The Pennsylvania State University
The Graduate School
College of Engineering
INNOVATIVE METHODS TO MITIGATE ALKALI-SILICA REACTION IN CONCRETE MATERIALS CONTAINING RECYCLED GLASS AGGREGATES
A Dissertation in
Civil and Environmental Engineering
by
Seyed-Mohammad-Hadi Shafaatian
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Doctor of Philosophy
December 2012
The dissertation proposal of Seyed-Mohammad-Hadi Shafaatian was reviewed and approved* by the following:
Farshad Rajabipour (Chair) Assistant Professor off the civil and environmental engineering Dissertation advisor, Chair of Committee
William Burgos Professor of Civil and Environmental Engineering
Carlo Pantano Distinguished Professor of Materials Science and Engineering
Barry Scheetz Professor of Civil and Environmental Engineering
*Signatures are on file in the Graduate School.
Peggy Johnson
Civil and Environmental Engineering Department Head
ABSTRACT
Application of recycled glass as a cement or fine aggregate replacement in concrete could result in major benefits towards a more sustainable design of concrete materials. This is however only possible if the main obstacle, the alkali silica reaction (ASR), is properly addressed and the ASR damage is mitigated.
ASR is a deleterious reaction that occurs between meta-stable silicate phases of aggregates and hydroxyl ions present in the pore solution of portland cement concrete. With the overall goal of developing effective methods to mitigate ASR in concrete containing recycled glass, the research objectives of this
PhD study are: 1- To understand the mechanisms through which fly ash controls ASR in accelerated mortar bar test (AMBT). 2- To investigate the beneficial effects of high alkali content glass powder towards ASR mitigation and elucidate the underlying ASR controlling mechanisms in AMBT 3- To perform a preliminary study of the feasibility of using Al(OH)3 as a cement replacement to control ASR .
As one potential mechanism, the effect of Al on the dissolution rate of glass aggregates in alkaline solutions is further researched.
To achieve the first objective, a series of experiments were designed to assess different potential ASR mitigation mechanisms with the aid of computer modeling coupled with advanced material characterization techniques. Various properties (e.g., pore solution composition, ion diffusion coefficient, pore size distribution, strength, ASR gel production and its composition) of mortars with the minimum required fly ash dosage were compared with 100% portland cement mortars. The findings revealed that fly ash mainly reduces the ASR expansion of mortars, by reducing the rate of ingress of the attacking hydroxyl ions from the external alkali bath. This was linked to the pore refinement occurred due to pozzolanic reaction of fly ash. With a numerical study, simulating the adsorption of alkalis by cement hydration products as a sink term, it was shown that alkali binding was also effective. Application of fly ash increased tensile strength of the mortars and thus, could enhance their resistance against cracking.
Variation in ASR gel composition and alkali dilution were found to have minor effect in AMBT conditions. Finally, a new mechanism was introduced and validated in which fly ash to reduce ASR by
iii repressing the dissolution rate of silicate aggregates which occurs due to reduction of the effective hydroxyl ion to aggregate surface ratio.
To achieve the second research objective, four types of glass powder (GP) with different sizes were experimented in AMBT. An approximately linear relationship was obtained between GP size and the required dosage to mitigate ASR. A set of material characterization techniques was designed to shed light on the ASR suppressing effects of GP. The findings revealed that cement replacement by glass powder leads to a decrease in the pH of the pore solution even if alkalis are released from the glass powder. Based on charge balance, it was speculated that some alkalis were present in non-ionic form which do not increase the pore solution’s pH and promote ASR. Application of glass powder leads to a significant reduction in the rate of ion transport from the external bath. During the test, portlandite content of GP mortar decreased due to the pozzolanic reaction which led to a reduction in the average pore size of the binder phase of GP mortar. GP enhanced the tensile strength of the mortars which could be beneficial in
ASR mitigation.
Finally, to achieve the third research objective, the replacement of 20% weight of cement with Al(OH)3 powder was shown to suppress ASR expansion during ASTM C1260 test. The study was then focused on one possible mechanism through which Al(OH)3 can mitigate ASR. The effect of soluble Al on the rate of silicate glass dissolution, at high pH was studied. Corrosion of glass slides in alkaline solutions significantly decreased in presence of Al in the solution. The cause of this phenomenon was further investigated and related to formation of a semi-crystalline Al-Si (zeolite) layer at the surface of glass slides as well as Al-poisoning of silica surface. Based on experimental results, it is speculated that the latter mechanism is more significant in reducing glass dissolution rate and the protective action of the zeolite layer is likely to be less effective. The presence of sufficient dissolved Al was found to be essential for minimizing the dissolution rate of silicate glass.
iv
TABLE OF CONTENTS
LIST OF FIGURES ……………………………………………………………..ix
LIST OF TABLES……………………………………………………………….xv
CHAPTER 1: INTRODUCTION……...……………………...…………………1
1.1 The Problem of Waste Glass ...... 1 1.2 Alkali Silica Reaction (ASR) ...... 3 1.3 Research Significance and Needs: ...... 4 1.4 Organization of Contents ...... 6 1.5 References: ...... 9
CHAPTER 2: BACKGROUND……………...………………………………………………….10
2.1 Using Recycled Glass in Concrete: ...... 10 2.2 Benefits of Using Recycled Glass in Concrete: ...... 14 2.3 Challenges to Use of Recycled Glass in Concrete: ...... 16 2.4 Chemical Mechanism of ASR ...... 17 2.4.1 Structure of soda-lime glass ...... 17 2.4.2 Glass dissolution ...... 18 2.4.3 Silica gelation and swelling ...... 23 2.5 ASR Gel: ...... 25 2.6 Controlling ASR: ...... 26 2.7 Tests to Evaluate the ASR risk of Aggregate-Cement Combinations ...... 28 2.8 Summary: ...... 33 2.9 References: ...... 34
CHAPTER 3: HOW DOES FLY ASH MITIGATE ALKALI-SILICA REACTION (ASR) IN ACCELERATED MORTAR BAR TEST (ASTM C1567)?...... 38 3.1 Introduction ...... 38 3.2 Existing Literature ...... 40 3.3 Materials and Methods ...... 44
v
3.3.1 Accelerated mortar bar test (ASTM C1567) ...... 45 3.3.2 Pore solution extraction and analysis ...... 46 3.3.3 Measurement of ion diffusivity using electrical impedance spectroscopy ...... 47 3.3.4 Tensile and compressive strength tests ...... 48 3.3.5 SEM/EDS imaging ...... 49 3.3.6 Aggregate dissolution rate measurements ...... 50 3.3.7 Numerical Model to Simulate Alkali Transport and Binding ...... 51 3.4 Results and Discussion ...... 53 3.4.1 Sufficient dosage of fly ash to mitigate ASR ...... 53 3.4.2 Pore solution composition ...... 55 3.4.3 Ion diffusion coefficient ...... 59 3.4.4 Significance of alkali diffusion versus binding ...... 61 3.4.5 Tensile and compressive strength ...... 63 3.4.6 Microstructural analysis (SEM/EDS) ...... 65 3.4.6.1 Source of alkalis: ...... 68 3.4.7 Aggregate dissolution rate ...... 70 3.5 Conclusions ...... 72 3.6 References ...... 74
CHAPTER 4: ASSESSING THE ROLE OF ION TRANSPORT IN MITIGATION OF ASR BY FLY ASH IN ASTM C1567…….………………78 4.1 Introduction ...... 78 4.2 Research Significance ...... 80 4.3 Materials and Methods ...... 81 4.3.1 More details on measuring the electrical conductivity of mortars: ...... 82 4.3.2 Porosity of pastes and mortars: ...... 85 4.3.3 Pore Size Distribution (PSD) characterization: ...... 85 4.4 Results and discussion ...... 87 4.4.1 Composition of the binder to mitigate ASR ...... 87 4.4.2 Pore solution composition ...... 90 4.4.3 Ion transport ...... 97 4.4.4 Porosity: ...... 102 4.4.5 Pore size distribution: ...... 105 4.5 Conclusions ...... 106
vi
4.6 References: ...... 108
CHAPTER 5: INVESTIGATING ASR MITIGATING MECHANISMS OF GLASS POWDER IN ASTM C1567 ACCELERATED TEST..………..111 5.1 Introduction ...... 111 5.2 Materials and Methods ...... 113 5.2.1 Accelerated mortar bar test (ASTM C1567) ...... 115 5.2.2 Pore solution extraction and analysis ...... 116 5.2.3 Pore size distribution characterization ...... 116 5.2.4 Pozzolanic reaction of glass powder: ...... 116 5.3 Results and discussion: ...... 117 5.3.1 Sufficient dosage of glass powders to mitigate ASR ...... 117 5.3.2 Tensile and compressive strengths ...... 120 5.3.3 Pore solution composition of ASTM C1567 mortars ...... 122 5.3.4 Long term pH of pore solution of cement pastes ...... 127 5.3.5 Ion diffusion coefficient of mortars ...... 129 5.3.6 Porosity, pore connectivity and pore size distribution ...... 131 5.3.7 Portlandite consumption ...... 133 5.3.8 Microstructural analysis (SEM/EDS) ...... 134 5.3.9 Reducing silica dissolution rate from aggregates ...... 137 5.4 Conclusions ...... 138 5.5 References: ...... 140
CHAPTER 6: INVESTIGATING THE USE OF Al(OH)3 AS AN ASR SUPPRESSOR: EFFECT ON THE DISSOLUTION RATE OF SODA-LIME GLASS……………………………………………..142 6.1 Introduction ...... 142 6.2 Research significance ...... 144 6.3 Materials and methods ...... 145 6.4 Results ...... 148 6.4.1 ASR expansion in ASTM C1260 ...... 148 6.4.2 Dissolution rate of silica glass slides ...... 150
6.4.2.1 Dissolution of Al(OH)3 in high alkaline aqueous solutions ...... 150 6.4.2.2 Mass loss measurement of glass slides and the solution analysis ...... 153
vii
6.4.2.3 Dissolution of glass slides in solutions containing finite [Al] ...... 159 6.4.3 Surface ccharacterization of the corroded glass slides ...... 162 6.5 Discussion ...... 168 6.6 Conclusion ...... 170 6.7 References ...... 172
CHAPTER 7: SUMMARY, CONCLUSIONS AND FUTURE DIRECTIONS……………………………………………………...173 7.1 Summary ...... 175 7.2 conclusions of the research on the role of fly ash in ASR mitigation ...... 176 7.3 conclusions of the research on the beneficial effect of glass powder against ASR ...... 179
7.4 conclusions of the study on the feasibility of using Al(OH)3 as an ASR suppressor ...... 180 7.5 Suggestions for future research ...... 181
Appendix A: Calculation of the sink term………………………………………………………………180
Appendix B: MATLAB code implementing the diffusion and
sink term as alkali binding……………………….………………………………………………….....182
Appendix C: Glossary……………………………………………………………….……………….....183
Appendix D: Procedure of standard deviation calculation of element concentrations
in the pore solution…………………………………………………………………………………....188
viii
LIST OF FIGURES
Figure 1.1: Stockpile of waste glass, Huntington, West Virginia, USA………………………....2
Figure 1.2: (a) Extensive map cracking in concrete barriers along State Highway 2 near Leominster, MA. (b) Gel staining around cracks in the parapet wall of a bridge structure affected by ASR …………………………………………….…….4
Figure 2.1: ASR Expansions for different sizes of glass particles used as fine aggregate during ASTM C1260 test ...…………………………………………………………...12
Figure 2.2: SEM Micrographs of (a) ASR and (b) pozzolanic reactions of glass particles of different sizes………………………………………………………………….. 13
Figure 2.3: Schematic 2D representation of amorphous silica……………………..……………17
Figure 2.4: Schematic 2D representation of amorphous binary soda-silicate glass……………..18
Figure 2.5: Proposed mechanism of dissolution of silica in water in the presence of hydroxyl ions……………………………………………………………………21
Figure 2.6: Solubility of amorphous silica versus pH at different temperatures………………..21
Figure 2.7: Schematic 2D representation of soda-lime-silicate glass structure…………………23
Figure 2.8: Exposure sites at the University of Texas in Austin (left) and CANMET-MTL
in Ottawa (right)……………… ………………………………….……………………...….29
Figure 2.9: container cured in ASTM C1293 conditions………………………………….……30
ix
Figure 2.10: ASTM C1260 mortar bars cured in 1 N NaOH @80°C for 14 days……………..32
Figure 3.1: Pore solution extraction die ………………………………………………………..46
Figure 3.2: HP-4194A impedance analyzer…………………………………………………….48
Figure 3.3: 3-point bending test performed on 25×25×250mm mortar bars with MTS machine: Displacement control mode…………………………………………..49
Figure 3.4: Rims of hydrated portland cement particles……………………………………….53
Figure 3.5: ASR expansion of control, 15%F1 and 35%C2 mortars during ASTM C1567……55
Figure 3.6(a): Pore solution composition of mortars during ASTM C1567 test: [OH-] ion...... 57
Figure 3.6(b): Pore solution composition of mortars during ASTM C1567 test: [Na] element..58
Figure 3.6(c): Pore solution composition of mortars during ASTM C1567 test: [K] element....59
Figure 3.7: Ion diffusion coefficient of the control (100% PC), 15%F1 and 35%C2 mortars…60
Figure 3.8: Simulation results showing the OH- concentrations at the conclusion of ASTM C1567 test for different diffusion and binding coefficients………………………62
Figure 3.9: Normalized (to the control mortar) tensile and compressive strengths of mortars 3 days after exposure to 1M NaOH solution………………………..…64
x
Figure 3.10: SEM micrographs of cross section of mortars at conclusion of ASTM C1567 test………………………………………………….……...... 67
Figure 3.11: Comparison of X-ray patterns of ASR gel and adjacent glass…………….……….68
Figure 3.12: (a) SEM micrographs of cross section of mortars at conclusion of ASTM C1567 test; mortars are submerged in 1M KOH at 80°C …………..……..…...... 70
Figure 3.12: (b) EDS mapping of potassium element………………………………..………….70
Figure 3.13: Mass loss of glass slides in 1M NaOH solution at 80ºC in the presence or absence of fly ash………………………………………………..……….71
Figure 4.1: Particle size distribution of portland cement and fly ashes used……………………81
Figure 4.2: Niqyuist plot of the control and 15%F1 mortars after 7 days cured in ASTM C1567 test……………………………………………………..84
Figure 4.3: Results of ASTM C1567 showing ASR expansions as a function of fly ash type and dosage……………………………………………...………88
Figure 4.4: Ternary phase diagram showing cement and fly ash compositions……………..…..89
Figure 4.5: Pore solution concentration during ASTM C1567 test: Hydroxyl ion……..…...... 91
Figure 4.6: Pore solution concentration during ASTM C1567 test: Sodium…………...... …….93
Figure 4.7: Pore solution concentration during ASTM C1567 test: Silicon………………...... 95
Figure 4.8: Pore solution concentration during ASTM C1567 test: Aluminum…………..……96
xi
Figure 4.9: Comparison of the electrical conductivity of the pore solution of control and fly ash mortars after submergence in 1M NaOH bath ………………………98
Figure 4.10: Comparison of the electrical conductivity of the control and fly ash mortars after submergence in 1M NaOH bath………………….…...... 99
Figure 4.11: Comparison of the ion diffusivity of the control and fly ash mortars during ASTM C1567 test ……………………………………..……….…..101
Figure 4.12: Porosity of the control and fly ash mortars
and pastes after submergence in 1M NaOH bath………………………………….…….….103
Figure 4.13: Pore refinement in control and fly ash pastes after 3 days exposure to NaOH solution at 80°C………………………………………………………...106
Figure 5.1: Particle size distribution of different classes of glass powder used……….…….…114
Figure 5.2: ASR expansion of mortars with different GP2 replacements over 14-day period of ASTM ……………………………………….………………...... 118
Figure 5.3: Results of ASTM C1567 showing ASR expansions as a function of glass powder size and dosage ………………………………………….…119
Figure 5.4: Compressive and tensile strengths of control and GP2 mortar after 3 days Curing in 1M NaOH solution at 80°C ………………………………..……….120
Figure 5.5: Compressive and tensile strengths of control and GP2 mortar after 3 days Curing in 1M NaOH solution at 80° ……………………………………..……122
Figure 5.6: OH- concentration in pore solution of mortars during ASTM C1567 test...... 124
xii
Figure 5.7: Concentration of Na and K in the pore solution of mortars during ASTM C1567 test…………………………………………………………..………...125
Figure 5.8: OH- concentration in the pore solution of pastes cured at 23°C, RH=100%.……...128
Figure 5.9: Comparison of the ion diffusion coefficient of the control and 40%GP2 mortars during ASTM C1567 test……………………………………..………130
Figure 5.10: Pore size distribution in 100%PC and 40%GP2 pastes after 3 days exposure to 1M NaOH solution at 80°C (total age=5days)……………………….….132
Figure 5.11: TGA analysis of the mortars after 0, 7 and 14 days NaOH exposure….…….…...134
Figure 5.12: SEM image of control and 40% GP2 mortars cross section after 14 days exposure to NaOH solution at 80°C……………………………………….….136
Figure 5.13: Mass loss of glass slides in 1M NaOH solution at 80ºC in the presence or absence of glass powder………………………………………....138
Figure 6.1: Reduction in ASR expansion with increasing
Al(OH)3 content in ASTM C1260………………………………………………………..…149
Figure 6.2: Al speciation in water at different pH levels at 25°C……….………………….…..152
Figure 6.3: Dissolution of Al(OH)3 powder in 1M NaOH at 80°C……………………………153
Figure 6.4: The effect of addition of Al on the rate of glass slide dissolution…………….……154
xiii
Figure 6.5: Si concentration due to glass slides dissolution in presence and absence of
20g Al(OH)3, variation in [Al] in presence of glass slides and 20g Al(OH)3 ……………….156
Figure 6.6: Variation in [OH-] during the glass corrosion test in 1M NaOH solution
with and without Al(OH)3 powder………………………………………………..….………158
Figure 6.7: Mass loss of glass slides in 1M NaOH solution in presence of
finite initial concentrations of aluminum ………………..………………………….……….160
Figure 6.8: Formation of a translucent alumino-silicate layer forms at the surface
of glass slides exposed to 1M NaOH in presence of Al…………………………………….163
Figure 6.9: SEM/EDS mapping of Al, Si and Na elements in the glass slides
surface reaction product in presence of Al(OH)3…………………………………….………165
Figure 6.10: XRD patterns of aluminosilicate layer formed
on glass slide surface treated in presence of Al…………………….………………………...166
Figure 6.11: Comparison of the mass loss of glass slides with and
without an aluminosilicate layer at the surface………...…………………………………...... 167
Appendix E: Detailed X-ray pattern of Aluminosilicate layer……………………….190
xiv
LIST OF TABLES
Table 3.1: Oxide composition (wt %) of Portland cement, glass, and fly ashes……………..….45
Table 3.2: Expansion of mortar mixtures at the conclusion of ASTM C1567 test…………..….54
Table 3.3: Average atomic composition (wt.%) of ASR gel measured by EDS………………...67
Table 4.1: Summary of the tests performed………………………………………………….….82
Table 5.1: Oxide composition (%) of Portland cement and glass powder………………..……115
Table 5.2: Average atomic composition (wt.%) of ASR gel measured by EDS……………….135
Table 6.1: Oxide composition of portland cement……………………………………………...147
Table 6.2: Atomic weight % of the elements from the Al-Si layer and glass…………………..164
Appendix D-Table 1: Standard deviation in acid titration measurement …………………...... 189 Appendix D-Table 2: Standard deviation in ICP measurement………………………………..189
xv
Acknowledgments
As my long journey to fulfill my PhD is finally coming to an end, I would like to thank all the people who helped me along the way. First of all, I am deeply indebted to my adviser, Dr.
Farshad Rajabipour. Without steady support, unending patience, and wise counsel this dissertation would not have come to fruition. I would also like to thank members of my dissertation committee for taking their precious time to review my thesis: Dr. William D.
Burgos, Dr. Carlo Pantano and Dr. Barry Scheetz. Special thanks are extended to Dr. Pantano, who set aside his busy schedule and kindly replied to my questions on numerous occasions. His help deepened my knowledge and greatly improved my work.
I must thank my friends and colleagues who provided me considerable support for experimental studies in the CITEL laboratory: Hamed Maraghechi, Alireza Akhavan, Nima Ostadi, Jared
Wright and Chris Cartwright. My special thanks to Dan Fura for his great assistance and support in the CITEL laboratory. I wish all the best for you in your life ahead. I am particularly grateful to my dear friend, Hamed Maraghechi, for his help, fruitful conversations and insightful discussions. I also wish to express my sincere gratitude to all the staff and members of material science complex for providing a friendly atmosphere. Their support is also appreciated.
I acknowledge the financial support which was provided for this project by the National Science
Foundation (NSF) under Grant No. CMMI 1030708 and Hawaii department of transportation
(DOT)/ federal highway administration (FHWA) under project No. HWY-L-2.6129 granted to my advisor. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science
Foundation.
xvi
I truly appreciate all the efforts of my parents. Many dreams of theirs were sacrificed to help me achieve my goals. They showed the meaning of endless love and support which cannot be expressed through words. Every second of past four years being far away from you was a torture and hard to survive. Thanks very much for all you have done for me.
Last but not least, I want to thank my kind beautiful wife, Dr. Ghazal Izadi, for her limitless love and support. Without you I would hardly be able to get through this. Thanks for passionately being with me in ups and downs of this journey. Being away from home was bearable just by having you by my side. I am blessed to have you.
xvii
To my parents... and to my lovely wife.
CHAPTER
ONE
INTRODUCTION
1.1 The Problem of Waste Glass
The growing of stockpiles of post-consumer recycled glass (e.g., bottles, window plates) is a
major concern for municipalities in many states and countries. This problem is more urgent in
locations with limited land (e.g., Hawaii) or large population density (e.g., North-East corridor of
US) and has led to increasing disposal costs for both consumers and local governments.
According to the most recent US-EPA report [1], from approximately 250 million tons of municipal solid waste generated in 2010 in the United States, 4.6% (approximately 11.53 million tons) was waste glass. Out of this volume, 27.1% (about 3.13 million tons of glass) was recovered by recycling facilities. Even so, ultimately ~600,000 tons out of that is not actually recycled into new glass [2]. The main obstacle against glass to glass recycling is the potentially high cost of shipping glass, from collecting points to remelting facilities (which could be few
1 hundred kilometers apart); and this has made the economics of glass recycling unattractive. In addition, waste glass stockpiles are often contaminated (e.g., by sugar, paper, and other organics) and mixed in color which makes them less suitable for reuse in glass making industry. These factors have led to continual stockpiling of waste glass and its subsequent environmental impacts
(Figure 1.1).
Other than the ideal use of recycling glass for manufacturing new glass, secondary applications of recycled glass such as in abrasives, glass wool, or as water filtration media have been developed [4]. Lately, it has been attempted to utilize waste glass in construction materials such as bricks and ceramics [5] and enhanced night visible asphalt [6].
Figure 1.1: Stockpile of waste glass, Huntington, West Virginia, USA [3]
Due to enormous volume of concrete produced annually (estimated worldwide at 11 billion tons/year) [7], incorporation of recycled glass in concrete has a great potential to convert large
2
quantities of collected glass into a value-added material. Substitution of waste materials will
conserve limited natural resources (used for production of cement and aggregate) and will
mitigate the environmental and ecological damages caused by quarrying and exploitation of raw materials for making concrete.
From an engineering standpoint, concrete is designed to fulfill certain mechanical properties and
satisfy specific durability performance during its service life. Fortunately, application of glass as concrete fine aggregate has not been found to have a major impact on mechanical properties of concrete [8]. However, the exposure of amorphous silicate in glass to high alkaline environment
of concrete can result in an alkali-silica reaction (ASR). As a result of this reaction, an alkali- silica gel is produced which absorbs water, swells and exerts tensile stresses inside concrete.
These stresses can result in cracks and damages in concrete which shortens the effective service life of the structure.
1.2 Alkali Silica Reaction (ASR)
Alkali silica reaction has long been a major durability problem in concretes containing reactive
natural aggregates (e.g., sandstone, argillite, chert, opal, greywacke). Aggregates containing
metastable silicates can react with the hydroxyl ions of the concrete pore solution. As a result of
this reaction, a silicate gel is produced. This gel is hygroscopic and imbibes available moisture
from the interior of concrete and expands. A deleterious expansion for concrete is regarded as
0.04% to 0.05% expansion [9]. The swelling pressure is sufficient to disrupt the fabric of concrete and induce expansions higher than the tensile strain capacity of concrete. Since the
3
reactive aggregates are distributed in the bulk of concrete, generally this reaction leads to a map
cracking at the member’s surface with gel exudation which is a common characteristic and
feature of ASR (Figure 1.2)
(a) (b)
Figure 1.2: (a) Extensive map cracking in concrete barriers along State Highway 2 near Leominster, MA. (b) Gel staining around cracks in the parapet wall of a bridge structure affected by ASR [10]
1.3 Research Significance and Needs:
Recycling glass is a major problem for municipalities across the world. Utilizing glass as
aggregate/cement replacement in concrete is a potential solution if the deleterious ASR is
controlled. The present study aims at expanding the knowledge related to the ASR mitigation in
concretes containing recycled glass. Study of ASR using glass as reactive aggregate has the
added benefit of simplifying the investigations by studying the physical and chemical reactions
using a homogeneous and isotropic reactive material comparing to natural reactive aggregates which are mostly composed of different phases and are orthotropic. For example, the effect of
4
aggregate composition on ASR performance can be examined by experimenting glasses with
various compositions (e.g. aluminosilicate glass, borosilicate glass,etc). In addition, it would be
possible to take advantage of the available literature on glass corrosion to gain a better
understanding of durability of silicate glasses in alkaline environment of concrete. It is also
possible to advance ASR knowledge towards developing new materials and admixtures to
mitigate ASR. After a review of the available literature on developing glass-based concrete
materials, the following research needs were identified to be further studied in the present work:
1. Previous research has recommended application of supplementary cementitious materials
(SCM) (e.g, fly ash) to control ASR. However, it is unclear how fly ash and other SCMs mitigate
ASR (i.e. what are the mechanisms involved) and what fly ash properties most significantly
determine its efficiency against ASR. There is specifically a knowledge gap with regards to role of fly ash and other SCMs during accelerated ASR performance tests. In these tests, ASR is artificially accelerated by submerging concrete or mortar bars in high alkaline solutions (e.g., 1M
NaOH) at high temperatures (e.g. 80°C). As such, the role of fly ash in mitigating ASR could be through reduction of mass transport properties of concrete, dilution and binding of alkalis, increasing the resistance of concrete to cracking, or a combination of these and other factors.
These mechanisms are studied in chapters 3 and 4 of this document.
2. Previous experiments have shown that soda-lime glass powder is pozzolanic and may be capable of controlling ASR. However, high sodium content of glass powder is a concern in
promoting rather than inhibiting ASR. Also, it is unclear if reducing the size of glass powder is beneficial towards improving its pozzolanic reactivity or in contrast, leads to higher alkali release
5
rate and promotes ASR. Chapter 5, attempts to clarify the underlying mechanisms in ASR
mitigation of glass powder to address these questions.
3. There is a need for development of more efficient and less costly admixtures to inhibit ASR in
concrete. Earlier research on glass corrosion suggests that the presence of Al can slow down the
dissolution rate of glass in alkaline media. This is important from an ASR perspective as Al has
the potential to cease ASR at its very first step. While there have been some recent studies
focusing on the effect of the Al content of different pozzolanic materials on their efficiency to
suppress ASR (e.g., metakaoline, fly ash), the feasibility of using Al compounds independently
as an ASR inhibitor is a novel idea. This is the main focus of chapter 6. Studying Al compounds not only sheds light on the exact role of Al content of pozzolans, but also may lead to the development of new additives or cement replacements to avoid ASR in future concrete structures.
1.4 Organization of Contents
This study focuses on characterizing and developing innovative methods to mitigate ASR in concrete, especially where recycled glass is used as fine aggregates. Chapter 2 provides a
background on the use of recycled glass in concrete. Challenges and benefits are introduced. The
alkali-silica reaction is reviewed and conventional methods for mitigation of ASR are explained.
Tests that have been suggested to predict the ASR potential of aggregates or to determine the
effectiveness of mineral and chemical admixtures in mitigation of ASR are presented.
6
Chapter 3 investigates the mechanisms by which coal fly ashes reduce ASR. A broad series of
analytical techniques are used to study the mechanisms of ASR mitigation in mortars containing
glass aggregates and tested under the environmental conditions specified by ASTM C1567, accelerated mortar bar test.
The findings in chapter 3 revealed that reducing ion transport is one of the dominant mechanisms which enables fly ash to reduce ASR during ASTM C1567 test. As such, chapter 4 focuses on characterizing the transport properties of mortars containing a sufficient dosage of fly ashes to mitigate ASR and compares these properties with those of 100% portland cement mortars or the mortars containing a lower dosage of the same fly ash. In addition, the dosage of different fly ashes that is sufficient in suppressing ASR in mortars with recycled glass aggregates are obtained and related to the chemical composition of fly ash.
Chapter 5 investigates the capacity of soda-lime glass powder as an ASR inhibitor. Glass powder exhibits pozzolanic behavior (more significant for smaller particle sizes) and can be used as partial cement replacement to mitigate ASR of glass aggregates or natural reactive aggregates.
Powders with different particle size distributions are studied according to ASTM C1567 and the proper contents to mitigate ASR are obtained. Further investigations are performed to understand the mechanisms through which glass powder reduces ASR expansions.
Chapter 6 presents a preliminary study to investigate the feasibility of using amorphous aluminum hydroxide to control ASR. Specifically, the focus is to understand the mechanisms by which aluminum slows down the dissolution rate of silicate glass is high alkaline environment.
7
Chapter 7 provides a summary of the findings and conclusions of this Ph.D. study. Based on the results, areas for future research are identified to improve the knowledge on ASR mechanisms and its mitigation.
8
1.5 References:
1. http://www.epa.gov/osw/nonhaz/municipal/pubs/msw_2010_rev_factsheet.pdf
2. J. Reindl, Reuse/recycling of glass cullet for non-container uses. 2003, Wisconsin Department of Public Works: Madison, WI (www.epa.gov/osw/conserve/rrr/greenscapes /pubs/glass.pdf).
3. http://www.corbisimages.com/stock-photo/rights-managed/JA004525/bucket-loader-at-green-glass- stockpile/?tab=details&caller=search.
4. R. Idir, M. Cyr and A. Tagnit-Hamou, Use of fine glass as ASR inhibitor in glass aggregate, Construction and Building Materials, 2010, 24, 1309-1312.
5. N. F. Youssef, M. F. Abadir and M. Shater, Utilization of soda glass (cullet) in the manufacture of wall and floor tiles, Journal of European ceramic society 1998, 18, 1721-1727.
6. R. D. Pascoe, R. W. Barley and P. R. Child, Autogenous grinding of glass cullet in a stirred mill. In: Proceedings of the International Symposium Recycling and Reuse of Glass Cullet. Dundee: Thomas Thelford; 2001, 15-27.
7. P. K. Mehta, P. J. M. Monteiro, concrete: Microstructure, Properties and Material, 3rd edition, McGraw Hill, New York , 2006.
8. I. B. Topcu and M. Canbaz, Properties of concrete containing waste glass, Cement and Concrete Research, 2004, 34, 267-274.
9. D.W. Hobbs, Deleterious expansion of concrete due to alkali-silica reaction: influence of Pfa and slag, Magazine of Concrete Research, 1986, 38(137), 191-205.
10. http://en.wikipedia.org/wiki/File:ASR_cracks_concrete_step_barrier_FHWA_2006.jpg.
11. Y. Shao, T. Lefort, S. Moras and D. Rodriguez, Studies on concrete containing ground waste glass, Cement and Concrete Research, 2000, 30(1), 91-100.
12. Y. Oka and M. Tomozawa, Effect of alkaline-earth ion as an inhibitor to alkaline attack on silica glass, Journal of Non-Crystalline Solids, 1980, 42(1-3), 535-543.
13. S.Y. Hong, F.P.Glasser, Alkali sorption by C-S-H and C-A-S-H gels: Part II. Role of alumina, Cement and Concrete Research, 2002, 32(7), 1101-1111.
