INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from aiy type of computer printer.

The quality of this reproduction is dqiendent upon the qnality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard marging, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note wiH indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, begiiming at the upper left-hand comer and contimiing from left to light in equal sections with «mall overk^ Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscrit have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for zny pbot<^rsq}hs or illustrations appearing in this copy for an additional charge. Contact UMI directfy to order.

UMI A Bell & Howell Information Company 300 North Zeeb Poad. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800/521-0600

OXYANION-MINERAL SURFACE INTERACTIONS IN ALKALINE ENVIRONMENTS:

ASO4 AND C1O4 SORPTION AND DESORPTION IN ETTRINGITE

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By Satish Chandra Babu Myneni, M.S., M.Tech.

The Ohio State University 1995

Dissertation Committee T.J.Logan S.J.Traina Approved By G.A.Waychunas J.M.Bigham G.Faure Advisor Y.Chin Environmental Sciences Graduate Program UMI Number: 9612249

UMI Microform 9612249 Copyright 1996, by UMI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 DEDICATION PAGE ACKNOWLEDGEMENTS

I am indebted to several people for their support and encouragement, who deserve special

recognition. I would like to express deepest gratitude to my advisers Drs. Terry J. Logan and

Samuel J. Traina for their support, and encouragement throughout the three and half years of my

stay in Kottmaa I would like to thank Dr. Logan for introducing me to the subject of oxyanions

and their play at the mineral-water interfaces. Special thanks to my coadviser (‘Sam’) who created

interests in microscopic studies and introducing me to the world of spectroscopy. His unselfishness

in sharing his time for discussions is commendable and appreciated. I would like to share with them what ever good that comes out of this dissertation.

Although Dr. Waychunas joined the ^execution committee' little later, he has influenced my research career greatly. I would like to thank him for teaching x-ray absorption spectroscopy and showing how to view the experimental data in several different ways. Dr. Bigham and Dr. Faure needs special mention for their encouraging me in this research. I would like to thank Dr. Bigham for introducing me to the mineral world of ‘ettringite’, without which this dissertation would have been directed in some other direction. Dr. Yo Chin showed me the other side of inorganic world and admire him for all the discussions we had in the journal club.

Although Dr. Krissek is not a member of my dissertation committee, he needs a special recognition for seeing a ‘potential researcher’ (euphanism for a Ph. D. student ?) in me. His

111 unselfishness in guiding a student is really appreciated. He gave me a new life in my research career.

Special thanks to my good friends, Gustavo for technical discussions and encouragement, and

Eric and Matt, for sharing my views and introducing me to the American society and culture. I would like to thank Quirine and Valerie (dectron-electron pair correlations) for keeping the life in

Rm 409 very social. Ubi, Doug, Brad boy, Billy and Jagat need a special acknowledgment for helping me in several different laboratory analyses. Thanks to all my other friends of Rm 409 for their support and encouragement. I would like to thank Dr. Paul Powliat (Smithsonian Museums),

Dr. Douglas Pride, and Dr. Tettenhorst for loaning me mineral samples; and Dr. Ingrid Pickering for EXAFS data collection and analysis.

Ultimately, a special acknowledgment goes to my school teacher Donepudi Janakiram, who invigorated my interests in science from my childhood. Now I realize that he is very optimistic about me becoming a scientist. In additioa, the unselfish people of my village, who together started a small school in the comermost untained village in India, are acknowldged, without which 1 would probably have been an illiterate.

