Roy, R.; and Coussot, P., “Thixotropy modelling at local and macroscopic scales,” Journal of Non-Newtonian Fluid Mechanics, V. 117, No. 2-3, 2004, pp. 85-95. doi: 10.1016/j.jnnfm.2004.01.001 2. Banfill, P. F. G., “The rheology of fresh cement and concrete-a review,” Rheology Reviews, V. 2006, 2006, pp. 61-130. 3. Kauppi, A. et al., Improved superplasticizers for high performance concrete. in Proc. 11th ICCC, Durban 11–16 (2003).

462 SP-302-34

4. Wallevik, O. H., and Wallevik, J. E., “Rheology as a tool in concrete science: The use of rheographs and workability boxes,” Cement and Concrete Research, V. 41, No. 12, 2011, pp. 1279-1288. doi: 10.1016/j.cemconres.2011.01.009 5. Roussel, N.; Ovarlez, G.; Garrault, S.; and Brumaud, C., “The origins of thixotropy of fresh cement pastes,” Cement and Concrete Research, V. 42, No. 1, 2011, pp. 148-157. doi: 10.1016/j.cemconres.2011.09.004 6. Toutou, Z., and Roussel, N., “Multi scale experimental study of concrete rheology: from water scale to gravel scale,” Materials and Structures, V. 39, 2006, pp. 167-176. 7. Ovarlez, G. & Coussot, P. Sédimentation dans les fluides à seuil en écoulement. in (2011). 8. Roussel, N., “A thixotropy model for fresh fluid concretes: theory, validation and applications,” Cement and Concrete Research, V. 36, No. 10, 2006, pp. 1797-1806. doi: 10.1016/j.cemconres.2006.05.025 9. Hot, J., and Roussel, N., Influence of adsorbing polymers on the macroscopic viscosity of concentrated cement pastes. in Proceedings of the 10th International Conference on Superplasticizers and Other Admixtures in Concrete SP-288, 223–233 (American Concrete Institute, 2012). 10. Saak, A. W.; Jennings, H. M.; and Shah, S. P., “A generalized approach for the determination of yield stress by slump and slump flow,” Cement and Concrete Research, V. 34, No. 3, 2004, pp. 363-371. doi: 10.1016/j.cemconres.2003.08.005 11. Roussel, N., and Coussot, P., “‘Fifty-cent rheometer’ for yield stress measurements: From slump to spreading flow,” Journal of Rheology, V. 49, No. 3, 2005, pp. 705-178. doi: 10.1122/1.1879041

SP-302-35

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency by Rachid Ait-Akbour, Christine Taviot-Guého, Fabrice Leroux, Pascal Boustingorry, and Frédéric Leising The interaction of methoxy-capped poly(ethylene glycol) polymers (MPEG) and a poly(methacrylic acid) anionic polymer (PMA) from water onto sodium Montmorillonite (Na-Mmt) particles untreated or treated by calcium chloride was studied at 20°C. In the absence of Ca2+, MPEGs are able to intercalate by displacing the water molecules present in the interlayer space, as shown by XRD and TGA analyses. In contrast, the adsorbed amount of PMA remains low. The saturation of Mmt with Ca2+ prevents MPEG intercalation through replacing sodium by a stronger water coordinator in the interlayer space, but slightly increases PMA adsorption possibly through a calcium bonding mechanism. This was confirmed with PCE superplasticizers and Na- and Ca-saturated Mmt clays. Whatever the PCE, a larger amount was consumed on Na-Mmt than on Ca-Mmt. This confirms the occurrence of two consumption mechanisms: (i) a superficial adsorption via cation bonding of the carboxylate groups with anionic sites on clay surfaces, (ii) intercalation of ether units of the grafts in the interlayer space by displacement of water molecules coordinated to the exchangeable cations. Keywords: superplasticizer; clay interaction; interlayer space; exchangeable cations; coordinated water; polymer intercalation. INTRODUCTION Concrete performance is affected by the presence of clay-minerals in aggregates. Clays are a part of the phyllosilicate family i.e. a family of minerals structured as stacks of silicate sheets or ‘platelets’ separated by interlayer gaps. Due to a similar ionic size, silicium ions may be easily replaced by aluminum ions in the platelets, leading to an electrical charge imbalance which is compensated by the insertion of cations in the interlayer spaces.1 A variety of clay 463

464 SP-302-35

Table 1-PCE polymer structures used in the study Polymer name PCE-A PCE-B PCE-C

Polymer average graft length (number of monomers) 45 31 114

Polymer grafting ratio (acid: graft) 4:1 3:2 4:1

Most abundant molecular weight (Mp) (g/mol) 27000 35000 51000

structures is very well known today, and molecular models were established that describe this equilibrium state of aluminosilicate sheets separated by cation and water-filled gaps.2-4 These stacks make up particles that possess properties of ion exchange (where the native cations in the interlayer are prone to be replaced by foreign cations from the surrounding solution5) or polymer adsorption/intercalation properties, opening way to a range of interesting hybrid materials.6-8 It may be inferred that they may interact with the superplasticizer present in concrete formulation therefore leading to a reduction in their dispersion efficiency. Solving this problem requires to unravel the interactions between the superplasticizer molecules and clay particles. This study focuses on the influence of Sodium-Motmorillonite (Na-Mmt), a clay which was shown to have an intense detrimental influence on polycarboxylate ether (PCE) superplasticizers.1,9 RESEARCH SIGNIFICANCE The increasing use of low-grade aggregates in the manufacturing of concrete leads to a more frequent occurrence of clay presence in the mixtures. Clays induce higher superplasticizer dosages, thus an increase in cost, without the usual benefit of longer workability retention times. This work sheds some light on the interaction mechanisms and a possible cure is proposed. EXPERIMENTAL PROCEDURES Materials Montmorillonite (Mmt) particles (KSF), specific surface area of 36 m2.g-1 and median particle size of d50=50 µm, were used as received from Sigma Aldrich. The model molecules used were poly(methacrylic acid) (Mw < 10000 g.mol-1) hereafter called PMA and methoxy-poly(ethylene glycol) series with weight average molecular weights of Mw=750, 2000 and 5000 g.mol-1 and called MPEG750, MPEG2000 and MPEG5000, respectively. In a second part, three different PCEs were used denoted hereafter as PCE-A, B and C. They featured very different molecular structures in terms of number of carboxylate groups, number and length of ether groups, as shown in Table 1. These PCE were supplied by CHRYSO (France) and their average molecular weights Mw were determined by Size Exclusion Chromatography (SEC) with a Malvern Viscotek TDA 305 chromatographer (eluent NaNO3 0,1N with pH buffered at 7, flow 1 ml/min, column temperature 30°C) equipped with a triple detector (Refractive index, Light Scattering and Viscometer). The average molecular masses ranged between 27000 and 51000g.mol-1 as shown also in Table 1. Aqueous Mmt solutions were prepared using fresh deionized water and chemical reagents of p.a. quality. Two sets of experiments were prepared: the ionic strength was

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  465 imposed using either sodium chloride (NaCl (10-2M)) and the pH was kept constant by adding a small amount of either sodium hydroxide (NaOH) or calcium hydroxide Ca(OH)2 (Carlo Erba [Ca2+]0=2.10-2M and pH0 > 12). Methods Adsorption experiments—Various series of Montmorillonite – polymer aqueous dispersions were prepared in order to elucidate the effect of Ca2+ ions on the adsorption isotherms of model polymers on Montmorillonite. The same experiments were conducted with real case molecules i.e. PCE-A, B and C In the first case (no Ca2+ ions), a dispersion of Montmorillonite at 8.5g/l in 10-2 M NaCl solution at pH=12 was prepared. Appropriate small volumes of polymer stock solution (20g/l) were added in order to obtain final concentrations ranging from 0 to 1g/l. In the second set of experiments (with Ca2+ ions), Montmorillonite particles were equilibrated with increasing quantities of calcium hydroxide Ca(OH)2 (initial calcium concentration, [Ca(OH)2]0=2.10-2 M, pH>12) during 24h under gentle magnetic stirring. Subsequently appropriate volumes of polymer stock solution (20g/l) were added in order to obtain the same range of concentrations (0 to 1 g/l) as in the first set. In all cases dispersions were stirred by a magnetic stirrer for 20h at 20°C (except if indicated). After sorption, the mixture was centrifuged for 30min at a speed of 10.000 rpm and the supernatant solution was analysed with a TOC analyzer (TOC 5050A from Shimadzu). The adsorbed amount of polymer on the mineral particles was calculated from the difference between the amount of polymer in the liquid before and after contact with the clay. X-ray diffraction—The basal spacing of the modified Montmorillonites was analyzed by using CuKα (λ=1.5418Å) radiation from an automated X-ray diffractometer (XRD) (X’Pert Pro Philips). The equipment was operated under a 40kV tension and a 30mA current in a continuous scan mode. The scanning speed was 0.02°/s. X-ray diffraction measurements were performed over a 2θ range of 2.5 -70°. Thermogravimetric analysis—Thermogravimetric analysis of the modified montmorillonite was carried out using a Setaram Instrumentation equipment (Setsys Evolution) operating at a ramp of 5°C/min from room temperature to 1100°C under a flowing air atmosphere. Mortar testing—Mortar compositions are summarized in Table 2. Two mix designs are described, with #2 being the same as #1 but with the addition of Na-Mmt. The mixing procedure was the following. 1. Both sands (Fulchiron and AFNOR) were dry-blended in a Perrier-type planetary mixer during 30 seconds then 2/3 of the water was added and the wet mixture underwent another 30-second mixing. 2. The pre-wet sands were left at rest for 4 minutes after which cement and filler were added and mixed at low speed for one minute. 3. The remaining water with superplasticizer was then added in 30 seconds under low speed mixing and the wet mixture was homogenized during 90 seconds at the same speed. 4. A 30 second stop allowed scraping the bowl in order to recover the whole material spread over by the mixer and then an ultimate mixing stage of one minute at high speed was applied. When Na-Mmt was present in the blend, it was added at the pre-wetting stage of the sands (stage 1).

466 SP-302-35

Table 2-Mortar compositions Component Type I Cement (Lafarge Saint Pierre La Cour, France) Limestone Filler 0/0.5 mm sand 0/2 mm AFNOR standard sand 0/4 mm unwashed riverbed sand (Lafarge Aggregates Lillion) Montmorillonite Total water

Mix design #1 in g (oz)

Mix design #2 in g (oz)

Mix design #3 in g (oz)

624.9

(22.04)

624.9

(22.04)

624.9

(22.04)

412.1 587.7 1350.0

(14.54) (20.73) (47.62)

412.1 587.7 1350.0

(14.54) (20.73) (47.62)

412.1 -

(14.54) (20.73) -

-

-

-

-

1946.5

(68.7)

375.1

(13.23)

19.4 375.1

(0.68) (13.23)

375.1

(13.23)

Fig. 1–Polymer consumption isotherms onto Na-Mmt clay EXPERIMENTAL RESULTS AND DISCUSSION Interaction of MPEGs and PMA with Na-Montmorillonites Fig. 1 represents adsorption isotherms of MPEG750, 2000, 5000 and PMA polymers from water onto Na-Mmt particles at ambient temperature, at fixed pH and ionic strength values (pH=12, NaCl=10-2M). As for PMA, a non monotonous trend with low adsorption values is observed. A possible cause is a change in adsorption regime: polycarboxylates are known to provide H-bonding with clay surfaces10 but beyond a given adsorbed amount electrostatic repulsion between molecules may take over and establishes a lower adsorbed amount. Complementary measurements (not included here for lack of space) show that the zeta potential of clay dramatically decreases towards more negative values when PMA dosage increases. For all MPEGs, polymer consumption increases with an increase of the polymer initial concentration and then reaches a maximum value. The maximum adsorption seems to obey a decreasing logarithmic law with the MPEG molecular weight as shown in Fig. 2, which shows that the longer the MPEG the fewer the molecules able to interact with this clay.