9
CHAPTER
TWO
BACKGROUND
2.1 Using Recycled Glass in Concrete:
Depending on particle size, crushed glass cullet has the potential to be utilized as different components of concrete; coarse aggregate, fine aggregate or cement replacement. Particles with a size larger than 4.75 mm are generally termed “coarse aggregate” in concrete industry. One of
10
the main problems in using glass coarse aggregates is their flat and elongated shape (dictated by
the thickness of original glass bottles) which results in poor workability and lower strength of
concrete [1]. Crushing glass to a size less than 4.75 mm (known as “fine aggregate” in concrete
industry) helps glass particles to gain a better geometry (e.g., aspect ratio close to 1) for use in concrete which results in better workability and strength. While application of glass fine aggregates has been reported to slightly reduce the early age strength of concrete, long term strengths have shown similar performance comparing to concretes with natural fine aggregate
[2]. The main problem, however, when glass is used as fine or course aggregate is the alkali
silica reaction (ASR) [2,3]. The main goal of the present study is to tackle this problem and gain
a better understanding of its underlying mechanisms and methods for its mitigation.
It has been reported that when glass particles are ground to a size less than 300m, their
deleterious ASR expansion is no longer observed [4]. Figure 2.1 shows the ASR expansion of
mortar bars containing different sizes of glass particles in ASTM C1260 environment. While
glass particles in a the range 1.18-2.36mm undergo excessive ASR expansion much beyond the
deleterious threshold, mortars with glass particles of a size less than 300m exhibit no ASR
expansions. The reason for the deleterious reaction of glass particles is production of expansive
ASR gel. Figure 2.2(a) shows SEM micrograph of ASR gel occurred in a glass particle of a size
~2mm. Some workers have suggested grounding glass to this size to prevent ASR in order to be
used in concrete. This approach has the disadvantage that only a portion of concrete aggregates
that fit within this limited size range can be replaced by glass. It also adds to the cost of
processing and production of concrete. Ideally, it is desirable to replace the entire fine aggregates
of concrete with recycled glass.
11
1‐2.38mm 0.3‐0.6mm 0.075‐0.15mm
ASR Deleterious limit Innocuous limit
Figure 2.1: ASR Expansions for different sizes of glass particles used as fine aggregate during ASTM C1260 test [4]
Interestingly, by reducing the glass particle size to less than 100m, glass particles start to react pozzolanically [5,6] and by further reducing the particle size the pozzolanic behavior intensifies.
This is a result of high silicate content of glass (~70% SiO2) which can react with available
Ca(OH)2 in concrete and form C-S-H gel (i.e. the main binding phase of concrete). Figure 2.2(a)
shows a small size glass particle which underwent pozzolanic reaction at its surface. It is unclear
why large glass particles undergo ASR while finer particles react pozzolanically.
12
(a) (b)
200m 200m
Figure 2.2: SEM Micrographs of (a) ASR and (b) pozzolanic reactions of glass particles of different sizes [4]
A pozzolanic reaction is a reaction between amorphous silicate (e.g., from fly ash or glass
powder) and portlandite, Ca(OH)2, which is the byproduct of portland cement hydration.
Pozzolanic reaction leads to formation of calcium-silicate hydrate gel (C-S-H), which is the main binder phase in concrete.
Pozzolan (S) + hydrated lime (CH) + Water → C-S-H (2-1)
For a material to be considered pozzolanic, three criteria should be satisfied: high equivalent silicate content (at least 50% depending on the class of pozzolan, SiO2eq= SiO2+Al2O3+Fe2O3), amorphous structure, and large surface area [5]. According to the chemical composition requirement of ASTM C618, soda-lime glass may be classified as a pozzolan provided that glass is ground to a sufficient level of fineness. The standard requires that the pozzolan powder amount, retained on 45μm (No. 325), when wet-sieved, should not exceed 34% by weight.
Since the pore solution in concrete has a high alkalinity with a pH generally above 12.5, it is expected to result in a high dissolution rate for amorphous silicate [7]. By further decreasing the
13
particle size and thus increasing the surface area in direct contact with high alkaline medium,
silicate dissolution is further promoted. This accelerates the pozzolanic reaction.
In addition to the use as aggregate or cement replacement, glass can be fed as raw material in
cement kiln to produce clinker. This has been done for in Hong Kong for 1.8t/h glass added to
280-290t/h raw materials [8]. However due to the high sodium content of glass, the alkali content
of portland cement will increase and thus increases the potential for alkali–silica reaction even
for aggregates with moderate reactivity. It could also result in flash setting due to the high alkali content and the formation of 2CaSO4.K2SO4 compound [9].
2.2 Benefits of Using Recycled Glass in Concrete:
If waste glass can be utilized in concrete construction while satisfying a mechanical and
durability performance comparable to conventional concrete, it can have many economical and
environmental benefits:
Prohibiting the expansion of present landfills or stockpiles
Adding to the value of waste glass instead of simply dumping it in landfills
Reducing the transportation costs of the waste material
Reducing the transportation cost of natural aggregate to regions with scarce good quality
aggregate sources
Reducing the environmental burden of irreversible mining of aggregate quarries and
processing of natural aggregates [10]
Due to its lower thermal conductivity concretes containing glass may help with better
insulation in buildings
14
Concretes containing glass may have a higher abrasion resistance due to harder nature of
glass particles comparing to natural aggregates [3]
Aesthetic benefits
Incorporation of recycled glass in concrete can earn credits towards LEED certification of
buildings in the materials and resources category [11].
Using glass in powder form as a replacement of cement is more desirable than replacement of
natural aggregates. It improves concrete’s durability through pozzolanic reactions and reduces cost/carbon footprint by replacing the costly and environmentally negative portland cement. As a
result of pozzolanic reaction (eq. 2-1) of glass powder with portlandite (Ca(OH)2) and converting
it to secondary C-S-H, the microstructure of concrete matrix is densified. As a result of this
conversion, the porosity and interconnection of the pores in concrete are reduced. Pore
refinement that is achieved by this mechanism leads to reduced rate of mass transport in concrete
and improves the durability [6]. Portland cement concrete with an annual production of
approximately 11 billion tons is the largest among all man-made materials [12]. Concrete
production relies on the use of energy/CO2 intensive portland cement (embodied energy=5.7
MJ/Kg, CO2 footprint=0.95kg/kg) [10]. It is estimated that over 5% of the global man-induced
CO2 emissions are caused by cement manufacturing [13]. Due to technical (engineering),
economical and environmental issues, researchers are seeking new alternatives to portland
cement. Examples of such efforts include development of high-volume fly ash [14] and
geopolymer concretes [15].
15
2.3 Challenges to Use of Recycled Glass in Concrete:
General aspects of using waste glass as different constituents of concrete and challenges to be tackled have been reviewed by [16,17]. The most important challenge when using glass as fine aggregates is the alkali-silica reaction. Unless ASR is effectively controlled by using mineral and/or chemical admixtures or through other methods, the applicability of glass aggregates in concrete will remain limited [17].
In an early study, Johnston [2] started to use crushed glass as coarse aggregates in concrete. But due to excessive ASR expansions, he concluded that glass cannot be used in concrete as coarse aggregate without addition of ASR controllers such as fly ash. Topcu and Canbaz, while investigating different properties of glass aggregates [1], found that the strength of concrete slightly decreases with the application of coarse glass. Polley et al. [3] linked this reduced strength to flat shapes, smooth glass surface comparing to that of natural aggregate and high friability of glass, specifically for particle sizes greater than 1.5 mm.
In addition, workability or slump variations have been reported due to angular shape of the glass aggregates [18,19]. Also the organics (e.g., sugar) and paper contamination may reduce the strength of concrete [3] and as such these contaminations should be removed as much as possible. While the reduction in strength with application of glass may be compensated by reducing the water to cement ratio (w/c), the major problem remains as the deleterious alkali- silica reaction.
The challenge with using glass powder as partial portland cement replacement is a lower early rate of strength gain and delayed time of setting [5]. Decreasing the glass powder particle size
16 has been shown to increase pozzolanic reactivity and thus higher rate of early-age strength gain.
However, the challenge then becomes the amount of energy required to grind glass powder to smaller particle sizes.
2.4 Chemical Mechanism of ASR
2.4.1 Structure of soda-lime glass
Silicon is an atom from group IV which tends to bond to four oxygen atoms and form a tetrahedral cell. These tetrahedral cells can share their corner oxygens and build up a 3D microstrucutre of glass network. A schematic 2D representation of the 3D glass structure is shown in Figure 2.3. In the amorphous form of such a network, Si-O-Si rings with different number of Si-O-Si bonds are present. Each oxygen atom is just shared between two tetrahedral cells. These oxygens are called bridging oxygens (BO). In a pure silica network, these oxygens are mostly present at the surface of the glass.
BO
NBO
Figure 2.3: Schematic 2D representation of amorphous silica 17
When Na is added to the glass structure, it modifies the Si-O network and thus, is called modifier. To compensate the positive charge of Na+ ion, a non-bridging oxygen atom (NBO) with a negative charge forms. The tetrahedral cells with NBOs still interact electrostatically. A schematic 2D structure of such a modified network is shown in Figure 2.4. The effect of glass composition on its corrosion rate will be discussed later.
Figure 2.4: Schematic 2D representation of amorphous binary soda-silicate glass
2.4.2 Glass dissolution
Of all its properties, great resistance of glass to most chemicals at normal working temperatures is one of its distinguished advantages. Without it, most of the applications of glass would be unthinkable [20]. Chemical resistance of glass is generally regarded against the attack of water or aqueous solutions. This is known in glass science literature as resistance of glass against
“corrosion”. The corrosion mechanism of soda-lime silicate glass includes alkali leaching which
18
is more dominant at low pH levels and congruent dissolution which is more dominant at high pH
levels. Usually the resulting compositional and structural changes occur only at glass-solution
interface and thus they are limited to the glass surface [21].
Three general classes of chemical reactions have been identified between glass and aqueous
solutions. Selective leaching of modifier cations (e.g., Na+) is often attributed to “ion-exchange”
reaction which is the dominant reaction at low pH solutions [22]:
Si O Na H 2O Si OH Na OH (2-2)
The rate and extent of this reaction depends on the solution chemistry as well as glass structure.
The concentration of alkalis and the pH of the solution have a major impact on the rate of ion-
exchange. In general, the rate of ion-exchange increases with an increase in the amount of silica that is replaced by alkali oxides in the glass structure. It also decreases with decreasing the alkali content and even ion radius of different alkali ions.
The second mechanism is called “hydration”. The water molecules can diffuse into the glass structure through the void space between non-bridging oxygens in the microstructure. This
penetration of water molecules into the network without attacking any Si-O bond is called hydration. It is controlled by the size of the voids between Si-O-Si rings. As such it occurs in less packed glass structures like sol-gel glasses. Sol-gel processing involves the generation of suspensions (“sols”) which are subsequently converted to viscous gels and then to solid materials
[21].
19
A third class of reaction which can occur simultaneously with hydration and ion-exchange is called “hydrolysis”. This reaction is more pronounced in pH>9 aqueous solutions and is not completely reversible. Silica tetrahedral is subjected to nucleophilic OH attack. Siloxane bonds are disrupted and negatively charge non-bridging oxygens (NBO) are generated:
Si O Si OH Si OH Si O (2-3)
The non-bridging oxygen formed may interact with water molecule and produce more silanol sites:
Si O H 2O Si OH OH (2-4)
The dissolution of silica results in the liberation of silicic acid into the aqueous solution with a general formula of [H2xSiO2+x]n (monomer or polymer forms). This is shown schematically in
Figure 2.5[23]. The solubility values for amorphous silica at acidic to neutral pH levels are in the range 70 to 150 ppm at 25°C [23]. Above pH 9, a significant apparent increase in the solubility
- of silica occurs due to formation of silicate ions from Si(OH)4 monomers by the attack of OH :
Si(OH ) 4 OH H 3 SiO4 H 2O (2-5)(a)
2 H 3 SiO4 OH H 2 SiO4 H 2O (2-5)(b)
Other di-, tri-, tetra-meric ions can form as well [24].
In acidic solutions, the silicate ions are instantly converted back to monomer, but in basic media, they are stable [23]. As a result of these ionization reactions (eqs 2-5), the solution remains unsaturated with respect to silicic acid [23] and further silica dissolution continues per Figure
20
2.5. As a result the entire solid structure of amorphous silicate gradually dissolves to form
soluble silicates. This represents itself as an apparent high solubility of silica at high pH [23].
Figure 2.5: Proposed mechanism of dissolution of silica in water in the presence of hydroxyl ions [23]
In summary, in an alkaline medium, the reaction results in congruent network dissolution of silica
[21]. The effect of pH on solubility of amorphous silica at different temperatures is depicted in
Figure 2.6. High pH levels not only provide higher concentrations of attacking hydroxyl ions leading to an increase of network hydrolysis, but also enhance the thermodynamic solubility of dissolved silicates [21]. Figure 2.6: Solubility of amorphous silica versus pH at different temperatures [23]
21
As previously mentioned, the glass composition affects its corrosion rate in alkaline environment. Vitreous silica with nearly 100% SiO2 and nearly no NBO has the highest
durability in alkaline environments. This is due to the fact that tetrahedral cells with four
bridging oxygens (called “Q4”) are more resistive against hydroxyl attacks. By adding alkalis to glass structure (e.g., sodium-silicate glasses, Figure 2.4) the network opens up for further hydration. This is a result of disconnection of the rings which facilitates further penetration of water molecules for hydration. Also, due to introduction of NBOs and conversion of Q4 to
- Qn(n=0,1,2,3) they become less resistive against OH attack and hydrolysis will occur at a higher
rate leading to less durability. The alkalis can leach out, leaving empty pores for hydration to
occur. As a result, a coupled hydration-hydrolysis reaction occurs. The pH dependence for
network dissolution in alkali-silicate glasses is similar to that seen for pure amorphous silica
[23].
Finally, the addition of alkali earth elements such as Ca has been reported to improve the
chemical durability of silicate glasses (Figure 2.7) [25]. Unlike alkalis, these divalent ions to two
Si atoms and help in resisting network dissolution. This dual connection also blocks the
interconnected void space inside the glass structure and mitigates hydration and ion diffusion
inside the network.
Not only glass composition affects its corrosion, but the presence of various ions in the attacking
aqueous solution also influences the corrosion rate. For example it has been shown [25] that the presence of Ca ion in the attacking solution decreases the corrosion rate of glass. This is linked to
the partial protection of silicate surface due to the formation of a layer of calcium silicates [21].
22
The presence of aluminum in the attacking solution is also reported to be beneficial in slowing down the rate of silicate dissolution. This is further studied in chapter 6.
+ ‐
+ ‐ +‐
Ca
Figure 2.7: Schematic 2D representation of soda-lime-silicate glass structure [26]
2.4.3 Silica gelation and swelling
As a result of glass dissolution, Si increases in the pore solution mostly in the form of mono-
silicic acid, Si(OH)4aq. However, other oligomers may also form with a general formula of
23
SinOa(OH)b where 2a+b=4n [23]. At high pH, ionization of silicic acid results in conversion to
highly soluble silicate ions. As long as pH and temperature are maintained to avoid super
saturation, the dissolved species remain in the solution; the aggregates dissolution slows and
eventually stops as the dissolved SiO2 approaches its solubility limit. However, in the presence
of polyvalent metal ions, (e.g. Ca, Al), silicate ions are linked to form poly-metal-silicates [27],
e.g.:
2 2((HO)3 Si O )eq Caaq ((HO)3 Si O Ca O Si (OH )3 ) sol (2-6)
A 2- or 3-dimensional polymerization results in formation of insoluble metal silicate particles;
which flocculate to form colloidal silica or silica gel. Further condensation forms silica (ASR)
gel. The surface of colloidal particles bears many –OH groups which are highly hydrophilic;
resulting in osmosis, adsorption of water and swelling of the gel. As the solution becomes under- saturated with respect to silicate ions, further dissolution of reactive aggregates proceeds.
There is a controversy on the role of Ca in the development of the gel. Calcium-rich gels have
low swelling capacity but some researchers [28] suggest that expansion only occurs in the
presence of high concentration of portlandite (CH). Blezynsky and Thomas [29] observed that
ASR accompanied with little to no expansion in samples with very limited portlandite content.
Thus it is believed that removal of free CH, either by leaching or pozzolanic reactions suppresses
ASR expansion. Chatterji [30] postulated that Ca2+ is the controller for out of grain diffusion of
silicate species and thus the expansion. On the other hand, the NMR/XRD results of Hou et al.
[31] show that ASR gel does not form as long as a readily-soluble Ca is locally available to react
24
with dissolved silicates to form pozzolanic C-S-H. Further investigation is required to shed light
on these conflicting results.
2.5 ASR Gel:
The hygroscopic gel formed as a result of alkali-silica reaction has significant specific volume
compare to the SiO2 structure which it replaces. This conversion leads to a swelling pressure and
thus expansion which is a characteristic of alkali-silica reaction. The extent of swelling has not been predictable since the corresponding specific volume is a function of water content and the composition of the gel (e.g., Na, Ca, Al content, etc.) [32]. The ASR gel can further react with
Ca2+ ions originating from calcium hydroxide (CH), available inside concrete. This ion exchange
reaction involves Ca2+ replacing 2Na+ in the gel. As the Ca content of the gel increases, it
becomes more viscous and less swelling, due to cross-linking of Si by Ca(as discussed earlier).
Given the low solubility of Ca(OH)2 in comparison with NaOH, this ion-exchange reaction,
results in increasing the pH of pore solution. It is suggested that ASR gel may be a two-phase
composite, a non-swelling high Ca gel being dispersed in a swelling low Ca gel component [33].
Field investigation of ASR gels has revealed that they have typical compositional molar ratios in
a range 0.05-0.6 for (Na2O+K2O)/SiO2 and 0-0.2 for (CaO+MgO)/SiO2 which can change with
time and distance from ASR gel production sites [34].
There are also variations in the viscosity of gels depending on their composition and water
content. Some gels are sufficiently fluid to flow easily along cracks and fill voids inside concrete
without exerting damaging stresses. In some cases, examples of half-filled voids indicate the
25
original orientation of the sample with respect to the gravity. The relationship between the gel
composition, its viscosity, and expansion pressure is complex and not clearly understood [35].
2.6 Controlling ASR:
The deleterious effect of ASR can be mitigated if one of its governing factors (i.e. presence of
hydroxyl ions, reactive meta-stable silicates, and moisture) is eliminated. By application of glass
in concrete an abundant source of amorphous silicate is already provided. Thus, the focus on
ASR mitigation should be on the other two factors. Since high alkalinity is the triggering mechanism for ASR, any method capable of lowering the pH of pore fluid is expected to be helpful in controlling this reaction. As such, it has been attempted to put restrictions on the alkali content of concrete or specifically its binder phase. Researchers have been trying to establish a threshold for total alkali content in concrete which corresponds to deleterious ASR. For example, it has been shown that abnormal ASR expansion in concretes containing United Kingdom’s reactive aggregates occurs for alkali levels in excess of about 5 kg/m3 of concrete [36]. Other
workers have focused on cement composition which is believed to be more responsible for alkali
release. The simplest way to reduce the alkali content of binder is to take advantage of low alkali
cement. It has been shown [37] that below 0.6% Na2O equivalent, deleterious expansion usually
does not occur (Na2Oeq=Na2O+ 0.658 K2O).
Cement alkalis are often found as Na2SO4 or K2SO4 on the surface of portland cement particles
[38]. Unfortunately, the alkali content of cement is difficult to lower since it is mostly dependent
on the original raw feed materials for its manufacturing. This is further dictated by the
26
composition of nearby quarries and sources. From a technological standpoint, it is very
expensive and energy consuming to extract cement alkalis.
A more economical and environmental friendly option to reduce the alkali content of concrete
binder is by partial cement replacement with supplementary cementitious materials (SCM).
SCMs are natural or byproduct materials with a common property of high amorphous silicate
content and less chemical reactivity than portland cement. Fly ash, ground granulated blast
furnace slag (GGBFS), silica fume and some natural materials (e.g., volcanic ash) exhibit
pozzolanic behavior due to their high silicate content.
Inclusion of SCMs results in diluting the alkali content of the binder. As a less reactive material, they release alkalis to the pore solution at a lower rate, even if they have a higher total alkali content comparing to portland cement [39]. Also, the products of hydration of cements blended with SCMs are known to have a higher capacity for binding alkalis. The combination of alkali dilution and binding of SCMs can be very effective for mitigation of ASR. This remedy however may not be sufficient for concretes exposed to external alkali sources (e.g., deicing salts).
Nevertheless, application of SCMs can benefit concrete and mitigate ASR by other mechanisms than are further discussed in chapters 3 and 4.
Moisture is an essential factor in ASR initiation and propagation. It carries the attacking hydroxyl ions and is further imbibed by the ASR gel, causing it to swell. Any technique which hinders water access to reaction sites is also expected to be successful in suppressing ASR. This can be achieved by coating of exposed surfaces of concrete (e.g., silanol coating) or densifying the microstructure of the binder with different techniques to reduce mass transport rate. Coating may be a good option for specific structures (e.g., highway barriers). For others, coating may not
27
be a reliable remedy. For example, in pavements the coating can be removed by traffic and
moisture can also be absorbed from the underlying soil. Weathering is another concern
considering the time scale of ASR (years). Coatings also need a continuous inspection and repair
which is impossible for inaccessible surfaces.
Another ASR mitigation technique is the application of lithium salts as additives inside concrete
[40]. The effectiveness of these additives appears to be based on the preferential formation of
non-swelling lithium silicate hydrates as opposed to the ASR gel [41].
2.7 Tests to Evaluate the ASR risk of Aggregate-Cement Combinations
Since the first diagnosis of ASR by Stanton [42], different tests have been developed to examine
the reactivity of various aggregates or the preventive effect of different supplementary
cementitious materials. Meta-stable silicate is an essential prerequisite of ASR. Therefore, it has
been proposed that by measurement of the volume of meta-stable silicate content of aggregates,
it is possible to estimate the risk of reactivity of that aggregate [43,44]. As such, petrographic
examination using a polarizing microscope with thin sections cut from aggregates can be one method in identifying aggregate components which may be potentially reactive [32]. Another test, standardized as ASTM C289, measures the chemical reactivity of aggregates. A sample of aggregates in question is stored in 80°C inside 1M NaOH solution for 24 hours. The aggregate reactivity is determined by measuring the concentration of dissolved Si and reduction in Na concentration of the solution. The main drawback of this test is that it does not represent the conditions to which aggregates are exposed to inside concrete. Not all aggregates with same silica react similarly at the same rate inside concrete. Also, the effect of cementitious materials
28
(e.g., fly ash) cannot be included in this test. As such, this method has not been suggested
reliable to evaluate aggregate reactivity in concrete.
Since expansion is an intrinsic characteristic of ASR reaction, it has been suggested to monitor
the length changes of concrete or mortar specimens to evaluate the reactivity of various
aggregates or aggregate-cement-pozzolan combinations. A deleterious expansion for concrete is
attributed to 0.04% to 0.05% expansion [45]. Thus, it is possible to monitor the ASR expansion
over time and have a judgment of the degree of reactivity of any aggregate. Ideally, preparing large concrete blocks with suspicious aggregates and exposing them to real service life conditions is the most reliable way to evaluate long-term ASR performance of a given concrete mixture. Such outdoor exposure sites have been established in the UK and expansions have been monitored for up to 10 years [32]. Similar field exposure studies are currently being conducted at the University of Texas at Austin and CANMET in Ottawa as shown in Figure 2.8 [46]. Since
ASR reactions are not intentionally accelerated, these types of tests are time consuming and cannot be easily and practically used for most civil engineering projects.
Figure 2.8: Exposure sites at the University of Texas in Austin (left) and CANMET-MTL in Ottawa (right) [45]
29
ASR is generally a slow reaction and its symptoms may only appear after 15 to 25 years of service. As such, an important factor, for development of useful laboratory performance tests is the test duration. Researchers have been trying to develop accelerated tests to simulate ASR in a practicable time limit. Acceleration can be achieved by increasing the temperature or pH of the test or both. The shortcoming of accelerated methods, however, is that they no longer simulate the exact real life exposure of concrete structures.
Figure 2.9: container cured in ASTM C1293 conditions
Concrete prism expansion test (ASTM C1293) is a suggested standardized experiment in which
ASR is promoted by continuous exposure of concrete prisms to high humidity (100%RH) and an
elevated temperature of 38°C. The concentration of alkalis in the concrete mixture is also
artificially raised to 1.25% Na2Oeq (based on cement weight) to further boost the reaction. The
prisms (75×75×275 mm) are stored over water in a sealed container (Figure 2.9) and held in these conditions for 1 to 2 years. It takes one year to identify the reactivity of an aggregate source and
30 two years to assess the mitigation of ASR by SCMs or chemical admixtures. The expansions are compared with a deleterious expansion threshold of 0.04%. The main shortcoming of this method is the one/two year period of the test. In addition, alkali leaching from the prisms due to water condensation on the surface of specimens is a problem [45].
An earlier similar test (ASTM C227) has been standardized by ASTM based on the methodology first established by Stanton [42]. The conditions and duration of this test are the same as C1293 except that it is performed on mortar bars instead of concrete using reactive fine aggregate with a specific gradation. The prism dimensions are also smaller, resulting in alkali leaching to be more significant in ASTM C227. Overall, this test is considered not reliable [45,47]. Due to alkali leaching, the alkali content of concrete required to produce expansion has been found to be much higher than what is required in ASTM C1293 test. This test has failed to correctly identify the potential reactivity of numerous well established reactive rock types. Slowly reactive rocks such as certain gneisses, greywackes, argillites, quartzites and metavolcanics may not expand in this test when combined with high-alkali cement [41]. Hence, this test is not suitable for measuring reactivity of aggregates or the preventive effects of different admixtures.
To reduce the duration of experiments to improve the practicality of the test, accelerated mortar bar test (AMBT) have been developed by Oberholster and Davies [48]. In this test, the reaction is further promoted by submerging mortar bars (with same fine aggregate gradation of ASTM
C227) inside a 1M NaOH solution at an elevated temperature of 80°C. Expansions are monitored over time and finally the 14-day expansion is compared with a threshold of 0.1%. This test is standardized as ASTM C1260 for aggregate reactivity and ASTM C1567 in case of assessment
31
of the efficiency of additives or pozzolans to mitigate ASR (Figure 2.10). This test is widely
accepted in concrete quality control practice in North America due to its short period. However, this test in turn has got criticized as it over-promotes ASR and questions the ASR performance of some aggregates that have been previously exhibited innocuous behavior in long-term tests and/or field exposures. In addition, it is not an independent test as even if an aggregate does not
meet its requirements to be considered innocuous, supplementary information should be
provided by other tests for a complete rejection. Nevertheless, it has been shown that in general
there is a good agreement between highly promotes (ASTM C1260) and moderately promoted
(ASTM C1293) tests. Thomas and Innis have experimented various combinations of
supplementary cementitious materials (SCM’s) and aggregates and demonstrated a reasonable
correlation between the 2-year expansion (ASTM C1293) and the 14-day expansion of
accelerated mortar bar test (ASTM C1260) [49]. These tests will be more discussed in section
3.1.
Figure 2.10: ASTM C1260 mortar bars cured in 1 N NaOH @80°C for 14 days
32
2.8 Summary:
This chapter provided a review on the application of recycled glass in concrete. Beneficial effects and challenges were explained. The main obstacle against use of recycled glass in concrete was introduced to be the deleterious alkali-silica reaction. The underlying chemical mechanism Chemical mechanism of the ASR reaction was reviewed. Properties of the ASR gel were described and different methods to control ASR in concrete were pointed out. Finally, standard performance test methods for evaluation of ASR risk were explained and their advantages and limitations were discussed.
33
2.9 References:
1. I. B. Topcu and M. Canbaz, Properties of concrete containing waste glass, Cement and Concrete Research, 2004, 34, 267-274
2. C. D. Johnston, Waste glass as coarse aggregate for concrete, Journal of Testing and Evaluation, 1974, 2(5), 344-350
3. C. Polley, S.M. Cramer and R.V. de la Cruz, Potential for using waste glass in portland cement concrete, ASCE Journal of Materials in Civil Engineering, 1998, 10(4), 210-219
4. F. Rajabipour, H. Maraghechi and G. Fischer, Investigating the alkali-silica reaction of recycled glass aggregates in concrete materials, ASCE Journal of Materials in Civil Engineering, 2010, 22(12), 1201-1208
5. Y. Shao, T. Lefort, S. Moras and D. Rodriguez, Studies on concrete containing ground waste glass, Cement and Concrete Research, 2000, 30(1), 91-100
6. V. Corinaldesi, G. Gnappti, G. Moriconi and A. Montenero, Reuse of ground waste glass as aggregate for mortars, Waste Management, 2005, 25(2), 197-201
7. C. Shi, P. V. Krivenko and D. Roy, Alkali-Activated Cements and Concretes, 2006, Taylor and Francis, New York
8. G. Chen, H. Lee, K. L. Young, P. L. Yue, A. Wong, T. Tao and K. K. Choi, Glass recycling in cement production-an innovative approach, Waste Management, 2002, 22, 747-753
9. Z. Xie,Y. Xi, Use of recycled glass as a raw material in the manufacture of portland cement, Materials and Structures, 2002, 35, 510-515
10. M. F. Ashby, Materials for the Environment, Butterworth-Heinemann (Elsevier), Burlington, MA 2009
11. J. Bourn, Building for the future: Sustainable construction and refurbishment on the government state, National audit office, 2007
12. S. Chandra, Waste Materials Used in Concrete Manufacturing, NOYES Publications, 1997
13. J.S. Damtoft, J. Lukasik, D. Herfort, D. Sorrentino, E.M. Gartner, Sustainable development and climate change initiatives, Cement and Concrete Research, 2008, 38, 115-127
14. A. Bilodeau and V. M. Malhotra, High-Volume Fly Ash System: Concrete Solution for Sustainable Development, ACI Materials Journal, 2000, 97(1), 41-48
15. J. Davidovits, Properties of geopolymer cements, Concrete Int 9 (1987), 23-35.
34
16. C. Shi and K. Zheng, A review on the use of waste glasses in the production of cement and concrete, Resources, Conservation and Recycling, 200, 52(2), 234-247.
17. F. Rajabipour, , G. Fischer, P. Sigurdardottir, S. Goodnight, A. Leake and E. Smith, Recycling and utilizing waste glass as concrete aggregate, TRB Annual Conference, 2009, CD-Rom, Paper# 09-2195, Transportation Research Board, Washington, DC
18. S.C. Koui, C.S. Poon, Properties of self-compacting concrete prepared with recycled glass aggregate, Cement and Concrete Composites, 2009, 31, 107-113
19. M. C. Limbachiya, Bulk Engineering and durability properties of washed glass sand concrete, Construction and Building Materials, 2009,23, 1078-1083
20. H. Scholze, Glass Nature, Structure and Properties, Springer-Verlag, New York, 1991
21. L. Armelaoa, A. Bassanb, R. Bertoncello, G. Biscontinc, S. Daoliod and A. Glisentib ,Silica glass interaction with calcium hydroxide: A surface chemistry approach, Journal of Cultural Heritage, 2000, 1, 375-384
22. B. C. Bunker, Molecular mechanisms for corrosion of silica and silicate glasses, B. C. Bunker, Journal of Non-Crystalline solid, 1994,179, 300-308
23. R. K. Iler,The chemistry of silica: solubility, polymerization, colloid and surface properties, and biochemistry, WILEY, 1985
24. I. L. svensson, S. Sjoberg and L. Ohman, Polysilicate Equilibria in Concentrated Sodium Silicate Solutions, Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases, 1986, 12, 3635-3646
25. Y. Oka, K. S. Ricker, M. Tomozawa, Calcium Deposition on Glass Surface as an Inhibitor to Alkaline Attack, Journal of the American ceramic society, 1979, 62(11-12), 631-632
26. Eric Le Bourhis, Glass: Mechanics and Technology, WILEY-VCH, Weinheim, 2007
27. F. Gaboriaud, A. Nonat, D. Chaumont, and A. Craievich, Aggregation and gel formation in basic silico-calco-alkaline solutions studied: A SAXS, SANS, and ELS study, Journal of physics and chemistry, 1999, 103, pp. 5775-5781
28. M. D. A. Thomas, The role of calcium in alkali–silica reaction, In: M. Cohen, Editor, Materials Science of Concrete, The Sidney Diamond Symposium, American Ceramics Society, Westerville, OH (1998), 325–337
29. R. F. Bleszynski, M. D. A. Thomas, Micro structural studies of alkali-silica reaction in fly ash concrete immersed in alkaline solutions. Advanced Cement Based Materials, 1998, 7(2), 66- 78
35
30. S. Chatterji , The role of Ca(OH)2 in the breakdown of Portland cement concrete due to alkali-silica reaction, Cement and Concrete Research, 1979, 9(2), 185-188
31. X. Hou, R. J. Kirkpatrick, L. J. Struble and P. J. M. Monteiro, Structural Investigations of Alkali Silicate Gels, Journal of the American ceramic society, 2005, 88(4), 943–949.