IV VITA

June 24,1966...... Bom- Mynenivaripalem, India

1987 ...... M.Sc., Indian Institute of Technology, Bombay, India

1990 ...... M.Tech., Indian Institute of Technology, Kharagpur, India

1990-199 1 ...... Research Scientist, Steel Authority of India Ltd., India

1991-199 5 ...... Teaching & Research Associate, The Ohio State University

FIELDS OF STUDY

Major Field: Environmental Sciences

Studies in oxyanion vibrational and XAS spectroscopy, mineral-water interface chemistry, solubility and mineral equilibria, and alkaline waste materials. TABLE OF CONTENTS

ACKNOWLEDGMENTS...... iii

VITA...... V

LIST OF TABLES...... ix

LISTOFHOURES...... xi

CHAPTER PAGE

I. INTRODUCTION

Ettringite: Occurrence, and Crystal Structure...... 2 Oxyanion Coordination Sites in Ettringite ...... 3 Resarch Objectives ...... 8 Approach ...... 8 Organization ...... 10

n. SOLUBILITY AND WEATHERING OF ETTRINGITE: GEOCHEMISTRY OF Ca-Al-S04-H20 SYSTEM

Introduction...... 11 Previous Studies on Ettringite Solubility and Geochemistry of the Ca-Al-S 0 4 -H2 0 system...... 12 Experimental Materials and Methods ...... 17 Results and Discussion...... 24 Solubility Product of Ettringite...... 24 Incongruent Dissolution of Ettringite and Geochemistry of Ca-Al-S0 4 -H2 0 System...... 32 Geochemistry of the Ca-Al-S 0 4 -H2 0 System: Open to Fe^*, M g^ Si( 0 H)4°, and COj^ ...... 46 Conclusions...... 48

m . INTERACTIONS OF ASO4 WITH ETTRINGITE

Introduction...... 49 vi Experimental Materials and Methods ...... 51 Results and Discussion...... 54 As0 4 Adsoiption ...... 55 As0 4 Coprecipitation ...... 6 6 ASO4 Desorption ...... 6 8 ASO4 Interactions with Ettringite ...... 72 Conclusions...... 76

IV. INTERACTIONS OF C1O 4 WITH ETTRINGITE

Introduction...... 78 Experimental Materials and Methods ...... 79 Results and Discussion...... 81 Cr0 4 Adsorption ...... 82 C1O4 Coprecipitation ...... 8 6 C1O4 Desorption ...... 89 Cr0 4 Interactions with Ettringite...... 89

V. VIBRATIONAL SPECTROSCOPY OF OXYANIONS IN ETTRINGITE

Introduction...... 92 Data Collection and Analysis ...... 95 Results and Discussion...... 99 Coordination Environment of Ca, Al, SO 4, OH and H 2O in Ettringite ...... 99 ASO4 Complexation in Ettringite...... 106 Semi Empirical Molecular Orbital Calculations of ASO4 Species ...... 107 ASO4 Spéciation in Water ...... 114 Structure of ASO4 in Crystalline Ca Arsenates...... 98 ASO4 Spéciation in Ettringite ...... 126 Summary of ASO4 Coordination in Ettringite ...... 137 Cr0 4 Complexation in Ettringite ...... 138 Cr0 4 Spéciation in Water, NazCr 0 4 ; and NajCraO?. 2 H2 0 ...... 140 Cr0 4 Spéciation in Ettringite ...... 142 Summary of Cr0 4 Coordination in Ettringite ...... 153

VI. EXAFS OF ASO4 AND Cr04 IN ETTRINGITE

Introduction...... 154 Materials and Methods ...... 156 Results and Discussion...... 162 Structure and Oxyanion Coordination in Ettringite ...... 162 As-EXAFS of ASO4 -Ettringite ...... 169 EXAFS of Ca Arsenate Model Compounds ...... 169 ASO4 Spéciation in Ettringite ...... 174 vii As0 4 Adsorption in Ettringite ...... 181 ASO4 in Coprecipitated Ettringite ...... 186 Desorption of ASO4 from Aisenated Ettringite ...... 189 ASO4 Interactions in Ettringite ...... 191 Summary...... 194 Cr-EXAFS of Cr0 4 -Ettringite ...... 195 Desorption of C 1O4 from Cr0 4 -Ettringite ...... 201 Cr0 4 Interactions in Ettringite ...... 201