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  467

Fig. 2—Relationship between MPEG uptake saturation onto Na-Mmt and MPEG molecular weight Moreover, the affinity of the poly(ethylene glycol) polymers for Mmt surface, as indicated by the initial slope of the isotherm, increases with molecular weight. The consumption of those polymers by Mmt particles may be explained by the combination of three mechanisms: (i) The hydrophobic interaction (between CH2-CH2- groups and siloxane surface). In fact, the neutral siloxane (-Si-O-Si-) surfaces located on clays with no isomorphous substitution have an overall hydrophobic character.11 (ii) The hydrogen bonding interactions between ether oxygen atoms (Lewis base) and OH groups (Bronsted acid) associated with aluminol and silanol groups of Mmt.9,12 (iii) A direct interaction between ethylene oxide units and sodium cations in the interlayer space. Some previous work claims the ethylene oxide segments may indeed coordinate around the sodium ions in a crown-ether type chelation.13,14 This means that MPEG750, 2000 and 5000 polymer ‘adsorption’ may occur partly through intercalation of poly(ethylene glycols) chains in the interlayer space of Mmt, promoted by Na+ interlayer cations. This intercalation would cause the displacement and/or replacement of water molecules forming the hydration shell of Na+ exchangeable cations. In order to assess polymer intercalation and water replacement by ethylene oxide groups treated, Mmt particles were analyzed by XRD and TGA. X-ray diffractograms are shown in Fig. 3 where a noticeable shift of the peak attributed to the (001) plane is observed in the 4 to 8° 2-theta range. This is a result of intercalation of polymers in the interlayer space and the conversion into distances yields an increase of basal spacing from 1.29 nm to 1.38, 1.59 and 1.86 nm for MPEG molecular weights of 750, 2000 and 5000 g/mol respectively. A rough volume balance may be computed by taking a specific weight of 2.45g/cm3 (thus 2.45.10-21 g/nm3) for Na-Mmt as claimed on the product datasheet from Sigma Aldrich and considering that the 4% interlayer water mass replacement (thus 9.8.10-23 g/nm3) measured above accounts for a proportional volume loss of around 10% (with the specific gravity of interlayer water taken as 1 g/cm3) whatever the MPEG molecular weight. The respective net volume increases from XRD are estimated at 7, 23 and 44% yielding a true volume

468 SP-302-35

Fig. 3—X-ray diffractograms of dried Na-Mmt samples treated with polymer solutions. Arrows show the peaks attributed to the main [001] basal spacing of the clay increase by polymer intercalation of 17, 33 and 54% for MPEG molecular weights of 750, 2000 and 5000 respectively. No noticeable shift was observed with PMA which shows that this polymer only adsorbs on the surface of clay, if it ever adsorbs at all. TGA curves are shown in Fig. 4 which shows that reference Na-Mmt undergoes two main weight loss events. The first event occurs below 150°C (302 °F) and is related to the departure of water from the interlayer space, roughly accounting for an 8% mass loss. In the 400-600°C range (752-1112°F), transformations of the mineral phase, namely surface dehydroxylation, lead to a secondary weight loss. In the presence of MPEGs, a third intense event is observed in the 300-400°C range (572752 °F), attributed to the calcination of organic matter (interlayer MPEG). Below 150°C (302°F), the weight loss is less than 4% which is a proof that MPEG-treated Na-Mmt contains half less water in the interlayer space than the reference. The PMA-treated Mmt curve lies very close to the reference curve, proving that no water substitution in the interlayer space occurs, a confirmation that this polymer only interacts with clay platelet surface through external adsorption. To sum up the results, it is expected that only the lateral MPEG grafts of a PCE would interact by intercalation in the interlayer space, expelling water molecules and replacing them by ethylene oxide units around sodium cations. It was previously shown that at least 5 to 6 ethylene oxide units are necessary to promote interlayer insertion of MPEG, due to the crown-ether-like conformation taken by such polymers.14 This led to a recent proposition to decrease the length of the lateral grafts below this value in order to obtain PCEs less sensitive to clays.9

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  469

Fig. 4—Thermogravimetric analysis of Na-Mmt samples treated with polymer solutions A different way was chosen here, where sodium ions were exchanged by other cations such as calcium, the water-binding and basal layer-binding energies of which are much stronger than those of sodium. Interaction of MPEGs and PMA with calcium-exchanged Montmorillonite The measured polymer sorbed amounts onto Mmt with sodium and with calcium are plotted on Fig. 5. Almost no change in PMA sorbed amount was observed whereas a decrease in MPEG consumption occurred in the presence of calcium, whatever the MPEG molecular weight. This decrease is larger for the highest molecular weights i.e. 2000 and 5000 g/mol. No noticeable increase of the basal spacing was observed, as shown on Fig. 6. Indeed the peaks of the (001) plane lie at the same 2-theta angle regardless of the polymer weight which shows that almost no intercalation occurs anymore in the presence of calcium. TGA curves plotted on Fig. 7 show that Ca-treated Mmt features a more intense weight loss in the interlayer-related temperature range (below 150°C/302°F) while a second weight loss around 600°C (1112 °F) may be related to dehydroxylation as shown in the work of Koster van Groos and Guggenheim.15 In the presence of MPEGs, the curves are remarkably close and parallel to the Ca-Mmt reference with only a slight extra weight loss in the 300–400°C (572-752°F) range coming from the combustion of polymer. The same combustion, slightly more intense, may be observed on the PMA-curve in the 400-500°C range (752-932 °F). However, no decrease in the interlayer water amount was observed below 200°C (392°F) which proves that the above explanation for Na-Mmt intercalation does not stand anymore for polymer sorption by Ca-Mmt. Surface adsorption may then be involved. This study shows then that provided the Na-Mmt is exchanged with calcium ions prior to or even simultaneously with polymer contact, MPEG intercalation is prevented and only surface adsorption seems to occur.

470 SP-302-35

Fig. 5—Compared sorbed amounts of polymers on Na-Mmt and Ca-exchanged Mmt

Fig. 6— X-ray diffractograms of dried Ca-Mmt samples treated with polymer solutions.

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  471

Fig. 7—Thermogravimetric analysis of Na-Mmt and Ca-Mmt samples treated with polymer solutions The consequences of this calcium-dependent behaviour were investigated with PCE comb copolymers, as described in the next section. Interaction of PCE comb polymers with Montmorillonites Three PCEs the structures of which are described in Table 1 were tested according to the protocols above. Polymer sorption onto Na-Mmt is represented on Fig. 8. It shows that polymer consumption decreases when graft ratio or length increases. Indeed, the lowest amount was measured for PCE-C bearing the longest grafts whereas PCE-B with short grafts and a higher grafting ratio features intermediate uptake values. PCE-A with the same grafting ratio as PCE-C but with shorter grafts, shows the strongest affinity for Na-Mmt. This may be explained by the combined contribution of simple adsorption onto the platelet surfaces and graft intercalation. This is quite consistent with the results obtained on the grafts alone and discussed in the previous section, where MPEG consumption saturation value decreases with an increase of the molecular weight. The influence of calcium ions is illustrated in Fig. 9 for all three PCEs where a general decrease in polymer consumption is observed, which again is consistent with the results obtained on the grafts. An intense decrease was observed for PCE-C which was expected since it bears the longest grafts, and these are the most sensitive to the presence of calcium. X-ray diffractograms on Fig. 10 again bring evidence that intercalation seems to be prevented by calcium exchange in the interlayer space, leading to the conclusion that polymer uptake by Ca-Mmt only occurs through surface adsorption. Investigation of a possible cure protocol According to the above results, any pre-treatment of a clay-containing aggregates with a calcium-containing solution should have a beneficial effect on superplasticizer behav-

472 SP-302-35

Fig. 8—Polymer consumption isotherms of three PCEs onto Na-Mmt

Fig. 9—Influence of calcium presence on the sorbed amounts of PCEs onto Mmt iour. This hypothesis was checked through mortar trials following the procedure described above, the mortar mix proportioning being recalled in Table 2. A solution of calcium acetate was prepared by adding acetic acid to a suspension of calcium hydroxide until complete dissolution. A subsequent filtration ensured that no particle remained in the solution the final concentration of which is 20% by weight of calcium acetate.

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  473

Fig. 10—X-ray diffractograms of Mmt samples treated with PCE. (a) Na-Mmt, (b) Ca-Mmt

Fig. 11—Mortar test results showing the influence of a calcium acetate salt solution on the performance of PCE-B superplasticizer. The reference mortar was prepared according to Mix design #1 and the others correspond to Mix design #2. It was chosen to use this solution either as a co-admixture i.e. by adding it with the superplasticizer in the batching water, or as a pre-treating additive for the aggregate, by spraying it onto the sand during the pre-wetting stage. The influence of this calcium solution is demonstrated in Fig. 11 where a constant dosage of 1,4% PCE-B by weight of total binder was used. The addition of 1% Na-Mmt to sand had a tremendous effect on PCE efficiency, the initial flow dropping from 315 mm (12.4

474 SP-302-35

Fig. 12—Mortar test results showing the influence of a calcium acetate salt solution on the performance of PCE-B superplasticizer. All mortars correspond to Mix design #3. in) to 205 mm (8.1 in). The subsequent flow measurements could not be carried out beyond 60 minutes due to a sudden decrease in workability. The addition procedure of the calcium solution (0.2% by weight of total sand) allowed to increase the initial slump flow almost back to the reference measurement: 270 mm(10.6 in) instead of 315 mm(12.4 in) ; it also retrieves the workability retention curve of the reference mix. In another series of experiments, Mix design #3 was used for which the sands were replaced by a riverbed unwashed sand provided by Lafarge Granulats France. In that case, the superplasticizer dosage was determined in order to reach a target initial flow. Fig. 12 shows that a superplasticizer dosage decrease from 2.20% to 1.75% was achieved whatever the addition procedure of the calcium solution. The results are then still valid on natural sand. CONCLUSIONS It was shown that the interaction mechanism between a PCE superplasticizer and sodium Montmorillonite clay relies on both adsorption onto the platelet surface and poly(ethylene oxide) graft intercalation between the aluminosilicate sheets. The latter phenomenon is promoted by a relatively strong interaction between sodium ions and ethylene oxide units which probably replace water molecules initially present in the coordination layer of the cations. Exchanging sodium by calcium ions allowed populating the interlayer space with cations featuring a stronger interaction energy with water and inducing a much lower, if ever at all, intercalation of the PCE grafts. Those results were transposed at the mortar scale, showing that either an improvement of workability at the same superplasticizer dosage or a decrease in dosage at constant workability may be achieved by adding a calcium acetate solution to the admixture system. This