32. R.N. Swamy, The Alkali-Silica Reaction in Concrete, Routledge, New York, 1992
33. R. Helmuth, D. Stark, S. Diamond and M. Moranville-Regourd, Alkali-silica reactivity: An overview of research, Strategic Highway Research Program, National Research Council, Washington DC,1993
34. X. Hou, L. J. Struble, R. J. Kirkpatrick, Formation of ASR gel and the roles of C-S-H and portlandite, Cement and Concrete Research, 2004, 34(9), 1683-1696
35. S. Diamond, ASR – Another look at Mechanisms, 8th International Conference on Alkali Aggregate Reaction, (Ed. K.Okada, S. Nishibayashi, and M. Kawamura), Kyoto, Japan, 83- 94
36. D.W. Hobbs, The effectiveness of fly ash in reducing the risk of cracking due to ASR in concretes containing cristobalite, Magazine of Concrete Research, 1994, 46(168), 167-175
37. H. Woods, Durability of Concrete Construction, Monograph, 1968, No. 4., American Concrete Institute, Detroit, Michigan
38. I. Jawed and J. Skalny, Alkalies in cement: A review I. Forms of Alkalies and their effect on clinker formation, Cement and Concrete Research, 1977, 7(6), 719-729
39. S. Diamond, Effects of two Danish fly ashes on alkali contents of pore solutions of cement-fly ash pastes, Cement and Concrete Research, 1981, 11, 383-394
40. X. Feng, M.D.A. Thomas, T.W. Bremner, B. J. Balcom, K. J. Folliard, Studies on lithium salt to mitigate ASR-induced expansion in new concrete: A critical review, Cement and Concrete Research, 2005, 35, 1789-1796
41. K.J. Folliard, M. D. A. Thomas and K. Kurtis, Guidelines for the use of lithium to mitigate or prevent ASR, 2003, Publication number FHWA-RD-03-047, Federal Highway Administration, Washington, DC
42. T. E. Stanton, The expansion of concrete through reaction between cement and aggregate, Proceedings ASCE, 66, 1940, 1781-1811
43. I. Sims and P. Nixon, RILEM Recommended Test Method AAR-I: Detection of potential alkali-reactivity of aggregates-Petrographic method, Materials and Structures, 2003, 36, 480-496,
44. ASTM C295, Standard Guide for Petrographic Examination of Aggregates for Concrete
36
45. D.W. Hobbs, Deleterious expansion of concrete due to alkali-silica reaction: influence of Pfa and slag, Magazine of Concrete Research, 1986, 38(137), 191-205
46. 1 M. D. A. Thomas, B. Fournier, K. Folliard, J. Ideker and M. Shehata, Test methods for evaluating preventive measures for controlling expansion due to alkali-silica reaction in concrete, Cement and Concrete Research, 2006, 36, 1842-1856
47. 1 R. F. Blanks and H. S. Meissner, The Expansion Test as a Measure of Alkali-Aggregate Reaction, ACI Journal Proceedings, 1946,42(4), 517-540
48. 1 R.E. Oberholster and G. Davies, An accelerated method for testing the potential alkali reactivity of siliceous aggregates, Cement and Concrete Research, 1986,16, 181–189
49. 1 M. D. A. Thomas and F. A. Innis, Use of the accelerated mortar bar test for evaluating the efficacy of mineral admixtures for controlling expansion due to alkali–silica reaction, Cement-Concrete and Aggregates, 1999, 21(2), 157–164
37
CHAPTER
THREE
HOW DOES FLY ASH MITIGATE ALKALI-SILICA REACTION (ASR)
IN ACCELERATED MORTAR BAR TEST (ASTM C1567)?
3.1 Introduction
Alkali-silica reaction (ASR) is a deleterious reaction in which amorphous or cryptocrystalline silicates contained in some aggregates (e.g., chert, greywacke, silicate glass) dissolve in the high alkaline pore solution of concrete and further coagulate in the form of a colloidal silicate gel.
This gel is highly hygroscopic and can swell significantly in exposure to moisture [1,2]. The resulting tensile stresses lead to cracking and deterioration of concrete. Literature on the mitigation of ASR by using supplementary cementitious materials (SCM) is extensive. For example, it has been shown that ASTM C618 coal fly ash is effective in controlling the ASR
38
expansions of different aggregates [3-14] and this has been linked to the ability of fly ash in
reducing the alkalinity of concrete’s pore solution [15].
A variety of test methods are available to examine the potential reactivity of an aggregate source
and to determine the required dosage of a certain SCM to be used in combination with a given
reactive aggregate [16,17] as explained in 2.7. Among these, ASTM C1293 (concrete prism
expansion test) has been suggested as the most reliable laboratory method to correctly replicate
ASR that occurs in concrete structures in service, albeit at a moderately accelerated pace. In this
test, the expansion of concrete prisms containing reactive coarse or fine aggregates is monitored
over a 12-month period (24 months if SCMs are used). ASR is accelerated by boosting the
mixture’s alkali content and by maintaining the samples at 38oC and 100%RH. The main
drawback of this test is its duration (up to 24 months); which has prevented the widespread use
of this test, especially for QC/QA applications to determine the durability of a concrete mixture
when a new source of aggregates or a new source of SCM must be used.
Alternatively, ASTM C1567 (accelerated mortar bar test) provides a highly accelerated test method to determine the potential ASR performance of combinations of SCMs and aggregates.
In this test, mortar bars containing potentially reactive aggregates are submerged in a bath of 1M
NaOH at 80oC; and their expansion measured over a 2-week period. This test too has received
its share of criticism, mainly because it exposes aggregates to an excessively harsh environment,
and since the underlying ASR mechanisms may be different than those in field conditions. For
example, unlike actual concrete structures whose pore solution alkalinity is dictated by the w/c,
cement alkali content, and presence of SCMs; ASTM C1567 exposes mortar bars to rapid
penetration of external alkalis, and as such, the results may be significantly affected by the mass
39 transport properties of mortars [18]. Despite these criticisms, ASTM C1567 has been shown to generally provide a conservative assessment of the ASR performance of mixtures containing
SCM [16]. Based on published data for 70 different combinations of SCM and aggregates,
Thomas and Innis [19] concluded that the required dosage of SCM to produce <0.1% expansion in C1567 test (and therefore passing this test) had a low risk of failing ASTM C1293. As such, and due to its time effectiveness, ASTM C1567 is currently widely used in North America and elsewhere to determine the required dosage of SCM for preventing ASR.
It is not clear why fly ash and other SCMs inhibit ASR in ASTM C1567 test. Past research using field exposure and ASTM C1293 tests have suggested that fly ash mitigates ASR primarily through reduction of concrete pore solution’s pH through alkali dilution and binding [15]. This conclusion, however, may not be directly applicable to ASTM C1567 mortars that are exposed to an “inexhaustible” source of external alkalis [20], which may erase the beneficial effects of alkali dilution and binding. As such, to better understand the mechanisms leading to mitigation of ASR in ASTM C1567, this study performs a systematic evaluation of mortars containing fly ash and a highly reactive aggregate (soda-lime-silica glass). A variety of analytical tools were used to study the mortar microstructure, its mechanical and transport properties, pore solution composition, aggregate dissolution rate, and ASR gel formation and composition.
3.2 Existing Literature
Prior studies have mentioned six potential mechanisms for mitigation of ASR by fly ash: alkali dilution, alkali binding, limiting mass transport, improving strength, modifying ASR gel, and consumption of portlandite. A brief description of each mechanism is provided below.
40
(1) Alkali dilution: the alkalinity of concrete’s pore solution is reduced by replacing cement with
fly ash [6,10]. Even when the Na2Oeq content of fly ash is higher than cement’s, the dilution
effect is observed as only a portion of fly ash alkalis are soluble per ASTM C311 test [2]; and
even that portion dissolves very slowly in the pore solution [6].
(2) Alkali binding: the additional pozzolanic C-S-H is able to bind alkali ions and remove them
from the pore solution [21,22]. In addition to increasing the volume fraction of C-S-H, the
pozzolanic C-S-H has a lower C/S, and shows a higher alkali binding capacity in comparison
with C-S-H from the cement hydration [15,23]. This may be attributed to increasing the acidity
of silanol (Si-OH) groups [24] or developing alkali attractive (i.e., negative) surface charges on
the low C/S C-S-H [25]. In addition, the alumina in fly ash can be incorporated into pozzolanic reactions to form C-A-S-H gel with a considerably higher binding capacity than C-S-H [26].
Based on the results of ASTM C1293 test, pore solution alkalinity of concrete has a strong
correlation with ASR prism expansions [27]. Fly ashes with high Na2Oeq and CaO contents have
been found to be less efficient in reducing pore solution alkalinity, and also less effective in
controlling ASR [22,28].
(3) Limiting mass transport: pozzolanic reactions can, over time, reduce the mass transport
properties of concrete [1,29]. A reduction in ion diffusion coefficient is significant especially
where an external source of alkalis is present. In addition, reduction in hydraulic permeability of
concrete can slow down the absorption of water and swelling of the ASR gel. Despite the
importance of mass transport for long-term measurements, the significance of this mechanism
for accelerated tests is unclear. Fly ash generally reacts slower than Portland cement and it is
41
reported that concrete containing fly ash has higher porosity and transport properties at early-
ages in comparison with 100% Portland cement concrete [30,31].
(4) Improving tensile strength: fly ash concrete, over time, develops a higher tensile strength
which aids in resisting internal stresses and cracking [1,32]. Again, it is unclear whether the
tensile strength is improved or reduced at early ages due to slow reactivity of fly ash.
(5) Modifying ASR gel properties: by changing the composition of ASR gel, fly ash may reduce
the swelling capacity, swelling pressure, and viscosity of ASR gel. Monteiro et al. [25] observed
that high CaO/Na2Oeq gels have lower swelling capacity than gels with low CaO/Na2Oeq. Struble and Diamond [33] reported that calcium-free sodium-silicate gels can become fluid (i.e., lose their viscosity) under moderate pressures; unlike calcium-rich gels which remain solid.
Similarly, Bleszynski and Thomas [20] observed that very low CaO/Na2Oeq gels can diffuse
freely into surrounding cement paste without exerting damage; which may indicate their low
viscosity and swelling pressure. Bonakdar et al. [34] related ASR gel composition to the number
of bridging oxygens in silica tetrahedra and reported that gels produced in a less basic environment (i.e., fly ash) have fibrous structure which results in limited swelling pressure,
comparing with the 3-dimensional gels form without fly ash.
(6) Consumption of portlandite: it has been suggested that the presence of free Ca(OH)2 is necessary for formation of expansive ASR gels at the perimeter of reactive aggregates [20,35].
As a result, consumption of portlandite by the pozzolanic reaction of fly ash may reduce the tensile stresses resulting from formation and swelling of the ASR gel. Portlandite can also serve as a pH buffer to maintain the pore solution’s pH above 12.6. However, the pH of pore solution is generally larger than 13.0 due to presence of alkali sulfates in Portland cement [36].
42
Given that recycled soda-lime-silica glass (e.g., crushed container or window glass) is used as
reactive aggregates in this study, a brief literature review on the alkali-silica reactivity of this
material is useful. Soda-lime glass contains a homogenous matrix of amorphous silicate. The
typical chemical composition includes SiO2 (>70%), Na2O (~13%), and CaO (~11%). Since
glass particles are crushed, they are angular and have residual intra-particle microcracks. ASR
induced by recycled glass aggregates shows a considerable size effect in which larger glass particles are significantly more reactive than smaller particles [37]. This trend is counter intuitive as increasing the particle surface area should accelerate the reactions; assuming that ASR occurs at glass-cement paste interface. SEM examination of ASTM C1293 and C1260 specimens affected by ASR revealed that glass particles do not undergo ASR at their surface [37,38].
Rather, ASR initiates inside intra-particle cracks which are originated during bottle crushing (i.e., they exist before glass is mixed in concrete). Larger particles contain wider cracks and a higher crack density which result in a higher alkali silica reactivity [38]. The hypothesis that residual bottle crushing cracks are responsible for ASR was validated by two further observations. First, soda-lime glass beads with same oxide composition but no internal cracks were found to be innocuous when tested by ASTM C1260 [39]. Second, annealing of crushed glass cullet to heal the residual cracks was found to mitigate ASR [38]. The reason that glass-paste interface is protected from ASR is not clearly understood, but may be related to deposition of portlandite
(CH) which favors a pozzolanic reaction [40]. Evidence of pozzolanic reaction and C-S-H formation at the interface has been observed using SEM/EDS [37].
This paper investigates the contributions of different mechanisms leading to mitigation of ASR by fly ash in ASTM C1567 test. Out of the six potential mechanisms described above, the first five are considered in this work. Since recycled glass aggregates do not undergo ASR at their
43 surface, where CH is confirmed by SEM to be present; it was decided to exclude the portlandite consumption mechanism from this study. Currently a parallel study is being performed in our research group dedicated to better understanding the role of CH on alkali-silica reaction.
In addition, a new hypothesis is proposed and evaluated in this work: “The presence of fly ash
(or other SCMs) significantly increases the available surface area of amorphous silicates in the system. Since such surfaces attract the damaging OH- ions; at a given pH, the concentration of
OH- ions attacking a unit surface area of siliceous aggregates will be reduced. This results in reducing the dissolution rate of the aggregates.” This hypothesis, if validated, can serve as a new mechanism for ASR mitigation by siliceous SCMs.
3.3 Materials and Methods
Mortar mixtures were prepared according to the proportions of ASTM C1567 using pulverized recycled glass sand and a mixture of portland cement and fly ash as binder. ASTM C150 type I
Portland cement was used. Six ASTM C618 fly ashes were studied including four class F
(identified as F1, F2, F3, F4), and two class C (identified as C1, C2). Fly ashes were used at different cement replacement levels (mass based) to evaluate their efficiency in controlling ASR.
A control mortar containing 100% portland cement was also tested. The oxide composition of cement, glass, and all fly ashes are presented in Table 3.1. Recycled glass sand was composed of three main colors: amber (~30%), clear (~30%) and green (~40%). Glass bottles were washed and crushed using a ball mill. All mortars were prepared with w/cm=0.47, and 53% volume fraction of sand with gradation in the range 150μm to 4.75mm according to ASTM C1567.
Mortars were mixed according to ASTM C305. Specimens were prepared and moist cured for 24
44
hours at 23°C. After 24 hours, specimens were demolded and cured submerged in water at 80°C
for another 24 hours. Finally, the specimens were transferred to a 1M NaOH bath at 80°C and maintained for 14 days. The following tests were performed subsequently.
3.3.1 Accelerated mortar bar test (ASTM C1567)
This test was used to determine the dosage of each type of fly ash needed to reduce the mortar
bar expansion below the ASTM threshold of 0.1%. A total of 56 different mixtures were tested.
For each mixture, four 25×25×250mm prisms were prepared with embedded gage studs at the
ends to facilitate length measurements. A digital comparator with accuracy ±0.0025mm was
used. Prism lengths were measured after demolding, 24 hours of water curing, and 1, 3, 5, 7, 10,
and 14 days submersion in NaOH bath.
Table 3.1: Oxide composition (wt %) of Portland cement, glass, and fly ashes
OXIDE CEMENT GLASS F1 F2 F3 F4 C1 C2 CaO 62.50 10.62 2.42 1.26 3.81 13.52 26.63 27.33
SiO2 19.90 73.13 51.75 59.93 49.20 52.23 33.48 34.02
Al2O3 5.44 1.99 33.70 24.97 23.34 16.36 18.59 18.74
Fe2O3 2.26 0.52 4.08 6.33 14.72 5.78 6.13 5.86
Na2O 0.30 13.74 0.40 0.36 0.69 2.82 0.37 1.50
K2O 0.89 0.34 1.16 1.90 1.78 2.16 0.72 0.35 MgO 2.31 0.53 0.60 1.00 1.03 4.30 1.48 5.00 MnO 0.09 N/D* 0.03 0.07 0.03 0.05 0.03 0.05
TiO2 0.29 N/D* 1.30 1.48 1.03 0.64 1.95 1.57
SO3 4.93 N/D* 0.25 1.33 1.47 1.17 7.65 3.67
P2O5 0.23 N/D* 0.30 0.28 0.35 0.18 0.26 1.03 LOI 0.86 --- 4.01 1.09 2.55 0.80 2.71 0.88 C/S 3.14 0.19 0.05 0.02 0.08 0.26 0.80 0.80
Na2Oeq 0.89 13.96 1.16 1.61 1.86 4.24 0.84 1.73 *N/D is not detected
45
3.3.2 Pore solution extraction and analysis
To investigate the impact of fly ash on pore solution ion concentrations, and to quantify the extent of alkali dilution, binding, and transport, pore solution of the control (100%PC), 15%F1 and 35%C2 mortars were extracted. The latter two mortars contained sufficient fly ash to pass
ASTM C1567 test (expansion<0.1%). Pore solution extraction was performed immediately after mixing, after demolding (24h), and at 0, 3, 7, and 14 days of NaOH bath exposure. Pore solution of plastic mortar was extracted by pressure filtration using nitrogen gas. For hardened samples, mortar prisms were removed from the baths, surface dried, broken into smaller pieces, and then pressurized inside a pore solution extraction die (Figure 3.1) up to a maximum stress of 550MPa at a controlled rate of 75MPa/min. The extracted solutions were collected in sealable
Polypropylene vials, filtered through 0.2 m PTFE membrane filters, and immediately tested by
HCl titration to measure [OH-]. The concentrations of other elements (Na, K, Ca, Si, Al and S) were measured using a Perkin-Elmer Optima 5300 ICP-AES.
Figure 3.1: Pore solution extraction die
46
3.3.3 Measurement of ion diffusivity using electrical impedance spectroscopy
The ion diffusion coefficient of mortars (D, m2/s) was determined using non-destructive
electrical conductivity measurements according to the Nernst-Einstein equation [41]:
D Do (3-1) 0
Where (S/m) is the electrical conductivity of mortar, o(S/m) is the conductivity of pore
2 solution, and Do (m /s) is the diffusion coefficient of ion in pore solution. For information on
scientific basis of this method and details of measurements and data interpretation, please see
[18,42,43]. Samples from 3 mortars were tested: control, 15%F1, and 35%C2. The electrical
conductivity of mortars () was measured between a pair of 3.2mm-diameter stainless steel electrodes embedded through the thickness of each mortar bar at 15mm distance between the centers of electrodes. The bulk resistance was measured using an HP-4194A impedance analyzer
(Figure 3.2) in frequency sweep mode (40Hz to 10MHz) and using 250mV voltage.
The results were converted to electrical conductivity using an experimentally established geometry factor [44]. The measurements were performed after 0, 3, 7, and 14 days of mortars exposure to NaOH bath. An average of four measurements obtained from four duplicate specimens was used to establish each data point. In parallel, pore fluids of mortars were extracted (as described above) and their electrical conductivity (o) was measured using a
commercial conductivity probe. The value of ion diffusivity of pore solution was assumed as
-9 2 o Do=1.065×10 m /s for NaOH in water at 80 C [45]. This value corresponds to effective self-
diffusion coefficient of NaOH that is calculated according to the formula by [45] using the
47
diffusivity of Na+ and OH- ions. The self-diffusion coefficient does not consider the ionic strength of the pore solution.
Figure 3.2: HP-4194A impedance analyzer
3.3.4 Tensile and compressive strength tests
To examine the effect of fly ash on tensile strength (i.e., modulus of rupture) of mortars,
25×25×250mm prisms were tested in 3-point bending. The set-up is shown in Figure 3.3.
Specimens were prepared according to ASTM C1567 and using a dosage of each fly ash that
controlled expansions to below 0.1%: 15%F1, 15%F2, 20%F3, 20%F4, 25%C1 and 35%C2.
Prisms were tested after 3 days exposure to NaOH bath. ASTM C1567 measurements showed
expansions starting in the controlled (100%PC) mortar after 3 days NaOH exposure. As such,
this time was chosen as the onset of cracking. Prisms were tested using a displacement-
controlled setup that applied a mid-span deformation of 5µm/sec until failure. The compressive
48
strength of similarly cured mortar cubes (50×50×50 mm) was tested according to ASTM C109.
For both tensile and compressive strength tests, an average of 3 measurements was used per each mixture.
Figure 3.3: 3-point bending test performed on 25×25×250mm mortar bars with MTS machine: Displacement control mode
3.3.5 SEM/EDS imaging
SEM/EDS imaging was used to study the microstructure of the mortars, and to study ASR gel
formation and composition. Imaging was performed on the control, 15%F1, and 35%C2 mortars.
After 7 and 14 days of NaOH exposure, cross sections were saw-cut from mortar bars with
49 approximate thickness of 1cm. Specimens were vacuum dried inside a desiccator for 48 hours and then impregnated with a low viscosity epoxy. After setting, the epoxy was polished off and the specimens were successively polished with 30, 15, 9, 6, 3 and 1 μm grits. Polishing oil was used instead of water to prevent leaching. Specimens were carbon coated prior to SEM. Image acquisition was performed in BSE mode using an FEI Quanta 200 ESEM instrument with a lateral resolution of 3.5nm. The ESEM was equipped with an X-ray energy dispersive spectroscopy (EDS) detector for compositional analysis. EDS spot analysis (20×20 windows) was performed on at least 20 ASR gel points for the control, 15%F1 and 35%C2 mortar cross sections to gain information regarding the composition of ASR gels. The gel composition was examined after 7 and 14 days of NaOH exposure to monitor changes in its composition over time.
3.3.6 Aggregate dissolution rate measurements
To assess the dissolution rate of glass aggregates in the presence or absence of fly ash, a glass corrosion experiment was performed. Soda-lime glass slides (75×25×1mm) were submerged in a
340mL 1M NaOH solution at 80°C for a period of 14 days. Solutions were not stirred during the experiments. Periodic mass loss measurements from glass slides were used to monitor the dissolution rate of glass. For each measurement, the glass slides were removed from the solution, the silica gel formed on the surface of each slide was carefully washed off by deionized water.
Mass measurements were performed using a balance with accuracy of ±0.0001g. For each data point, measurements were obtained from two duplicate slides. The corrosion experiments were performed in the absence of fly ash, and also for systems where 10g or 20g of F1 fly ash was added to the solution. The corrosion tests were performed inside sealable plastic containers to
50
minimize evaporation and carbonation. The [OH-] was periodically monitored by sampling the solution and performing acid titration.
3.3.7 Numerical Model to Simulate Alkali Transport and Binding
A numerical model was developed to assess the simultaneous effects of ion diffusion and binding
on pore fluid concentration of mortars during ASTM C1567 test. Several simplifying
assumptions were made in development of this model, as described below. As such, the results
only provide a semi-quantitative evaluation of the relative significance of ion diffusion and
binding; and should not be directly/numerically compared with pore fluid compositions of
mortars that were measured experimentally (section 3.5.2).
A 1D finite-difference model was developed to simulate the diffusion of NaOH from the soak
solution by solving Fick’s 2nd law (Appendix B). The alkali binding effect was accounted for by
introducing a sink term in the model. Ion diffusion was simulated within a mortar cross section
of 25×25mm2 that was exposed to 1M NaOH from two opposite faces (Figure 3.8). The other
two faces were considered as sealed (1D model). Mortar diffusion coefficients were obtained experimentally (section 3.5.3) at the time of submerging samples into NaOH bath (2 days after casting). For the 100%PC mortar, D=1.16×10-11 m2/s and for the 15%F1 mortar, D=3.02×10-12 m2/s was used. Diffusion coefficients were assumed to remain constant during the simulations.
o -9 2 Diffusivity of NaOH in pore solution (Do) at 80 C was used as 1.065×10 m /s [45], without
considering the effect of ion activity coefficients. The model included a sink term (Appendix A)
to account for alkali binding by C-S-H according to the distribution ratio [24]:
51
[Na] in solid C-S-H (Mol/kg) Rd = (3-2) [Na] in pore solution (Mol/kg)
The values of Rd used in the model were a function of Ca/Si of C-S-H according to the work of
Hong and Glasser [24]. For 100%PC mortar (Ca/Si=1.87), Rd=0.5; and for 15%F1 mortar
(Ca/Si=1.39), Rd =1.25 were used. This suggests a better binding capacity for mortars containing
fly ash. The Ca/Si was determined by EDS of C-S-H rims in 100%PC and 15%F1 mortars. The rims of hydrated portland cement particles are shown in micrographs of Figure 3.4. The rims are embracing the unhydrated phase of cement particles. At least 20 EDS spot analyses were
perfomed on C-S-H rims around cement and fly ash particles; and the average Ca/Si was
obtained. The mass fraction of C-S-H was assumed to be the same for both 100%PC and 15%F1
pastes (mC-S-H ≈ 50%wt. of each paste). At every time step (Δt=15min), [Na] in pore solution was
calculated based on Fick’s law. Subsequetntly, [Na] in pore solution was adjusted by subtracting
the amount that can be absorbed by the solid phase according to Eq. (3-2). The model then
proceeded to the next time step and the simulation continued.
The pore solution was assumed to only contain Na+ and OH- ions; as such multi-ion diffusion
was ignored. At every point within the pore solution, charge neutrality was maintained meaning:
[Na+]=[OH-]. Initial condition was defined as [OH-]=0.23M across the thickness of mortar for
both control and fly ash mortars. This value reflects the [OH-] for the control mortar at 48 hours
after casting and before exposure to NaOH (Figure 3.6a). The 15%F1 mortar showed smaller
[OH-] at this time; however the same value of [OH-]=0.23M was chosen in the simulation of both
mortars so the final results exclusively represent the contribution of NaOH diffusion and binding
during the 14-day bath exposure period. Further OH- release due to continued hydration of 52
cement and fly ash beyond 48 hours was assumed negligible. The model’s boundary condition
was [OH-]=1.0M at 0 and 25mm (i.e., the bath concentration).
Figure 3.4: Rims of hydrated portland cement particles
3.4 Results and Discussion
3.4.1 Sufficient dosage of fly ash to mitigate ASR
The average expansion of all mortar mixtures at the conclusion of ASTM C1567 test (i.e., 14 days NaOH exposure) is presented in Table 3.2. Based on these results, all fly ashes can mitigate
53
ASR but at different dosage levels. The minimum cement replacement levels to achieve expansions below 0.1% were as follows: 15%F1, 15%F2, 20%F3, 20%F4, 25%C1 and 35%C2.
A comparison suggests that fly ashes with higher CaO content are less effective in mitigating
ASR. This could be due to inefficiency of such ashes in maintaining a low alkalinity of pore fluid due to higher ion penetrability of the matrix or lower alkali binding capacity of the matrix
[46].
Table 3.2: Expansion of mortar mixtures at the conclusion of ASTM C1567 test
Cement replacement Fly Ash level (%) F1 F2 F3 F4 C1 C2 0 0.60% 0.60% 0.60% 0.60% 0.60% 0.60% 10 0.19% 15 0.02% 0.06% 0.11% 0.15% 20 0.01% 0.03% 0.03% 0.04% 0.12% 0.26% 25 0.01% 0.01% 0.03% 30 0.00% 0.00% 0.02% 0.09% 35 0.06% Highlights represent dosage of each fly ash that resulted in expansion well below the 0.1% threshold
Figure 3.5 shows the expansion of the control, 15%F1 and 35%C2 mortars during the entire period of ASTM C1567 test. Expansions in all three mortars were small for the first 3 days exposure to NaOH. Afterwards, while the expansions in fly ash mortars increased slowly for the remainder of the test, the control mortar showed a significant increase in expansion at 5 days and beyond. The rate of expansion increased after 3 days and remained relatively constant after 5 days. This may suggest that it took ~3 days for the concentration of OH- ions to build up to sufficient levels in the control mortar to initiate ASR (see Figure 3.6a). Comparison with pore solution analysis data suggests that after 5 days, the rate of expansion may not be directly related 54
to [OH-]. This may further imply that ASR kinetics is no longer controlled by the rate of OH- diffusion, but by the rate of ASR gel formation and swelling.
0.70% Control 15%F1 0.60% 35%C2
0.50%
0.40%
0.30% Expansion (%)
0.20%
ASTM Threshold 0.10%
0.00% 0 2 4 6 8 10 12 14 Days exposed to NaOH
Figure 3.5: ASR expansion of control, 15%F1 and 35%C2 mortars during ASTM C1567
3.4.2 Pore solution composition
The results of pore solution analysis are presented in Figure 3.6(a-c). Figure 3.6(a) shows [OH-] over the duration of ASTM C1567 test. At 0d, all 3 mortars have [OH-]≈100mM; despite higher
Na2Oeq of both fly ashes comparing with PC. Note that fly ash alkalis may dissolve slower than
cement’s due to slow reactivity of fly ash. During the first 24 hours, concentrations rise mainly
55
due to cement hydration. At 24h, [OH-] of fly ash mortars is 22% and 34% less than the control
mortar; which shows the effect of alkali dilution. During the next 24 hours, mortars are
submerged in a water bath at 80oC which promotes leaching of alkalis. The rate of leaching is
proportional to the ion diffusivity of mortars and is the highest for 100%PC. At 48h, the [OH-] has dropped to 170 to 270mM and the effect of alkali dilution is nearly totally erased. During the next 14 days, [OH-] of mortar pore solutions increase steadily due to exposure to NaOH bath.
The control mortar shows the highest concentrations which agrees with its high ion diffusion coefficient (section 3.5.3). As the dissolution rate of glass aggregates is strongly related to [OH-], the highest rate of ASR is observed in the 100%PC mortar. At the test’s conclusion, the [OH-] in the control mortar reaches and surpasses the OH- concentrations of the soak solution (1M). This
is due to total breakdown of soda-lime glass structure and the release of Na to the pore solution
[47]:
+ - Si-ONa + H2O → Si-OH + Na + OH (3-3)
This reaction results in hydrolysis of water, and an increase in both [Na+] and [OH-] of pore solution. This is further discussed in section 3.5.7.
Figure 3.6b shows variations in elemental [Na] in the pore solution of the three mortars. It must
be noted that ICP measures the total [Na] which is not the same as ionic [Na+]. In pore solutions
studied here, silica gel globules (<0.2 μm) are present in the pore solution and these can contain
non-ionic Na in their structure. These globules have been dissolved from the soda-lime glass aggregates. As a result, it should not be surprising that [OH-] does not necessarily equal
[Na]+[K]. Immediately after mixing, [Na] is in the range 78 to 138mM, with the 100%PC mortar
56
1200 (a)
[OH-] in soak solution 1000 Demolding; submerging in water
800 Submerging in 1M NaOH
600 Concentration (mM) - 400 OH
Control(100% cement) 200 15%F1 35%C2 0 0246810121416 Days since casting
showing the highest concentration. During the first 48 hours, [Na] in all three mortars increase
steadily due to reaction of cement and of fly ash. At 48hrs, [Na] in 35%C2 mortar is 333mM in
comparison with 231 and 202mM for the other mortars. This is likely due to a significantly
higher Na2O of C2 fly ash (see Table 3.1). Interestingly, the water curing period (between 24 and
48 hours) did not result in reduction of [Na]; unlike the trend observed for [OH-] and [K]. After submersion of the mortar prisms inside 1M NaOH, the Na content of pore solutions increase by a factor of up to 5.8 times. This increase was more rapid in the 100%PC mortar which agrees well with its higher ion diffusion coefficient as discussed in the next section.
57
1400 (b)
1200
[Na] in soak solution 1000
800
600 Submerging in 1M NaOH in 1M Submerging Na Concentration (mM) Na Concentration
400 water in submerging Demolding;
Control(100% cement) 200 15%F1 35%C2 0 0246810121416 Days since casting
Finally, the concentrations of K in the three pore solutions are shown in Figure 3.6(c).
Significant differences in the initial [K] is observed due to potassium dilution of the fly ashes.
Although the K2O content of F1 fly ash is higher than PC (1.16% vs. 0.89%), the 15%F1 mortar shows 43% less [K] at 0 day, due to a slower reaction and release of alkalis by fly ash. During the first 24 hours, [K] increases in all three mortars. During the water curing period (between 24 and 48 hours), potassium content drops by leaching out of the mortars. At 48 hours, [K] in the
100%PC mortar is 224mM comparing with 171mM and 129mM for the fly ash mortars. During the next 14 days, where mortars are submerged in NaOH bath, potassium leaching continues,
58 albeit at a slower rate. The control mortar shows a faster rate of K leaching, in agreement with its higher ion diffusivity.