VII. CONCLUSIONS

Ettringite Solubility, Weathering and Geochemistry of Ca AI-SO4-H2O System...... 202 Arsenate and Chromate Interactions with Ettringite ...... 203 Recommendations for Future Research...... 207 Environmental Applications ...... 208

APPENDICES

A. SOLUBILITY PRODUCTS OF MINERALS...... 209

B. ARSENATE INTERACTIONS WITH CaO AND GIBBSITE

Introduction...... 210 Materials and Methods ...... 210 Results and Discussion...... 211 ASO4 Interactions with CaO ...... 211 Characterization of johnbaumite ...... 218 ASO4 Interactions with Gibbsite ...... 224 Summary...... 225

C. ETTRINGITE SURFACE SITE DENSITY CALCULATIONS...... 226

D. EXAFS DATA FITS...... 228

LIST OF REFERENCES...... 238

vm LIST OF TABLES

TABLE PAGE

2.1 Solubility product constant measurement for ettringite ...... 25

2.2 Chemical analysis of synthetic ettringite and its reaction products after solubility measurement...... 27

2.3 Chemical analysis of synthetic ettringites ...... 30

2.4 Effect of pH and suspension concentration on “X p ...... 30

2.5 Ettringite weathering experiments ...... 37

3.1 ASÜ4 adsoiption by ettringite at low initial ASO 4 concentrations...... 57

3.2 ASO4 coprecipitation in ettringite ...... 67

3.3 ASO4 activity in equilibrium with ettringite, Ca3(As04)2.6 H2 0 and johnbaumite ...... 74

4.1 Cr0 4 adsorption in ettringite ...... 83

4.2 Cr0 4 coprecipitation in ettringite ...... 87

5.1 Site symmetry analysis of ASO4 coordination in different minerals ...... 97

5.2 Normal modes of SO4 in ettringite and related systems...... 104

5.3 Semi empirical calculations on different species of ASO4...... 109

5.4 ATR-FTIR spectra of aqueous ASO4 species ...... 118

IX 5.5 FTIR spectra of As-0 vibrations in different crystalline arsenates...... 118