Interaction of Montmorillonite with Poly(ethylene Glycol) and Poly(methacrylic Acid) Polymers. Consequences on the Influence of Clays on Superplasticizer Efficiency  475 protocol was efficient either as a pre-treatment of the aggregates minutes before mixing or as a co-admixture in the batch water. The assumed mechanism was interlayer cation exchange as well as an adsorption of acetate ions onto the platelet surface competing with PCE adsorption, which should be checked in any subsequent work. AUTHOR BIOS Dr Rachid Aït-Akbour received his PhD in the Université Ibnou Zohr in Agadir (Morocco). He was a post-doctoral student in the Inorganic Materials Team in the Université Blaise Pascal in Clermont-Ferrand (France) at the time of the present study. Pr Christine Taviot-Guého is a Researcher in the Inorganic Materials Team in the Université Blaise Pascal in Clermont-Ferrand (France). Dr Fabrice Leroux has a CNRS senior research position in the Chemical Institute of Clermont-Ferrand and is the Head of the Inorganic Materials Research Team in the same university. Their professional interests revolve around Layered Double Hydroxides (LDH) and their interaction with organics and minerals for the design of new materials with original properties. Dr Pascal Boustingorry is the Head of the Interface Physical Chemistry Team in the main research and development laboratory of CHRYSO in France. He received his PhD from the INP Grenoble along with the Ecole des Mines in Saint Etienne (France). Dr Frédéric Leising received a PhD from the Université Louis Pasteur in Strasbourg (France) and led a career in the R&D Departments of Rhône-Poulenc then Rhodia in the fields of latices and silicon chemistry. He is now an Associated Researcher in the main research and development laboratory of CHRYSO in France. Their main research interests are the interaction of organic molecules with cement suspensions and the links between superplasticizer polymer architecture and the flow properties of building materials. REFERENCES 1. Ng, S., Interactions of Polycarboxylate based Superplasticizers with Montmorillonite Clay in Portland Cement and with Calcium Aluminate Cement. (Technische Universität München, 2012). 2. Cygan, R. T.; Liang, J. J.; and Kalinichev, A. G., “Molecular models of hydroxide, oxyhydroxide, and clay phases and the development of a general force field,” The Journal of Physical Chemistry B, V. 108, No. 4, 2004, pp. 1255-1266. doi: 10.1021/jp0363287 3. Cygan, R. T.; Greathouse, J. A.; Heinz, H.; and Kalinichev, A. G., “Molecular models and simulations of layered materials,” Journal of Materials Chemistry, V. 19, No. 17, 2009, pp. 2470-2481. doi: 10.1039/b819076c 4. Kalinichev, A. G.; Kirkpatrick, R. J.; and Cygan, R. T., “Molecular modeling of the structure and dynamics of the interlayer and surface species of mixed-metal layered hydroxides; chloride and water in hydrocalumite (Friedel’s salt),” The American Mineralogist, V. 85, 2000, pp. 1046-1052. 5. Bansal, O. P., and Bansal, V., “Influence of time, pH, temperature, organic matter and exchangeable cations on the adsorption of oxamyl on illites and kaolinites,” The Journal of Agricultural Science, V. 94, No. 03, 1980, pp. 557-563. doi: 10.1017/S0021859600028562

476 SP-302-35

6. Plank, J.; Keller, H.; Andres, P. R.; and Dai, Z., “Novel organo-mineral phases obtained by intercalation of maleic anhydride–allyl ether copolymers into layered calcium aluminum hydrates,” Inorganica Chimica Acta, V. 359, No. 15, 2006, pp. 4901-4908. doi: 10.1016/j.ica.2006.08.038 7. Burchill, S. et al., “Smectite-polymer interactions in aqueous systems,” Clay Minerals, V. 18, No. 4, 1983, pp. 373-397. doi: 10.1180/claymin.1983.018.4.04 8. Merlin, F.; Lombois, H.; Joly, S.; Lequeux, N.; Halary, J.-L.; and Van Damme, H., “Cement-polymer and clay-polymer nano- and meso-composites: spotting the differenceBasis of a presentation given at Materials Discussion No. 5, 22???25 September 2002, Madrid, Spain,” Journal of Materials Chemistry, V. 12, No. 11, 2002, pp. 3308-3315. doi: 10.1039/b205279m 9. Ng, S., and Plank, J., “Interaction mechanisms between Na montmorillonite clay and MPEG-based polycarboxylate superplasticizers,” Cement and Concrete Research, V. 42, No. 6, 2012, pp. 847-854. doi: 10.1016/j.cemconres.2012.03.005 10. Mpofu, P.; Addai-Mensah, J.; and Ralston, J., “Investigation of the effect of polymer structure type on flocculation, rheology and dewatering behaviour of kaolinite dispersions,” International Journal of Mineral Processing, V. 71, No. 1-4, 2003, pp. 247-268. doi: 10.1016/S0301-7516(03)00062-0 11. Charnay, C.; Lagerge, S.; and Partyka, S., “Assessment of the surface heterogeneity of talc materials,” Journal of Colloid and Interface Science, V. 233, No. 2, 2001, pp. 250-258. doi: 10.1006/jcis.2000.7259 12. Su, C.-C., and Shen, Y.-H., “Adsorption of poly (ethylene oxide) on smectite: Effect of layer charge,” Journal of Colloid and Interface Science, V. 332, No. 1, 2009, pp. 11-15. doi: 10.1016/j.jcis.2008.12.024 13. Aranda, P., and Ruiz-Hitzky, E., “Poly (ethylene oxide)-silicate intercalation materials,” Chemistry of Materials, V. 4, No. 6, 1992, pp. 1395-1403. doi: 10.1021/cm00024a048 14. Lu, Y.; Kong, S.-T.; Deiseroth, H.-J.; and Mormann, W., “Structural Requirements for the Intercalation of Polyether Polyols into Sodium—Montmorillonite: The Role of Oxyethylene Sequences,” Macromolecular Materials and Engineering, V. 293, No. 11, 2008, pp. 900-906. doi: 10.1002/mame.200800155 15. Van Groos, A. K., and Guggenheim, S., “Dehydroxylation of Ca-and Mg-exchanged montmorillonite,” The American Mineralogist, V. 74, 1989, pp. 627-636.

SP-302-36

Plasticizing Geopolymer-Type Auspensions: A Challenge by L. Nicoleau, M. Pulkin, and T. Mitkina The composition of the aqueous solution in alkali-activated binders, i.e., the high alkalinity and the high ionic strength challenge chemists to design molecules exhibiting the same plasticizing effects as in cementitious materials. The highest difficulty probably lies in alkali-silicate activated systems due to the presence of multivalent silicate oligomers in solution. Reported here are new insights about the adsorption of polymers in presence of various concentrated electrolyte solutions in order to mimic the harsh conditions present in geopolymer pastes. In order to eliminate the problem of the reactivity of such systems, TiO2 nanoparticles were used as a model substrate. The adsorption of polymer molecules as well as the specific adsorption of monovalent and divalent ions is revealed. Those results are compared to the rheological characteristics of alkali-hydroxide or alkali-silicate activated geopolymers. The conclusions which can be drawn from the model system fit qualitatively very well with the classical slump tests done on real systems. Keywords: geopolymer; high-alkalinity; silicate adsorption. INTRODUCTION Geopolymers are alkali-activated alumino-silicates and an interesting alternative to portland cement because of their lower CO2 footprint, their excellent acid and heat resistance and their low pore-connectivity.1 In spite of these attractive properties, geopolymer-based materials are still only marginally present in the construction market due to three main drawbacks, which seriously hamper their use when compared to more traditional concretes. The first one is the lack of durability investigations, the second one is shrinkage and the last one, being the subject of this paper, is the poor rheological properties. Unlike cement, the development of highly efficient plasticizers for geopolymers has revealed strong difficulties. In this regard, we have recently developed new highly charged polycondensates which outperform the typical polycarboxylate ethers used in cement mixtures and which significantly enhance the flowability of alkali-activated geopolymer systems. Nevertheless, the adsorption of these molecules, i.e. their efficiency as plasticizers, strongly depends on solution conditions. A study of the main relevant parameters in solution for the adsorption of these polymers is presented in this paper. It is questionable to try to produce adsorption isotherms on reactive materials like aluminosilicates, hence the 477

478 SP-302-36

decision was made to use an unreactive model substrate, such as titanium dioxide (TiO2), in order to mimic the adsorption on aluminosilicates. TiO2 also offers the advantage that the calcium, hydroxide and silicate concentrations can be independently varied. The polymer adsorption, the ion adsorption and the rheological properties of suspensions are discussed, which lead to a better understanding of the parameters controlling the efficiency of anionic plasticizers in typical geopolymer solutions. RESEARCH SIGNIFICANCE To the author’s best knowledge, no study has been reported so far that investigates in detail the adsorption behavior of plasticizers on systems representative of geopolymer materials, and the mechanism leading to this adsorption. With this paper, the authors present a rationale about the factors limiting the adsorption of anionic polymers in such systems, and why the fluidification of geopolymers represents a challenge for the construction industry. EXPERIMENTAL SECTION Materials A new polycondensation polymer (called as PGP) was used for this study. This polymer consists of a comb structure with phosphate and carboxylate anchor groups likely to adsorb on inorganic surfaces. The negative charge density borne by these groups measured by acid-base titration is 0.003 e-/g (0.085 e-/oz); this charge corresponds to the complete deprotonation. The effect of this new polymer has been also compared to a regular and commercial “Polycarboxylate Ether” provided by BASF and marketed under the name MasterGlenium 51®. In order to achieve adsorption isotherms, an unreactive model substrate, i.e. which does not significantly dissolve, has to be used. At the same time, it has to exhibit a representative surface similar to the one of geopolymer raw materials. To ensure these requirements, aluminosilicates or silica should be avoided since they are too reactive in alkaline solutions. The authors decided to work with TiO2 and in particular with the anatase-rich (90% anatase-10% rutile) powder P25 (Evonik Industries AG). This powder presents the advantage to be very fine, and thus features a large interface area with the solution. The specific surface area obtained by the BET method is 60.2 m2/g (18400 ft2/oz). An analysis by ICP spectrometry reveals that a suspension of this powder does not dissolve any significant amounts of impurities and in particular negligible amounts of sodium, calcium, potassium, aluminum or silicon have been detected. As it will be explained below in detail, the surface charging of TiO2 is similar to that of other alumino-silicate materials, i.e. the zeta potential evolution over pH is comparable and the overcompensation in presence of calcium ions is evident. As the adsorption of charged polymers onto inorganic substrates at high pH is mainly driven by electrostatics, the first important physical property to be determined is the surface charge density of the substrate. Upon the increase of pH, the surface of minerals is getting charged. In the particular case of TiO2, due to the deprotonation of the Ti-OH groups, the surface is getting negatively charged. These protons can be titrated with NaOH. Without any additional salt, the charge density is 0.3 e-/nm2 (2.77E16 e-/ft2) and in presence of a background 1:1 electrolyte (NaCl) around 1.1 e-/nm2 (1.019E17 e-/ft2). The presence of