700 (c) Control(100% cement) 15%F1 600 35%C2
500
400 Submerging in 1M NaOH
300 K Concentration (mM) 200
100 Demolding; submerging in water 0 0246810121416 Days since casting
Figure 3.6: Pore solution composition of mortars during ASTM C1567 test: (a) [OH-] ion, (b) [Na] element, (c) [K] element
3.4.3 Ion diffusion coefficient
The ion diffusion coefficient of mortars was measured using electrical impedance spectroscopy and according to the Nernst-Einstein equation (3-1). Figure 3.7 shows the results for the control,
15%F1, and 35%C2 mortars over the duration of ASTM C1567 test. The control mortar shows a
59
significantly larger (by a factor of 4 to 7) ion diffusivity than the two fly ash mortars. This means
that NaOH can penetrate much faster through the 100%PC mortar (in agreement with pore fluid analysis data of Figure 3.6); which in turn, results in significantly larger ASR expansions (Figure
3.5). As reported in [18], and in agreement with the results of [30,31], while the fly ash mortars
show slightly higher porosity, their lower ion diffusivity is the result of a reduction in the pore
size and a significant increase in the tortuosity of the pore network.
1.6E-11 Control 15%F1 35%C2 1.4E-11
/s) 1.2E-11 2
1E-11
8E-12
6E-12
Mortar Diffusion coefficient (m 4E-12
2E-12
0 0 2 4 6 8 10 12 14 16 Days since casting
Figure 3.7: Ion diffusion coefficient of the control (100% PC), 15%F1 and 35%C2 mortars
60
These findings strongly support the role of fly ash in reducing ion transport as a major mechanism by which fly ash mitigates ASR during ASTM C1567 test. It is observed that even after 48 hours since casting, fly ash mortars can show a considerably lower diffusion coefficient.
Note that these mortars are water cured for 24 hours at 80oC which can drastically boost the reactivity of fly ash. It is also observed in Figure 3.7 that the diffusivity of the control mortar increases from 2 to 5 and 9 days but subsequently decreases. To make sure that this is not an experimental error, the tests were repeated and similar results were obtained. The initial increase in diffusivity is probably due to microcracking caused by ASR. As these cracks are filled by
ASR gel (with much lower electrical conductivity and ion diffusivity than the pore solution); the overall conductivity/diffusivity of the mortar decreases. The diffusivity of the fly ash mortars remained relatively constant during the test period. This further implies a low level of microcracking in fly ash mortars.
3.4.4 Significance of alkali diffusion versus binding
It was observed in Figure 3.5(a) that fly ash mortars have a lower concentration of OH- ions in their pore solution; which helps in reducing the magnitude of ASR. Yet, it is not clear to what extent this reduction in [OH-] is due to the lower ion diffusivity of mortars versus the mortars’ improved alkali binding capacity. The numerical modeling results provided in Figure 3.8 is aimed at addressing this question. The figure shows the simulated [OH-] profiles as a result of the penetration and binding of NaOH during ASTM C1567 test. Here, the triangles represent the control (100%PC) mortar; where the OH- front has reached the centerline of the specimen
(position = 12.5mm) within 14 days of NaOH bath exposure. Larger [OH-] would be anticipated at the specimen’s interior had a more realistic 2D model been used, or if the model accounted for
61 the effect of cracking on increasing the mortar diffusion coefficient. SEM imaging confirmed the presence of ASR gel across the mortar’s cross section, with higher severity of ASR attack at the specimen’s surface.
1
0.9 sealed 1M NaOH 0.8 25 mm
0.7 NaOH 1M
sealed 0.6
0.5 ] (M) -
[OH 0.4
0.3
0.2 D,Rd ≡ 100%PC D ≡ 100%PC ; Rd ≡ 15%F1 0.1 D ≡ 15%F1; Rd ≡ 100%PC D,Rd≡ 15%F1 0 0 5 10 15 20 25 Position (mm)
Figure 3.8: Simulation results showing the OH- concentrations at the conclusion of ASTM C1567 test for different diffusion and binding coefficients
The squares in Figure 3.8 show the simulated [OH-] profile corresponding to 15%F1 mortar. This mortar shows lower [OH-] due to a combination of lower ion diffusivity and higher C-S-H binding capacity. In comparison with the 100%PC mortar, the NaOH penetration depth is limited
62
to approximately 5mm from the surface; which accounts for ~60% of the specimen’s cross
sectional area. SEM results confirm the absence of ASR gel in this mortar, except within few
millimeters from the surface.
To evaluate the contributions of alkali binding versus ion diffusion on [OH-] profiles, two hypothetical mortars were simulated. The first mortar (shown as diamonds in Figure 3.8) had the same ion diffusivity as the 100%PC mortar but had a higher alkali binding capacity similar to
15%F1 mortar. The second mortar (shown as circles in Figure 3.8) had the same alkali binding capacity as the 100%PC mortar but had lower ion diffusivity similar to 15%F1 mortar. By comparison, it is evident that the effect of reduced ion diffusivity is more significant than the effect of improved alkali binding in reducing [OH-]. Overall, both mechanisms delay the penetration of NaOH which favors lesser ASR.
3.4.5 Tensile and compressive strength
The results of the tensile and compressive strength measurements of the mortars after 3 days
submersion in NaOH solution is shown in Figure 3.9. The results are normalized by dividing by
corresponding strengths of the control (100%PC) mortar. Commonly, it is assumed that replacing
Portland cement with fly ash results in a reduction in the early-age strength [31]. In the
environment of ASTM C1567 test, however, it is observed that mortars containing fly ash show
15% to 38% increase in tensile strength comparing to the 100%PC mixture. Improvement in the
compressive strength can be as high as 54%. Such strength improvements could be due to
reduced poe size (i.e., flaw size) in fly ash mortars as reported in [18]. The results of TGA and
63
porosity measurements (not included here) show continuous consumption of portlandite and reduction of porosity with age in all fly ash mortars; which contributes to higher strengths.
1.8 Compressive Strength Max fc= 59MPa Tensile Strength 1.6
Max ft=7.6Mpa 1.4
1.2
1
0.8 Normalized strength (-) strength Normalized 0.6
0.4
0.2
0 Control 15%F1 15%F2 20%F3 20%F4 25%C1 35%C2
Figure 3.9: Normalized (to the control mortar) tensile and compressive strengths of mortars 3 days after exposure to 1M NaOH solution
An increase in the tensile strength of mortars in ASTM C1567 test has two benefits. First, it prevents or delays the formation of cracks. Cracking provides immediate access of the NaOH solution to the interior of the specimen, which significantly accelerates ASR. Second, by delaying ASR, fly ash provides additional time to allow the hydration of the binder to proceed,
64
which results in a denser and less permeable paste matrix. This can further mitigate alkali
transport, formation and swelling of the ASR gel, and cracking.
It should be noted that the increase in the early-age tensile strength of fly ash mortars may be an
artifact of ASTM C1567 test which exposes specimens to high temperatures and alkalinities.
Such an environment significantly promotes the pozzolanic reactions of fly ash. In real-life
service exposures, the pozzolanic reaction of fly ash may result in higher tensile strengths only
after a long term.
3.4.6 Microstructural analysis (SEM/EDS)
SEM/EDS imaging was performed to answer two main questions. First, does ASR gel form in
large quantities in fly ash mortars? Second, does the presence of fly ash alter the composition of
ASR gel and as such, change its viscosity and swelling pressure as suggested by [25,34].
Previous research by [20] on ASTM C1567 prisms containing reactive flint showed that
significant dissolution of aggregates and formation of gel could occur in mortars containing fly ash. However, probably due to its low viscosity, this ASR gel could flow freely through the cement paste matrix without exerting large stresses and cracking.
Figure 3.10 (a) and (b) show SEM images of the 100%PC and 15%F1 mortars at the conclusion of ASTM C1567 test. While the control specimen is severely distressed by ASR, only minor traces of ASR gel is detected in the fly ash mortar. Other SEM images (not included) reveal that
ASR is more severe at the perimeter versus the interior of mortar prisms; which is in agreement with the alkali transport mechanism driving the ASR.
65
A summary of the EDS compositional analysis of the ASR gel from the control and fly ash
mortars is presented in Table 3.3. A typical X-ray pattern for an ASR gel and the adjacent glass
obtained with EDS spot analysis is also shown in Figure 3.11.The results show that the gel
compositions are approximately similar; no significant differences on the Ca/Si or Ca/Na
between the control and fly ash mortars are detected. This could mean that ASR gel has similar
properties in all mortars. It is the massive volume of gel produced in the control mortar that leads
to its deterioration, in comparison with small traces of gel in fly ash mortars. In addition, it was found that although the volume of gel formed is increasing with time in all mortars, there is only a slight variation in the gel composition based on EDS analysis between 7 and 14 days NaOH exposure. Also variations in the gel composition from the perimeter to the interior of 100%PC mortar prisms were small.
(a) Control mortar
Glass
ASR gel
66
(b) 15%F1mortar
Figure 3.10: SEM micrographs of cross section of mortars at conclusion of ASTM C1567 test
Table 3.3: Average atomic composition (wt.%) of ASR gel measured by EDS
Na Ca Si K Al Mg Ca/Si Ca/Na Control 7.76 4.91 24.81 1.29 0.64 0.37 0.2 0.63 at 7 15%F1 9.04 5.33 24.37 1.66 1.56 1.44 0.22 0.59 days 35%C2 8.95 5.23 24.49 2 1.19 - 0.21 0.58 Control 10.63 6.73 22.39 0.31 0.77 0.36 0.3 0.63 at 14 15%F1 7.24 5.5 24.89 0.64 0.73 0.37 0.22 0.76 days 35%C2 9.28 5.86 23.42 1.02 0.65 0.37 0.25 0.63
67
ASR Gel Glass
Figure 3.11: Comparison of X-ray patterns of ASR gel and adjacent glass
3.4.6.1 Source of alkalis:
In the previous studies [38,39], it was revealed that glass aggregates undergo ASR from their pre-existing cracks. These cracks are generated during crushing of glass bottles to the required smaller sizes. Since soda-lime glass has high sodium content (10-15%), the source of alkalis in the ASR gel developed in the cracks was unclear. In other words, if Na in the ASR gel was originating from the external soak solution or Na from glass sand particles contributed to ASR gel formation. To answer this question a similar ASTM C1567 test was run, but this time soaking the mortar bars in 1M KOH solution. At the conclusion of the test, SEM samples were prepared according to 3.3.5. The SEM micrograph of KOH exposed mortar bar is shown in Figure 3.12(a).
Similar ASR gel patterns originated from pre-existing cracks are observed here. However, as
EDS mapping of the cross section suggests (Figure 3.12b), the concentration of potassium is very
68 high in the regions which ASR gel has developed. This confirms that the alkalis in the ASR gels are sourced from the external alkaline bath. This further confirms the role of ion transport. It takes time for the alkalis to penetrate the mortars and finally get consumed in the ASR gel in the particles’ cracks. Therefore, a less penetrable system can resist ASR for a longer period of time and retard the corresponding ASR expansion.
Glass
ASR gel
Glass
69
Concentration of K in the gels
Figure 3.12: (a) SEM micrographs of cross section of mortars at conclusion of ASTM C1567 test; mortars are submerged in 1M KOH at 80°C (b) EDS mapping of potassium element
3.4.7 Aggregate dissolution rate
To evaluate the dissolution rate of glass aggregates and the potential benefits of fly ash, a glass
corrosion experiment was performed in which the dissolution (i.e., mass loss) of soda-lime glass
slides exposed to a constant volume (V=340mL) of 1M NaOH solution was monitored over time.
Each corrosion cell, contained two glass slides with a total surface area of SA=79×10-4m2.
Duplicate experiments were performed in the absence of fly ash or by adding 10g or 20g of F1 fly ash (specific surface area = 0.2624 m2/g) to the system.
70
Exposure time (day) 02468101214 0.00
-0.02
-0.04
-0.06 ) 2 -0.08
-0.10
-0.12 Mass loss (mg/mm Mass loss
-0.14 2 glass slides 2 glass slides+10g F1 -0.16 2 glass slides+20g F1 -0.18
-0.20
Figure 3.13: Mass loss of glass slides in 1M NaOH solution at 80ºC in the presence or absence of fly ash
The results are presented in Figure 3.13. The rate of mass loss from slides is higher in the system without fly ash. At 14 days, the mass loss of slides was 48% or 62% lower when 10g or 20g fly ash was present. This is due to a drastic increase in the silicate surface area to solution volume ratio (SA/V) as a result of the silicate area provided by the fly ash (note that fly ash is mainly a silicate glass that also contains some crystalline silicate phases). While the solution volume remained at V=340mL, the silicate surface area increased to SA=2.63 m2 or SA=5.26 m2 for 10g
71
or 20g fly ash addition. As a result, the concentration of OH- ions (which are responsible for
dissolving both glass slides and fly ash) decreases significantly per unit silicate surface area. For
340mL1M NaOH solution, the OH- concentration per unit silicate surface area is 43,038 mM/m2 when no fly ash is present. This value drops to 129.3 or 64.6 mM/m2 when 10g or 20g F1 fly ash
is added. The rate of glass slide corrosion increases with time, mainly due to an increase in the
pH of the solution as a result of the ion exchange reaction presented by equation (3). Starting
from the initial [OH-]=1000 mM, the [OH-] in the solution reached 1060 mM at 14 days when no
fly ash was present. For the system with 10g fly ash, [OH-]=1150 mM at 14 days.
The results presented in Figure 3.13 suggest a new mechanism by which SCMs can mitigate
ASR. By increasing the accessible silicate surface area, the SCM reduces the effective
concentration of OH- at the surface of aggregates; and as such, reduces the aggregate dissolution
rate. Even if no subsequent pozzolanic reaction occurs (e.g., due to a local absence of portlandite), the mere presence of SCM increases the accessible silicate surface area and reduces aggregate dissolution rate. To our knowledge, this mechanism has not been previously discussed in the ASR literature.
3.5 Conclusions
The results of this study suggest that fly ash can effectively mitigate ASR in ASTM C1567 test
through the following mechanisms:
Fly ash reduces the alkalinity ([OH-]) of pore solution by significantly reducing the ion
diffusion coefficient of mortars. A diffusivity reduction by a factor of 4 to 7 was recorded, as
72
early as 48 hours after casting, when sufficient dosage of fly ash replaced portland cement.
As such, the external NaOH penetrates slower into fly ash mortars, resulting in a lower pore
fluid alkalinity, and significantly slower ASR.
Fly ash reduces the alkalinity ([OH-]) of pore solution through alkali binding. Fly ash
reduces the C/S of C-S-H gel which in turn, improves its alkali binding capacity. In addition,
more C-S-H is produced by pozzolanic reactions. As such, a considerable fraction of the
penetrated NaOH is removed from the pore solution. The results of a simple numerical model
suggested that the contribution of transport reduction is more significant than the effect of
improved alkali binding.
Fly ash increases the tensile strength of mortars and prevents or delays the onset of cracking.
This also prevents an accelerated transport of NaOH through cracks to the interior of mortar
specimens.
Fly ash can reduce the dissolution rate of siliceous aggregates even when the pH of pore
solution is maintained constant (e.g., near the perimeter of mortar prisms that are exposed to
external NaOH bath). Fly ash provides a large silicate surface area that is accessible to the
corrosive OH- ions. As such, the concentration of OH- per unit surface area of silicate is
markedly reduced. In other words, for a unit volume of pore solution at a given pH, a
significant fraction of hydroxyl ions are involved in dissolving fly ash instead of attacking
the reactive aggregates.
The results of this work also suggest that alkali dilution and modifying ASR gel composition are
not major contributors to mitigation of ASR by fly ash in ASTM C1567 test.
73
3.6 References
1. R.N. Swamy, The Alkali-Silica Reaction in Concrete, Routledge, 1992.
2. D.W. Hobbs, Alkali-Silica Reaction in Concrete, Thomas Telford, London. 1988.
3. M. Berra, T. Mangialardi and A.E. Paolini, Application of the NaOH bath test method for assessing the effectiveness of mineral admixtures against reaction of alkali with artificial siliceous aggregate, Cement and Concrete Composites, 1994, 16(3), 207-218.
4. R.L. Carrasquillo and P.G. Snow, Effect of fly ash on alkali-aggregate reaction in concrete, ACI Materials Journal, 1987, 84 (4), 299–305.
5. H. Chen, J.A.Soles and V.M. Malhotra, Investigations of supplementary cementing materials for reducing alkali-aggregate reactions, Cement and Concrete Composites, 1993, 15(1-2), 75-84.
6. S. Diamond, Effects of two Danish fly ashes on alkali contents of pore solutions of cement-fly ash pastes, Cement and Concrete Research, 1981,11(3), 383-394.
7. J. Duchesne, M.A. Bérubé, Long-term effectiveness of supplementary cementing materials against alkali–silica reaction, Cement and Concrete Research, 2001, 31(7), 1057–1063.
8. E.R. Dunstan, The effect of fly ash on concrete alkali-aggregate reaction, Cement Concrete and Aggregates, 1981, 3, 101-104.
9. D.W. Hobbs, Deleterious expansion of concrete due to alkali-silica reaction: influence of Pfa and slag, Magazine of Concrete Research, 1986, 38(137), 191-205.
10. D.W. Hobbs, Influence of pulverized-fuel ash and granulated blast furnace slag upon expansion caused by the alkali-silica reaction, Magazine of Concrete Research, 1982, 34(1), 83-94.
11. D.W. Hobbs, The effectiveness of fly ash in reducing the risk of cracking due to ASR in concretes containing cristobalite, Magazine of Concrete Research, 1994, 46(168), 167-175.
12. S. Nagataki, H. Ohga and T. Inoue, Evaluation of fly ash for controlling alkali-aggregate reaction, Proc. 2nd Int. Conf. on Durability of Concrete, Montreal, Canada, 1991, 955-972.
13. A. Shayan, R. Diggins and I. Ivanusec, Effectiveness of fly ash in preventing deleterious expansion due to alkali-aggregate reaction in normal and steam-cured concrete, Cement and Concrete Research, 1996, 26(1), 153-164.
14. M.D.A. Thomas, Field studies of fly ash concrete structures containing reactive aggregates. Magazine of Concrete Research, 1996, 48(177), 265-279. 74
15. M.D.A. Thomas, The effect of supplementary cementing materials on alkali-silica reaction: A review, Cement and Concrete Research, 2011, 41, 209-216.
16. M. D. Thomas, B. Fournier, K. Folliard, J. Ideker, M. Shehata, Test methods for evaluating preventive measures for controlling expansion due to alkali–silica reaction in concrete, 2006, Cement and Concrete Research, 36(10), 1842–1856.
17. J. Lindgård, Ö. Andiç-Çakır, I. Fernandes, T. F. Rønning, M. D. A. Thomas, Alkali–silica reactions (ASR): Literature review on parameters influencing laboratory performance testing, Cement and Concrete Research, 2012, 42, 223–243.
18. S.M.H. Shafaatian, H.Maraghechi and F. Rajabipour, Assessing the Role of Ion Transport in Mitigation of ASR by Fly Ash in Accelerated Mortar Bar Test, Submitted to ASCE Journal of Materials in Civil Engineering, May 2012.
19. M. D. A. Thomas and F. A. Innis, Use of the accelerated mortar bar test for evaluating the efficacy of mineral admixtures for controlling expansion due to alkali-silica reaction, Cement, Concrete and Aggregates, 1999, 21(2), 157-164.
20. R.F. Bleszynski, M.D.A. Thomas, Microstructural studies of alkali-silica reaction in fly ash concrete immersed in alkaline solutions, Advanced Cement Based Materials, 1998, 7(2), 66- 78.
21. J. Duchesne and M.A. Bérubé, The effectiveness of supplementary cementing materials in suppressing expansion due to ASR; II: Pore solution chemistry, Cement and Concrete Research, 1994, 24(8), 1579-1581.
22. I. Canham, C.L. Page and P.J. Nixon, Aspects of the pore solution chemistry of blended cements related to the control of alkali silica reaction, Cement and Concrete Research, 1987, 17(5), 839-844.
23. P.L. Rayment, The effect of pulverised-fuel ash on the c/s molar ratio and alkali content of calcium silicate hydrates in cement, Cement and Concrete Research, 1982, 12(2), 133-140.
24. S.Y. Hong, F.P. Glasser, Alkali binding in cement pastes: Part I. The C-S-H phase, Cement and Concrete Research, 1999, 29(12), 1893-1903.
25. P.J.M. Monteiro, K. Wang, G. Sposito, M.C. dos Santos and W.P. de Andrade, Influence of mineral admixtures on the alkali-aggregate reaction, Cement and Concrete Research, 1997, 27( 12), 1899-1909.
26. S.Y. Hong, F.P.Glasser, Alkali sorption by C-S-H and C-A-S-H gels: Part II. Role of alumina, Cement and Concrete Research, 2002, 32(7), 1101-1111.
75
27. M.H. Shehata, M.D.A. Thomas, The effect of fly ash composition on the expansion of concrete due to alkali–silica reaction, Cement and Concrete Research, 2000, 30(7), 1063- 1072.
28. M.H. Shehata, M.D.A.Thomas, Alkali release characteristics of blended cements, Cement and Concrete Research, 2006, 36(6), 1166-1175.
29. B. Lothenbach, K.Scrivener, R.D. Hooton, Supplementary cementitious materials, Cement and Concrete Research, 2011, 41, 217–229.
30. Y. Li, Y. Chen, J. Wei, X. He, H. Zhang, W. Zhang, A study on the relationship between porosity of the cement paste with mineral additives and compressive strength of mortar based on this paste fla, Cement and Concrete Research, 2006, 36(9), 1740-1743.
31. S.P Pandey, R.L Sharma, The influence of mineral additives on the strength and porosity of OPC mortar, Cement and Concrete Research, 2000, 30(1), 19-23.
32. Y. Shao, T. Lefort, S. Moras and D. Rodriguez, Studies on concrete containing ground waste glass, Cement and Concrete Research, 2000, 30(1), 91-100.
33. L.J. Struble, S. Diamond, Swelling properties of synthetic alkali silica gels, Journal of the American Ceramic Society, 1981, 64(11), 652–655.
34. A. Bonakdar, B. Mobasher, S.K. Dey and D.M. Roy, Correlation of reaction products and expansion potential in alkali-silica reaction for blended cement materials, ACI Materials Journal, 2010, 107(4), 380-386.
35. S. Chatterji, The role of Ca(OH)2 in the breakdown of Portland cement concrete due to alkali-silica reaction, Cement and Concrete Research, 1979, 9(2), 185-188.
36. F. Rajabipour, G. Sant and J. Weiss, Interactions between shrinkage reducing admixtures (SRA) and cement paste's pore solution, Cement and Concrete Research, 2008, 38(5), 606– 615.
37. F. Rajabipour, H. Maraghechi and G. Fischer, Investigating the alkali-silica reaction of recycled glass aggregates in concrete materials, ASCE Journal of Materials in Civil Engineering, 2010, 22(12), 1201-1208.
38. H. Maraghechi, S.M.H. Shafaatian, G. Fischer, and F. Rajabipour, The role of residual cracks on alkali silica reactivity of recycled glass aggregates, Cement and Concrete Composites, 2012, v.34, 41-47.
39. F. Rajabipour, H. Maraghechi, S.M.H. Shafaatian, Mechanisms of ASR and Its Mitigation by Fly Ash in Mortars Containing Recycled Glass Aggregates , 2012, 14th International Conference on Concrete Alkali Aggregate Reactions (ICAAR), Austin, Texas.
76
40. X. Hou, L.J. Struble, R.J. Kirkpatrick, Formation of ASR gel and the role of C-S-H and portlandite, Cement and Concrete Research, 2004, 34, 1683-1696.
41. E. J. Garboczi, Permeability, diffusivity, and microstructural parameters: a critical review, Cement and Concrete Research, 1990, 20(4), 591-601
42. B.J. Christensen, T. Coverdale, R. A. Olson, S. J. Ford, E. J. Garboczi, H. M. Jennings, T. O. Mason, Impedance Spectroscopy of Hydrating Cement-Based Materials: Measurement, Interpretation, and Application, Journal of the American Ceramic Society 1994, 77(1), 2789–2804.
43. A. Akhavan, F. Rajabipour, Evaluating ion diffusivity of cracked cement paste using electrical impedance spectroscopy, Submitted to Materials and Structures (Apr 2012).
44. F. Rajabipour, Insitu electrical sensing and material health monitoring in concrete structures, Ph.D. Dissertation, Purdue University, West Lafayette, Indiana, Dec. 2006, 193.
45. CRC Handbook of chemistry and physics, 90th Ed., CRC Press, Boca Raton, Florida, 2010.
46. L.J. Malvar, L.R. Lenke, Efficiency of fly ash in mitigating alkali-silica reaction based on chemical composition, ACI Materials Journal, 2006, 103(5), 319-326.
47. D.E. Clark, E.L. Yen-Bower, Corrosion of glass surfaces, Surface Science, 1980, 100, 53-70.
77
CHAPTER
FOUR
ASSESSING THE ROLE OF ION TRANSPORT IN MITIGATION OF ASR BY FLY ASH IN ASTM C1567
4.1 Introduction
From its first introduction by Stanton [1], ASR has been extensively studied from different standpoints. Specifically, different tests have been suggested to evaluate the ASR potential of reactive aggregates as well as examining the efficacy of preventive measures (e.g. portland cement replacements or additives). Although long term test imitates service life of concrete structures and yield more reliable results [2], accelerated tests are more preferred from a practical perspective. These tests are required to accommodate the rapid pace of concrete construction.
From different accelerated tests, ASTM C1567 accelerated mortar bar test (AMBT) is designated to assess the ASR risk in a 16-day period. In this test, ASR is promoted by boosting the temperature (80ºC) and submerging mortar bars in 1M NaOH solution. This test is an option in
78
AASHTO [3] and PCA [4] guide specifications and is currently being used by numerous agencies.
A common application of ASTM C1567 is to measure the proper content of supplementary cementitious materials (SCMs) (e.g. fly ash) to mitigate ASR expansion of reactive aggregates. It has been shown that in general there is a good agreement between SCMs safe replacement level
in this accelerated test and other field or long term tests [5]. However, this agreement is
suggested to be largely “fortuitous” [6] due to different mechanisms involved in these tests.
While different materials have been used to mitigate the deleterious alkali-silica reaction in
concrete construction, fly ash has always been a prime candidate. The application of fly ash against ASR has been well articulated [7] and was reviewed in detail in the previous chapter. It was shown that fly ash is effective in ASTM C1567 test environment, through a combination of different mechanisms such as alkali binding, increasing strength, reducing ion transport and aggregate dissolution. Specifically, reducing the ion transport was found to be a major contributing factor which capacitates different fly ashes to suppress ASR in this test environment. The effect of fly ash on the transport of different ions such as sodium and chloride
[8-12] and sulfate [13] ions has been extensively studied in the literature. The concerns were corrosion of steel reinforcement bars and sulfate attack, respectively. However, the fly ash effect on the transport of [OH-] ions as the ASR triggering factor, specifically in ASTM C1567 test
environment, has not been previously investigated.
The focus of the current chapter is to explore the effect of fly ash on the ion transport in mortars
exposed to ASTM C1567 environment. This sheds light on the link between the ion diffusion of
79 mortars and their corresponding ASR expansion. The variation in the composition of the pore fluid of the mortars was monitored. The ion diffusion coefficient of different mortars was compared. To understand the exact effect of fly ash on the microstrucutre of the binder phase of the mortars, pore size distribution were obtained with mercury intrusion porosimetry. The results may help to explain why no consistent safe replacement levels are achieved for some fly ashes
(or other pozzolans) in accelerated and long term tests.
4.2 Research Significance
It is well known that fly ash mitigates ASR. While previous research shows its efficacy both in accelerated and long term tests, the dominant mechanism may be completely different in these tests. The dosage of fly ash which is required to repress ASR obtained from these tests is generally in good agreement. However, there are still some exceptions as a result of the distinctive main mechanisms. Although using long term test is more recommended, due to limited time the accelerated tests are much preferred. Thus, there is a possibility of using insufficient fly ash dosage or a conservative fly ash content. These may ultimately affect the
ASR durability performance or cause early age strength problems, respectively. In addition, the role of developed ASR microcracks and their consequent accelerating effect on the ion penetration from the external bath have not been previously investigated in this test environment.
It is required to clarify the relation between the ability of fly ash in reducing ion transport and the ultimate effect on ASR expansion in ASTM C1567.
80
4.3 Materials and Methods
Unless it is mentioned specifically, the details of the experiments and materials used are similar to section 3.3. The same portland cement and fly ashes were used with oxide compositions presented in Table 3.1. The particle size distribution of four of the fly ashes, which were studied in more detail, was obtained by using laser diffraction method and is shown in Figure 4.1.
7
Cement 6 F1 F3 5 C2
4
3 Volume (%) Volume
2
1
0 0.01 0.1 1 10 100 1000 Particle Diameter(m)
Figure 4.1: Particle size distribution of portland cement and fly ashes used
The minimum required dosage of each fly ash which met the ASTM threshold was obtained by
ASR expansion tests according to 3.4.1. To investigate the role of ion transport, the ion diffusion coefficient of mortars with the proper content of fly ash (15%F1, 20%F3 and 35%C2) were
81
compared with the control mortar containing 100% portland cement. In addition, for one of the
fly ashes, a lower replacement level (20%C2) which was not able to mitigate the ASR expansion
was considered. For the details of the experiment please see 3.3.3. To have a better idea, the tests
performed are summarized in Table 4.1.
Table 4.1: Summary of the tests performed
Control F1 F2 F3 F4 C1 C2 Particle size distribution yes yes yes yes ASTM C1567 10,**15,20,30 15,20,25 15,20 15,20,25,30 20,25 20, 30, 35 Pore solution analysis yes (0,1,2,5,9,16*) 15% 35% (2,5,9,16) 20% 20% Electrical Resistivity(2,5,9,16) yes yes(15%) yes(20%) yes(20%,35%) Porosity (2,5,9,16) yes yes(15%) yes(35%) Pore size distribution (5) yes yes yes
*Days after casting **Minimum safe replacement level
4.3.1 More details on measuring the electrical conductivity of mortars
Electrical impedance spectroscopy is a non-invasive non-destructive material characterization
technique which requires no sample preparation with high accuracy. In 3.3.3 the procedure for
measuring the ion diffusion coefficient of the mortars were explained. In this section more details are provided on the procedure and analysis method. The method is based on electrically stimulating the mortar bars and then measuring their electrical resistivity (Rb,). Using equation
4-1 the measured electrical resistivity is converted to the electrical conductivity (S/m). The
result can then be plugged in to equation 3-1 for calculating the ion diffusivity coefficient.
Electrical conductivity and electrical resistance are related according to equation 4-1: 82
k (4-1) Rb
k is a correction factor which was obtained experimentally according to the following procedure.
First with plexiglass a container with exactly the same geometry of the mortar bars (25
×25×250mm) was prepared. Then it was filled with a NaCl solution and two studs with a
distance of 15mm (i.e. distance considered in the mortar bars) were fixed in the center of the
plexiglass container. Bulk electrical resistance of the same solution was determined with HP-
4194A impedance analyzer (Figure 3.2) in frequency sweep mode (40Hz to 10MHz) and using
250mV voltage. Electrical conductivity of the solution was also measured with the aid of
conductivitymeter. These values for an arbitrary NaCl solution prepared for this study were
found to be: =7.26S/m and Rb=4.5. Plugging these numbers in to the equation 4-1 the
geometry factor was obtained to be 32.67(1/m). This value is independent of the solution
properties and has been used to measure the electrical conductivity of the mortar bars by first
measuring their electrical resistivity.
The electrical stimulus can be DC or AC. While different drawbacks are associated with DC
resistance measurements due to the polarization, AC stimulus is more desirable. In this technique sinusoidal voltage (V) is applied with known voltage and frequency. Due to a phase shift generated in the current (I) response, electrical resistance would be a complex number
(Z=Z’+iZ”, electrical impedance). The real part is corresponded to electrical resistance while the
imaginary part reflects the energy dissipation due to generated magnetic and electric fields and is
frequency dependent. For different frequencies Z” values can be plotted against Z’ to achieve so
called “Nyquist plot”. For cement paste it is a typical two arc graph. The intersection of these
83 arcs is associated with the bulk resistance of paste. For more details the readers are referenced to
[14]. Figure 4.2 shows the Nyquist plots of the control and 15%F1 mortars after 7 days exposure to ASTM C1567 conditions. As it suggests after 7 days the bulk electrical resistivity of control mortar is much lower comparing to 15%F1 mortars. This will be further discussed in 4.4.3.
140
Control (100%PC) 120 15%F1
100
80 ) Z"( 60
40
20
0 0 400 800 1200 1600 Z'()
Figure 4.2: Niqyuist plot of the control and 15%F1 mortars after 7 days cured in ASTM C1567 test
84
4.3.2 Porosity of pastes and mortars
Porosities of mortars and pastes exposed to ASTM C1567 test conditions were measured prior to
being submerged (0 day) and after 3, 7, and 14 days submersion in 80ºC 1M NaOH bath. Three
5mm-thick mortar cross sections were saw-cut from ASTM C1567 prisms. The porosity of these
mortars was determined by measuring the buoyant, saturated surface dry, and oven dry masses
according to ASTM C127. To exclude the effect of ASR cracking of the mortars on the porosity,
similar procedure were followed on pastes, made with w/c of 0.47 and cured in the same
conditions. The porosity measurements were performed on the control, 15%F1 and 35%C2
mortars and pastes.