5.6 FTIR spectra of As-0 vibrations in coprecipitated As0 4 -ettringite ...... 136

5.7 IR spectra of Cr 0 4 -ettringite ...... 139

5.8 Vibrational spectra of Cr 0 4 inNaiCr0 4 ...... 139

6.1 Oxyanion reactive sites in ettringite ...... 163

6.2 Comparison of crystal refinement and EXAFS fits of Na & Ca arsenates...... 167

6.3 Comparison of haidingerite EXAFS fits collected at 298 and 10 K ...... 171

6.4 EXAFS analysis of As0 4 -adsorbed ettringite ...... 175

6.5 EXAFS analysis of ASO4 - coprecipitated ettringite 177

6 .6 EXAFS analysis of C1O4 - coprecipitated ettringite 197

7.1 Summary of ASO4 spéciation in ettringite 204

B. 1 EXAFS analysis of ASO4 reacted portlandite, gibbsite, and johnbaumite ...... 217

B.2 Solubility of johnbaumite...... 221 LIST OF FIGURES

FIGURE PAGE

1.1 Crystal structure of ettringite...... 4

1.2 Oxyanion coordination sites in ettringite ...... 6

1.3 Experimental Design ...... 9

2.1 Ettringite stability in alkaline environments...... 15

2.2 Ettringite solubility...... 28

2.3 Ettringite dissolution in mvironmental pH conditions (total measured concentrations)...... 33

2.4 Ettringite weathering in environmental pH range ...... 34

2.5 Stability diagram for theCa-Al-S 0 4 -H2 0 system...... 39

2.6 XRD patterns of ettringite weathering products ...... 40

2.7 SEM micrographs of ettringite weathering products ...... 42

2.8 EDX analysis of different mineral phases of ettringite reaction products in the pH range of4.0-6.0...... 43

3.1 Efficiency of ASO4 removal during adsorption ...... 58

3.2 Variation in concentrations of ions during As04 adsoiption in ettringite ...... 60

3.3 XRD of ASO4 reacted ettringites ...... 62

x i 3.4 SEM micrographs of ASO 4 reacted ettringites ...... 63

3.5 Effect of suspension density on equilibrium ASO4 concentrations...... 65

3.6 Kinetics of ASO4 desorption from coprecipitated As-ettringite...... 69

3.7 Ionic strength dependent desorption of ASO 4...... 71

4.1 C 1 O 4 adsorption in ettringite ...... 84

4.2 SEM micrographs of C 1O4 reacted ettringites ...... 85

4.3 X-ray diffraction patterns of coprecipitated Cr0 4 -ettringite ...... 88

4.4 CTO4 desorption from coprecipitated ettringite ...... 90

5.1 Arsenate in different symmetries and a cartoon of the IR spectra of V 3 vibrations...... 94

5.2 FTIR spectral analysis ...... 98

5.3 Vibrational spectra of ettringite and aqueous SO4...... 102

5.4 As04^ and its vibrational modes in Td symmetry 110

5.5 Symmetric stretching of As-0 vibrations in (As04^)(H20)4...... I l l

5.6 Symmetric stretching of As-O.X (X = H 2O) vibrations in HAs0 4 ^, H2ASO4 , and hydrogen bonded As 0 4 ^'...... 112

5.7 Symmetric stretching of As-O-Cd vibrations in bidentate mononuclear CdAs0 4 '(H20)4...... 115

5.8 Comparison of PM3 computed results with the experimental data for solvated and protonated ASO4 species ...... 116

5.9 ATR-FTIR spectra of a) HAsÜ 4^ and b) H2ASO4 ...... 119

Xll 5.10 OH stretching of H 2O in oxyanion solvation shells 121

5.11 FTIR spectra of As-O vibrations and the local symmetry of arsenate tetrahedron in haidingerite ...... 122

5.12 FTIR spectra of As-O vibrations and the local symmetry of arsenate tetrahedra in rauenthalite 124

5.13 FTIR spectra of As-O vibrations and the local symmetry of arsenate tetrahedron in Na arsenate...... 125

5.14 OH stretching in adsorbed ASO 4 - ettringite ...... 127

5.15 OH stretching in coprecipitated ASO 4 - ettringite ...... 129

5.16 FTDR spectra of adsorbed ASO 4.ettringite (1800-720 cm-')...... 130

5.17 V4 vibrations of SO4 in ettringite ...... 131

5.18 FTIR spectra of coprecipitated ASO 4.ettringite (1800-680 cm-', pH = 12.4) ...... 133

5.19 FTIR spectra of coprecipitated As 0 4 .ettringite (1800-680 cm ', pH = 11.8) ...... 134

5.20 ATR-FTIR spectra of aqueous Cr0 4 ^, and CTiCh^...... 141

5.21 Vibrational spectra of Cr 0 4 ^, and CrzO?^ in solids ...... 143

5.22 OH stretching bands of coprecipitated Cr 0 4 ettringite ...... 145

5.23 OH bending vibrations of coprecipitated Cr 0 4 - ettringite ...... 146

5.24 SO4 stretching vibrations (as.) in coprecipitated Cr0 4 .ettringite ...... 148

5.25 S-0 bending vibrations (as.) in coprecipitated Cr0 4 .ettringite ...... 149

xiu 5.26 Cr-0 vibrations in coprecipitated Cr 0 4 . ettringite ...... 150