Plasticizing Geopolymer-Type Auspensions: A Challenge  479

Ca2+ ions favors the deprotonation of the surface and the charge density reaches the, a priori, maximum value of 2.1 e-/nm2 (1.94E17 e-/ft2). Methods All adsorption experiments were performed on the same type of suspension. 0.25g (0.0088 oz) of TiO2 powder was mixed with 50g (1.76 oz) of solution (CTiO2 = 5 g/L (0.026 oz/in3)). Samples were prepared with increasing amount of adsorptive species and stirred for 24 hours at controlled room temperature (23°C (73.4°F)). Then, suspensions were ultra-centrifuged and the supernatant was collected to be filtered and analyzed. When the adsorptive species are ions, the analysis was always done by ICP spectrometry (Ciros Vision, SPECTRO Analytical Instruments GmbH). In case of the polymer, the analysis was preferentially done by TOC analysis (TOC II, Elementar Analysensysteme GmbH). However, this method was not always possible since the solutions are not always stable after acidification (a necessary step before the analysis). In this case, the determination of the polymer content in solution was done by ICP by measuring the C content. The calibration was made over a series of standards containing a known quantity of the same polymer, leading to a minimization of the matrix effects. Nevertheless, it has to be noted that ICP spectrometry on carbon was not as accurate as the TOC method. In some suspensions, portlandite was present or precipitated. A series of adsorption experiments on portlandite have been performed, and it was found that the PGP polymer does not significantly adsorb on portlandite, thus not causing an interference on the adsorption results on TiO2 (Data not reported in this paper). The zeta potential was measured in solutions containing 1g/L (0.000527 oz/in3) of TiO2 powder with a Zetasizer Nano ZS apparatus (Malvern Instruments Ltd). In order to vary pH, concentrated NaOH or HCl solutions were titrated. In some cases, different TiO2 suspensions were prepared over the desired pH-range with additions of Na2SiO3 or Ca(OH)2 solutions, in order to determine for example the effect of calcium or silicate ions on surface charging. EXPERIMENTAL RESULTS AND DISCUSSION Rheology of geopolymer suspensions. From this investigation, the rheological behavior of different suspensions is reported. These suspensions consist of two types of raw materials, either blast furnace slag (Table 1a) or fly-ash (Table 1b), and are activated with two types of solutions, either with a KOH solution or with a metasilicate solution. Though the polymer studied in this paper demonstrates a much higher efficiency than classical polycarboxylate ethers, its plasticizing efficiency varies according to (1) the composition of the geopolymer raw material and (2) the composition of the activator. Indeed, the plasticizing efficiency decreases when the binder is a class F fly-ash instead of a slag, and, drops dramatically when an alkali-silicate activator is used instead of KOH. Adsorption of polymer on TiO2 The adsorption isotherms of PGP are reported in this section. In order to qualitatively evaluate the efficiency of the molecule as a plasticizer from the adsorption results, a crosshatched zone, corresponding to the required adsorption level for a valuable plasticizing

480 SP-302-36

Table 1a - Mortar spread values measured according to the norm DIN EN 1015-3 on blast furnace slag-based mortar activated with KOH or Na2SiO3 solutions. 1% of polymer is added to the mixture. Dosages of activator/ polymer are expressed in weight of activator/dried polymer by weight of binder. The water to binder ratio is 0.5. Admixture none Polycarboxylate Ether PGP

7% KOH 19 cm (7.5 in) 20 cm (7.9 in) >30 cm (>11.8 in)

7% Na2SiO3 15 cm (5.9 in) 15 cm (5.9 in) 16 cm (6.3 in)

Table 1b - Mortar spread measured on low calcium fly ash-based mortars (class F in European classification). Same proportions are used as in Tab. 1a. Admixture none Polycarboxylate Ether PGP

7% KOH 19 cm (7.5 in) 19 cm (7.5 in) 27 cm (10.6 in)

7% Na2SiO3 17 cm (6.7 in) 17 cm (6.7 in) 17 cm (6.7 in)

effect, is drawn. This zone is estimated from adsorption curves realized with typical polycarboxylate ethers in cementitious systems and by comparison of the respective surface areas between TiO2 and cements. In addition, a dotted straight-line is given and represents a degree of adsorption equal to 50%. First, the results relative to the adsorption of PGP in different alkaline conditions are presented in Fig. 1. There is a very low adsorption of PGP at low pH (water) and at relatively low ionic strength (100 mM NaOH). The adsorption is slightly increased when 400 mM of NaCl is added to 100 mM NaOH or when the particles are mixed in 500 mM NaOH. Even if the critical level of adsorption may be reached, it requires substantially more polymer than typically for cement mixtures and the adsorption degree is only 10% which reveals a low affinity for the TiO2 surface in these conditions. Similarly to the addition of 400 mM NaCl in 100 mM NaOH (Fig. 1), the addition of low amounts of calcium chloride favors the adsorption of PGP (Fig. 2a). Under condition of a saturated solution with calcium ions, which can be realized for instance using a reservoir of portlandite (Ca(OH)2), the adsorption is drastically enhanced (Fig. 2b). It turns out that calcium ions likely mediate the adsorption of PGP as it has been already reported in many works related to calcium-rich systems (like cements).2 The difference of effect between the CaCl2 and the Ca(OH)2(sat) experiments is the saturation in calcium. Indeed, assuming also a complex formed with calcium and polymer molecules, the concentration of free calcium significantly decreases upon the addition of polymer. It thus leads to the defection of calcium close to TiO2 surface and calcium ions cannot mediate the adsorption anymore. Such an effect was already shown elsewhere.2,3 Further support for this hypothesis is found looking at the curve representing the addition of 5 mM CaCl2 (Fig. 2a). At the concentration of PGP above 4500 mg/L (0.002372 oz/in3), the adsorption slightly decreased as indicated by the bending of the curve (marked with an arrow). The charge density of this polymer is about 3 meq/g (85 meq/oz), 4500 mg (0.158 oz) or 1L of 4500 mg/L solution of this polymer can therefore complex about 6.75 mmol of Ca2+ assuming that 100% of the polymer charges are counter-balanced only by Ca2+. It is typically in the same order

Plasticizing Geopolymer-Type Auspensions: A Challenge  481

Figure 1 - Adsorption isotherm of PGP onto TiO2 in presence of different salts.

Figure 2 - (a) Adsorption isotherm of PGP onto TiO2 in equilibrium with 100mM NaOH and in presence of different concentrations of calcium chloride. (b) Adsorption isotherm of PGP onto TiO2 in equilibrium with Portlandite and in presence of different salts. of magnitude as the initial calcium concentration (5 mM) used in this case. A competition between the complexation of calcium by the polymer in solution and the adsorption of calcium onto the mineral surface can be assumed. The addition of NaCl does not alter the adsorption of polymer in saturated conditions with respect to portlandite, but the high pH conditions (500 mM NaOH) do. In 500 mM NaOH, the solubility of portlandite is about 1 mmol/L of Ca2+ which is insufficient to saturate the polymer molecules and the surface of TiO2 with calcium. Nevertheless, even at such a low concentration of Ca2+, it seems to be possible to reach the critical level of adsorption with high polymer concentration. Unfortunately, the measurement of adsorption isotherms in solutions combining calcium and silicate ions is not reliable because of the possible precipitation of calcium silicate hydrates, especially after 24 hours equilibrium time. Also, based on the adsorption isotherms performed so far, the main conclusion is that without calcium ions the adsorption of polymer is difficult. Calcium ions seem to be the main driver for the adsorption.

482 SP-302-36

Figure 3 - Adsorption of calcium ions on TiO2 at different pHs. pHs are maintained constant at pH=11 and pH=12 and considered as constant in 100 mM and 500 mM of NaOH, the pH is 12.91 and 13.48 respectively. Without calcium ions, no significant difference is seen between polymer adsorptions in alkali-hydroxide or in alkali-silicate activated solutions. Adsorption of anions and cations on TiO2 1) Ca2+—Inspired by their key-role in cementitious systems and in the adsorption of PGP, the adsorption of Ca2+ onto TiO2 particles was investigated and is reported in the following section. Fig. 3 shows the adsorption of Ca2+ at different pHs. The vertical adsorption slope at very low Ca concentrations reveals that the first calcium ions added to the solution are immediately adsorbed. At pH=11, all Ca ions are adsorbed up to the coverage of 1 Ca2+ per nm2 of TiO2. Above pH=11, the coverage of strongly adsorbed Ca2+ ions increases up to 1.5 Ca2+ per nm2 and over the whole alkaline pH range, the maximum Ca2+ coverage is about 2 Ca2+ per nm2 which is achieved when [Ca2+] = 8 mM at pH=11 and [Ca2+] = 1.5 mM at pH=13. It means that one calcium ion is complexed by one TiO2 negative site, which should lead to an overcompensation value equal to the absolute charge density. In turn, the electroneutrality is ensured by hydroxyl ions, the only anions present here. It is worthwhile to point out the significant variation of the calcium concentration next to the surface at low calcium concentration range. The concentration of calcium present in the solution can be a prime performance parameter. Actually, in such concentrated salt solutions, representative of geopolymer systems, one should not refer to the concentration but rather to ion activity. Other ions and especially the anions strongly modify the activity of calcium. In order to estimate the possible influence of these anions, different concentrations of sodium sulfate were added in the equilibrium solution and the calcium adsorption isotherms produced. The sulfate ions are divalent anions and form ion pairs with Ca2+ decreasing therefore the activity of Ca2+. It would have been difficult to use another polyvalent anion since all of them precipitate with calcium already at very low calcium concentration and it would have hinder the adsorption

Plasticizing Geopolymer-Type Auspensions: A Challenge  483

Figure 4 - Adsorption isotherms of calcium ions on TiO2 in 100 mM NaOH and in presence of different concentration of sodium sulfate, in function of (a) the calcium concentration in solution and (b) in function of the activity in Ca2+.