4.3.3 Pore Size Distribution (PSD) characterization
While porosity is reported to affect the ion transport in the cementitious systems, the distribution
and connectivity of pores in the overall porosity of pastes is reported to be much more important.
Therefore, the pore structure of two fly ash pastes (15%F1 and 35%C2) was compared with the control sample (100% cement) to investigate the effect of fly ash on the pore size distribution of pastes. Paste prisms were prepared with the same w/c ratio and cured with the same conditions of
ASTM C1567 test. After 3 days exposure to the alkaline bath, the pastes were taken out from the bath and 1mm thick samples were cut from the prisms. Then, they were vacuum dried for 48 hours. To remove any excess water the samples were then oven-dried at 105ºC for another 24
hours and preserved in bottle containing desiccants until the Mercury Intrusion Porosimetry
(MIP) test. The writers are aware that the oven-drying is not the ideal pretreatment method for
MIP sample preparation [15], but for the sake of comparison since all the samples were treated
85
similarly the adopted method is assumed to be acceptable. Specifically, Oven-drying is reported
to have a significant effect on the pore sizes in the range 100-200nm. In this study the comparison is based in the pore sizes in the range of 10-40nm [16].
The MIP test was performed with a Pascal 240 porosimeter. Pressure was increased in a continuous mode from sub-atmospheric pressure up to 200 MPa and correspondingly the pore size distribution was characterized in the range of ~7.4 to 10000nm, theoretically. The contact angle between mercury and paste was assumed to be 140º and surface tension was considered to be 480 (Dyne/cm). The number of output data was selected to be sufficient enough to achieve reproducible differential curve.
The 3 days curing period was chosen due to several facts: 1) after 3 days curing the difference in the ion diffusion coefficients of the mortars with and without fly ash was highest, so this led to question how different the pore sizes were, 2) This age was considered as a relatively lower stage of the paste microstrucutre development, 3) The fact that until 3 days exposure in the alkaline bath, expansion of the mortar prisms was not actually measurable by comparator, indicates that no major change in the microstructure of the mortars occurred due to ASR cracking. Thus, it is reasonable to assume that up to this age ASR cracks do not affect the pore structure development. Therefore the results of paste could be extended to the binder phase of the mortars
4) considering the pastes’ porosities no significant variation in the porosities of pastes occurred after 3 days.
86
4.4 Results and discussion
4.4.1 Composition of the binder to mitigate ASR
The results of 14-day expansion of mortar bars in ASTM C1567 test is presented in Figure 4.3.
This figure can be used to determine the minimum replacement level of each fly ash to control
the expansion of the mortars below the 0.1% threshold: 15%F1, 20%F2, 20%F3, 25%F4, 25%C1
and 35%C2.
The oxide compositions of the cement and six fly ashes are shown in the ternary phase diagram
of Figure 4.4 (oxide contents were normalized by subtracting the values for SO3 and LOI).
Considering Figure 4.3 and Figure 4.4, it is evident that fly ashes with higher contents of CaOeq are less effective in mitigating ASR. In general, this was linked to inefficiency of such ashes in reducing the alkalinity of pore fluid of mortars as suggested by [17,18]. However, as will be
shown later, in this test, since the alkali source is external, it may be due to inability of lower
contents of these fly ashes to reduce the ion diffusion coefficient as a result of their lower
pozzolanic reactivity.
Figure 4.4 shows the oxide contents of composite binders that were able to mitigate ASR (e.g.,
85%PC+15%F1, 80%PC+20%F2, etc.). Interestingly, all composite binder points are clustered
in one area of the phase diagram below a CaOeq value of 60%. It would be interesting to test
other combinations of different cements and fly ashes to determine if a maximum CaOeq threshold of ~60% is sufficient in controlling ASR for mortars containing recycled glass sand. In addition, it is important to determine the role of Al2O3eq: at similar CaOeq, would ashes with
87 higher contents of Al2O3 and Fe2O3 be more efficient in ASR mitigation? Of course, other parameters such as fly ash particle size and crystal content also affect its efficiency against ASR.
0.7 F1 F2 0.6 F3 F4 0.5 C1 C2 0.4
0.3 Expansion after 14 days (%) after 14 days Expansion 0.2
ASTM threshold 0.1
0.0 0 5 10 15 20 25 30 35
Cement replacement (wt.%)
Figure 4.3: Results of ASTM C1567 showing ASR expansions as a function of fly ash type and dosage
88
SiO 2eq =SiO +P O +TiO 2 2 5 2
hase diagram showing cement and fly ash compositions
F2
F4 F1 F3
C1 C2 15%F1 25%F4 20%F2 20%F3 30%C1 35%C2
CaO eq Al O =CaO+Na O+K O+MgO+MnO 2 3eq 2 2 =Al O +Fe O 2 3 2 3 Figure 4.4: Ternary phase diagram showing cement and fly ash compositions used in ASTM C1567 mortar prisms
89
4.4.2 Pore solution composition
The results of pore solution analysis are presented in Figure 4.5-4.8:
[OH-] concentration:
ASR initiates from attacking the silicate structure of glass or reactive aggregates so called corrosion. Thus, the concentration of [OH-] in the pore liquid of the mortar prisms soaked in 1M
NaOH solution has a major impact on the first stage of ASR.
Figure 4.5 compares the variation of [OH-] in the pore solution of mortars. For control, 15%F1 and 35%C2 mortars pore solution was extracted and analyzed in the first 48 hours and for the other ages 20%F3 and 20%C2 mortars were monitored as well.
Right after mixing the [OH-] content is nearly the same (~100mM) in the control, 15%F1 and
35%C2 mortars. During the next 24hr moist room curing, due to higher reactivity of portland cement, concentrations increase at a higher rate in the control mortar. At the end of this period, pore liquid of the control has 28 to 52% higher [OH-] comparing to 15%F1 and 35%C2 mortars.
This is more than a dilution effect as 15%(F1) and 35%(C2) of the portland cement is replaced and alkali binding has a contribution as well. However, the next 24hr 80ºC water curing, totally erases this combined dilution/binding effect as [OH-] sharply drops due to the leaching of ions to the external water bath. The rate of leaching is higher in the control mortar indicating that even in this early age, fly ashes reduce the ion diffusivity of the binder phase; in agreement with 0-day ion diffusion coefficients (see Figure 4.11).
90
1200 (a)
1 M NaOH solution 1000 After 24hr moist room curing
800 After 24hr water curing
600 Concentration (mM) - 400 OH Control(100% cement) 15%F1 200 35%C2 20%C2 After mixing 20%F3 0 0 2 4 6 8 10121416 Days after casting
Figure 4.5: Pore solution concentration during ASTM C1567 test: [OH-]
At the time of exposure to 1M NaOH bath, [OH-] concentration is in the range of 170mM
(15%F1) and 279mM (20%C2). The concentration increases steadily in the pore solution of all mortars to the end of the test. However, this increase occurs at a faster rate in the control and
20%C2 mortars due to a higher ion diffusion coefficient. Since the dissolution rate of glass aggregates strongly depends on the pH of the solution, the higher rate of [OH-] diffusion leads to significant ASR expansion in these mortars in the time limit of the test. At the conclusion of the test the [OH-] of the control mortar (1.15M) surpasses the OH- concentration of the soak solution
(1M). This has been explained in detail in 3.4.7. The mortar with 20%C2 is less successful in
91
reducing the ion diffusion from the external bath and at the test’s conclusion the concentration
(970mM) gains equilibrium with the external bath. All the mortars with proper fly ash content
end up to OH- concentrations between 60 to 70% of the control at the end of the test.
Comparison of the two different replacement levels of the same fly ash more emphasizes the
transport effect. While 20%C2 had a slight positive effect in reducing [OH-] diffusion and thus
final ASR expansion, it is not successful in passing ASTM C1567 expansion requirement. This
indicates that in ASTM C1567 test, a replacement level is successful in mitigating ASR that
reduces the rate of ion ingress into mortars below a certain level.
[Na] Concentration:
The presence of alkali is essential to produce the expansive alkali silica gel. Variation of
elemental [Na] in the pore solution of control, 15%F1, 20%F3, 35%C2 (controlled ASR) and
20%C2 (ASR not controlled) are presented in Figure 4.6. In contrast to their higher Na content
comparing to PC (0.4%F1 and 1.5%C2 Vs. 0.3%PC), the immediate Na release to the mixing
water is lower in 15%F1 and 35%C2 mortars comparing to the control mortar due to less
reactivity of fly ash. A similar trend continues to the end of 24 hour moist room curing.
However, during the next 24 hour water curing period, the [Na] content of the pore solution of
35%C2 mortar accumulates at a higher rate and surpasses both the control and 15%F1. This is
likely due to the initial higher Na2O content of C2 (see Table 3.1). Note that this higher early age
reactivity of fly ash is not surprising as the curing temperature is increased to 80ºC. Unlike [OH-]
Na concentration of the pore solution increases in all three mortars during this period. It may be
concluded that the rate of Na release to the pore solution is higher than the rate of Na leaching to
the curing water bath. This is in agreement with 0-day ion diffusion coefficients (Figure 4.11) as less Na can diffuse out from 35%C2 mortar. While the initial Na content of the 35%C2 mortar is
92
higher, even if it was the same in all mortars, the buildup would be higher due to its lower ion
diffusion coefficient.
1400
1200
1 M NaOH solution 1000
800
600 water curing After 24hr
Na Concentration (mM) Na Concentration Control(100% cement) After curing room moist 24hr 400 15%F1 35%C2 20%C2 200 20%F3
After mixing 0 0 2 4 6 8 10 12 14 16 Days after casting
Figure 4.6: Pore solution concentration during ASTM C1567 test: Sodium
Prior to submergence in 1M NaOH bath, [Na] is in the range of 156mM for 20%F3 to 333mM for 35%C2 mortar. Interestingly, 35%C2 mortar having an initial Na content higher than the control mitigated the ASR expansion at the conclusion of the test. After submerging the prisms into alkaline bath, [Na] increased steadily in all mortars alike [OH-]. However, due to higher
diffusivity of control and 20%C2 mortars, it occurs at a higher rate in the control and 20%C2
93
mortars comparing to mortars with proper fly ash content. At the conclusion of the test, the [Na]
of fly ash mortars passing ASTM C1567 threshold is nearly 75% of the control. The rate of
increase in the Na content of pore solution is higher in the 1st week of the test comparing to the
2nd. The increase in the 1st period is associated mostly to the diffusion of Na from the external
bath while in the 2nd period the lower rate increase may be attributed to total dissolution of glass
network and the release of alkalis from the glass particles. Overall at different ages in all mortars
the concentration of [Na] is higher comparing to [OH-]. The ICP measures the total [Na] which is
not the same as ionic [Na+] as will be explained in 5.3.3 .
[Si] concentration: It has been shown [19] that in general the dissolution rate of silicate-based glasses accelerates
exponentially for a pH above 11.5. As the [OH-] concentration increases in the pore solution of
the mortars, they further attack the Si-O-Si bonds in the glass structure and thus dissolve the total
silicate network of glass particles. Therefore, [Si] increases with a corresponding higher rate in
higher pH levels of the pore solution (Figure 4.7). Keep in mind that this is the first stage of
ASR.
The sensitivity of the corrosion rate of glass particles of mortar prisms which were exposed to
the high alkaline solution is well pronounced in the results of [Si] concentrations in the pore
liquid. The concentration of [OH-] in the controlled mortar increases from the time of exposure
to the end of the test from 230 to 1130mM. This leads to an increase in [Si] by a factor of 266
with a final concentration of 42mM. However, in the mortars with proper fly ash content with a maximum [OH-] of 800mM (15%F1) at the conclusion of the test, [Si] increases not more than
49 times leading to a maximum [Si] of 13mM. Comparing Figure 4.5 and Figure 4.7, it is evident
94 that [OH-] concentration above 600mM is a triggering point of acceleration in this environment.
If the pore solution in any of these mortars hits this threshold the dissolution rate of glass accelerates. The main difference is the retarding effect of safe fly ash contents which delayed the diffusion of hydroxyl ions to hit this threshold.
50 Control(100% cement) 15%F1 35%C2 40 20%C2 20%F3
30
20 Si Concentration (mM) Si Concentration
10 After curing room moist 24hr water curing After 24hr
0 0246810121416 Days after casting
Figure 4.7: Pore solution concentration during ASTM C1567 test: Silicon
[Al] concentration:
As Figure 4.8 shows, the higher Al content of fly ashes leads to a higher concentration of Al in the pore solution of fly ash mortars. After 5 days no significant change occurs in the [Al] of the control mortar as the initial Al content of portland cement is much lower than all fly ashes.
95
2.4 Control(100% cement) 15%F1 2 35%C2 20%C2 20%F3 1.6
1.2
Al Concentration (mM) Al Concentration 0.8 After 24hr water curing After 24hr After curing room moist 24hr 0.4
0 0246810121416 Days
Figure 4.8: Pore solution concentration during ASTM C1567 test: Aluminum
In all fly ash mortars, [Al] increases continuously to the end of test leading to a considerably higher concentration (up to 4 times) at the conclusion of the test. The presence of aluminum has been reported to reduce the dissolution rate of silicate glasses (Chapter 6). The mechanism was speculated to be the formation of negatively charged sites on the surface which could repel the negatively charged hydroxyl ions and thus inhibit the glass corrosion [20]. In addition, the Al
incorporated in C-S-H formed in fly ash mortars has a higher capacity for binding alkalis [21,22].
An important question is the extent of effectiveness of these low Al contents of pore solution
(maximum 2mM at the end of the test) in reducing the dissolution rate of silicate glasses. In
96
chapter 6 the effect of aluminum on reducing the rate of glass dissolution and thus ASR
suppression is discussed in detail.
4.4.3 Ion transport
As previously depicted in Figure 4.5 and Figure 4.6, the ingress of [Na] and [OH-] into the
mortars in ASTM C1567 test occurred at different rates. This incentivized quantification of the
ion diffusivity of the mortar prisms. In order to use the Nernst-Einstein equation (3-1), first, the electrical conductivity of the pore solution of different mortars was measured at various ages and
is presented in Figure 4.9. Prior to exposure to the alkaline bath, the pore solution of all mortars
has approximately similar electrical conductivities (~5S/m). At later ages due to penetration of
[OH-] and [Na+] ions from the external bath and the release of different ions due to hydration, the electrical conductivity of the pore solutions of the control and 20%C2 increase at a higher rate.
This is expected as [OH-] is the main contributing factor to the pore solution conductivity of
cementitious systems [23]. At the end of the test, the electrical conductivity of fly ash mortars
that passed ASTM C1567 requirement is 85% to 90% of the control (11.6S/m). For 20%C2
mortar with a higher diffusivity the conductivity of pore solution is very close (11.1S/m) to the
control.
97
14
12 Time of submergence in 1M NaOH bath
10
8
6 Control 15%F1 4 35%C2 20%C2 Pore Solution Electrical conductivity(S/m) Pore Solution 2 20%F3
0 0 2 4 6 8 10121416 Days after casting
Figure 4.9: Comparison of the electrical conductivity of the pore solution of control and fly ash mortars after submergence in 1M NaOH bath
While minor differences are observed between the electrical conductivity of mortars’ pore liquid,
the electrical conductivity of the mortars are substantially different. Figure 4.10 shows the
variation of the electrical conductivity of mortars over ASTM C1567 test period. Prior to exposure to the alkaline bath, the control mortar is 4 to 6 times more electrical conductive than the mortars with ASR controlling amounts of fly ash. This may not be anticipated as fly ash is expected to react at later ages due to lower reactivity. However, the initial 80ºC water curing presumably has a significant effect on early age pozzolanic reaction of fly ash.
98
0.16 Control 15%F1 35%C2 0.14 20%C2 20%F3
0.12
0.1
0.08
0.06 Time of submergence in 1 M NaOH bath
0.04 Electrical conductivity of mortar bars(S/m) mortar of Electrical conductivity 0.02
0 0 2 4 6 8 10 12 14 16 Days after casting
Figure 4.10: Comparison of the electrical conductivity of the control and fly ash mortars after submergence in 1M NaOH bath
Up to 3 days the electrical conductivity of the control mortar increases at a higher rate comparing to fly ash mortars which leads to a significant difference in the electrical conductivities. This occured through different mechanisms: 1) In the control mortar, due to the faster ion diffusion from the external bath and accumulation of ions, the pore liquid of control mortar would be more electrically conductive (Figure 4.9); although it was discussed to have only a minor impact. 2)
Due to the simultaneous ASR cracking, the open porosity of the control mortar increased at a higher rate in the control mortar. Subsequently a conductive liquid penetrates into the newly
99
developed cracks substituting the relatively non-conductive solid phase. 3) In fly ash mortars,
due to the pozzolanic reaction CH is converted to C-S-H and thus the microstrucutre more
densifies. Pores are refined (Figure 4.13) and pore connections are disconnected. Overall, this
continues up to 7 days and yields 6 to 9 time’s greater electrical conductivity of control mortar
comparing to the mortars with proper fly ash content to control ASR.
The counter-intuitive result obtained for the electrical conductivity of the control mortars was
previously discussed in 3.4.3. The cracks in the control mortars are filled with ASR gel (with
relatively much lower electrical conductivity than the pore solution) and thus, the overall
conductivity of the mortar starts to decrease after 7 days. This is not the case for 20%C2 mortar
as it still in the lower stages of ASR, not enough gel is produced to have such impact.
The electrical conductivity of the ASR controlling fly ash mortars remains relatively constant
during the test period. The 20%C2 mortar has approximately similar behavior comparing to the control. It is initially 4 times more electrical conductive comparing to 35%C2 (e.g. the ASR controller amount of the same fly ash). As a result it was not able to sufficiently alleviate the penetration of [OH-] ions and thus could not suppress ASR during the test period. It should be
noted that once the reaction initiates due to the following ASR cracking, the diffusion of the
attacking ions into the mortars significantly increases; in other words, there is a momentum in
this reaction that once it is gained the reaction could not be mitigated in the time scale of the test.
With the assumption of constant self-ion diffusivity (D0) for the pore liquid of all mortars at
different ages and using Nernst-Einstein equation, the ion diffusion coefficient of mortars was
calculated. Figure 4.11 compares the variation of the ion diffusion coefficient at different ages for all mortars.
100
2.E-07 Control 15%F1 35%C2 20%C2 20%F3 1.E-07
1.E-07 /s) 2 1.E-07
8.E-08
6.E-08 Diffusion coefficient(cm 4.E-08 Time of submergence in 1 M NaOH bath
2.E-08
0.E+00 0 2 4 6 8 10 12 14 16 Days after casting
Figure 4.11: Comparison of the ion diffusivity of the control and fly ash mortars during ASTM C1567 test
The diffusion coefficient of different mortars follows approximately similar trends observed for
electrical conductivity of mortars as relatively minor variations in electrical conductivity of pore solutions occurred during the test period. The ion diffusivity of the control mortar was greater compared to mortars with safe fly ash content by a factor of 3.8 to 7.5 at different ages. This means that alkali and hydroxyl ions can penetrate much faster into the control mortar as observed in Figure 4.5 and Figure 4.6. The reduction in ion diffusivity has a significant impact on the rate of alkali-silica reaction in ASTM C1567 mortars as previously discussed in chapter 3. It is
101
worthwhile to mention that transport of ions occurs though cracks and pores of the binder.
Cracking leads to a higher porosity which is evidenced by mortar porosities (Figure 4.12).
However, the generated ASR cracks may not have a significant effect on ion diffusion in
comparison to the pore refinement effect. Contrary to the continuous increase in the porosity of
the control mortar during the test, the ion diffusion coefficient started to decrease after 7 days. As
previously discussed this reduction may be linked to the development of the ASR gels which led
to disconnection of the pores. Thus, it is speculated that decreasing pore size and pore
connectivity in the binder phase are the main factors in reducing the ion transport comparing to
the cracking effect.
These findings suggest that the transport properties of mortars have a profound impact on how
they perform in ASTM C1567 test. With the same reactive aggregates, the binders with lower
ion transport coefficients can perform significantly better in this test, even if they do not actively
reduce pore solution alkalinity. Such binders could be produced by reducing the w/cm, use of
accelerators (e.g., CaCl2), micro-fillers, or permeability reducing (i.e., water proofing)
admixtures.
4.4.4 Porosity:
The results of mortar and paste porosities after 0,3,7 and 14 days of submergence in 1M NaOH
solutions are provided in Figure 4.12. The concern here was that by increasing the temperature to
80ºC and exposure to high alkaline solution the precedence or recency of the cement hydration
and fly ash pozzolanic reaction would change in such environment. Due to rapid hydration of
cement paste and water consumption, the porosity decreased faster in the control paste comparing to the pastes with partial fly ash replacement. During the test at all ages, the porosities
102 of both fly ash pastes (15%F1 and 35%C2) were higher than the control. The increase in the porosity due to application of fly ash is well documented in literature [24] and ASTM C1567 curing conditions lead to similar trend in increasing the porosity.
50
45
40
Control Mortar 35 15%F1 Mortar 35%C2 Mortar 30 Control Paste 15%F1 Paste Porosity(%) 25 35%C2 Paste
20
15
10 0 2 4 6 8 10 12 14 16 Days after casting
Figure 4.12: Porosity of the control and fly ash mortars and pastes after submergence in 1M NaOH bath
Incorporation of fly ash with an average particle size even smaller than cement is reported to yield a higher porosity [25]. The porosity increase has been explained to be a result of higher water to solid ratio (volume) in fly ash pastes having the same water to solid ratio (weight) comparing to 100%PC pastes. For a detailed explanation please see [8]. The higher porosities in
103
fly ash pastes may suggest higher mass transport coefficients (i.e., permeability, diffusivity) for
fly ash mortars. However, as presented in previous section, the transport coefficients are actually
lower for fly ash mortars. Note that at the conclusion of the test the porosity of fly ash pastes is at most 11% higher than the control.
The same trend was observed for the porosity of mortars comparing to pastes with fly ash at the
early stage of the test (prior to exposure to 1M NaOH bath). At 0 day, F1 and C2 mortars had 7%
and 17% higher porosities comparing to the control. In the control mortar, the porosity of NaOH exposed specimens increased with exposure time due to progressive cracking and damage by
ASR. For fly ash mortars exposed to NaOH solution, porosity increased gradually, which is consistent with gradual ASR expansion and microcracking in these mortars during ASTM C1567 test. At the conclusion of the test, F1 and C2 mortars had 14% and 10% lower porosities comparing with the control, respectively.
The increase in the porosity is indicative of cracking with a consequent direct exposure of internal reactive aggregates to high alkaline bath. In this regard, fly ash not only reduces the rate of ion ingress and retards the reaction, but also it postpones the ASR cracking and direct exposure of internal aggregates to the external bath. However, it might be falsely concluded that higher porosity essentially means a higher rate of ion transport. A poor correlation has been reported between the porosity of concrete and diffusion of chloride ions [26]. The same trend is observed here as [OH-] and [Na+] ions penetrate into fly ash mortars with a lower rate. The pore
size distribution is reported to have a significant impact on the ion diffusion as has been observed
for [Cl-] before [26].
104
4.4.5 Pore size distribution:
The porosities of the binder phase of fly ash mortars (e.g. pastes) are found to be higher comparing to the control. However, the pore size distribution of cementitious pastes cured in
ASTM C1567 test environment indicates that the porosity increase occurred along with a decrease in the average size of the pores in fly ash pastes. Figure 4.13 shows that by application of fly ash, a pore refinement occurs soon after 3 days curing in ASTM C1567 test environment.
This is in contrast to previous research [27] which reports an increase in mean pore diameter even after 90 days curing in room temperature with application of fly ash. The cumulative pore
size distribution graphs (not presented here) show similar trends for the porosity results obtained
via a different method, indicating a higher total porosity for fly ash pastes.
Overall the pore size distribution graph has shifted to the smaller size pores in both fly ash pastes
which were able to suppress ASR. Application of 15%F1 has led to a decrease in the average
pore size distribution from 32nm in the control to 14nm (56%). For 35%C2 pores have partially
decreased from a size of 32nm (control) to 9nm (72%). Interestingly, the two peak curve of the
original particle size distribution (Figure 4.1) of C2 fly ash is reflected in the pore size
distribution of 35%C2 paste. Overall the particle size distribution of fly ashes in micro-scale was
mirrored in their corresponding pore size distribution of the pastes in nano-scale. The pore
refinement achieved is a result of a combined effect of pozzolanic reaction and alkali activation
[28,29] of fly ashes. As a result of this reaction larger pores are disconnected (reduced pore
connectivity) or converted in to many smaller size pores. It has been suggested that the ion
transport through smaller pores are much more affected by surface charge of the matrix’ wall [9].
This may also contribute to the lower diffusivity of fly ash cementitious system which was
previously observed in Figure 4.11.
105
45
40
Control 35 15%F1 35%C2 30 /g) 3 25
20 Volume(mm
15
10
5
0 1 10 100 1000 10000 100000 Pore radius(nm)
Figure 4.13: Pore refinement in control and fly ash pastes after 3 days exposure to NaOH solution at 80°C
4.5 Conclusions
Based on the observations of this study the following conclusions can be made:
Application of fly ash does not significantly affect the electrical conductivity of the pore
solution of mortar prisms in ASTM C1567. However, they significantly reduce the
electrical conductivity of the mortar bars.
106
The pastes prepared with minimum required dosage to control ASR expansion of mortar
bars; cured in the similar conditions; show a higher porosity comparing to 100%PC
pastes at all ages during the test period. An initially higher porosity was observed for
mortars with proper content of fly ash. No major change occurred in their porosity during
the test period. In contrast, the porosity of 100%PC mortar bars steadily increased due to
ASR cracking. After 3 days exposure to the alkaline bath, it surpassed the porosity of fly
ash mortars.
A content of fly ash was able to mitigate ASR expansion that reduced the ion diffusion
coefficient below a certain value. In the present study, fly ash contents that were able to
reduce mortar diffusivity by a factor of 3 to 7 had the ability to control ASR in this test
environment.
Partial replacement of portland cement with fly ash leads to a significant reduction in the
average pore size of the binder phase of the mortars. The pore size distribution curve has
an overall shift to the smaller sizes. The lower ion diffusion coefficient of fly ash mortars
can be linked to this lower average pore size.
An interesting observation was made suggesting that cement-fly ash binders containing
less than approximately 60% CaOeq=CaO+Na2O+K2O+MgO+MnO may be able to
mitigate ASR expansions below the standard threshold of 0.1%. This preliminary result
indicates that it is worth trying combinations of other cement and fly ashes to see if
similar trend would be observed.
107
4.6 References:
1. T.E. Stanton, Expansion of concrete through reaction between cement and aggregate, Proceedings, American Society of Civil Engineers, 1940, 1781-1811.
2. L.J. Malvar, G.D. Cline, D.F. Burke, R. Rollings, T.W. Sherman, and J.L. Greene, Alkali- silica reaction mitigation: State of the art and recommendations, ACI Materials Journal, 2002, 99(5), 480-489.
3. AASHTO, Guide Specification for Highway Construction, Section 56X, American Association of State Transportation Officials, http://leadstates,transportation.org/asr/library/gspec.stm
4. PCA, Guide Specification for Concrete Subject to Alkali-Silica Reactions, IS 415, Portland Cement Association, Skokie, IlI, 1998, 7.
5. M.D.A. Thomas, F. Innis, Use of the accelerated mortar bar test for evaluating the efficacy of mineral admixtures for controlling expansion due to alkali-silica reaction, Cement Concrete and Aggregates, 1999, 21(2), 157-164.
6. M.D.A. Thomas, B. Fournier, K.J. Folliard, M.H. Shehata, J.H. Ideker, C. Rogers, Performance limits for evaluating supplementary cementing materials using accelerated mortar bar test, ACI Matererials Journal, 104 (2007), 115–122.
7. S.M.H. Shafaatian, A. Akhavan, H. Maraghechi, and F. Rajabipour,How Does Fly Ash Mitigate Alkali-Silica Reaction (ASR) in Accelerated Mortar Bar Test (ASTM C1567)?, Submitted to Cement and Concrete Composites, May(2012).
8. S. Li and D. M. Roy, Investigation of relations between porosity, pore structure, and Cl- diffusion of fly ash and blended cement pastes, Cement and Concrete Research, 1986,16, 749-759.
9. S. W. Yu and C. L. Page, Diffusion in cementitious materials: 1. Comparative study of chloride and oxygen diffusion in hardened cement pastes, Cement and concrete Research, 1991, 28, 581-588.
10. S. Goto and D. M. Roy, Diffusion of ions through hardened cement pastes, Cement and Concrete Research, 1981, 11, 751-757.
11. Y. Elakneswaran, T. Nawa, K. Kurumisawa,Influence of surface charge on ingress of chloride ion in hardened pastes, Materials and Structures, 2009, 42, 83-93.
12. K.O. Ampadu, K. Torii, M. Kawamura, Beneficial effect of fly ash on chloride diffusivity of hardened cement paste, Cement and Concrete Research, 1999,29, 585–590.
108
13. K.Torii, M. Kawamura, Effects of fly ash and silica fume on the resistance of mortar to sulfuric acid and sulfate attack, Cement and Concrete Research, 1994, 24(2), 361–370.
14. F. Rajabipour, Insitu electrical sensing and material health monitoring in concrete in concrete structures, Ph.D. Dissertation, Purdue University, West Lafayette, Indiana, Dec. 2006, 193.
15. J. Zhang, G.W. Scherer, Comparison of methods for arresting hydration of cement, Cement and Concrete Research, Cement and Concrete Research, 2011, 41, 1024–1036.
16. K. K. Aligizaki, Pores structure of cement-based material, testing, interpretation and requirements, Modern concrete technology, Taylor& Francis, 2006.
17. M.H. Shehata, M.D.A. Thomas, The effect of fly ash composition on the expansion of concrete due to alkali–silica reaction, Cement and Concrete Research, 2000, 30(7), 1063-1072.
18. L.J. Malvar, L.R. Lenke, Efficiency of fly ash in mitigating alkali-silica reaction based on chemical composition, ACI Materials Journal, 2006, 103(5), 319-326.
19. R.K. Iler, Chemistry of silica - Solubility, polymerization, colloid and surface properties and biochemistry’, 1979, John Wiley & Sons, New York.
20. Y. Oka and M. Tomozawa, Effect of alkaline-earth ion as an inhibitor to alkaline attack on silica glass, Journal of Non-Crystalline Solids, 1980, 42(1-3), 535-543.
21. B. Lothenbach, K.Scrivener, R.D. Hooton, Supplementary cementitious materials, Cement and Concrete Research, 2011, 41, 217–229.
22. S.Y. Hong, F.P.Glasser, Alkali sorption by C-S-H and C-A-S-H gels: Part II. Role of alumina, Cement and Concrete Research, 2002, 32(7), 1101-1111.
23. K.A. Snyder, X. Feng, B.D. Keen and T.O. Mason, Estimating the electrical conductivity of cement paste pore solutions from OH−, K+ and Na+ concentrations, Cement and Concrete Research,2003, 33(6), 793–798.
24. D C Hughes, Sulphate resistance of opc, opc/fly ash and srpc pastes: pore structure and permeability, cement and concrete research, 1985,15,1003-1012.
25. P. Chindaprasirt, C. Jaturapitakkul, T. Sinsiri ,Effect of fly ash fineness on compressive strength and pore size of blended cement paste, Cement and Concrete Composites, 2005,27(4),425–428.
109
26. H. Y. Moon, H. S. Kim and D. S. Choi, Relationship between pore diameter and chloride diffusivity in various concretes, Construction and Building Materials, 2006, 20, 725-732.
27. S.P. Pandey, R.L. Sharma, The influence of mineral additives on the strength and porosity, Cement and Concrete Research, 2000, 30, 19–23.
28. Y. Cai, J. T. Ding and Y. Bin, Activation of fly ash by the high temperature and high alkalinity in ASR tests, Science China, Technological series, 2011, 54(1), 28-32.
29. Della M. Roy, Weimin Jiang, M.R. Silsbee, Chloride diffusion in ordinary, blended, and alkali-activated cement pastes and its relation to other properties, Cement and Concrete Research, 2000, 30, 1879-1884.