5.27 Raman spectra of Cr-0 vibrations of C 1O4 in adsorbed and coprecipitated ettringite ...... 152

6.1 EXAFS data reduction...... 159

6.2 Structure of an ettringite column...... 165

6.3 Fourier transforms of haidingerite, rauenthalite, Na2HAs04.7H20...... 172

6.4 Raw EXAFS of ASO4 adsorbed and coprecipitated ettringite ...... 179

6.5 RSF of coprecipitated arsenate ettringite ...... 180

6 .6 RSFs of adsorbed As 0 4 -ettringite EXAFS spectra 182

6.7 RSFs of coprecipitated ASO 4 ettringite EXAFS 187

6 .8 Desorption of ASO4 from adsorbed ASO 4- ettringite ...... 190

6.9 Desorption of ASO 4 from coprecipitated ASO 4- ettringite ...... 192

6.10 Coprecipitation of C 1O 4 in ettringite ...... 198

6.11 Desorption of C 1O4 from coprecipitated Cr 0 4 - ettringite ...... 20 0

B. 1 Sorption of ASO4 on portlandite and gibbsite ...... 212

B.2 XRD patterns of ASO4 reacted CaO ...... 214

B.3 FTIR spectra of ASÜ4 reacted CaO ...... 215

B.4 XRD of synthetic and reported johnbaumite ...... 219

B.5 FTIR spectra of johnbaumite ...... 220

D.l EXAFS fits for Rauenthalite...... 229

D.2 EXAFS fits for Hiarmacolite...... 230

XIV D.3 EXAFS fits for ASO4 adsorbed ettringite ASO4 = 3.24 mmol kg"'...... 231

D.4 EXAFS fits for ASO4 adsorbed ettringite ASO4 = 6.48 mmol kg'* ...... 232

D.5 EXAFS fits for ASO4 adsorbed ettringite ASO4 = 0.29 mol kg'* ...... 233

D.6 EXAFS fits for ASO4 adsorbed ettringite ASO4 = 0.65 mol kg'* ...... 234

D. 7 EXAFS fits for ASO4 coprecipitated ettringite ASO4 = 1.44 mol kg'* ...... 235

D. 8 EXAFS fits for ASO4 coprecipitated ettringite As0 4 = 0.144 mol kg'* ...... 236

D.9 EXAFS fits for ASO 4 coprecipitated ettringite ASO4 = 5.75 mmol kg'* ...... 237

XV CHAPTER I

INTRODUCTION

The concept of the chemical bond is the central idea for all chemical reactions that we see in

nature (PAULING, 1960). Characterization of the bond by measuring its strength, length, and

force constants by using spectroscopic tools will help to better understand the chemical behavior of

an element or a molecule. Although simple macroscopic studies do offer a great deal of indirect

information, they often lack specificity on the chemical nature of a system. Natural systems are

heterogeneous and the application of different ion specific spectroscopic methods to identify the

behavior of a chemical species is essential to understanding the environmental geochemistry of

toxic pollutants. Even with the applicatim of these macroscopic and microscopic tools, there is no clear understanding of oxyanion complexation, viz., inner- (IS) and outer-sphere (OS), on the surfaces of minerals, which is the main focus of this dissertation.

Oxyanions are highly mobile in soils, sorb very little to mineral surfaces above neutral pH and thus are of major environmental concern. Knowledge of the types of interactions of these ionic complexes will help to understand their mobility, redox behavior, kinetics, and solubility

(SPOSITO, 1994), information which is crucial to the engineering aspects of remediation of oxyanion contaminated waste sites. Previous studies of oxyanion sorption on Fe oxides have shown that closeness of the system’s pH to the first pKa of the oxyacid results in IS complexation 2 (HAYES, 1987). On the basis of this central idea, oxyanions such as phosphate, arsenate and selenite are thought to form IS complexes; and sulfate, chromate, selenate, nitrate and perchlorate always form OS complexes. However, recent spectroscopic and macroscopic studies of chromate sorption by Mn oxides (JOHNSON and XYLA, 1991; CHARLET and MANCEAU, 1992), and

Fe oxides (PETERSON et a i, 1994); and selenate sorption on Fe oxides (HAYES et a!., 1987;

CHARLET and MANCEAU, 1994) showed contradictory results. These studies indicate that the

IS/OS nature of an oxyanion is much more complicated than the previously assumed simple relationship to the first pKa of the oxyacids. To better understand these mechanisms, arsenate

(ASO4) and chromate (Cr 0 4 ) and an environmentally significant mineral, ettringite, were selected for this study because:

1) ASO4 and Cr 0 4 are observed to form different types of complexes at neutral pH, and the structure of ettringite allows the formation of both IS and OS complexes on its surfaces because of its column and channel like stmcmie (discussed below).