Figure 5 - Adsorption isotherms of silicate ions on TiO2 at different pHs. study on a meaningful calcium concentration range. The results with sulfates are shown in Fig. 4a. The presence of SO42- ions does not drastically change the adsorption of calcium, except at very low calcium concentration where a slight decrease is observed. This effect has almost vanished when the activity of Ca2+ is considered (Fig. 4b) 2) H3SiO4-/H2SiO42-—Other important anions in geopolymer systems are the silicate ions as concentrated silicate solutions (>1 M), which can also be used for the activation of aluminosilicate binders. In high alkaline conditions, the silicate ions are present as monomeric species once and twice deprotonated at low concentration of Si and as oligomers at higher concentrations of Si. It is assumed in this study that the adsorption of monomer species is representative of the adsorption of all silicate oligomers. The adsorption of silicate ions at different pHs is presented in Fig. 5. Silicate ions, negatively charged, adsorb on TiO2 also negatively charged in this pH range. This adsorption seems to be little depending

484 SP-302-36

Figure 6 - Zeta-potential evolution of TiO2 particles over pH in presence of different amounts of NaCl. The pH is given by addition of concentrated NaOH or HNO3 solutions. of the alkaline pH conditions. It would indicate a weak influence of electrostatics in this case. In silicate activated geopolymers, the concentration of silicate in solution is often higher than 1 M, i.e. higher than the concentration used for the adsorption isotherms in this study. Presuming that polymers and silicates can adsorb on the same TiO2 surface sites, the adsorption results above indicate a possible competition between both. This behavior is akin to the role of sulfate in cement paste with respect to the adsorption of polycarboxylate ethers. Electrokinetic measurements In order to corroborate the adsorption results, electrokinetic measurements have been carried out. These ones have the advantage to be easier and faster than the adsorption studies but have a disadvantage of stronger artifacts related to the method and the calculus of zeta-potential. In addition, it has to be pointed out that the technique measures the potential resulting from the surface charge of the particles plus the ions closely adsorb next to the particle. The definition of the ion cloud next to the surface which moves with the particle is not uniquely defined, and is a source of possible different interpretations. 1) Influence of ionic strength—A first series was carried out in different salt concentrations (Fig. 6) and illustrates well the purpose and the difficulty to get unique interpretation of the results. In the salt-free solution, the isoelectric point of TiO2 is found to be at pH=6 which appears to be a reasonable value4 for the anatase polymorph. The absolute zeta plateau values at low pH and high pH are quite similar. This indicates that, without additional salt, the number of sites being extra protonated and the number of sites being once deprotonated are similar and most probably the same considering the TiO2 surface. It is also clear that the addition of NaCl decreases the absolute zeta values at high pHs. This can be explained by two phenomena: (1) the specific adsorption of sodium (Na+) on the surface and (2) the so-called breaking forces depending on the ionic strength of the

Plasticizing Geopolymer-Type Auspensions: A Challenge  485

Figure 7 - Zeta-potential evolution of TiO2 particles over pH, when the pH is given by addition of NaOH or by addition of Ca(OH)2 or by addition of NaOH in a solution containing 2mM of CaCl2. solution which limits the displacement of the particles. Though quantitative description of both phenomena is not possible from solely electrokinetic measurements, some details of the curve help us to suggest qualitative statements. The isoelectric point is shifted to the higher pH with the addition of NaCl which indicates that sodium adsorbs. However, the shift is not proportional to the addition of salt. The absolute value of the pseudo plateau at high pH is decreasing with the increasing salt concentration, which can be attributed to the screening effect or to the adsorption of sodium ions. This variation is lower at very acidic pHs, i.e. the addition of salt influences only a little the positive zeta value due probably to the sole screening effect. This indicates that the variation at high pH is mostly due to the adsorption of sodium ions. At a given salt concentration, it seems also that the absolute zeta value is decreasing at very high pH (>12). It could be associated to the increase of ionic strength, i.e. a salt screening effect, but, it even leads sometime to a reversal of the zeta values (zeta>0) which cannot be attributed to a screening effect. The conclusion of this simple examination is that alkali ions adsorb but no qualitative information is given (ξ−values are not proportional to the charge next to the surface because of the screening effect). 2) Addition of Ca ions—As previously highlighted, calcium ions strongly adsorb on TiO2 at high pH. This phenomenon can be also revealed by electrokinetic measurements. Instead of an addition of NaOH to increase the pH, a solution saturated with respect to the portlandite is used ([Ca] ~ 22 mM at 25°C) and this way the pH and the calcium concentration simultaneously increase without adding any alkali. It leads to a reversal of the zeta potential at pH=10 as the curve with squares in Fig. 7 shows. Another manner is to add certain amount of Ca salt and to increase step-wise the pH with NaOH additions. At pH=10-11, the addition of 2 mM CaCl2 is sufficient to fully reverse the zeta potential, i.e. ζ ~ -30mV without salt and ζ ~ 30 mV with 2 mM CaCl2. It thus indicates that about one Ca2+ is bound to one negative site of TiO2. When pH further increases to pH = 12-13 the zeta potential decreases. Such a phenomenon is not observed with Ca(OH)2 solutions, and

486 SP-302-36

Figure 8 - Zeta-potential evolution of TiO2 particles over pH when the pH of the solution results of addition of NaOH or Na2SiO3. is therefore due to the presence of sodium ions added with NaOH. Sodium ions compete with calcium for the site adsorption, if one Na+ replaces one Ca2+ the positive charge next to the surface decreases. 3) Influence of silicate ions—The same approach has been applied to silicates ions. The results presented in Fig. 8 reveal undoubtedly that the silicate ions are adsorbed on the surface of TiO2. The zeta potential is even more negative when pH is given by the step-wise addition of a metasilicate solution (curve with squares) than by the usual addition of NaOH (curve with triangles). The authors recall that even more sodium ions are added with the silicate than with NaOH in order to achieve a same pH. With the addition of NaOH in a solution containing only 1mM Na2SiO3, the zeta values are also more negative, indicating also an adsorption of silicate ions even at low concentrations of silicate. The influence of calcium ions on the silicate adsorption was also investigated. The precipitation of C-S-H has to be avoided and it forced us to work at low concentration of calcium. Fig. 9 shows that the presence of silicate ions clearly influences the zeta potential leading to negative values even if calcium ions are in solution. From these sole results, it is difficult to conclude whether the silicate ions yield desorption of calcium or whether they adsorb. Thermodynamic calculations Calcium ions play a key role in the adsorption of anionic polymers. Whereas their influence in alkali-hydroxide solutions is readily measurable, the same measurements in alkali-silicate solution are more prone to interference since the formation of calcium silicate hydrates is possible and can thus produce erroneous results. In typical alkali-silicate activated geopolymers, the silicate concentration in solution is very high (>1M) and the supersaturation with respect to calcium silicate hydrates is rapidly reached as soon as some calcium is released by the dissolution of geopolymer raw materials. In addition, although poorly defined and never experimentally evidenced, the formation of CaH2SiO4, CaH3SiO4+ and CaHSiO4- ion pairs can be assumed.5 Both factors decrease the activity of

Plasticizing Geopolymer-Type Auspensions: A Challenge  487

Figure 9 - Zeta-potential evolution of TiO2 particles over pH in presence of calcium chloride. The pH is given with Na2SiO3 additions. Ca2+ which is the adsorption mediating agent for anionic polymers. In order to estimate this decrease, thermodynamic calculations with the software Phreeqc using the DebyeHückel’s approximation for the calculus of activities were carried out. Alkali-hydroxide or alkali-silicate solutions containing a fixed calcium concentration are put in equilibrium with calcium silicate hydrates. Equilibrium constants of calcium-silicate ion pairs are not known and therefore not considered in the calculus. It also means that this calculus maximizes the activity of Ca2+ since any other complexes involving calcium would further decrease the Ca2+ activity. A starting calcium concentration of 2 mM is chosen and the results are reported in Table 2. It can be concluded that, either for a same pH or for a same concentration of activator, the activity of Ca2+ is drastically reduced when a metasilicate activator is used instead of a pure alkali-hydroxide activator. This simple calculus indicates that the activity of calcium in solution and thus its adsorption on any mineral surface has to severely decrease in presence of silicates which should clearly disfavor the adsorption of anionic polymers. CONCLUSIONS The experimental observations related to the adsorption of PGP are summarized below: 1) The surface of TiO2 is highly negatively charged and at high pH, up to one calcium ion is adsorbed per surface charge, which leads to a large overcompensation next to the surface if hydroxide ions are the co-anions. 2) In the presence of only Na+ ions, a low adsorption of PGP is exhibited, but nevertheless, the higher the concentration of alkali ions, the greater the adsorption of polymer. The adsorption of PGP is getting higher by addition of calcium, but it is strongly depending of the concentration of free Ca2+ ions. With a reservoir of calcium, the adsorption of polymer is high. 3) The electrokinetic measurements indicate that in presence of silicate ions, the charge next to the surface is negatively charged, even in presence of calcium. There are two possible explanations: either the calcium desorbed in presence of silicate ions, or the silicate ions are adsorbed on top of calcium ions.

488 SP-302-36

Table 2 - Evolution of the activity of OH- and Ca2+ in different activator solutions when 2 mmol/L of calcium is added to a solution in equilibrium with respect to C-S-H.

NaOH NaOH Na2SiO3 Na2SiO3

Concentration [mmol/L] 500 1000 500 1300

[Ca2+] [mmol/L] 2 2 2 2

pH 13.46 13.71 13.04 13.46

(OH-) 2.87E-01 4.97E-01 1.06E-01 2.71E-01

(Ca2+) 1.94E-04 3.07E-05 6.01E-08 3.81E-09

4) The activity of calcium in solution drastically decreases in the presence of solubilized silicates, either due to the precipitation of calcium silicate hydrates and/or to the likely formation of calcium silicate ions. The results presented in this paper elucidate the difficulty to obtain a sufficient polymer adsorption, and thus a reasonable fluidification, in geopolymer systems. All factors influencing the calcium concentration at the interface between the mineral and polymer molecules are of prime importance. First, the dissolution of the various raw materials release more or less calcium according to their composition: class F fly-ashes deliver less calcium than blast furnace slags and are potentially less favorable substrates for anionic polymers. Second, the activity of Ca2+ decreases with the concentration of activator, and strongly drops in presence of silicates: silicate solutions prevent the adsorption compared to pure soda or potash solutions. Third, silicate ions, and in particular divalent H2SiO42- ions, adsorb on minerals, meaning that there is a potential competition between silicate and polymer which still makes the polymer adsorption in alkali-silicate activated systems more difficult. Taking into account this competition and the low concentration of Ca2+ in solution, polymers bearing a lot of negative charges, and especially anchor groups having a high affinity for calcium, are preferential candidates as anionic plasticizers for alkali-activated geopolymer systems. AUTHOR BIOS Luc Nicoleau is Senior Scientist within BASF Advanced Materials & Systems Research division, working at the global research center for construction of BASF. He received his PhD in 2004 from the University of Dijon. His research interests include the interactions cement-admixtures, the hydration of cement and the alkali-silica reaction. He was the main developer of a new accelerator technology based on suspended C-S-H nuclei: Master XSEED®100. Maxim Pulkin is a researcher within BASF Advanced Materials & Systems Research division. He received his PhD in 2011 from the University of Bremen (Germany). His research interests are in area of materials science and currently include the interactions between inorganic particles and polymers as well as the chemistry and properties of alternative (non-OPC) inorganic binders. Tatiana Mitkina is a researcher within BASF Advanced Materials & Systems Research division. She received her PhD in 2008 from the Nikolaev Institute of Inorganic Chem-

Plasticizing Geopolymer-Type Auspensions: A Challenge  489

istry SB RAS (Novosibirsk, Russia). Her current research interests are focused on development of admixtures for alternative inorganic binders, in particular for alkali activated binders, and, on additives for mining applications. REFERENCES 1. Geopolymers Structure, processing, properties and industrial applications, Edited by J.L. Provis & J.S.J. van Deventer, Woodhead Publishing Limited and CRC Press LLC Cambridge 2009 2. Turesson, M.; Labbez, C.; and Nonat, A., “Calcium mediated polyelectrolyte adsorption on like-charged surfaces,” Langmuir, V. 27, No. 22, 2011, pp. 13572-13581. doi: 10.1021/la2030846 3. Flood, C.; Cosgrove, T.; Espidel, Y.; Howell, I.; and Revell, P., “Sodium polyacrylate onto anionic and cationic silica in the presence of salts,” Langmuir, V. 23, No. 11, 2007, pp. 6191-6197. doi: 10.1021/la070047z 4. Bourikas, K.; Hiemstra, T.; and van Riemsdijk, W. H., “Ion Pair Formation and Primary Charging Behavior of Titanium Oxide (Anatase and Rutile),” Langmuir, V. 17, No. 3, 2001, pp. 749-756. doi: 10.1021/la000806c 5. Taylor, H. F. W., Cement Chemistry 2nd Edition, Thomas Telford edition, 1997