110
CHAPTER
FIVE
INVESTIGATING ASR MITIGATING MECHANISMS OF GLASS POWDER IN ASTM C1567 ACCELERATED TEST
5.1 Introduction
When glass is ground to <100m, it exhibits a pozzolanic behavior [1] and can be used as partial cement replacement. Pattengil and Shutt (1973) [2] were the first who suggested the possible use of soda-lime glass as a pozzolanic material considering its high amorphous silicate content. The idea, however, was not pursued until recently when it was brought up by Shao et al.[1] proposing to produce a more environmentally friendly concrete by partially replacing cement with glass powder (GP). The use of glass as cement replacement is more desirable than its use as fine aggregate (chapter 4); not only because of a larger profit margin of replacing cement over sand,
111
but also because of producing a “greener” product by reducing the embodied energy and carbon
footprint associated with portland cement [3]. There are also other potential benefits associated
with the use of glass powder: improving cement hydration [4], workability [5] and long term
strength and durability of concrete [6]. Glass powder may be readily available as a byproduct of
spherization process of flat glass to produce glass spheres. It may also be produced during the
recycling process of waste glass when it is ground to coarse particles known as cullet.
According to ASTM C618 a material is considered to be pozzolanic if it meets certain chemical
and physical requirements. Glass powder with high silicate content meets all the chemical
requirements, although some physical criteria, specifically early age strength may not be
satisfied. The pozzolanic behavior of glass powder has been studied by measuring the
compressive strength of concretes with partial replacement of portland cement [7]. It has been
shown that reducing the particle size of GP is effective in increasing its pozzolanic reactivity [6-
9]. Glass powder has also been reported to repress ASR expansion of reactive aggregates by its
pozzolanic reaction [1,4,7,10,12] . However, research on this subject has been limited and
resulted in contradictory conclusions. For example, Dyer and Dhir [13] believed that glass
powder is not a suitable option for controlling ASR based on long term expansion tests. In contrast, glass powder showed a very good ASR performance in long term tests of Idir et al. [12]
and Taha and Nounu [8]. Also soda-lime glass powder contains very high alkali content
(Na2Oeq>10%) and it is unclear if these alkalis will leach into pore solution and exacerbate ASR.
Understanding of the mechanisms through which GP may mitigate ASR of reactive aggregate is
important. Finally, the correlation between the fineness and the required dosage of glass powder
to control ASR needs further attention. Finer glass can react pozzolanically at a faster rate [1];
but it may also release alkalis more rapidly. In addition, the required fineness of glass powder
112 particle size can be important from a cost and energy standpoint. Fine GP may only be available in limited locations and further processing and energy would be required to grind glass to a finer particle size.
The current study provides preliminary investigations to examine why glass powder may benefit
ASR performance; or in other words, what mechanisms lead to mitigation of ASR when glass powder is used as partial cement replacement. A set of complementary experiments is used to investigate the potential mechanisms that glass powder may be beneficial against ASR.
Excluding the fact that glass powder has relatively high alkali content comparing to conventional pozzolans, it may be considered as a pozzolanic material. In section 3.2, the proposed mechanisms which enable pozzolans to mitigate ASR expansion were presented in detailed description: 1) Alkali dilution 2) Alkali binding 3) reducing transport 4) modifying ASR gel 5) reducing aggregate dissolution 6) portlandite consumption. As previously explained in section
3.2, the latter will be not a focus of the current investigation. The study is based on ASTM
C1567, accelerated mortar bar test. However, further investigations using concrete prism test
(ASTM C1293) or even outdoor exposure of concrete blocks [14] would be beneficial in improving the understanding of the pozzolanic and ASR mitigation performance of glass powder.
5.2 Materials and Methods
Four different sizes of soda-lime glass powder were studied in this work, all having an average particle size smaller than 105m. This range is considered as glass powder was shown to have minor pozzolanic effect for particle sizes larger than 100m [1]. The four types are named GP1 113 to GP4. Their particle size distributions were measured by laser diffraction and shown in Figure
5.1. The figure also shows the average particle size distributions of the glass powders. GP1 and
GP2 are byproducts of glass bead manufacturing from post-industrial and post-consumer window plate glass. GP3 is obtained by crushing recycled glass bottles with a ball-mill and passing through sieve #100 (150m). The first three glass powders have a median particle size greater than portland cements (19m). GP4 is obtained by wet sieving of GP3 through #400
(38m) sieve, which delivered a particle size distribution similar to that of the type I portland cement. The oxide compositions of portland cement and glass powder are presented in Table 5.1.
100
90 Cement Avg 18 um GP1 Avg 102um 80 GP2 Avg 78um
70 GP3 Avg 39um GP4 Avg 20um 60
50
40
Cummulative volume(%) Cummulative 30
20
10
0 0.01 0.1 1 10 100 1000 Particle Size(m)
Figure 5.1: Particle size distribution of different classes of glass powder used 114
Table 5.1: Oxide composition (%) of Portland cement and glass powder
Oxide CaO SiO2 Al2O3 Fe2O3 Na2O K2O MgO TiO2 SO3 LOI C/S Na2Oeq Cement 62.5 19.9 5.44 2.26 0.3 0.89 2.31 0.29 4.93 0.86 3.14 0.89 Gp1/GP4 13.5 72.7 N/D N/D 13.8 N/D N/D N/D N/D --- 0.19 13.8 GP2/GP3 11.0 74.5 N/D N/D 14.1 N/D N/D N/D N/D --- 0.15 14.1 N/D is not detected
5.2.1 Accelerated mortar bar test (ASTM C1567)
This test was performed to determine the required dosage of each type of glass powder that was
able to control the 14-day ASR expansion below the threshold value of 0.1%. Mortars were
prepared and cured according to the test requirements. Details are provided in section 3.3.1. A
total of 22 mortar mixtures were tested containing different types (sizes) and dosages of glass
powder. All mortars contained recycled glass sand as reactive aggregate with a particle size
distribution specified by the standard ASTM C1567. The expansion of each prism was monitored
up to 14 days after submersion in 1M NaOH bath at 80°C. The subsequent tests (described
below) were performed on the control mortars (100% PC) as well as the mortar with 40% GP2
replacement of cement which had a 14-day ASR expansion of 0.05%. These tests were based on
the methodology used in the previous two chapters. Ion diffusivity of mortars was obtained with
the same procedure described in section 3.3.3 to assess the effect of GP on the ion transport of the mortars. In addition, tensile and compressive strengths were measured according to section
3.3.4 to compare the strength of mortars with and without GP. SEM/EDS with sample preparation procedure explained in section 3.3.5 was performed to examine the extent of ASR gel production and its composition. 115
5.2.2 Pore solution extraction and analysis
Pore solution of 100%PC and 40%GP2 mortars were extracted and analyzed according to the
procedure described in section 3.3.2. In addition, to monitor alkali release from glass powder in
the binder phase of concrete at ambient conditions, paste specimens were also prepared with
100% cement and 40%GP2 replacement of cement with the same w/cm ratio of 0.47. Performing
independent tests on paste specimens helps with measuring the alkali release from portland
cement and glass powder and exclude the contribution from aggregates. The paste cylinder
specimens with a diameter of 30mm and 60mm height were sealed in plastic wraps and cured in
the moist room with 100% RH at 23°C until the time of pore solution extraction. The mass of the specimens were measured after demolding and at the time of pore solution extraction. The ion concentration of pore solutions was monitored after 0, 15, 30, 65, 125 and 220 days. The pore
solution was extracted with similar procedure followed for mortars with a maximum pressure of
80MPa.
5.2.3 Pore size distribution characterization
To examine the effect of GP on pore size distribution and total porosity of the binder phase, paste
specimens with the same geometry of ASTM C1567 prisms were prepared with 100% PC and
40%GP2 formulations. The paste bars were cured and prepared for mercury intrusion
porosimetry according to section 4.4.5.
5.2.4 Pozzolanic reaction of glass powder
116
Thermo-gravimetric analysis (TGA) was performed to monitor the pozzolanic reaction of glass
powder in ASTM C1567 test environment by quantifying the portlandite (CH) content as a
function of time. Paste bars were prepared from the control and 40% GP2 with a w/cm=0.47.
The bars were exposed to ASTM C1567 test conditions. At 0, 7 and 14 days of exposure to
NaOH bath, a section was cut from the center of each bar, oven dried at 105°C for 24 hours and
pulverized to powder. Approximately 50mg of the powder was inserted inside an SDT-2960
TGA instrument with a programmed temperature rise at a constant rate of 10°C/min up to 800°C.
The portlandite (CH) content was calculated based on a sharp mass loss in the temperature range
435-500°C [15].
5.3 Results and discussion:
5.3.1 Sufficient dosage of glass powders to mitigate ASR
Figure 5.2 shows the ASR expansion of mortar bars with different GP2 replacement levels. It is
observed that in general, the ASR expansion of mortar bars reduce by increasing the glass powder content. A minimum replacement level of 40% is sufficient to lower the ASR expansion
below the 0.1% threshold. The 14-day expansion of all mortar mixtures in ASTM C1567 test is
shown in Figure 5.3 as a function of glass powder content. The error bars represent the standard deviation based on the measurements on for duplicate samples. The minimum replacement level of each size of glass powder which was able to mitigate expansions below 0.1% ASTM threshold is found to be 45%GP1, 40%GP2, 35%GP3, and 30%GP4. It is evident that as the average particle size of glass powder decreases, a lower replacement level is required to control ASR expansion. Although finer glass powder is probably releasing alkalis at a faster rate, this did not
117 result in enhancing ASR. Rather, expansions reduce by using finer glass powder due to faster pozzolanic reactions (Figure 5.11). An increase in the ASR expansion of mortar bars at 10% cement replacement is observed. A similar behavior is usually observed for some other pozzolans and is called “pessiumum” pozzolan content which promotes rather than mitigates
ASR. The reason for such a behavior could not be explained before [16].
0.80%
Control 0.70% 10%GP2 20%GP2 30%GP2 0.60% 40%GP2 50%GP2 0.50%
0.40% Expansion(%) 0.30%
0.20%
0.10%
0.00% 02468101214 Day
Figure 5.2: ASR expansion of mortars with different GP2 replacements over 14-day period of ASTM C1567
118
In the GP size range considered (20-102m), there is an approximately linear correlation
(R2=98%) between the average particle size of glass powder and the required dosage to control
ASR (Figure 5.4). From an economic standpoint, this may help material engineers to
compromise between the required energy for grinding glass and its effectiveness against ASR.
0.8 GP 1: avg 102 um 0.7 GP 2: avg 78 um GP 3: avg 39 um GP 4: avg 20 um 0.6
0.5
0.4
0.3 Expansion after 14 days(%) Expansion
0.2
ASTM threshold 0.1
0.0 0 5 10 15 20 25 30 35 40 45 50 Weight % cement replacement with glass powder
Figure 5.3: Results of ASTM C1567 showing ASR expansions as a function of glass powder size and dosage
119
50
45
R² = 0.9837
40
35 Minimum content(%) controller ASR Minimum
30
25 0 20406080100120 Avg. Particle size(um)
Figure 5.4: Relation between glass powder size and required dosage to mitigate ASR
In the remained part of this chapter, it was focused on the mitigating mechanisms of ASR by
glass powder as the main objective of this research. This was accomplished by replacing 40% of
portland cement with GP2 in the ASTM C1567 mortar bars and comparing different properties
with mortars containing 100% portland cement.
5.3.2 Tensile and compressive strengths
The 3-day tensile and compressive strengths of mortars with and without glass powder are compared in Figure 5.5. The error bars represent the standard deviation based on strength 120
measurements on four duplicate samples. The reason that the test is performed on specimens
after 3-day exposure to NaOH bath was previously explained in section 3.3.4. After 3 days
exposure to 1M NaOH bath at 80°C, the tensile strength was found to be ~18% higher for the
GP2 mortar (6.40MPa) comparing with the control (5.41MPa). The increase in compressive
strength was even higher with GP mortar having 40% higher strength than the control
(100%PC). The higher strength is somewhat surprising since glass powder is known to have a
lower reactivity at ambient temperatures comparing to PC and thus relatively lower early-age
strengths were anticipated for glass powder mortar. However, high temperature (80oC) and a
high alkali content (pH=14 in the soak solution) significantly accelerates the pozzolanic activity
of glass powder which leads to a denser microstrucutre just after 3-days of curing in this test
environment. An evaluation of the pore size distribution of the binder phase for the two materials
is provided in Figure 5.10 and will be discussed later.
The higher early-age tensile strength could have a beneficial impact on ASR performance of
mortars. The improved strength leads to a better resistance against cracking caused by swelling of the ASR gel. Formation of cracks provides a direct access of the high pH soak solution to the
interior of mortar bars. As such, by preventing or retarding crack formation, glass powder reduces the ion transport which in turn, lessens ASR. The effect of higher tensile strength of
glass powder mortars is also predicted for concretes exposed to ambient conditions. In service
life, increase in the strength of glass powder concretes occurs in relatively higher rates (months)
[7,9] comparing to the time scale of ASR (years). Application of glass powder could possibly
lead to concretes with higher tensile strength and a correspondingly higher ASR cracking
resistivity.
121
40 8
35 7
30 6 esl Strength(MPa) Tensile Control 25 5 40% GP2
20 4
15 3 Compressive Strength(MPa) Compressive
10 2
5 1
0 0
Figure 5.5: Compressive and tensile strengths of control and GP2 mortar after 3 days Curing in 1M NaOH solution at 80°C
5.3.3 Pore solution composition of ASTM C1567 mortars
As the driving force for ASR is the concentration of OH- ions in mortar pore solution,
monitoring the composition of the pore fluid of the mortars over time during the ASTM C1567
122
test delivers valuable information. A comparison between the composition of pore solutions from
the control and 40%GP2 mortars are provided in Figure 5.6 and Figure 5.7:
The variation of [OH-] in the pore solution of mortars during the ASTM C1567 test period is
shown in Figure 5.6. Immediately after mixing, [OH-] is 25% lower in GP mortar, although the
alkali content of glass powder is much higher comparing with PC (Table 5.1). Due to further
cement hydration during 24hr moist room curing period, [OH-] increases dramatically in the
control mortar leading to a concentration of 744mM. The corresponding value for GP mortar is
395mM, indicating approximately 50% decrease in OH- ion as a result of 40% GP2 replacement.
Not only fewer alkalis are released to the pore due to relatively much lower reactivity of glass
powder, but also some alkalis are probably bound in due to pozzolanic effect of glass powder.
However, the initial dilution effect of glass powder is completely erased during the following
24hr water curing at 80°C. Similar [OH-] was measured for both mortars (~230mM) 2 days after
casting and prior to submerging them in the NaOH bath. It seems that OH- ions diffuse out of
mortars into the water bath at a faster rate in the control mortar. This is further evaluated by
measurement of the mortars’ diffusivity as discussed later (Figure 5.9). After submerging in the
NaOH bath, [OH-] increases in both mortars for the remaining of the test period. However, this
increase was faster for the control mortar presumably due to its higher ion diffusivity. After only
3 days, the [OH-] of control mortar has increased by 156% (from 230mM at 2 days to 670mM at
5 days). This value reaches to 1130mM after 16 days of NaOH exposure. For 40%GP2 mortar,
[OH-] is 430mM at 5 days and 570mM at 16 days. In other words, 40% glass powder
123
replacement of cement leads to nearly 50% reduction in [OH-] at the conclusion of ASTM C1567 test.
1200 [OH]- Control(100% cement) [OH]- 40% GP2 1100 OH- concentrtin in the external bath 1000 After 24 hr Moist room curing 900
800 After 24 hr water curing 700
600
500 Concentration (mM) -
OH 400
300
200
100
0 0 2 4 6 8 10 12 14 16 Days after casting
Figure 5.6: OH- concentration in pore solution of mortars during ASTM C1567 test
124
1600
1400 After 24 hr Moist room curing 1200
1000
800 After 24 hr water curing
Na Control(100% cement) 600 Concentration (mM) Na 40% GP2
K Control(100% cement) 400 K 40% GP2
200
0 0246810121416 Days after casting
Figure 5.7: Concentration of Na and K in the pore solution of mortars during ASTM C1567 test
Figure 5.7 compares the concentration of Na in the pore solution of the control and GP mortars
during ASTM C1567 test. It should be mentioned that ICP measures the total elemental Na
content of the pore solution. The elemental Na includes both the Na ions and the Na element in dissolved or colloidal silicates which as were explained previously in section 3.4.2. Considering
that Na element can be contributed by cement, glass aggregates, and glass powder, it should not
be surprising that [OH-] does not necessarily equal the elemental concentration of [Na]+[K].
125
Note that electrical charge balance still suggests that ionic concentrations must balance:
[Na+]+[K+]=[OH-].
Unlike the concentrations of [OH-] and [Si], the Na concentration of the pore solution in both
mortars, increased steadily from the mixing time to the end of the test. For the controlled
mixture at 0 day, [Na]=231mM and [K]=234mM. Given that [OH-]=230mM, this may suggest that the majority of Na at this age is non-ionic. At 14 days exposure to NaOH bath, the
- concentrations are [Na]=1335mM, [K]=37mM, and [OH ]=1130. This may suggest that [Nanon- ion]≈1335+37-1130=242mM which agrees with the non-ionic Na concentration at 0 day. Note
that K is diffusing out of mortar during the test.
For the pore solution of 40%GP2 mortar, the following concentrations are obtained at 0 day:
[Na]=435mM, [K]=102mM, [OH-]=230mM. Potassium is contributed by cement only is
expected to be lower in 40%GP2 mortar due to cement dilution. The non-ionic Na can be
estimated as [Nanon-ion]≈435+102-230=307mM. This increase in [Nanon-ion] could be attributed to
the presence of glass powder which dissolves and contributes Na into pore solution. This may
suggest that during the first 48 hours of curing, glass powder had dissolved into the pore
solution, but this dissolution did not affect the alkalinity (i.e., pH) of pore solution. This is an
important concept which deserves further investigation. But preliminary results suggest that
although glass powder releases Na to the pore solution, this is not necessarily in the form of Na+ ion which increases the alkalinity of the medium and promotes ASR.
At 14 days, the pore solution alkali concentrations of 40%GP2 mortar are: [Na]=1386mM,
[K]=72mM, [OH-]=570mM; which leads to two notable observations. First, [K] in GP2 mortar is
126
higher than the control mixture. Given that [K] is controlled by the outward diffusion of
potassium, this agrees well with the lower ion diffusivity of the GP2 mortar. Second,
[Nanonion]≈1386+72-570=888mM. This suggests that although some ionic Na penetrates from the
soak solution and raises the pore solution’s pH, the majority of Na available in the pore solution
is in the non-ionic form which is contributed by dissolution of glass powder. This non-ionic Na does not exacerbate ASR.
5.3.4 Long term pH of pore solution of cement pastes
The composition of pore solution of paste specimens prepared with and without glass powder
was monitored over a 220 day period. The purpose was to obtain a preliminary understanding of
the impact of glass powder on the pH of pore solution when replaces cement. The mass of sealed
paste cylinders were nearly the same at the time of pore solution extraction comparing to the
time they were plastic wrapped. This means that no mass is either lost or gained during the moist
curing period. The pH of pore solution of the pastes cured in 100%RH at 23°C is compared in
Figure 5.8. The only source of alkalis is cement and glass powder. At the time of demolding after
24 hours from casting (0-day), the [OH-] is 550mM in 40%GP2 paste, nearly 70% of that of the
control (100%PC) paste’s pore solution (776mM). This is a result of replacement of cement with
relatively less reactive glass powder. However, not exactly a 40% dilution occurs due to cement replacement by glass powder. This may be partially from an alkali release of glass powder. In addition, it may be partially be a result of promotion effect of glass powder on cement hydration
[4].
127
1200
1000
800 OH- contributed by GP
600 Concentration (mM) -
OH 400 Control 100% cement
40% dilution
200 40% GP2 replacement
0 0 50 100 150 200 250 Days
Figure 5.8: [OH-] concentration in the pore solution of pastes cured at 23°C, RH=100%
It is observed that after 125 days no significant increase occurs in the [OH-] of 100%cement paste comparing as nearly similar value is observed after 220 days. However, OH- concentration
increases slightly with a 7% linear increase from 65 to 220 days. This indicates that in a longer
run more alkalis are anticipated to be released from the glass powder in this test environment.
While the time scale of the test is very limited, it brings up the concern of more alkali release
from the glass powder in long term tests. Based on this preliminary study it is suggested to
investigate alkali release of glass powder in long term tests such as ASTM C1293.
128
5.3.5 Ion diffusion coefficient of mortars
To understand the effect of glass powder on the rate of ion transport of mortars, the variation of
ion diffusivity is compared in Figure 5.9 during the ASTM C1567 test. The ion diffusion
coefficient was calculated with similar method explained in section 3.4.3. It is evident that even
with an average particle size of 4× that of Portland cement, GP2 has been effective in reducing
the ion diffusion coefficient of the mortar. At the early stage of the test prior to submersion in
NaOH bath, the diffusion coefficient of the control mortar is 2.5× higher comparing to 40%GP2 mortar. Up to 9 days, the diffusivity of the control mortar further increases unlike the steady
decrease in the diffusivity of GP2 mortar. At 9 days, the diffusivity of the control is 14× larger
which suggests much faster ion penetration from the soak solution into the control mortar. After
9 days, the ion diffusivity of the control mortar starts to decrease as explained before in section
3.4.3. However, the ion diffusivity of the 40%GP2 mortar slightly increased towards the end of
the test period.
The reduction in the diffusivity of 40%GP2 mortar up to 9 days is linked to a pore structure
refinement as a result of pozzolanic reaction of glass powder (Figure 5.11). The slight increase in
the diffusivity of the GP2 mortar after 9 days is potentially a result of the near surface ASR
micro-cracking as evidenced by SEM micrographs.
129
2.E-11 Control(100%PC)
1.E-11 40%GP2
1.E-11 /s) 2
1.E-11
8.E-12
6.E-12
Mortar Diffusion Coefficient (m 4.E-12
2.E-12
0.E+00 0246810121416 Day after casting
Figure 5.9: Comparison of the ion diffusion coefficient of the control and 40%GP2 mortars during ASTM C1567 test
The rate of transport of ions from the external bath extensively affects the rate of alkali-silica
reaction in ASTM C1567 mortars. The observations include in this section long with the pore
fluid analysis data of Figure 5.6 and Figure 5.7 strongly support the role of the pozzolanic
reaction of glass powder in reducing the ion transport of the mortar. The results show that inclusion of glass powder with an average particle size of nearly 4× larger than cement’s is able to reduce the ion diffusivity of the mortar and thus reduce ASR in ASTM C1567 test.
130
5.3.6 Porosity, pore connectivity and pore size distribution
Schwarz et al. [5] have reported relatively lower early age (2-day) porosities for pastes
containing different GP contents comparing to 100% cement pastes. In contrast, later age
porosities were reported to be higher for GP pastes. They explained that at early ages, lower
density glass powder (2.5g/cm3) occupies a greater volume for the same mass of cement
(3.15g/cm3) and thus reduces the pore volume. At later ages, cement hydrates and produces
lower density C-S-H which occupies a greater volume. As glass powder replaces cement a lower
content of C-S-H is produced at later ages. This results in a higher porosity in GP pastes at later
stages of hydration. In the current study, for the mortars exposed to ASTM C1567 conditions,
the relatively lower early age porosity of 40%GP2 paste is not observed. Due to the elevated
temperature of 80°C, cement hydration is promoted, enough to produce a C-S-H content with
relatively higher volume to counteract the effect of glass powder. From the cumulative pore size
distribution curves, it is observed that the 5-day total porosity of 100% PC paste (31.83%) is
lower comparing to the GP paste (35.54%). Considering the ion diffusion coefficients from
Figure 5.9 and using values obtained from cumulative pore size distribution curves, pore connectivity (is calculated to be ~10 times higher in the control paste comparing to 40%GP
paste 5 days after casting.
131
50 Control, Avg. pore radius=25nm 45 40%GP2, Avg. pore radius=20nm
40
35
30 /g) 3
25
20 Volume(mm
15
10
5
0 1 10 100 1000 10000 100000 Pore radius(nm)
Figure 5.10: Pore size distribution in 100%PC and 40%GP2 pastes after 3 days exposure to 1M NaOH solution at 80°C (total age=5 days)
The pore size distribution of the control and 40%GP2 (pastes) cured in the same conditions of
ASTM C1567 is compared in Figure 5.10. Not only the pore connectivity () decreases with application of glass powder, but also the average pore size decreased from 25nm in the control paste to 20nm in the 40%GP paste. There is a shift of the pore size distribution curve to the lower pore sizes indicating a pore refinement effect in the 40%GP2 paste. This has been previously observed for other pozzolans during ASTM C1567 test (chapter 4) and also in long term for normal curing conditions [17]. As larger pores convert to many smaller pores (pores refined) and
132
their connection are reduced, the ions will find it harder to pass through the mortars and their ion
diffusion coefficient drops. An indirect effect is a corresponding increase in the tensile strength
due to the improved interconnection of mortars’ microstructure.
5.3.7 Portlandite consumption
The pore refinement mentioned in the previous section, can be explained to be a result of glass
powder’s pozzolanic reaction. In this reaction, portlandite(CH) is consumed to produce
secondary C-S-H according to equation (2-1). Thus, by measuring the CH content of the pastes
with TGA, the pozzolanic behavior of glass powder can be monitored for pastes cured in ASTM
C1567 environment. The glass powder’s pozzolanic reaction rate has been previously measured
as a function of age for concretes that were cured in water bath at 20ºC [18]. While lower
portlandite contents were reported for cement replacement levels up to 20%, surprisingly higher
contents were observed for greater dosages. This is not the case in the current study as 40% GP2 replacement leads to a steady decrease in CH content. As Figure 5.11 suggests, due to cement hydration, CH content of the control paste (100% PC) increased steadily to the end of the test period. The 14-day CH content is ~28% higher in the control comparing to its initial 0-day value.
By replacing 40% of PC and assuming cement hydration rate is not altered, 40% dilution in the
CH content is expected (dashed line). Therefore, an initial exactly a 40% reduction in the CH content of 40%GP2 is observed. However, glass powder has more than a dilution effect as the difference in CH content of the control and 40%GP2 pastes increase to the end of the test.
133
25 Control (100% Cement) 40% Dilution effect 40% GP2 Replacement 20
15
10 Portlandite Consumption (Wt. %) Consumption Portlandite
5
0 0 2 4 6 8 10 12 14 Day
Figure 5.11: TGA analysis of the mortars after 0, 7 and 14 days NaOH exposure
5.3.8 Microstructural analysis (SEM/EDS)
SEM/EDS imaging was performed to answer two main questions. First, does ASR gel form in
large quantities in glass powder mortars? Second, does the presence of glass powder alter the composition of ASR gel and as such, potentially change its properties (e.g. viscosity and
swelling pressure)? Figure 5.12 shows backscatter SEM micrographs of the control and
40%GP2 mortar at the conclusion of ASTM C1567 test. The images were taken from areas of
134 similar distance to the mortars surface. The control mortar undergoes ASR and shows extensive gel formation and cracking at different locations of mortar cross section. In contrast, no ASR gel was observed in the central locations of the 40%GP2 mortar cross section and only small traces of gel were observed near the surface of mortar bars where the material was exposed to the soak solution. This is confirmed by higher magnification images (not included here) of the glass aggregate particles and aggregate-paste interface. The more severe near surface ASR gels further supports the dominance of ion transport as the mechanism governing the rate of ASR in the accelerated mortar bar test.
Table 5.2 provides the results of EDS compositional analysis of ASR in the two mortars. A total of 25 EDS spot analysis were performed on the ASR gel in the control mortar and due to much lower volume of gel, 7 spots were analyzed for the GP2 mortar. The compositions were averaged and reported in Table 5.2. Overall, ASR gel has a relatively consistent composition in the two mortars. The main difference is the lower Si content of the gel in GP2 mortar possibly due to a lower rate of silica dissolution. A lower CaO/SiO2 ratio could hint towards a less expansive gel
[19,20]. Based merely on composition, ASR gel in the two mortars is likely to exhibit similar swelling properties.
Table 5.2: Average atomic composition (wt.%) of ASR gel measured by EDS Na Ca Si K Al Ca/Si at 14 days Control 15.46 17.67 63.13 0.68 1.84 0.28 St. Dev. 3.38 3.38 3.95 0.14 0.48 40%GP2 16.40 19.04 59.08 0.71 1.44 0.32 St. Dev. 3.51 7.75 4.47 2.79 0.51
135
ASR gel
Glass sand particle
Figure 5.12: SEM image of control and 40% GP2 mortars cross section after 14 days exposure to NaOH solution at 80°C
136
5.3.9 Reducing silica dissolution rate from aggregates
The dependence of glass corrosion rate to the ratio of its surface area to the volume of the
attacking solution (SA/V) has been studied in [21-23] for acidic to neutral pH levels. Iler [24]
claims that for nano-size silica particles in pH levels 9~10, larger particles grow in presence of
more dissolvable smaller ones. The solubility of larger glass particles in presence of smaller ones
has not gained considerable attention in literature. This is specifically important from an ASR
perspective as large (150m-4.75mm) glass particles are exposed to high alkaline pore solution
of concrete when cement is replaced with glass powder (<100m).
Figure 5.13 shows the reduced rate of dissolution of glass slides (SA=79×10-4m2) in presence of
20g GP2 (SA=1.69 m2) in 340cc 1M NaOH solution. After 14 days, the mass loss of glass slides
is ~25% lower at the conclusion of the test due to presence of 20g of glass powder with relatively
high surface area (847cm2/g). Adding 20g GP2 to the corrosion cell means increasing SA/V by a
factor of 215. Due to the increase in the “accessible” surface area the effective concentration of
OH- (attacking ions which initiate ASR) drops from 4.3mM/cm2 to 0.02mM/cm2. Considering
that smaller glass particles (<600m) does not produce deleterious ASR expansion [25] the
reduction in effective OH- content will partially contribute to ASR mitigation. It should be
mentioned that the pH of the solution slightly increased (~6%) during the test period. Feng and
Pegg [23], discuss that in low pHs (~7 or more), hydroxyl ions are regenerated due to ion
exchange. Thus, pH of the solution tends to increase with a higher rate when SA/V is higher; as
more “accessible” reaction sites are available for unit volume of the solution. This is speculated
to be extendable in higher pH levels.
137
Exposure time (day) 0 2 4 6 8 10 12 14 0
-0.02
-0.04
) -0.06 2
-0.08
-0.1
Weight Loss(mg/mm -0.12 Glass slides
Glass slides+20 gr GP2 -0.14
-0.16
-0.18
Figure 5.13: Mass loss of glass slides in 1M NaOH solution at 80ºC in the presence or absence of glass powder
5.4 Conclusions
Based on the results of experiments performed on ASTM C1567 mortars in this study, the following conclusions can be drawn:
138
Glass powder can mitigate ASR in the accelerated mortar bar test. Finer glass powders are
more efficient, meaning that their required dosage to control ASR decreases by reducing the
particle size of the powder.
Glass powder is very effective in reducing ion diffusion coefficient of mortars, at least for
conditions imposed by ASTM C1567 test. Up to 14× reduction in the diffusivity of 40%GP2
mortar was observed in comparison with the control 100%PC mortar. The findings suggest
that glass powder has a great impact on the ASR mitigation by reducing ion transport.
Pore fluid analysis showed a significant reduction in the [OH-] concentration of 40%GP2
mortar during the test period which is consistent with low ASR expansion in this mortar.
This reduction in [OH-] is primarily due to lower diffusion coefficient of 40%GP2 mortars.
Preliminary results suggest that although glass powder releases Na to the pore solution, this
is not necessarily in the ionic form (Na+) which increases the alkalinity and promotes ASR.
Addition of glass powder has a considerable impact on compressive and tensile strengths of
mortars. The improved resistance against cracking also results in reduced transport through
cracks.
In presence of glass powder, the effective ratio of OH- to accessible surface area dramatically
decreases. As a result, larger glass aggregates are being attacked with a lower rate. This
contributes to ASR mitigation.
139
5.5 References:
1. Y. Shao, T. Lefort, S. Moras, D. Rodriguez, Studies on concrete containing ground waste glass, Cement and Concrete Research, 2000, 30(1), 91-100.
2. M. Pattengil, T.C. Shutt, “Use of ground glass as a pozzolan”, In: Albuquerque Symposium on Utilization of Waste Glass in Secondary Products, January 24–25, Albuquerque, New Mexico, USA (1973), 137–153.
3. J.S. Damtoft, J. Lukasik, D. Herfort, D. Sorrentino, E.M. Gartner, Sustainable development and climate change initiatives, Cement and Concrete Research, 2008, 38(2),115-12.
4. N. Schwarz, N. Neithalath., Influence of a fine glass powder on cement hydration: Comparison to fly ash and modeling the degree of hydration. Cement and Concrete Research, 2008, 38(4), 429- 436.