2) Ettringite is stable at high pH, and thus, the nature of ASO 4 and Cr 0 4 complexes at pHs above the first pKa of the oxyacids can be studied.

3) Ettringite is abundant, has high selectivity for oxyanions, and controls the geochemistry of several trace elements in the early stages of weathering of alkaline wastes such as fiyashes.

Therefore, the proposed work can help to predict trace element geochemistry in ettringite-bearing wastes.

Ettringite: Occurrence and Crystal Structure

Ettringite is a calcium aluminum hydroxy sulfate with several molecules of crystalline water

[Cag AI 2 (804)3 (0 H)i2. 2 6 H2O]. It is commonly found in alkaline materials such as cements, cement based waste solidification byproducts, mine spoils, alkaline fiyashes and flue-gas 3 desulphurization (FGD) wastes. The name “ettringite” is used to identify a mineral family with a general composition Xe Yz (Z >3 (OH)n. 26HzO (where X = Ca^*, Mg^^, Sr^^, Ba^*, N a\ Y =

Al^*, Fe^\ Cr*^, S i \ Mn''*; and Z = oxyanions such as S O i \ and also the SO4 end member of this family. Other known minerals of this group are sturmanite, thaumasite, charlesite, bentorite and jouravskite with different ionic substitutions in X, Y and Z positions (TAYLOR,

1973; DUNN et al., 1983; PEACOR et al., 1983). In this dissertation, ettringite refers to the SO4 end member.

Ettringite consists of columns of {Cae [Al(0 H)s]2 24HzO}^, and the inter-column space is occupied by [S04^']3 with water molecules between the oxyanion (as an OS complex) and the columns (MOORE and TAYLOR, 1970) (Fig. 1.1). Column Al is in octahedral coordination with six OH and is enclosed by distorted Ca antiprisms, linked to four OH and four H 2O molecules.

The H 2O of the Ca polyhedra are projected into the channels and form H-bonds with channel oxyanions, which hold these columns together by electrostatic interactions. Thus, the channel anions are apt to form OS complexes (since H 2O molecules surroimd the columns); however, they may also exhibit IS complexes (by replacing H 2O and directly complexing with Ca polyhedra).

Thus, the ettringite column stability may control the type of oxyanion complex inside the channels or vice versa. In addition, depending on the nature of the oxyanions, these may also form either IS or OS complex on ettringite external surfaces, in the absence of any major structural control.

Oxyanion Coordination Sites in Ettrinsite

Oxyanion sorption in ettringite may result in formation of complexes with Al/Ca polyhedra.

Preliminary structural analysis based on Pauling’s electrostatic valence rule indicates that Ca should exhibit stronger affinity for oxyanions as compared to Al because of smaller cationic charge on O attached to Ca. However, all possible oxyanion reactive surface sites of ettringite water

Calcium

Sulfate Aluminum

CtMumns Hydroxyl

Figure 1.1 Structure of Ettringite. (A): perpendicular to c-axis (001 plane), showing columns and channels. Channels are filled with oxyanion and water. (B); structure of a single column showing Al polyhedra and part of the Ca polyhedra. 5 are presented briefly here and a thorough discussion of the same is offered in Chapter VI. The

polyhedral dimensions of oxyanion (As-0 = 1.68 and 0 -0 = 2.75 Â for ASO 4, Cr-0 =1.64 and O-

O = 2.68 Â for C1O4), and Al and Ca in ettringite are significant to form certain IS complexes.

Inner-sphere, oxyanion edge- or face- sharing complexes may form with Al and/or Ca polyhedra when their edge and face dimensions are similar to those of oxyanion polyhedra. Although edge

sharing complexes are common, face sharing complexes have not been observed, which may be due to the very close approach of cations in face-shared complexes. On the other hand, comer-sharing oxyanion complexes can form on ettringite surfaces without any restrictions of polyhedral dimensions.