490 SP-302-36

SP-302-37

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials by Wolfram Schmidt, Nsesheye S. Msinjili, Herbert C. Uzoegbo, and John K. Makunza The economic use of chemical admixtures depends on supply chains. Therefore, in most regions ins sub-Saharan Africa (SSA), the use of admixtures is not common practice. This amplifies the unfavorable framework for concrete construction such as fragmentary supply chains, high local cement prices, and unfavorable construction site facilities in this region significantly. The use of superplasticizer (SP) and stabilizing agents (STA) can enhance the concrete technology in SSA, since they can disassociate the concrete quality from external boundary influences. After providing a general overview of the peculiarities of the SSA boundary framework, economic concepts are provided, how existing material solutions can be significantly improved by the use of SPs and STAs based on locally available materials such as lignosulphonates and cassava starch. Finally a three step optimization process is described that helps developing flowable concrete based on materials that can be accessed in most locations in SSA. Keywords: admixtures; lignosulphonate; polycarboxylate ether; robustness; selfcompacting concrete; starch ether; sub-Saharan Africa. INTRODUCTION The sub-Saharan African environment for concrete casting varies greatly from what can be found in most industrialized countries of the northern hemisphere. For instance, the price of cement in SSA ranges between 
8 - 16 USD per 50 kg (110 lbs) bag, which is the standard delivery. Regionally, the sale prices can be relatively higher depending mainly upon the logistic effort. If these prices are related to the gross national income, identical volumes of cement can be several hundred times more expensive in SSA than in Europe or Northern America.1,2 Considering that labor costs are extremely low in SSA, it is obvious that the basis for constructional cost calculation is exactly the opposite of Europe, Northern America, or Japan.3 Hence, it is worthwhile for engineers in SSA not to follow the existing trend of concrete regulations in Europe or Northern America but to develop customized concepts. 491

492 SP-302-37

The current situation for casting concrete in the majority of SSA is done on-site. Pre-cast concrete or ready-mix concrete is uncommon outside of South Africa. The disadvantageous consequences are often the lack of quality control and homogeneity of the final product. This is often accompanied by unsteady supply chains and expensive thus often unaffordable rental fees for adequate equipment, which often does not belong to the contractor. Most processes (especially concrete mixing) are conducted manually, and often the only possibility to adjust an appropriate consistency and workability retention is by addition of water on haptic and visual assessment. Nevertheless, the advantage of having on-site concrete casting in SSA is the availability of cheap labor. Therefore, the solution is not to further incorporate pre-cast or ready-mix concrete, but rather to take advantage of what is readily available in the market, i.e. local available materials and cheap labor. Solutions for the casting of concrete in sub-Saharan Africa The existing situation in SSA as described before adversely affects concrete performance and durability. In order to enhance the situation, it is desirable to exclude as many factors as possible that result in a negative impact on concrete casting. A similar consideration was the trigger for the invention of self-compacting concrete (SCC) in the 1980s in Japan. Some leading scientists observed a correlation between the decreasing skills of laborers and decreasing structural durability.4-6 They found that the only solution to counteract this trend was a durable concrete that performs reliably regardless of the laborers’ skills. Although Japan and SSA have greatly differing characteristic in terms of their geographic and economic conditions, the situation that existed in Japan before the concept of SCC, is somewhat comparable to the current situation in SSA. The difference is that beyond the laborers’ skill level SSA exhibits additional disadvantageous influencing parameters on construction sites. Furthermore, SSA exhibits a clear necessity to save costs, and the construction needs in SSA are more related to low-cost housing, multi-story buildings and infrastructure rather than to skyscraper towers. Therefore, if SCC is opted as a solution in SSA, it is important to discuss about tailored concepts for the local boundary framework. Flowable concrete solutions for high temperatures Due to its accelerating effect on the hydration, hot climate conditions cause rapid loss of workability.7,8 Short workability retention is disadvantageous for construction sites in SSA due to the unsteady casting logistics caused by unaffordable equipment such as pumps and machinery. The major influencing factors of flowable concrete that determine the flow retention are the charge density of the SP and the water to powder ratio (w/p).7,8 Apart from that the use of a set retarder can significantly enhance the workability retention. Realistically, a SP-based solution to customize flowable concrete for high temperatures can only be achieved by using polycarboxylate ether (PCE) due to the high versatility. In this case, it is beneficial to use low charge agents, since they provide better workability retention due to their lower adsorption tendency on cement hydration phases. Higher charge SPs are adsorbed on surfaces rapidly, but in return become inefficient rapidly, due to morphologic changes of the particles caused by the high temperature induced rapid hydration reaction. Low charge SPs adsorb at a lesser degree initially, causing a higher polymer content in the solution which allows for ongoing adsorption upon growth of new hydrates with time (Fig. 1).7-9

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  493

Fig. 1–Approaches to obtain workability retention for concrete at high temperatures. Since PCE is hard to purchase outside of the country of South Africa, a temperature optimization based on the polymeric architecture is a less feasible solution for the moment. Therefore, flowable concrete concepts need higher w/p values. As illustrated in Fig. 1, low w/p mixtures stiffen quickly due to rapidly occurring morphological changes on the surfaces that cause lower mobility of particles.10 Consequently, SCC optimized for hot weather conditions should not be designed for ultimately low w/p values (w/p ratios between 0.5 and 0.65 might be realistic).8 In this case, good flow properties can also be achieved with SPs based on lignosulfonates (LS) or poly naphthalene sulphonates (PNS). However, the higher w/p may also induce a tendency to segregation, which may need to be compensated by STA, which in return adds a sophisticated component into the mixture composition. Applicable solutions for flowable concrete based on local raw materials Major concerns about flowable concrete mixtures in SSA are the affordability, the availability of construction chemicals, as well as how to bring them into practice in an environment, which is not optimized for robust concrete proportioning and casting. Since cement is the most cost-driving factor in SSA at the moment, any method that helps saving cement helps reducing material costs. Appropriate mixture composition with optimized particle packing help significantly in reducing the costs. Furthermore an economically reasonable implementation can only be conducted by incorporating locally available materials such as rice husk or bagasse ashes, limestone fillers or natural pozzolans.

494 SP-302-37

Table 1–Raw materials and admixture solutions for sub-Saharan African concrete Constituent Cement extenders

Typically used in SCC Limestone filler, groung granulated blast furnace slag, fly ash

Fillers

Limestone filler, fly ash, silica fume

Superplasticizer Stabilizing agent

Polycarboxylate based Cellulose, potato starch, sphingans

Locally available alternatives in SSC Natural pozzolans, limestone filler, ricehusk ash, bagasse ash Natural pozzolans, limestone filler, ricehusk ash, bagasse ash Lignosulfonates, naphthalene sulfonates Cassava/maize/potato starch, cellulose

Since the demand for daily concrete practice in SSA is mainly normal concrete, it is worthwhile taking into consideration SPs such as LS and PNS, which are cheaper and readily available in SSA. Particularly LS seems to be an attractive solution, since it does not need sophisticated processing and occur as a waste of the cellulose production, which exists all over Africa. However, since LS is available at a low price, the calorific value of LS also makes it attractive as a source of energy in areas of high energy prices. This is the reason, why some admixture companies project better market chances for PNS products. In conjunction with a moderately high w/p, LS and PNS can function rather effectively as well as for self-compacting or very flowable concrete. However, a moderately high w/p may cause segregation. This can be counteracted by STAs, e.g. based on Cassava, which grows nearly everywhere on the continent. The starch of Cassava is very similar to the starch of potatoes, which are used effectively as STAs for concrete. Using materials from local sources, some of which are listed in Table 1, can help in developing economically efficient SCCs for the SSA markets. To overcome the problem field of unaffordable equipment and supply chains for proper proportioning and casting, a concept as illustrated in Fig. 2 can be adopted. Well-proportioned and pre-homogenized dry compounds including cement, fillers, SPs, STAs, and sand (e.g. up to 2 mm/0.08 in.) can be delivered to the construction sites, as opposed to the individual constituents delivered separately. Once on-site, such pre-homogenized dry compounds solely need to be amended by water and coarse aggregates in order to provide reproducible flow performance. The feasibility of such a concept has previously been proven for SCC.11 RESEARCH SIGNIFICANCE Markets in SSA are on the rise. This exhibits a high pressure on the local cement and concrete industry to build up state of the art houses and infrastructure. However, the boundary framework shows a significant number of drawbacks in casting of concrete. Flowable concrete or SSC can significantly contribute in enhancing the durability and safety of concrete structures in SSA. However, in order for it to be feasible in SSA, it needs to be customized for the local boundary framework and supplies. A study of how this can be practically done is presented in this paper. EXPERIMENTAL PROCEDURE The aim of the study was to prove that a pre-mixed binder compound can obtain highly flowable fresh concrete properties and function well despite the lack of sophisticated raw materials. Therefore preferably materials from SSA were used. Other materials

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  495

Fig. 2–Comparison of state of the art in SSA and a pre-mixed mortar concept. were replaced by materials from Germany, which could be obtained in SSA with similar characteristics. Materials under investigation The materials taken into account for the development of a pre-mixed dry binder compound are listed in Table 2. The admixtures were commercial powder type products for the use in dry mortar technology. They are listed in Table 3. Since the focus of the study was on the entire concrete concept, the admixtures were not further specified. Water demand of binder pastes and binder-sand mixtures In order to reduce the cement content, the water demand of cement with different replacement ratios of powders was determined using the Puntke-method. The lower the water demand, the higher the packing density becomes. The principle also functions for finding the optimum adjustment of binder paste and sand to obtain the highest possible packing density. For the test, which is described in detail elsewhere,8,12,13 water is added gradually to a powder under constant stirring and agitation until the powder or powder compound tends to settle with a remaining surface texture and a light shimmer, but without bleeding water. The water demand can be derived from equation (1). For the investigations

496 SP-302-37

Table 2–Mineral binder materials, sand and aggregate properties Material Ordinary portland cement Limestone filler Rice husk ash Fly ash Natural pozzolan Sand Aggregate

Provenience Germany Germany Tanzania Tanzania Uganda Germany Germany

Specification CEM I 42.5 R Burnt under random conditions Burnt under random conditions Ground at BAM laboratory Fractions 0-0.5/0.5-1.0/1.0-2.0 Fractions 2.0-4.0/4.0-8.0/8.0-16.0

Specific gravity 3.12 2.74 2.07 2.06 2.50 2.60 2.60

Table 3–Commercially available admixtures used in the experiments Admixture type Polycarboxylate ether Lignosulfonate Naphthalene sulfonate Cassava starch

Provenience Germany Germany Germany Nigeria

Specification Powder type Powder type Powder type Powder type, cold-water soluble

in sand-binder mixtures, the sand fractions 0-0.5, 0.5-1.0, 1.0-2.0 were mixed in a ratio of 25:25:50. .nW = VW / (VW+VP)

(1)

where: nW = water demand [-]; VW = water volume [ml]; VP = volume of solids [ml] Determination of the optimum admixture dosage For the determination of the necessary admixture dosage a method was used, which was recommended by Schmidt8 in order to determine the adsorption properties of PCE without sophisticated equipment. Spread flow tests were conducted with increasing SP dosages at powder water mixtures determined according to equation (1). This low water dosage is necessary to avoid segregation at high SP dosages and to make sure that the paste definitively does not flow without SP addition. The resulting curve provides information about the dosage of a SP, which is required to first induce flow. It also shows the dosage above which no further yield stress reduction can be obtained.8,14,15 After 0.5 minutes of dry mixing, the water was added and the wet compound was mixed for 2 minutes at a high rotational speed. After that, the mixture was left to rest for seven minutes before the SP was added, and it was mixed again for one minute at a low rotational speed. The reason for the resting period was to avoid intercalations and influences of the formation of ettringite and monosulfate during the first couple of minutes. Rheometric investigations of the temperature sensitivity Rheometric investigations on mortars were conducted with a Couette type viscosimeter (Schleibinger Viskomat NT) using a basket cell.8,16 The measurement setup is shown in Table 4. The investigated mortars were tested at 5, 20, 30, and 40 °C. The measurements were conducted at 5 and 20 minutes following the respective mixing.