5. N. Schwarz, M. DuBois ,N. Neithalath, Electrical conductivity based characterization of plain and coarse glass powder modified cement pastes, Cement and Concrete Composites, 2007, 29(9), 656-666.
6. N. Schwarz, H. Cam, N. Neithalath, Influence of a fine glass powder on the durability characteristics of concrete and its comparison to fly ash. Cement and Concrete Composites, 2008, 30(6), 486-496.
7. C. Shi, Y.Wu, C. Riefler, H. Wang, Characteristics and pozzolanic reactivity of glass powders, Cement and Concrete Research, 2005, 35(5), 987-993.
8. B. Taha, G. Nounu, Utilizing waste recycled glass as sand/cement replacement in concrete, ASCE Journal of Materials in Civil Engineering, 2009, 21(12), 709-721.
9. V. Corinaldesi, G. Gnappi, G. Moriconi and A. Montenero, Reuse of ground waste glass as aggregate for mortars, Waste Management, 2005, 25(2), 197-201.
10. Shayan, A.M. Xu, Value-added utilization of waste glass in concrete, Cement and Concrete Research, 2004, 34(1), 81-89.
11. A.J. Carpenter, S.M. Cramer, Mitigation of ASR in pavement patch concrete that incorporates highly reactive fine aggregate, Transportation Research Record, 1999, 1668, Paper# 99-1087, 60-67.
12. R. Idir, M.Cyr, A. Tagnit-Hamou, Use of fine glass as ASR inhibitor in glass aggregate mortars, Construction and Building Materials, 2010, 24(7), 1309-1312.
13. D. Dyer, R. K. Dhir., Evaluation of powdered glass cullet as a means of controlling harmful alkali-silica reaction, Magazine of Concrete Research, 2010, 62(10), 749-759.
140
14. M. D. A. Thomas, B. Fournier, K. Folliard, J. Ideker and M. Shehata, Test methods for evaluating preventive measures for controlling expansion due to alkali-silica reaction in concrete, Cement and Concrete Research, 2006, 36, 1842-1856
15. J. Duchesne and M.A. Bérubé, The effectiveness of supplementary cementing materials in suppressing expansion due to ASR, Another look at the reaction mechanisms: I. Concrete expansion and portlandite depletion, Cement and Concrete Research, 1994, 24(8), 1574-1576.
16. L.J. Malvar, G.D. Cline, D.F. Burke, R. Rollings, T.W. Sherman, and J.L. Greene, Alkali-silica reaction mitigation: State of the art and recommendations, ACI Materials Journal, 2002, 99(5), 480-489.
17. Della M. Roy, Weimin Jiang, M.R. Silsbee, Chloride diffusion in ordinary, blended, and alkali- activated cement pastes and its relation to other properties, Cement and Concrete Research, 2000, 30, 1879-1884.
18. T. D Dyer, R. K. Dhir, Chemical reactions of glass cullet used as cement component, Journal of materials in civil engineering, 2001, 13(6), 412–417.
19. P.J.M. Monteiro, K. Wang, G. Sposito, M.C. dos Santos, W.P. de Andrade, Influence of mineral admixtures on the alkali-aggregate reaction, Cement and Concrete Research, 1997, 27(12), 1899-1909.
20. Bonakdar, B. Mobasher, S.K. Dey, D.M. Roy, Correlation of reaction products and expansion potential in alkali-silica reaction for blended cement materials, ACI Materials Journal, 2010, 107(4), 380-386.
21. E. C. Ethridge, D. E. Clark and L. L. Hench, Effects of glass surface area to solution volume ratio on glass corrosion, Physics and Chemistry of glasses, 20(2), 1979, 35-40.
22. L. L. Hench and D. E. Clark, Physical chemistry of glass surfaces, Journal of Non-Crystalline Solids, 1978, 28, 83-105.
23. X. Feng and I. L. Pegg, A glass dissolution model for the effects of S/V on leachate pH, Journal of Non-Crystalline solids, 1994, 175, 281-293.
24. R.K. Iler, Chemistry of silica - Solubility, polymerization, colloid and surface properties and biochemistry, 1979, John Wiley & Sons, New York.
25. F. Rajabipour, H. Maraghechi, G. Fischer, Investigating the alkali-silica reaction of recycled glass aggregates in concrete materials, ASCE Journal of Materials in Civil Engineering, 2010, 22(12), 1201-1208.
141
CHAPTER
SIX
INVESTIGATING THE USE OF Al(OH)3 AS AN ASR SUPPRESSOR: EFFECT ON THE DISSOLUTION RATE OF SODA-LIME GLASS AGGREGATES
6.1 Introduction
Many concrete structures suffer from the deleterious alkali-silica reaction (ASR), which significantly affects their durability performance and reduces their anticipated service life.
Alkali-silica reaction initiates from dissolution of meta-stable silicate phases contained in many natural aggregates in the high alkaline pore solution of concrete. These silicates dissolve by the attack of OH- ions and the dissolution products repolymerize to produce an alkali-lime-silica
(ASR) gel [1,2]. This gel is highly hygroscopic, imbibes water, swells and cracks the concrete
[1,2]. A number of supplementary cementitious materials (SCMs) and chemical admixtures
142
(mostly lithium compounds) [3] are commonly used to mitigate ASR damage. However, effective ASR inhibiting admixtures could be costly and may not be available in some locations.
As such, there is a continuing need to develop cheap and highly effective additives to avoid ASR in new and possibly even in existing concrete structures. It is also of paramount importance to correctly identify the mechanisms by which these admixtures mitigate ASR, so their performance and cost can be improved.
In an attempt to find new materials to mitigate ASR, Natesaiyer and Hover [4], performed a study on the feasibility of using different chemicals against alkali aggregate reaction. They showed the positive effect of aluminum sulfate, zinc sulfate and pyrogallol (a benzene compound) in reducing the dissolution rate of amorphous silicates of spratt and opal aggregates.
It is known from the literature that the presence of alkali earth metals (e.g. Be, Ca), zinc, iron and aluminum [5,6] reduces the dissolution rate of silicate glasses in alkaline solutions. Aluminum is of special interest from an ASR perspective, given that it is one of the main elements in many
SCMs (e.g. metakaolin, fly ash and natural zeolite) [7-11]. SCMs are considered to be beneficial against ASR mainly due to their high amorphous silicate contents which lead to the pozzolanic reaction and binding of alkalis [12]. The aluminum content of SCM is also believed to contribute to the pozzolanic reaction [13]. There is evidence [14] that SCMs with considerable aluminum content are more effective in controlling ASR than those containing higher purity silica(???).
Aluminum also converts the C-S-H (i.e. the main binder phase of concrete) to C-A-S-H (also known as the Al-modified C-S-H). This gel has been reported to have a higher capacity for incorporating alkalis in its microstrucutre [15] and may better reduce the alkalinity of concrete pore solution. Finally, aluminum may have a retarding effect on the polymerization of silicic acid
143
(i.e. dissolution product of silica in aqueous solutions) [16] and may repress gelation of silicate species and formation of the ASR gel. It should also be noted that aluminum may be present in some reactive aggregates and the simultaneous presence of silicate and aluminate phases may affect the ASR reactivity of such aggregates.
There is limited quantitative information on the role of Al against ASR. This chapter attempts to advance the knowledge on this subject. Specifically, the effect of Al on dissolution of soda-lime- silica glass as a model reactive aggregate is studied. In addition, amorphous aluminum hydroxide powder is admixed in ASTM C1260 mortars as a partial cement replacement and its effect on
ASR expansions is monitored. Aluminum hydroxide, Al(OH)3, is used as the source of aluminum in order to reduce the complexity of the system by avoiding the addition of an extra anion. The preliminary results confirm that Al(OH)3 is successful in controlling ASR, at least in this test environment.
6.2 Research significance
Supplementary cementitious materials (SCMs) are known to mitigate ASR mainly by reducing the pH of the pore solution [17]. However, reducing the alkalinity has the potential negative side effect of increasing the risk of carbonation and corrosion of reinforcing steel bars. In addition, there is a new trend to produce more environmentally positive concretes based on alkali activation of industrial byproducts such as blast-furnace slag or fly ash. The pore solution of these concretes is inevitably highly alkaline which increases their risk of ASR. Therefore, it
144
would be beneficial to study materials/admixtures which are able to depress the dissolution rate
of reactive silicate aggregates, even at high pH.
Recently, the importance of the Al content of SCMs against ASR has gained some attention [18].
However, the exact role of aluminum in ASR mitigation has not been well established.
Moreover, the feasibility of using aluminum compounds independently, e.g., as partial cement
replacement, to mitigate ASR, has not been investigated. The current research aims to elucidate
the beneficial effect of aluminum against ASR. The findings of this study may provide initial
steps towards development of new admixture materials to avoid ASR in future concrete
structures. In addition to the effect of Al on ASR, the mechanisms of participation of the
admixed Al in hydration reactions of cement and its effect on concrete properties (e.g., strength,
risk of sulfate attack, etc.), must be independently studied. These, however, are outside the scope
of current investigation.
6.3 Materials and methods
To characterize the role of aluminum in reducing the risk of ASR, two series of experiments
were performed. In the first series, ASTM C1260 test was used to assess the feasibility of using
aluminum hydroxide as partial cement replacement to mitigate ASR. These mortars contained
recycled soda-lime glass aggregates, which are known to be highly reactive. Ideally, these test
results should be supplemented with the results of the more reliable ASTM C1293 (concrete prism expansion) test, which is considered to better represent the ASR performance of concrete
during its service life. This 2-year test is currently underway and its results will be presented in future publications. In the second series of experiments, the effect of aluminum on the
145
dissolution rate of soda-lime glass was studied. This was accomplished first by measuring the
mass loss of glass slides during a corrosion experiment inside a NaOH bath with or without Al.
In the next step, the species concentrations in the bath solutions were monitored as a function of time using inductively coupled plasma-atomic emission spectrometry (ICP-AES) and acid titration. Finally, the reaction products which precipitated on the surface of glass slides in the
presence of aluminum were characterized with the aid of scanning electron microscopy (SEM),
energy dispersive spectroscopy (EDS) and X-ray diffraction (XRD) techniques.
Mortar prisms were prepared according to ASTM C1260 standard with different Al(OH)3 contents as partial cement replacement. Type I ordinary portland cement was used with the oxide composition presented in Table 6.1. Amorphous aluminum hydroxide powder with an average particle size of 58m was procured from J. T. Baker with a purity of 98%. The glass aggregate
was obtained by washing and crushing recycled glass bottles. They were then sieved to produce
the standard ASTM C1260 aggregate size gradation. All mortars were prepared with w/cm=0.47, and cured according to the ASTM C1260 specified conditions. After mixing according to ASTM
C305, the mortars were molded and sealed cured in a moist room for 24 hours at 23°C. They were then demolded and cured under water at 80°C for another 24 hours. They were then submerged inside 1M NaOH solution at 80°C for a period of 14 days and their expansion was monitored over time. The expansion of the prisms was monitored using a digital comparator with the accuracy of ±0.0025mm.
146
Table 6.1: Oxide composition of portland cement (mass %)
OXIDE CaO SiO2 Al2O3 Fe2O3 Na2O K2O MgO MnO TiO2 SO3 P2O5 LOI CEMENT 62.5 19.9 5.44 2.26 0.3 0.89 2.31 0.09 0.29 4.93 0.23 0.86 GLASS 10.62 73.13 1.99 0.52 13.74 0.34 0.53 N/D* N/D* N/D* N/D* --- *N/D is not detected
The dissolution rate of Al(OH)3 powder and glass slides were individually measured inside
340cc solution of 1M NaOH at 80°C for 14 days. This ambient condition was considered to
duplicate the conditions of the ASTM C1260 test. (1) The dissolution of Al(OH)3 in these conditions was experimentally monitored by providing an ample source (20g) of Al(OH)3 powder in the container with 1M NaOH solution (without the presence of glass slides). The results were compared with thermodynamic predictions. (2) The dissolution corrosion rate of soda-lime glass slides (75×25×1mm) was monitored by submerging two glass slides in 340cc of 1M NaOH solution in the absence of a source of Al. No agitation or stirring was applied as static dissolution similar to concrete pore solution was attempted. (3) The effect of aluminum on the dissolution rate of glass slides was studied by submerging glass slides in a 1M NaOH bath that also contained Al(OH)3. Two cases were studied; one in which an inexhaustible quantity (20g) of
Al(OH)3 powder was added to the solution, and the next, where a finite quantity of dissolved
Al(OH)3 was present in the solution. To measure the glass corrosion rate in each case, the mass
loss of glass slides was measured at specific times using a balance with an accuracy of ±0.0001g.
For each mass loss data point, measurements were obtained from two duplicate slides. Slides
were carefully removed from the solution, gently washed with deionized water, dried using a
wipe and weighed. They were then placed back into the solution. The solution and glass slides
were kept inside polypropylene (PP) plastic containers with sealable lids to minimize any
147
carbonation or evaporation. (4) The concentration of metallic elements (Al, Si, Na, Ca, K) in the solution was determined by sampling the solution at regular time intervals and measurements, using a Perkin-Elmer Optima 5300 ICP-AES. Samples were taken from the bath with syringe and passed through a 0.2µm PTFE filter. They were stored in sealed PP vials until the ICP test.
The concentration of OH- in the solution samples was also determined using acid (HCl) titration.
(5) During the test, a layer of silica gel precipitated on the surface of glass slides. When no Al
was present in the bath, this layer could easily be removed by washing of slides with hand wipe using deionized water. This was not possible for the layer formed in the presence of Al which
strongly adhered to the surface of slides. This reaction layer at the surface of Al-treated glass slides was characterized with the aid of X-ray diffraction (XRD) with Panalytical Xpert Pro mpd theta-theta diffractometer and SEM/EDS with FEI NanoSEM 630 FESEM. After 14 days exposure to 340cc 1M NaOH bath at 80°C, the slides were removed from the bath, rinsed with deionized water and dried in ambient temperature under N2 purge. They were then stored in
sealed plastic bags until testing. XRD was performed on the undisturbed surface product of the
aluminum treated glass slides. This layer was also scraped off from similarly treated glass slides
and collected in powder form, and analyzed by XRD. In addition, similarly treated glass slides
were broken by hand and the exposed cross section was carbon coated and characterized using
SEM/EDS.
6.4 Results
6.4.1 ASR expansion in ASTM C1260
148
The ASR expansion of mortar prisms exposed to 1M NaOH solution at 80ºC with different
Al(OH)3 contents are shown in Figure 6.1. The error bars represent ±1 standard deviation based on length measurements of four duplicate samples. When Al(OH)3 replaces 15% of portland cement by mass, the ASR expansion of mortars is reduced by 64%. At 20% replacement,
Al(OH)3 powder further reduces mortar expansion to 0.03% at 14 days; which is far below the
ASTM innocuous threshold of 0.1%.
0.7%
Control 0.6% 15% Al(OH)3 20% Al(OH)3 0.5%
0.4%
0.3% Expansion %
0.2%
ASTM Threshold 0.1%
0.0% 02468101214 Days of exposure to 1M NaOH @80°C
Figure 6.1: Reduction in ASR expansion with increasing Al(OH)3 content in ASTM C1260
149
It could be speculated that, Al(OH)3 is reducing ASR in this test, possibly through one or more of
the following mechanisms. (1) Alkali dilution: by replacing 20% of portland cement with
- Al(OH)3 which is capable of absorbing rather than releasing OH ions (as discussed in the next
section), the pH of pore solution is reduced and thus ASR is mitigated. (2) Alkali binding: by
increasing the initial Al content, it is likely that more C-A-S-H gel is produced to bind alkalis out
of pore solution [16]. (3) Reducing mass transport: aluminum hydroxide may participate in
chemical reactions [13] leading to formation of C-A-S-H, C-A-H, and/or A-S-H products that
may reduce porosity and mass transport inside mortars. As such, the penetration rate of external
OH- from the soak solution is reduced. 4) Decreasing aggregates dissolution rate: presence of
dissolved aluminum can reduce the dissolution rate of silicate glass aggregates in alkaline
solutions [5,6]. While all these mechanisms may contribute to ASR mitigation and deserve
research attention, this study focuses on the last mechanism to better clarify the effect of Al on
the dissolution rate of soda-lime glass in aqueous alkaline solutions.
6.4.2 Dissolution rate of soda-lime glass slides
6.4.2.1 Dissolution of Al(OH)3 in alkaline aqueous solutions
Before studying the dissolution kinetics of soda-lime-silica glass in presence of Al, a brief evaluation of dissolution of Al(OH)3 at high pH is useful to determine its solubility and the type
of dissolved species formed. As solid Al(OH)3 is introduced in water, different ionic species form depending on the pH of the solution:
150
3 3+ - 3 Al(OH )3(s) Al 3OH K1=[Al ][OH ] log K1=-33.5 (6-3a)
3 2 2 3+ - Al OH Al(OH ) K2=[ Al(OH) ]/[Al ][OH ] log K2=9.03 (6-3b)
2 2 - Al(OH ) OH Al(OH ) 2 K3=[ Al(OH ) 2 ]/[ Al(OH) ][OH ] log K3=9.67 (6-3c)
- Al(OH ) 2 OH Al(OH )3(aq) K4=[ Al(OH)3(aq) ]/[ Al(OH) 2 ][OH ] log K4=8.3 (6-3d)
- Al(OH )3(aq) OH Al(OH ) 4 K5=[ Al(OH) 4 ]/[ Al(OH)3(aq) ][OH ] log K5=6 (6-3e)
The thermodynamic equilibrium constants were obtained from [19] and correspond to the temperature 25°C. Based on the thermodynamic data, speciation of Al as a function of pH can be calculated as presented in Figure 6.2. The total dissolved Al content is calculated by adding the concentration of the 5 different Al species introduced above. As the figure suggests, the solubility of Al changes by several orders of magnitude, depending on the pH of the aqueous
- solution. At pH>10, Al is dominantly present in a tetrahedral Al(OH)4 form and this has been experimentally confirmed by NMR studies of Moolenaar et al. [20]. The total solubility limit of
Al in a solution with pH=14 and at 25°C is obtained to be ~316mM according to this graph.
- Higher temperatures promote the solubility of Al(OH)4 species [21,22]. As such, the test temperature of 80°C used in the current study is anticipated to elevate the solubility limit. It
- should be noticed that Al(OH)4 not only doesn’t contribute to increasing the pH of the solution, but also removes OH- ions from the solution according to eq. (6-3e).
151
1.E+00
Al3+ 1.E-01 AlOH2+ 1.E-02 Al(OH)2+ Al(OH)3 1.E-03 Al(OH)4-
1.E-04 Al(Total)
1.E-05
1.E-06 Concemtration(M) 1.E-07
1.E-08
1.E-09
1.E-10 2 4 6 8 10 12 14 16 pH
Figure 6.2: Al speciation in water at different pH levels at 25°C
Figure 6.3 shows the time dependent Al concentration during the static dissolution test in which
340cc 1M NaOH solution at 80°C was slowly poured on 20g amorphous Al(OH)3 powder. As the figure suggests, Al(OH)3 dissolves steadily with time. As Al concentration grows toward the solubility limit, the rate of dissolution decreases. Similar time dependent dissolution of Al species has been observed for highly crystalline gibbsite [21] and AlCl3.6H2O [23] in 1M NaOH solutions at different temperatures. In this study, dissolution of 20g amorphous Al(OH)3 powder, yields an Al concentration of 225mM after 14 days. According to [23] when the concentration of
- Al(OH)4 surpasses 100mM some polyaluminates may also be present in the solution.
152
350
Solubility limit of Al(OH)3 at pH=14, 25°C 300
250 Al Concentration
200
150
100 Al Concentration (mM) Al Concentration (mM) in the solution 50
0 02468101214 Days of exposure
Figure 6.3: Dissolution of Al(OH)3 powder in 1M NaOH at 80°C
6.4.2.2 Mass loss measurement of glass slides and the solution analysis
Figure 6.4 compares the dissolution rate of glass slides in 1M NaOH at 80°C in the presence or absence of 20g Al(OH)3 powder. In the absence of Al(OH)3, the rate of dissolution of glass slides slightly accelerate during the test. Due to high Na content of soda-lime glass (Table 6.1), the pH of the solution increases during the experiment as shown in Figure 6.6, which results in increasing the dissolution rate of glass at later ages of the test. The presence of Al(OH)3 dramatically depresses the rate of mass loss of glass slides. At the conclusion of the test, the
153 mass loss in Al containing solution was lower by a factor of 11 (172.23g/m2 in NaOH solution vs. 15.78g/m2 in Al containing solution).
200
180 1M NaOH solution
160 1M NaOH solution+ 20 grAl(OH)3
140
120 ) of glass surface
2 100
80
60 Mass Loss (g/m Mass Loss
40
20
0 02468101214 Days in solution
Figure 6.4: The effect of addition of Al on the rate of glass slide dissolution
Figure 6.5 shows the results of the solution analysis with respect to Si and Al concentrations determined using ICP-AES. The figure shows Si concentrations ([Si]) in glass corrosion tests where 20g Al(OH)3 powder was or was not present. The figure also shows [Al] in the solution during the glass corrosion test where 20g Al(OH)3 powder was present. The following observations can be made:
154
(1) An approximately linear increase in the Si concentration of the solution is observed during the test when no Al was present. The Si concentration obtained at the conclusion of the test is
67.5mM. It should be noted that solution samples were collected by passing through a 0.2m filter. Thus, 67.5mM is the total concentration of dissolved silicate oligomers with a size smaller than 0.2m.
(2) The presence of Al powder leads to a dramatic decrease in the [Si] of the solution. During the test, the Si concentration remains small and varies in a range 0.1-5.3mM (avg.=2.2mM). EDS analysis shows that ~35% of the weight of soda-lime glass is composed of Si (Table 6.2). Mass loss measurements show that 0.25mg of glass from both slides combined has dissolved. If all the corresponding Si is passed into the solution, [Si] of the solution should be approximately 9.2mM.
The difference between 2.2 and 9.2mM suggests that some dissolved Si undergoes a chemical reaction (possibly with Al) and precipitates out of the solution. A likely chemical reaction involving Si and Al in NaOH solution is that of zeolite formation according to equation 6-4 [24].
xNa yAl(OH ) 4 zSiO2 (aq) Na x Al y Si2( yz) 2yH 2O (6-4)
155
75
Si concentration, No Al Si concentration,20g Al(OH)3 present 60 Al concetration, Si present
45
30 Concentration (mM)
15
0 0 2 4 6 8 10 12 14 Days of exposure
Figure 6.5: Si concentration due to glass slides dissolution in presence or absence of 20g Al(OH)3; Al concentration due to Al(OH)3 powder dissolution in presence of glass slides
(3) Al and Si can mutually reduce each other’s dissolution rate. Kvech and Edwards [25] have observed a reduction in Al content of alkaline solutions with increasing Si concentration for a pH of 9.5. In the present study, although 20g Al(OH)3 powder was available in the container, the concentration of dissolved Al effectively plateaues at 45 to ~50mM after 4 days of experiment.
In comparison with Figure 6.3, it is evident that dissolution of Al(OH)3 is considerably depressed in the presence of silicate glass. This may be caused by one of the following mechanisms: (i) The
- solubility of Al(OH)4 is decreased in presence of Si. (ii) Al is consumed to form aluminum-
156 silicates (e.g., zeolite) at the same rate as it is dissolving from Al(OH)3 powder. As discussed later in Figure 6.7, both mechanisms (i) and (ii) are unlikely. (iii) A protective aluminosilicate layer forms on the surface of Al(OH)3 powder which mitigates further dissolution. It has been suggested that silicate species can adsorb on the Al(OH)3 surface in consecutive layers and provide a physical barrier against out-diffusion of aluminum ions to the solution [26]. This requires further verification, but it is beyond the scope of the current study. The limited dissolution of Al in solutions containing Si has a practical significance in that simply adding solid Al compounds to concrete may not guarantee sufficient concentration of dissolved Al to mitigate silica dissolution.
The time-dependent variation of the OH- ion content of the corrosion cell solutions in the presence or absence of Al powder is compared in Figure 6.6. When Al is not present, [OH-] increases during the corrosion test. In this case, the soda-lime glass or the alkali-lime-silica
(ASR) gel produced as a result of glass dissolution and gelation, can undergo an ion-exchange reaction which releases OH- to the solution [27]:
( Si O Na) H 2O ( Si OH) Na OH (6-5)
157
1400 1 NaOH solution
1200 1 NaOH solution+Al(OH)3
1000
800
600 Concentration(mM) - OH 400
200
0 0-day 7-day 14-day
Figure 6.6: Variation in [OH-] during the glass corrosion test in 1M NaOH solution
with and without Al(OH)3 powder
The slight increase in the mass loss rate of glass slides during the corrosion test may be attributed
- to this effect (Figure 6.4Figure 6.7). In the presence of Al(OH)3, [OH ] remains constant during the experiment. Al(OH)3 dissolution may be expected to repress the pH of solution due to the consumption of OH- in eq. 6-3e. However, considering the limited solubility of Al (~50mM) in presence of glass slides, this would be a minor effect as Figure 6.6 suggests. As the pH is not increased in presence of Al, the corrosion rate of glass slides remain constant during the test.
158
6.4.2.3 Corrosion of glass slides in solutions containing finite [Al]
The previous experiment was performed in presence of an inexhaustible quantity (20g) of
Al(OH)3 powder added to the corrosion container. An important question is “what is the minimum dissolved Al content that is needed to significantly reduce the glass (SiO2) dissolution rate?”. This is important for proper selection of the type and dosage of the potential Al admixture to mitigate ASR. To answer, 340cc 1M NaOH solution with a temperature of 80°C was poured on 20g Al(OH)3 powder and after 3 days was passed through a 1m filter paper to remove the solid Al(OH)3 powder. The initial dissolved Al content of the solution was determined by ICP to be 89.3mM. The solution was then diluted with Al-free 1M NaOH at different levels to obtain different initial concentrations of dissolved Al. These solutions were then used to perform glass corrosion experiment similar to those described earlier.
Figure 6.7 compares the mass loss of glass slides at 14 days (at 80°C) for different initial dissolved Al contents. For comparison, the mass loss for the case with 20g Al(OH)3 powder added to the corrosion container is also included. In addition, the final (at 14 days) concentrations of dissolved Al are presented on a secondary axis on the right side of the graph.
The straight diagonal line shows the initial Al concentrations. The difference between the initial and final dissolved Al concentrations represents how much Al is consumed in the reaction with silica. It should be noted that no visible precipitate was observed at the bottom of the containers at the end of the test. As such, the majority of Si-Al reactions are expected to have occurred on the glass surface, leading to formation of an alumino-silica layer. This is further investigated in section 6.4.3.
159
200 90 Mass loss Al Consumption 180 Final Al content 80 Series3 160 )
2 70
140 Final Al concentration(mM) 60 120 50 100 40 80 30 60 20g Al(OH)3 powder added to the corrosion container 20
Mass Loss of glass slides at 14days(g/m 40
20 10
0 0 0 153045607590 Initial Al concentration(mM)
Figure 6.7: Mass loss of glass slides in 1M NaOH solution in presence of finite initial concentrations of Al
It is observed that with increasing the initial [Al] of 1M NaOH solution, the mass loss of glass slides decreases. An initial dissolved Al content of ~50mM is sufficient to reduce the mass loss
2 2 from 170.53g/m to 12.37g/m ; similar to the case when ample Al(OH)3 powder was present.
This is in agreement with dissolved [Al]≈50mM in the latter case in Figure 6.5. The initial drop in the mass loss curve is also of importance as the presence of only ~5mM Al yields nearly 50% drop in the final mass loss of glass slides. In a previous study [28], presence of 10mM of
- Al(OH)4 decreased the dissolution rate of quartz by 87% at pH=11.3. The explanation provided
160
- in [28] was that Al(OH)4 poisoned quartz surface and reduced the dissolution rate. This is further discussed in section 6.5.
Final Al contents of the solution (Figure 6.7) show that Al consumption due to reaction with silica has been small. When initial Al content was 89.3mM, after 14 days it drops to 85.0mM
(loss of 4.3mM). The amount of Al that has been removed from the solution in the next three data points is in the range 0.7 to 2mM. Overall, relatively small amount of Al has been consumed from the solutions in presence of glass slides, even when mass loss of glass is dramatically depressed. Al consumption can be estimated using mass balance calculations. From SEM micrographs, it is observed that the thickness of the aluminosilicate layer on the glass surface in presence of 20gr Al(OH)3 powder is approximately 6-7m (Figure 6.9). A similar thickness is assumed for the aluminosilicate layer for the corrosion test with initial [Al]=89.3mM. Given that
Al constitutes 18% of the weight of this layer (EDS results of Table 6.2) an estimate of the Al uptake from the solution can be made to be 2.5mM. This is based on the assumption that the reacted layer has a uniform thickness and the Na content of the reacted layer does not vary from one point to the other. The actual consumed value based on experiment is obtained to be 4.3mM.
Considering the available surface area of glass slides (0.0079m2) the Al uptake is 8.5g/m2 and
14.7g/m2 for theoretical calculation based on the thickness of the layer and actually experimentally measured cases, respectively.
The results of Figure 6.7 also shed light on the effect of Si on the dissolution rate of Al(OH)3 previously discussed in Figure 6.5. It was observed that [Al] of the solution in the presence of glass slides and ample (20g) source of solid Al(OH)3, plateaued at 45 to ~50mM after 4 days of
161 the test and three potential mechanisms that may have caused this effect were mentioned. In the test described in Figure 6.7, the initial dissolved [Al] (89.3mM) is only reduced to 85.0mM after
14 days exposure to glass slides. If the presence of dissolved Si had decreased the
- thermodynamic solubility of Al(OH)4 , as suggested by mechanism (i), it would be expected that the concentration of Al would drop to ~50mM during the test. Also, the low consumption of Al in this test compared with dissolution rate of Al(OH)3 (Figure 6.3), suggests that mechanism (ii) is also unlikely. If Al(OH)3 constantly dissolves and reacts with silica at the same rate, the pH must continuously drop according to equations (6-3e) and (6-4). This does not occur as presented in Figure 6.6.
6.4.3 Surface characterization of the corroded glass slides
When soda-lime glass slides are immersed in Al-free 1M NaOH solution, a layer of silica gel precipitates on the surface of slides. This gelation is likely catalyzed [29] by the presence of calcium which is contributed by the soda-lime glass (Table 6.1). In the absence of Al, this gel layer is soft and detaches easily from the glass slide by washing. After this gel is removed, the substrate glass is transparent and resembles the surface of original glass slide before exposure to
NaOH (Figure 6.8). However, when the glass slides are treated in 1M NaOH solution in presence of Al(OH)3, a translucent layer forms at the surface. This layer is not readily removable and requires using grinding papers. This layer forms at different levels of translucency when glass slides are treated in solutions with different finite Al contents.
The surface layer was further characterized with the aid of SEM/EDS and XRD techniques.
Glass slides treated in 1M NaOH+Al(OH)3 powder solution for 14 days were split in two by 162 hand a5.9nd the exposed cross section was carbon-coated and analyzed with SEM/EDS. A high resolution SEM micrograph with the corresponding EDS maps are shown in Figure 6.9. A layer with a thickness of ~6-7m has formed on the surface of glass slides. Some crystal traces are observed in the micrograph which motivated XRD characterization. In comparison with the substrate glass, EDS maps show a considerable higher concentration of Al and lower concentrations of Si and Na in this surface layer.
Treated in 1N NaOH + Al(OH) Treated in 1N NaOH 3
Figure 6.8: Formation of a translucent alumino-silicate layer at the surface of glass slides exposed to 1M NaOH in presence of Al
Table 6.2 presents the average elemental composition of the surface layer and the substrate glass slide based on EDS analysis of 15 spots from each phase. The average atomic Al weight of the layer is nearly 17.9% which is much higher comparing to the Al content of the glass slides
(0.9%). The concentration of Si and Na elements in the surface layer is lower comparing to the
163 glass slides. Also, the surface layer contains very little Ca. The ratio of Al:Si is close to 1 which further helps during the XRD characterization. EDS mapping suggests that the reaction in presence of Al in the solution is not an ion exchange reaction which usually occurs for divalent elements such as Ca or Mg. Both Na and Si concentrations are lower in the surface layer which indicates congruent dissolution of glass rather than an ion exchange reaction. In the case of ion- exchange, alkali depletion without the reduction in Si content of the original glass would be observed.
Figure 6.10 shows the XRD results for the surface layer. The crystal peaks correspond mainly to two types of zeolites. Some minor traces of phillipsite, concrinite and even tobermorite also do exist. In addition, an amorphous hump in the range 2θ=15º to 35º is observed, which is believed to correspond to alumino-silicate glass. An unidentified phase is present which is speculated to be also from zeolite family. Considering the fact that EDS analysis indicates an Al:Si=1 for the surface layer, it is believed that this layer mostly contains zeolite with the formula:
Na92.71(Si96.96Al96.01O384)(H2O)254.64 as well as Al-Si glass.
Table 6.2: Atomic weight % of the elements from the Al-Si layer and glass determined by EDS
Si Al Na Ca K O Substrate glass 34.3 0.9 10.3 4.3 0.9 48.3 Al-Si surface Layer 18.5 17.9 5.9 0.6 49.4
164
Na
~6m
Si Al
Figure 6.9: SEM/EDS mapping of Al, Si and Na elements in the glass slides
surface reaction product in presence of Al(OH)3
165
9000
8000 Zeolite: Na92.71(Si96.96Al96.01O384)(H2O)254.64 Zeolite: Na3.552(Si3.6Al12.4O32)(H2O)10.666 Phillipsite-Ca: KCa(Si5Al3)O16.6H2O 7000 Z Cancrinite: (NaCaAl)(SiAl)O(CO).3H2O Z 6000 Z Z Z 5000 Z Z 4000 Z Z Intensity (Counts) Intensity Z Z Z 3000 Z Z ZZ Z 2000 Z Z Z
1000
0 5 101520253035404550 2-º
Figure 6.10: XRD patterns of the aluminosilicate surface layer formed on glass slide surface treated in 1M NaOH in presence of Al (Appendix E)
166
400 Treated in 1M NaOH for the entire experiment Both treated in 1M NaOH solution 350 Treated for 14 days at 1M NaOH+Al and sebsequently at 1M NaOH without Al"
300 Both taken out from the solutions and submerged 250 in a new1M NaOH
/g)) solution 2
200
Treated in 1M Mass Loss(m 150 NaOH solution
100 Treated in 1M NaOH solution in presence of Al 50
0 0 4 8 1216202428 Days
Figure 6.11: Comparison of the corrosion rate of glass slides in the presence or absence of dissolved Al in the solution
Formation of this aluminosilicate surface layer may be the reason that the corrosion rate of glass slide is significantly depressed in NaOH solutions containing dissolved Al. The aluminosilicate layer is presumed to be irreversibly sorbed onto the silicate (glass) surface and produce a more stable, less soluble phase which depresses the glass dissolution [30]. According to Iler [31], silica samples treated with aluminum compounds were found to be completely insoluble in a pH of 8.3 at 25°C. In this study, to investigate the stability of the aluminosilicate layer glass slides were treated for 14 days in 1M NaOH solution at 80ºC in presence of 20g Al(OH)3 powder. These
167 slides corroded very slowly and formed a 6~7m thick aluminosilicate surface layer by 14 days, as discussed before. Thereafter, the slides were moved into a fresh 1M NaOH solution free of Al and their mass loss was monitored for additional 14 days (Figure 6.11). After this treatment, the surface of the glass slides was transparent indicating removal of the Al-Si layer. There is a slight reduction in the rate of mass loss during the first 3 days, which indicates a minor barrier effect of this layer against corrosion. However, this layer is not stable in such medium and disappears after 3 days confirmed by visual inspection. Thereafter, the rate of glass dissolution during the remained period was similar to slides which were never exposed to Al. These observations could suggest that: (1) the Al-Si layer is not stable and dissolves in 1M NaOH solution in the absence of Al; (2) The presence of this layer has a small effect on the rate of silicate glass dissolution this layer may have no major retarding effect on the rate of glass dissolution as long as dissolved Al is not present in the solution. Looking back at Figure 6.4, it is observed that the corrosion rate of glass in the solution containing Al, remains practically constant during the experiment, even though the thickness of aluminosilicate layer increases with time. If the Al-Si layer provides a barrier action, it is expected that the corrosion rate of glass should slow down with time.
6.5 Discussion
Based on review of earlier research on the effect of Al on aqueous corrosion of silicates
[25,31,32], two potential mechanisms could be identified through which Al may impact the dissolution of silica glass in alkaline environments. (1) Al reacts with the silicate glass and form an aluminosilicate layer on its surface which may be more stable (i.e., corrosion resistant) and capable of protecting the substrate silica glass against further attack of hydroxyl ions [30]. 168
(2) Presence of Al in the solution, “passivates” or “poisons” the surface of silica glass through
- surface adsorption of the Al(OH)4 ions according to equation 6-1. The negatively charged aluminosilicate surface sites thus formed, repel OH- ions and thus depress their attack on the silica structure [31]. However, when the concentration of positive alkali ions (e.g, Na+) are high in the solution, they could simultaneously co-adsorb on the surface of silica and neutralize the
- negative charge of Al(OH)4 (eq. 6-2) [33].
( Si OH) (Al(OH)4 ) ( Si O Al(OH)3 ) H 2O (6-1)
( SiOH Al(OH)4 ) (Na ) ( Si O Al(OH)3 Na ) H 2O (6-2)
Iler [31] speculated that each single negatively charged aluminosilicate site could cover a maximum surrounding area of silica, enough to repel hydroxyl ions and protect the substrate.
Based on this theory, a minimum dissolved aluminum content of 0.5mM in the solution could minimize the dissolution rate of glass slides. However, considering the fact that relatively harsh conditions applied in this study (pH=14,80°C vs. pH=8,25°C), a much higher Al content
(50mM) was required to minimize the dissolution rate of glass slides. This may be partially due to the fact that the solubility of Al drastically increases by variation of pH from 8 to 14. In the conditions imposed, a semi crystalline layer formed on the glass surface which mostly constitutes from zeolite family. Its Al content was calculated to be 4 times of that Iler’s theory suggests minimizing the dissolution rate. Based on the observations of the current study, this layer could possibly have minor effect on repressing the glass dissolution rate in the considered pH level. If
169 it had, a change in the rate of glass dissolution should be observed (Figure 6.4). This was seen before for initially Al treated silica exposed to moderate alkaline solutions [31]. In addition, the rate of mass loss should decrease as the layer thickens over time and its barrier action is expected to improve. Finally, such a layer with different levels of translucency forms on the surface of glass slides treated in solutions with different initial dissolved Al. This layer is speculated to be just an inferior byproduct. While sufficient initially dissolved Al is available to produce a similar layer in most of the solutions, they mostly undergo mass loss, unless the Al content is beyond
50mM. For the glass slides with minimum dissolution rate, the lost thickness calculated based on mass loss data (6m), is very close to the thickness of the layer formed on glass layer which is observed in SEM micrograph (7m). Based on the results, it is speculated that a sufficient content of Al is required to poison the surface enough to propel the reaction to promote the production of aluminosilicate layer rather than dissolving the glass in a much higher rate. Even if the surface is poisoned, the presence of sufficient Al in the solution is an essential factor to maintain the dissolution rate in its minimum range. Bickmore et al. [28] tried to establish an
- empirical model of the effect of Al(OH)4 on the dissolution rate of quartz. They postulated that if the quartz’ surface is 100% covered with aluminum no dissolution would occur. Whether or not there is a link between the sufficient dissolved Al in the solution and 100% Al cover suggested by Bickmore should be more investigated. Overall, it is speculated that for a specific area of soda-lime glass, a minimum content of dissolved Al in the solution is an essential factor to minimize the rate of glass dissolution.
6.6 Conclusion
170
In this study, the effect of using amorphous aluminum hydroxide powder on the ASR risk of soda-lime glass aggregates was evaluated. Based on the experimental observations, the following conclusions can be made:
For ASTM C1260 mortars containing soda-lime glass aggregates, replacing 20% of portland
cement with Al(OH)3 powder led to significant reductions in the ASR expansions from 0.6%
to 0.03%.
The presence of dissolved aluminum in the highly alkaline solution significantly represses
the corrosion rate of soda-lime glass (model reactive aggregate used in this work). This may
benefit towards developing additives which could be able to reduce the dissolution rate of
metastable silicate phases of aggregates and thus help with ASR mitigation.
SEM imaging reveals formation of a layer on the surface of soda-lime glass slides during the
corrosion test. XRD characterization revealed that this layer is semi-crystalline composed
mostly of zeolite families.
The protective barrier action of such a layer in reducing the corrosion rate of glass is less
certain. This layer is not stable in high alkaline solutions unless a minimum dissolved Al is
present.
Based on the observations of this study, approximately a minimum of 170g dissolved Al is
required to minimize the dissolution rate of 1m2 of soda-lime glass surface.
The addition of Al(OH)3 has not a major impact on the pH of the solution.
The dissolution rate of Al(OH)3 decreases in the presence of dissolved Si.
171
6.7 References
1. R.N. Swamy, The Alkali-Silica Reaction in Concrete, Routledge, 1992.
2. D.W. Hobbs, Alkali-Silica Reaction in Concrete, Thomas Telford, London. 1988.
3. M. D. A. Thomas, D. B. Stokes, Use of a Lithium-Bearing Admixture To Suppress Expansion in Concrete Due to Alkali-Silica Reaction, Journal of the Transportation Research Board, 1999, 1668, 54-59.
4. K. C. Natesaiyer and K. C. Hover, chemical agents for reducing solubility of silica in 1 N sodium hydroxide, Cement and Concrete Research, 1992, 22, pp. 653-662.
5. Y. Oka and M. Tomozawa, Effect of alkaline-earth ion as an inhibitor to alkaline attack on silica glass, Journal of Non-Crystalline Solids, 1980, 42(1-3), 535-543.
6. R. K. Iler,The chemistry of silica: solubility, polymerization, colloid and surface properties, and biochemistry, WILEY, 1985.
7. W. Aquino, D. A. Thomas, J. Olek, The influence of metakaolin and silica fume on the chemistry of alkali-silica reaction products, Cement & Concrete Composites, 2001, 23, 485-493.
8. F. Naiqian and H. Tingyu, Mechanism of natural zeolite powder in preventing alkali- silica reaction in concrete, Advances in cement research, 1998, 10(3), 101-108.
9. S. M.H. Shafaatian, A. Akhavan, H. Maraghechi, and F. Rajabipour, How Does Fly Ash Mitigate Alkali-Silica Reaction (ASR) in Accelerated Mortar Bar Test (ASTM C1567)?, Submitted to Cement and Concrete composites, May(2012).
10. T.Ramlochan, M. Thomas, K. A. Gruber, The effect of metakaolin on alkali-silica reaction in concrete, Cement and Concrete Research, 2000,30, 339-344.
11. N. Quanlin and F. Naiqian, Effect of modified zeolite on the expansion of alkaline silica reaction, Cement and Concrete research, 2005, 35, 1784-1788.
12. S.Y. Hong, F.P. Glasser, Alkali binding in cement pastes: Part I. The C-S-H phase, Cement and Concrete Research, 1999, 29(12), 1893-1903.
172
13. L. J. Malvar, L. R. Lenke, efficiency of fly ash in mitigating alkali silica reaction based on chemical composition, ACI Materials Journal, 2006, 103(5), 319-326. 14. http://lmc.epfl.ch/files/content/sites/lmc/files/6%20-%20Scrivener%20- %20EPFL%20studies%20on%20ASR.pdf.
15. S.Y. Hong, F.P.Glasser, Alkali sorption by C-S-H and C-A-S-H gels: Part II, Role of alumina, Cement and Concrete Research, 2002, 32(7), 1101-1111.
16. C. Yamanaka, T. Yokoyama and T. Tarutani, Retarding effect of aluminum on the polymerization of silicic acid particles, Journal of chromatography, 1987, 367, 419-422.
17. M.D.A. Thomas, The effect of supplementary cementing materials on alkali-silica reaction: A review, Cement and Concrete Research, 2011, 41, 209-216.
18. T. Chappex and Karen Scrivener, Controlling alkali-silica reaction by understanding the contribution of aluminum provided by supplementary cementitious materials, ICAAR, Austin, 2012.
19. W. Stumm, J.J. Morgan, Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters, 1995,Wiley-Interscience, New York.
20. R. J. Moolenaar, J. C. Evans and L. D. McKeever, The structure of the aluminate ion in solutions at high pH, The journal of physical chemistry, 1970, 74(20), 3629-3636.
21. D.J. Wesolowski, Aluminum speciation and equilibria in aqueous solution: I. The solubility of gibbsite in the system NaK-Cl-OH-A1(OH)4 from 0–100°C,1992, Geochim. Cosmochim. Acta, 56, 1065–1092.
22. K. H. Gayer, L. C. Thompson, and O. T. Zajicek, The solubility of aluminum hydroxide in acidic and basic media at 25°C, Canadian journal of chemistry, 1958, 36,1267-1271.
23. H. A. Gasteiger, W. J. Frederick and R. C. Streisel, Solubility of aluminosilicates in alkaline solutions and a thermodynamic equilibrium model, Industrial & Engineering Chemistry Research, 1992, 31, 1183-1190.
24. P. Lu, Q. Li, J. Zhai, Mineralogical Characterizations and Reaction Path Modeling of the Pozzolanic Reaction of Fly Ash–Lime Systems, Journal of American Ceramics society, 2008, 91(3), 955-964.
173
25. S. Kvech and M. Edwards, Solubility controls of aluminum in drinking water at relatively low and high pH, Water Research, 2002, 36, 4356-4368.
26. F.J. Hingston and M. Raupach, The reaction between monosilicic acid and aluminium hydroxide. I. Kinetics of adsorption of silicic acid by aluminium hydroxide, Australian Journal of Soil Research, 1967, 5(2), 295 – 309.
27. L. L. Hench and D. E. Clark, physical chemistry of glass surfaces, journal of non- crystalline solids, 1978, 28, 83-105.
28. B. R. Bickmore, K. L. Nagy, A. K. Gray and A. R. Brinkerhoff, The effect of Al(OH)4- on the dissolution rate of quartz, Geochimica et Cosmochimica, 2006, 70, 290-305.
29. A. Leemann, G. Le Saout, F. Winnefeld, D. Rentsch, and B. Lothenbach, Alkali-silica reaction: The influence of calcium on silica dissolution and the formation of reaction products, Journal of American ceramic society, 2010, 94(4), pp. 1243-1249.
30. J. C. Sang, R. F. Jakubic, A. Barkatt and E. E. Saad, The interaction of solutes with silicate glass and its effect on dissolution rates, Journal of non-crystalline solids, 1994, 167, 158-171. 31. R. K. Iler, Effect of adsorbed alumina on the solubility of amorphous silica in water, Journal of colloids and interface science, 1973, 43(2), 399-408.
32. E. A. Pfannkoch and B. S. Switzer, Aluminum ion mediated stabilization of silica-based anion-exchange packings to caustic regenerants, Journal of chromatography, 1990, 503, 385-401.
33. B. R. Bickmore, K. L. Nagy, A. K. Gray and A. R. Brinkerhoff, The effect of Al(OH)4- on the dissolution rate of quartz, Geochimica et Cosmochimica, 2006, 70, 290-305.
174
CHAPTER
SEVEN
SUMMARY, CONCLUSIONS AND FUTURE DIRECTIONS
7.1 Summary
Recycled glass can be potentially used in concrete with many economical and environmental benefits. However, the main obstacle is the alkali-silica reaction (ASR). It is a deleterious reaction that occurs between the meta-stable silicate phase of aggregates and hydroxyl ions present in the pore solution of concrete. The reaction product is a hygroscopic gel which swells and cracks the concrete. The goal of this research was to address some of the challenges associated with using recycled glass in concrete. Specifically, the following three main objectives were pursued: (1) this study attempted to gain a better understanding of the mechanisms involved in ASR mitigation of fly ashes. This was accomplished by an extensive literature review of the fly ash effect against ASR. Based on that, six possible mechanisms were identified through which fly ash may mitigate ASR. A systematic research approach was
175 designed and implemented to investigate validity and contributions of these mechanisms. (2)
This study investigated the feasibility of using soda-lime glass powder as a supplementary cementitious material (SCM) to mitigate ASR when reactive aggregates are used in concrete.
Glass powder has high amorphous silicate content and can potentially be used as a pozzolanic material. However, unlike most pozzolans, glass powder also has a high alkali content which may counteract its pozzolanic effect against ASR. To address this concern, first the ability of glass powder in controlling ASR was confirmed using ASTM C1567 test. Then, a research plan utilizing a series of materials characterization techniques was designed and executed to investigate how application of glass powder impacts and mitigates ASR. (3) The last objective of this research was to investigate the potential for using soluble aluminum as a new material to mitigate ASR. The feasibility of using aluminum hydroxide powder as cement replacement was assessed with ASTM C1260 test. Thereafter, the repressing effect of Al on the corrosion rate of silicate glass as a potential mechanism for ASR mitigation was examined.
7.2 Conclusions of the research on the role of fly ash in ASR mitigation
The following observations were made by investigating the mechanisms of ASR mitigation of fly ash in ASTM C1567 test environment:
Fly ash reduces the alkalinity ([OH-]) of pore solution by significantly reducing the ion
diffusion coefficient of mortars. A diffusivity reduction by a factor of 4 to 7 was recorded, as
early as 48 hours after casting, when sufficient dosage of fly ash replaced portland cement.
176
As such, the external NaOH penetrates slower into fly ash mortars, resulting in a lower pore
fluid alkalinity, and significantly slower ASR.
Fly ash reduces the alkalinity ([OH-]) of pore solution through alkali binding. Fly ash reduces
the CaO/SiO2 (abbreviated as C/S) of the C-S-H gel which in turn, improves its alkali
binding capacity. In addition, more C-S-H is produced by pozzolanic reactions. As such, a
considerable fraction of the penetrated NaOH is removed from the pore solution. The results
of a simple numerical model suggested that the contribution of transport reduction is more
significant than the effect of improved alkali binding, when fly ash is used.
Fly ash increases the tensile strength of mortars and prevents or delays the onset of cracking.
This also prevents an accelerated transport of NaOH through cracks to the interior of mortar
specimens.
Fly ash can reduce the dissolution rate of siliceous aggregates even when the alkalinity of
pore solution is maintained constant at pH=14 (e.g., near the perimeter of mortar prisms that
are exposed to external NaOH bath). Fly ash provides a large silicate surface area that is
accessible to the corrosive OH- ions. As such, the concentration of OH- per unit surface area
of silicate is markedly reduced. In other words, for a unit volume of pore solution at a given
pH, a significant fraction of hydroxyl ions are involved in dissolving fly ash instead of
attacking the reactive aggregates.
The results of this work also suggest that alkali dilution and modifying ASR gel composition
are not major contributors to mitigation of ASR by fly ash in ASTM C1567 test.
177
Application of fly ash does not significantly affect the electrical conductivity of the pore
solution of mortar prisms in ASTM C1567. However, they significantly reduce the electrical
conductivity of the mortar bars.
The pastes prepared with minimum required dosage to control ASR expansion of mortar
bars; cured in the similar conditions; show a higher porosity comparing to 100%PC pastes at
all ages during the test period. An initially higher porosity was observed for mortars with
proper content of fly ash. No major change occurred in their porosity during the test period.
In contrast, the porosity of 100%PC mortar bars steadily increased due to ASR cracking.
After 3 days exposure to the alkaline bath, it surpassed the porosity of fly ash mortars.
A content of fly ash was able to mitigate ASR expansion that reduced the ion diffusion
coefficient below a certain value. In the present study, fly ash contents that were able to
reduce mortar diffusivity by a factor of 3 to 7 had the ability to control ASR.
Partial replacement of portland cement with fly ash leads to a significant reduction in the
average pore size of the binder phase of the mortars. The pore size distribution curve has an
overall shift to the smaller sizes. The lower ion diffusion coefficient of fly ash mortars can be
linked to this lower average pore size.
An interesting observation was made based on the results of ASTM C1567 testing,
suggesting that cement-fly ash binders containing less than approximately 60%
CaOeq=CaO+Na2O+K2O+MgO+MnO may be able to mitigate ASR expansions induced by
recycled glass aggregate.
178
7.3 Conclusions of the research on the beneficial effect of glass powder against ASR
The following observations were made by investigating the beneficial effect of glass powder in suppressing ASR in ASTM C1567 test environment:
Glass powder can mitigate ASR in the accelerated mortar bar test. Finer glass powders are
more efficient, meaning that their required dosage to control ASR decreases by reducing the
particle size of the powder.
Glass powder is very effective in reducing ion diffusion coefficient of mortars, at least for
conditions imposed by ASTM C1567 test. Up to 14× reduction in the diffusivity of 40%GP2
mortar was observed in comparison with the control 100%PC mortar. The findings suggest
that glass powder has a great impact on the ASR mitigation by reducing ion transport.
Pore fluid analysis showed a significant reduction in the [OH-] concentration of 40%GP2
mortar during the test period which is consistent with low ASR expansion in this mortar.
This reduction in [OH-] is primarily due to lower diffusion coefficient of 40%GP2 mortars.
Preliminary results suggest that although glass powder releases Na to the pore solution, this
is not necessarily in the ionic form (Na+) which increases the alkalinity and promotes ASR.
Addition of glass powder has a considerable impact on compressive and tensile strengths of
mortars. The improved resistance against cracking also mitigates transport through cracks.
179
In presence of glass powder, the effective ratio of OH- to accessible silica surface area
dramatically decreases. As a result, glass aggregates are being attacked at a lower rate. This
contributes to ASR mitigation.
7.4 Conclusions of the study on the feasibility of using Al(OH)3 as an ASR suppressor
The following observations were made based on the investigation of the feasibility assessment of using Al(OH)3 as an ASR controller:
Replacing 20% of portland cement with Al(OH)3 powder led to ASR expansions (0.03%)
well below the ASTM C1260 threshold (0.1%).
The presence of dissolved aluminum significantly represses the corrosion rate of soda-lime
glass (model reactive aggregate used in this work), which could be a major contribution to
ASR mitigation.
Two different mechanisms have been identified that could explain the cause for reducing the
dissolution rate of silica at high pH in the presence of dissolved aluminum: (1) formation of
an aluminosilicate barrier layer on the silica surface (2) passivation of the silica surface due
to Al-poisoning. SEM imaging reveals formation of a semi-crystalline zeolite layer on the
surface of soda-lime glass slides during the corrosion test. The barrier action of such a layer
in reducing the corrosion rate of glass is less certain. Based on the observations of this study,
it is speculated that the second mechanism is probably more dominant.
180
It is believed that for a specific area of soda-lime glass (79cm2 in 340 CC 1M NaOH), a
minimum dissolved Al content (50mM) must be present in the solution to minimize the
dissolution rate of glass.
7.5 Suggestions for future research
ASR investigation by application of recycled glass cullet as fine reactive aggregate has the benefit of studying ASR using a homogeneous and isotropic aggregate, unlike natural reactive aggregates which are heterogenous. Soda-lime glass provides an opportunity to investigate the long-term effects of alkalis and lime contributed by the aggregates on the ASR performance of concrete. In addition, to the accelerated mortar bar test (ASTM C1567), utilized in the current study, it is beneficial to investigate ASR induced by recycled glass and the efficiency of fly ashes, glass powder, and Al compounds using the long-term ASTM C1293 test. In addition, it may be beneficial to prepare large concrete blocks containing glass aggregates and expose them to real service life outdoor exposure conditions to monitor the progress rate of ASR and the efficiency of the proposed mitigation mechanisms.
There is still a lack in understanding of the long-term mechanism(s) through which glass powder is able to mitigate ASR. A similar methodology to that established in this research, may be implemented to understand the beneficial effect of glass powder in ASTM C1293 test. Specially, the problem of alkali contribution by aggregates must be further investigated.
181
The aluminum hydroxide powder was shown to control ASR expansion in mortar bars exposed to ASTM C1567 test conditions. The repressing effect of aluminum in the alkaline solution on the dissolution rate of glass was studied in detail. However, the pore solution of concrete is more complex. Different ions and elements in various forms are generally present in the pore solution which may affect the solubility and thus repressing effect of the aluminum on glass dissolution.
Specifically, simultaneous interactions between Ca, Al, and Si deserve further investigation. The mechanisms involved in suppressing ASR expansion in ASTM C1567, when Al(OH)3 is used as admixture, need to be addressed. In addition, the effect of aluminum on the strength and other properties of fresh and hardened concrete should be considered. Finally, feasibility of using other
Al salts in reducing ASR may also be investigated. This is specifically important if their solubility limit and dissolution rate are higher comparing to aluminum hydroxide.
182
APPENDICES
183
Appendix A: Calculation of the sink term
The sink term is calculated as following:
mM ( C S H ) [Na] g R Bound (1) d [Na] mM Solution ( Solution) g
Introducing absorption and C-S-H content of the paste as:
Absorption:
m g Abs solution (2) mtotal g
Mass of C-S-H produced due to cement hydration:
mCSH g M CSH (3) mtotal g
Where mCS H is the actual mass of C-S-H produced from a unit total mass of cement. thus:
msolution msolution Abs M CS H (4) mCS H mCS H
M CS H
Combining equations (1), (2) and (3) results in:
R [Na] m [Na] (mM ) d solution solution (5) bound Abs mCS H M CS H
184
In the next step, the alkali binding effect was combined with the diffusivity of the mortars:
M M C (x,t) C (x,t) C (x,t)R CS H C (x,t 1) CS H (6) a b a d Abs s Abs
After rearrangement:
M CS H Cb (x,t) C s (x,t t) C (x,t) Abs (7) a M 1 C (x,t)R CS H a d Abs
Cs: includes the diffusion of alkali from the external bath at previous time step
Ca: the concentration of alkali after including the binding effect (sink term)
Cb: the concentration of alkali before including the binding effect (sink term)
This has been introduced as the sink term simulating alkali-binding in the MATLAB code presented in appendix B.
185
Appendix B: MATLAB code implementing the diffusion and sink term as alkali binding clc;clear; dt=0.01;dx=0.5; nt=14/dt+1;nx=25/dx+1; initialBC=1 initialC=0.23 cb=zeros(nt,nx);cb(1,1:nx)=initialC;cb(1:nt,1)=initialBC;cb(1:nt,nx)=initialB C; ca=zeros(nt,nx);ca(1,1:nx)=initialC;ca(1:nt,1)=initialBC;ca(1:nt,nx)=initialB C; d0=2.343e-9;MCSH=0.5;ABS=0.25;fibeta0=0.0103;Rd=0.34; for i=1:nt-1 for j=2:nx-1
fb=fibeta0; diffusion= fb*d0;
if j else a1=cb(i,j-1); a2=cb(i,j); cb(i+1,j)= cb(i,j)+coeff *(initialBC-2*a2+a1); ca(i+1,j)= (cb(i+1,j)+Rd*ca(i,j)*MCSH/ABS)/(1+Rd*MCSH/ABS); end cb(i+1,j)= ca(i+1,j); end end xc=[0:dx:nx*dx-dx]; xc'; size(xc); yc=cb(nt:nt,:); yc'; size(yc); plot(xc,yc, '.-r') ylim([0 1]) 186 Appendix C: Glossary ASR: Alkali-Silica-Reaction CH: Portlandite, Ca(OH)2 EDS: Energy Dispersive x-ray Spectroscopy EIS: Electricsl Impedance Spectroscopy GP: Glass Powder MIP: Mercury Intrusion Porosimetry OPC: Ordinary Portland Cement PC: Portland Cement PSD: Particle Size Distribution PSD: Pore Size Distribution SCM: Supplementary Cementitious Materials SEM: Scanning Electron Microscopy TGA: Thermo-Gravimetric Analysis XPS: X-ray Photoelectron Spectroscopy XRD: X-Ray Diffraction 187 Appendix D: Procedure of standard deviation calculation of element concentrations in the pore solution Pore solution analysis provides valuable information regarding the variation of different elements in the pore liquid of the mortars. This test has been extensively used in this dissertation. However, this experiment is tedious and requires extensive time and labor-work. As such, in order to gain reliable data, the following procedure was established to include different sources of errors that could potentially affect the results. These were further extended to all the pore solution analysis results: 1) Pore solution from one mortar sample will not be identical to the pore solution from a duplicate sample. 2) Pore solutions may get contaminated or carbonated during handling and this may be different for different samples. 3) The ICP equipment and operator (especially for acid titration) may generate some variability/errors. To address this concern, mortar prisms were prepared from 100% portland cement and glass aggregates and cured according to the ASTM C1567 environment. The pore solution of three duplicate samples (from each batch) was extracted immediately after mixing (fresh mortar) and after 3 days exposure to the 1M NaOH bath (hardened mortars). The pH of these solutions was measured by HCl acid titration to account for variations which could possibly occur in measurements of [OH-] from duplicate samples from the same mortar. The concentration of other elements was also obtained with ICP to include the variations which could be originated from 188 ICP measurements from duplicate samples of the same batch. The pH and OH- concentrations are provided in Table 1 obtained via acid titration: Appendix D-Table 1: Standard deviation in acid titration measurement Sample ID V base (mL) H+ V acid (mL) OH- StDev Mix1 0.5 0.1 3.075 0.61 Mix2 0.5 0.1 3.000 0.60 0.010 Mix3 0.5 0.1 2.950 0.59 Mortar1 0.5 0.1 0.8 0.16 Mortar2 0.5 0.1 0.85 0.16 0.004 Mortar3 0.5 0.1 0.8 0.16 The standard deviation of the concentration of different elements present in the pore solution of duplicate samples from the same batch, immediately after mixing and after 3 days exposure to 1M NaOH bath, is depicted in Table 2: Appendix D-Table 2: Standard deviation in ICP measurement Al Ca Fe K Mg Na S Si Sample (mM) (mM) (mM) (mM) (mM) (mM) (mM) (mM) Mix1 0.014 13.8 0.0 397.0 0.007 131.5 204.2 0.48 Mix2 0.018 14.0 0.0 398.4 0.014 144.2 206.0 0.72 Mix3 0.023 13.0 0.0 406.0 0.013 133.5 208.2 0.85 StDev of fresh mortar 0.004 0.419 0.000 3.971 0.003 5.599 1.658 0.151 COV 0.194 0.031 1.801 0.010 0.260 0.041 0.008 0.220 Mortar1 0.053 1.50 0.000 247.8 0.012 793.0 238.3 2.04 Mortar2 0.153 1.18 0.001 243.6 0.014 723.7 222.9 1.45 Mortar3 0.147 1.29 0.001 254.9 0.009 798.1 232.5 1.41 StDev of hardened 0.046 0.132 0.000 4.670 0.002 33.919 6.378 0.290 mortar COV 0.387 0.100 0.220 0.019 0.174 0.044 0.028 0.177 189 Appendix E: Detailed X-ray pattern of Aluminosilicate layer 190 Seyed M. H. Shafaatian Curriculum Vita Email: [email protected], Tel: (814) 404-6840 206 Cunningham Hall, University Park, 16802, PA Education Doctor of Philosophy (PhD), Civil and environmental Engineering department, The Pennsylvania State University (December 2012) Dissertation title: Developing innovative methods to mitigate alkali-silica reaction (ASR) in concrete Master of Science (MSc), Civil and environmental Engineering department Sharif university of Technology, Tehran, Iran (December 2002) Thesis subject: Feasibility of using active control in retrofitting of concrete structures Bachelor of Science (BSc), Civil and environmental Engineering department Shiraz University, Shiraz, Iran, May 2000 Publications A. Akhavan, S. M. H. Shafaatian, F. Rajabipour, Quantifying the effects of crack width, tortuosity, and roughness on water permeability of cracked mortars, Cement and Concrete Research 42 (2012) 313–320 H Maraghechi, S. M. H. Shafaatian, G Fischer, F. Rajabipour, The role of residual cracks on alkali silica reactivity of recycled glass aggregates, Cement and Concrete Composites, 34(2012) 41-47 Ooi, P.S.K., Rajabipour, F., Shafaatian, S.M.H., and Joo, S., Forensic Investigation of a Distressed Pavement Supported on a Base Course Containing Recycled Concrete Aggregate, Transportation Research Record, (2011), 2253, 22-31 S.M.H. Shafaatian, A. Akhavan, H. Maraghechi, and F. Rajabipour2012, How Does Fly Ash Mitigate Alkali-Silica Reaction (ASR) in Accelerated Mortar Bar Test (ASTM C1567)?, submitted to Cement and Concrete Composites, May (2012) S.M.H. Shafaatian, H. Maraghechi, and F. Rajabipour, Assessing the Role of Ion Transport in Mitigation of ASR by Fly Ash in ASTM C1567, Manuscripts in preparation S.M.H. Shafaatian, H. Maraghechi, and F. Rajabipour2012, Investigating ASR mitigating mechanisms of glass powder in ASTM C1567 accelerated test, Manuscripts in preparation S.M.H. Shafaatian, H. Maraghechi, and F. Rajabipour, Aluminum hydroxide: a potential replacement to mitigate alkali silica reaction: effect on dissolution rate of soda-lime glass, Manuscripts in preparation Teaching and Research experience 2009-Present, Teaching and research assistant, Civil Eng. Dep. PennState University, 2008-2009, Research assistant, Civil Eng. Dep. University of Hawaii,