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  497 Table 4–Test protocol for the rheometric investigations of mortars at varied temperature Shear rate [s-1] Duration [s]

24.5 60

21.4 30

18.4 30

15.3 30

12.2 30

9.2 30

6.1 30

3.1 30

Table 5–Test protocol for the rheometric investigations of concrete Shear rate [s-1] Duration [s]

0.47 10

0.4 5

0.33 5

0.27 5

0.20 5

0.13 5

0.7 5

These investigations were only conducted with the PCE type SP and the compound including natural pozzolan. Since the aim of the study was the development of a dry powder compound, for these investigations, the powders including the powder type SP were mixed for one minute before water addition, thereafter mixing was continued for another four minutes at a low rotational speed. Determination of the maximum aggregate volume After the development of a mortar compound rheometric investigations were conducted with a concrete rheometer (Rheometer 4-SCC) on mortar with increased aggregate content. The aggregates fractioned 2-4 mm (0.08-0.16 in.), 4-8 mm (0.16-3.15 in.), and 8-60 mm (0.31-0.63 in.), were blended in the proportion 30:40:30, respectively. A reference mortar mixture of 10 l with a maximum aggregate size of 2 mm (0.08 in.) was amended by aggregates in increments of 2 kg (4.4 Lbs), which is equal to a supplementary volume of 0.77 l. The measurement setup can be taken from Table 5. OBSERVATIONS Powder optimisation and binder-sand mixtures As long as the powders do not significantly vary in their water absorption, the water demand of a powder mixture is a good indicator for the packing density of powders. If two powders are blended, the blend exhibiting the lowest water demand could thus be considered as the mixture with the highest packing density. Fig. 3 shows the water demand of cement (at 0% replacement) and the varied fillers (at 100% replacement) as well as their blends in different replacement ratios. It can be found that the water demands of pure limestone filler and pure natural pozzolan were lower than the water demands of pure cement, and that with increasing replacement of cement by these fillers the water demand can be reduced. The water demand of the fly ash (FA) as well as the rice husk ash (RHA) was significantly higher than the water demand of cement. These materials were burnt under uncontrolled conditions in non-industrialized processes. The LOI for both materials was very high and the water absorption was also very high due to the porous structure. Above a replacement ratio of 20% the water demand increased steadily. However, it could be found that up to a replacement ratio of about 20% the water demand in case of RHA maintained at a similar level as for pure cement and was even lower in case of FA. For further investigations a volumetric replacement ratio of 20% was chosen. Fig. 4 shows the respective compressive strength at varied replacement ratios and at water content as determined from the water demand experiments. After 7 days there

498 SP-302-37

Fig. 3–Water demand of cement filler mixtures at varied replacement ratio.

Fig. 4–Compressive strength for different replacement ratios after 7 and 28 days. was only a negligible effect of the replacement of cement with limestone filler, and the replacement with RHA showed an increased strength. The replacement of cement with FA and natural pozzolan had negative effects on the compressive strength. Nevertheless at 20% replacement the natural pozzolan showed a similar behavior as the limestone filler. Whether the increase in strength between 15% and 20% replacement with natural pozzolan

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  499

Fig. 5–Water demand of binder (80% cement, 20% filler) sand (max grain size 2 mm) mixtures at varied replacement ratio. can be explained by the higher packing density, a beneficial chemical interaction between the binder components, a modified curing process or a testing error cannot be derived from the available data. It is interesting to observe that both, the FA and the RHA have a distinct negative effect after 28 days, while the performance loss of mixtures incorporating natural pozzolan or limestone filler show reduced strength properties in the order of magnitude of the replacement rate. Both ashes were derived from an incomplete combustion process with a high loss on ignition, containing a high content of sulfates and chlorides. Therefore it is assumable that the chemical composition of the ashes may have had a compromising effect on the mechanical performance of the specimens. Subsequently, mortars were generated with different replacement ratios of binder, and the water demand was similarly determined according to equation (1). The results are presented in Fig. 5. It can be seen that for all fillers, the minimum of the water demand occurs between 15% and 30% binder volume. In order to support the self-compacting properties, for further investigations a volumetric powder fraction of 30% was chosen for all powders. Admixture adsorption Based on the investigations presented in Fig. 5 the mixture compositions presented in Table 6 were derived. These mortar mixtures were further investigated on their interactions with the SP types as listed in Table 3. For these investigations the mortars in Table 6 were mixed with different SP dosages and spread flow tests were conducted. The aim was to identify the SP dosage at which maximum spread flow occurs. It was found that the mixtures containing FA and RHA hardly showed any flowability regardless of the SP type or dosage. By increasing the water content, flowability could be obtained, most effectively by using polycarboxylate ether superplasticizer. However, at the same time, the systems with these two filler materials showed distinct segregation.

500 SP-302-37

Table 6–Mortar mixture compositions derived from binder-sand optimizations Water Binder Sand demand Vol.-% of Vol.-% of Vol.-% of Filler type solids solids total Limestone filler 30% 70% 22% Rice husk ash 30% 70% 30% Fly ash 30% 70% 24.5% Natural pozzolan 30% 70% 20.5%

Cement [kg/m3] (lb/ft3) 580(36.2) 521(32.5) 562(35.1) 591(36.9)

Filler [kg/m3] (lb/ft3) 128(7.8) 87(5.4) 93(5.8) 119(7.4)

Water Sand w/b [kg/m3] [kg/m3] (lb/ by (lb/ft3) ft3) mass 220(13.7) 1420(88.6) 0.31 300(18.7) 1274(79.5) 0.49 245(15.3) 1374(85.8) 0.37 205(12.8) 1447(90.3) 0.29

Fig. 6–Slump flow with increased dosage of PCE, PNS, LS in powder mixtures containing limestone filler and natural pozzolan. The reason for the negative influence of the FA and RHA used can be found in the incomplete and uncontrolled incineration conditions. They exhibited a high loss on ignition, very porous surfaces and high water absorption. Furthermore their high gypsum content is likely to cause interactions with adsorbing SPs, since SPs and sulfate ions adsorb on hydration phases like ettringite and monosulfate competitively.17 Based on these observations, it can be concluded that without further processing and without a controlled burning process, these raw materials cannot be considered to be feasible materials for flowable concrete. The slump flow evolution of the limestone filler mixture at w/b = 0.31 (Table 6) can be seen in Fig. 6. PCE improved the flow properties significantly, while PNS and LS were required at higher dosages and could not achieve the same slump flow level. If the test was repeated for the natural pozzolan at w/b = 0.29 (Table 6) it was found that all SPs showed poor performance. Therefore, the water was increased stepwise at a PCE dosage of 1.6% solids related to cement (Fig. 7). The PCE dosage is equal to the dosage that yielded the widest slump flow diameter with limestone filler. It can be observed that increasing water demands enhance the slump flow value as expected. During the test, up to a w/c of 0.5, no serious segregation could be observed, indicating that the material has high water binding capacity.

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  501

Fig. 7–Increased slump flow with increasing w/p for the mixture containing natural pozzolan and a PCE solid dosage of 1.6% of the cement. Therefore, the slump flow tests with increased SP dosages were repeated at a w/b of 0.5 for the natural pozzolan. These results can also be seen in Fig. 6. While for PNS dosages above 0.4% the paste showed distinct segregation, with LS, the dosage could be increased and a maximum flow was observed at a dosage of 1.2% with slight signs of segregation. The flow performance compares to the limestone filler paste at w/b = 0.31 with PCE. The experimental results clearly indicate that the highest efficiency can be achieved by using PCE superplasticizers. In addition the formerly discussed versatility of PCE and the possibility to customize the performance according to the ambient temperature influences clearly advocate PCE for ultimate performance. However, in the sub-Saharan African environment the majority of construction needs are in the field of housing and infrastructure where the challenge is more related to cost issues rather than ultimate performance. The supply chains for PCE in SSA are fragmentary. Hence it is worthwhile looking at concepts that can be implemented on the short term. Therefore, the aim of the investigation was to obtain SCC without sophisticated or expensive constituents like PCE and without the need for ultimate performance. As a consequence, the final mortar with a moderate w/b of 0.5 was chosen, as well as the lignosulphonate plasticizer. In order to avoid segregation, the mixture was amended by adding 1.0% STA based on cassava starch related to the water or 0.58% related to the cement. Cassava starch is easily available at a low price in many countries in Africa. Temperature behavior The obtained mortar was investigated in a rheometer at varied temperatures. The results for yield stress and plastic viscosity after 5 and 20 minutes can be seen in Fig. 8. There is no significant influence of the temperature on the yield stress of the mortar, and the values maintain stable over the first 20 minutes. There is similarly no significant influence of the temperature on the plastic viscosity, but the viscosity increases at all temperatures between 5 and 20 minutes. Since yield stress is mainly affected by the SP adsorption

502 SP-302-37

Fig. 8–Yield stress and plastic viscosity of the derived mortar at varied temperature.

Fig. 9–Torque gradients and ordinate intercepts at varied aggregate contents. and plastic viscosity predominantly by the solid volume fraction, this indicates that the natural pozzolan does not withdraw active SPs over the course of time, but rather obviously absorbs water, thus increasing the plastic viscosity. Aggregate dosage The aim of the study was to develop a pre-mixed dry mortar compound to be amended by coarse aggregates and water, to obtain concrete with good flow properties. Therefore, in the final step it was evaluated, which volume of aggregates can be added to the mortar composite. Fig. 9 shows that increasing additions of aggregates mainly increase both yield stress and plastic viscosity. At an addition of 6 kg, which is equal to the total coarse aggregate volume of 18.8%, the yield stress was too high to obtain good self compacting proper-

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  503 Table 7–Mixture composition of the derived self-compacting concrete Homogenized, pre-mixed dry mortar compound

Resulting SCC

Cement 385 kg/m3 24.0 lb/ft3

Natural pozzolan 78 kg/m3 4.8 lb/ft3

Sand 0-2 mm 944 kg/m3 58.9 lb/ft3

Ligno sulfonate 1.78% of cement

Cassava starch 0.58% of cement

To be added on-site Aggregate2-16 Water mm 232 kg/m3 650 kg/m3 14.5 lb/ft3 40.6 lb/ft3

ties. Therefore, the dosage of LS was increased, which signifiantly reduced the yield stress but also increased the plastic viscosity. At an aggregate addition of 8 kg, which is equal to a total coarse aggregate volume of 23.5% after further addition of LS, a good SSC could be obtained with a sufficiently low yield stress and high plastic viscosity. A further increase of the aggregate was considered to be critical. COMPARISON OF THE CONCEPT WITH DATA FROM PRACTICE Based on the stepwise development, a pre-mixed dry mortar compound could be developed which can be amended by coarse aggregates and water according to the mixture composition in Table 7. In experiments with the aggregates specified in Table 2, a slump flow of 635 mm (25 in.) could be obtained. The 28-d compressive strength was 56.2 MPa, and the 90-d compressive strength was 68.4 MPa. The authors found a high number of construction sites in SSA, where the concrete strength was lower than 20 MPa despite high cement contents of more than 385 kg/m3 (24.0 lb/ft3) and w/c below 0.5. Furthermore, the scatter of the performance was significant. The pre-mixed binder compound can therefore significantly enhance the casting situation on construction sites in SSA, without the need to change the existing technology. FURTHER RESEARCH It was found that the observed RHA and the FA were not feasible for concrete. This, however, is not related to the material itself, but rather the processing technology. Further studies will focus on how the performance of these ashes can be enhanced through tempering or adequate processing. The cement content in the developed mixture was 385 kg/m3 (24.0 lb/ft3) and the strength after 28 days was 56.2 MPa. Since for most daily concrete works in SSA lower strength values are sufficient and the need to save cement is high, further studies will focus on reducing the cement by replacement with inert materials. The study could show that under laboratory conditions a functioning SCC could be derived with admixtures and additions of low level of sophistication. However, further studies under practical conditions need to be conducted to show the robustness of the concept as well as the robustness of the used admixtures. SUMMARY AND CONCLUSIONS The framework for casting concrete in SSA shows a number of peculiarities with negative effect on the safety and durability of structures. SSC or flowable concrete can be a feasible way to uncouple the concrete quality from the unfavorable boundary framework.

504 SP-302-37

SSC can be achieved from resources that are available in SSA and without sophisticated admixtures. The mortar compound presented in this study consists of ordinary portland cement, natural pozzolan, sand, lignosulphonate and cassava starch. In order to generate SSC with these materials, the w/p has to be moderately high, but the strength results show that the mechanical performance is much higher than the normal design strength for concrete in SSA. AUTHOR BIOS Dr. Wolfram Schmidt is a researcher at the BAM Federal Institute for Materials Research and Testing in Berlin. He received a Dipl.-Ing. from RWTH Aachen and a PHD from TU Eindhoven. His research focuses on SCC, admixtures, and rheology. He is member of the RILEM committee 228-MPS and the fib task group 8.8. MSc. Nsesheye S Msinjili is a researcher at BAM Federal Institute for Materials Research and Testing in Berlin. She received a MSc. in Structural & Fire Safety Engineering at University of Edinburgh and a BSc. (Hons.) in Civil & Structural Engineering at University of Dar es Salaam. She is a Structural Engineer with experience in concrete design in Africa. Prof. Herbert C Uzoegbo of the University of the Witwatersrand in South Africa graduated as Dipl. Ing. in Civil Engineering at the University of Bucharest. He also obtained MSc and DIC in concrete structures from Imperial College, University of London and a PhD in concrete structures from Kings College, University of London. He is a fellow of the South African Institution of Civil Engineers and of the International Masonry Society, UK. Dr. John K Makunza is senior lecturer at the University of Dar es Salaam in the Department of Structural and Construction Engineering. He obtained his PhD from the Technical University of Dortmund. He is a Civil and Structural Engineer expert with local and international experience in his field. He has worked with UN-WFP as a Consultant Civil Engineer for both civil and structural works. REFERENCES 1. Schmidt, W.; Hirya, N. N. M.; Bjegovic, D.; Uzoegbo, H. C.; and Kumaran, S. G.American Ceramic Society Bulletin, V. 91, 2012, p. 82 2. Schmidt, W.; Radlińska, A.; Nmai, C.; Buregyeya, A.; Lai, W. L.; and Kou, S., International Conference on Advances in Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013. 3. Schmidt, W.Concrete Trends, V. 16, 2013, p. 18 4. Okamura, H., and Ouchi, M., 1st International RILEM Symposium on Self-Compacting Concrete, Stockholm, Sweden, 1999. 5. Ozawa, K.; Maekawa, K.; Kunishima, M.; and Okamura, H., The second East-Asia and Pacific Conference on Structural Engineering and Construction, 1989. 6. Nagataki, S., International Symposium on Mineral and Chemical Admixtures in Concrete, Toronto, Canada, 1998.

Admixture Concepts for the Sub-Saharan African Environment with Indigenous Raw Materials  505 7. Schmidt, W.; Brouwers, H. J. H.; Kühne, H.-C.; and Meng, B., “Influences of superplasticizer modification and mixture composition on the performance of self-compacting concrete at varied ambient temperatures,” Cement and Concrete Composites, V. 49, 2014, pp. 111-126. doi: 10.1016/j.cemconcomp.2013.12.004 8. Schmidt, W., Eindhoven University of Technology, 2014. 9. Schmidt, W.; Brouwers, H. J. H.; Kühne, H.-C.; and Meng, B., in Design, Production and Placement of Self-Consolidating Concrete - Proceedings of SCC2010, Montreal, Canada, September 26-29, 2010, eds. K. H. Khayat and D. Feys, Springer, 2010, pp. 65-77. 10. Schmidt, W.; Msinjili, N. S.; and Kühne, H.-C., International Conference on Advances in Cement and Concrete Technology in Africa, Johannesburg, South Africa, 2013. 11. W. Schmidt, H.-C. Kuehne, D. Rosignoli and B. Meng, Concrete Plant + Precast Technology, 2008, 4-11. 12. Hunger, M., Eindhoven University of Technology, 2010. 13. W. Puntke, beton, 2002, 2002, 242-248. 14. Schmidt, W., 22. Workshop und Kolloquium Rheologische Messungen an Baustoffen, Regensburg, Germany, 2013. 15. Schmidt, W.; Brouwers, H. J. H.; Kühne, H.-C.; and Meng, B., in 7th International RILEM Symposium on Self-Compacting Concrete in Conjunction with the 1st International RILEM Conference on Rheology and Processing of Construction Materials - Supplementary Papers, ed. N. Roussel, Paris, France, 2013. 16. R. Vogel, Concrete Plant + Precast Technology, 2008, 124-126. 17. Plank, J., and Vlad, D.Zement-Kalk-Gips International, V. 59, 2006, pp. 28-39.

506 SP-302-37

Index A additive 211 admixture 63, 155, 359, 401 admixtures 145, 491 adsorbed layer thickness 63 adsorption 113, 145, 169, 183, 227, 243, 299, 333, 349, 401 adsorption of superplasticizer 425 alite 227 aluminate 349 amine 11 anionic charge amount 169 antifreezing admixtures 279

compressive strength 289 contact angle 211 coordinated water 463 copolymerization 25 crosslinking 199

D diester 199 dispersability retention 243 dispersing force 25 dispersing performance 199 dispersion 155 diutan gum 39

B

E

backbone 265 bentonite 333 blast-furnace slag 113 blast-furnace slag cement 113 blended admixtures 279 blended cements 299, 387 booster 359 brown coal 63

early compressive strength 359 ettringite 401 exchangeable cations 463 expansive agent 289 extreme freezing-point 279

C C3A 227 calcined clay 299 calcined marl 387 calcium oxide 359 Calorimeter 93 cement 11, 53, 199, 211, 315, 349 cement dispersion 401 cement hydration 145, 371 cement paste 371, 425 chemical admixture 113 civil engineering concrete 77 clay 333 clay interaction 463 clinker 11, 211 comb copolymers 227 complexation 349

F fiber reinforced mortars 289 flexural strength 289 fluidity 53, 183, 425 fluidity retention ability 113 fluoride ion 425 fly ash 359, 387 formwork filling 449 free length-change 289

G gel permeation chromatography 227 geopolymer 477 glass fibers 289 glycol 11 graft copolymer 63, 155 grafting 25 grinding aid 211 grinding aids 11

induction period 227 intercalation 333 interlayer space 463 internal curing 371 IPEG 265

polycarboxylate based superplasticizer 425 polycarboxylate ether 39, 53, 199, 491 polycarboxylate superplasticizer 183, 333 polymer 53 polymer intercalation 463 polymerization mechanism 155 polynaphthalene 93 polyols 349 polysaccharides 39 poly(vinyl alcohol) 333 polyvinyl-alcohol fibers 289 pore solution 169, 253, 299 portland cement 371 portland slag cement 169 pumping 77

K

R

kinetics 349

rate of heat liberation 437 reactivity 349 restrained length-change 289 retarder 93 rheology 93, 299, 315 robustness 491

H hardening accelerator 359 heat liberation 437 high-alkalinity 477 high range water reducer 401 high solid-content 125 hydration 227, 349 HydroxyPropyl Guar 315

I

L lignite 63 lignosulphonate 491 limestone filler 359

M macromonomer 199 manufactured sand 333 microstructure 25 molecular design 183 morphology 401 mortar 315, 333

N nano-size 401 natural zeolite 415 neutron radiography imaging 371

P PCE 265, 387 PCE superplasticizers 227 phenol 11 phosphonate 77 plasticizers 387 polycarboxylate 25, 93, 125, 155, 169, 243, 401

S segregation 53 self-compacting concrete 449, 491 set accelerators 145 set retarder 145, 437 setting control 145 setting time 145 side chain 265 silica fume 371 silicate adsorption 477 slow-release 243 sludge water 437 slump 53 slump retention 53 sodium gluconate 437 SRA 289 starch 39 starch ether 491 star-shaped 183 static yield stress 11 steel fibers 289 stereochemistry 349

structural parameters 265 sub-Saharan Africa 491 sulfonated acetone formaldehyde 93 sulphate ion 425 superabsorbent polymer 371 superplasticizer 63, 77, 169, 199, 279, 299, 449, 463 superplasticizer adsorption 253 superplasticizers’ combination 415 superplasticizer’s demand 415 supplementary cementitious material 299 surface tension 211 synthesis 155

V

T

X-ray powder diffraction 227

temperature 93 thixotropy 449

Z

viscosity 77 volume fraction 39

W water reduction 53 water retention 315 wetting 211 workability 125, 299 workability loss 415

X

zeta potential 93, 169, 243, 253, 299

9 781942

727224