Aluminum and Ca polyhedra are exposed on ettringite edges parallel to the (001) plane, but only Ca on all other edges (Fig. 1.2). Thus, comer-sharing, mononuclear oxyanion complexes may form with Al polyhedra only on surfaces parallel to (001); and with Ca on all surfaces. These sites are identified in this dissertation as ‘A’ - type (has no relation to the A, B and C type sites on mineral surfaces, SPOSITO, 1989). Comer-sharing binuclear complexes (bridging or sharing of two apices of oxyanion with comers of two Al/Ca polyhedra) are not possible with Al, since neighboring Al polyhedra are distantly separated. However, ASO 4/C1O4 can exhibit such complexes with Ca in two different ways: 1) as bidentate binuclear (‘B’ - type), and 2) monortentate binuclear ('C - type) complexes (Fig. 1.2). Edge-sharing between the oxyanion polyhedra, and Al (‘D’ - type) and Ca (‘E’ - type) are possible at the surfaces when the polyhedra edge lengths are similar (bidentate mononuclear complex). Thus formation of ‘B’, ‘D’ and ‘E’ type complexes are controlled by oxyaniwi, Ca and Al polyhedral dimensions. In addition, oxyanions can also share two of their apices with Al on one side and Ca on the other at ‘F’ - type 'A' - type

A o 0

ASO4 Td

& y # I O O Al-Octahedra D' - type ^ a. • @ 6 r 'B' - type

®®0s®oe®o0®o

® _ - | s —Ca Antiprisms Q iï G O © f i y > 'C - type

Fi2 ure 1.2. Oxyanion coordination sites in ettringite.

(continued on next page....) Fig. 1.2. Continued.

'F’-type^ Û O 0 9 o

• 'S @ § 0 , O.. ■0 * § ■o Ù

'G' - type o®0 ©®00® 0 O 2 0 < — H2O

® — I - 8 _ s . o :.A OH ¥ 8 sites, and inside the channels as OS complex (‘G’ - type) (Fig. 1.2). Although macroscopic techniques may not clearly distinguish these sites (A - G), with the application of ion specific spectroscopic tools it is possible to identify the nature of oxyanion coordination complexes.

Research Objectives

The available literature (Chapter II) on ettringite showed that its solubility and thermodynamic relation to other associated minerals in the Ca-Al-S 0 4 -H2 0 system are not well characterized. This information is crucial for understanding oxyanion-ettringite interactions, and for predicting oxyanion behavior in alkaline environments. In addition, studies of oxyanion- ettringite interactions have not been attempted previously.

Thus, the major objectives of this research were to:

1) Measure the solubility of ettringite and predict its chemical stability regime.

2) Investigate the weathering of ettringite at neutral pH, and interpret the geochemical

relations among minerals in the Ca-Al-SOrHzO system.

3) Explore the extent of arsenate and chromate sorption in ettringite.

4) Examine the nature of ASO4 and Cr 0 4 complexes in ettringite and other related metal

arsenates and chromâtes using vibrational and X-ray absorption spectroscopic techniques.

5) Propose a working hypothesis on oxyanion-mineral surface interactions with the data

obtained from the above experiments.

Approach

The overall research plan is shown in Figure 1.3. Ettringite solubility and its control over

Ca^*, Al^* and S 04 ^ activities were established by conducting simple solubility and weathering experiments. Arsenate and Cr 0 4 interactions with ettringite were studied using sorption and RESEARCH OBJECTIVES

Ettringite Synthesis & Solubility Studies

Ettringite Weathering in Environmental pH Range

Thermodynamic Equilibrium and Reaction Path Calculations

Oxyanion Behavior in Alkaline Environments

Macroscopic and Microscopic Investigation

r Oxyanion Sorption Vibrational Spectroscopic Studies

Adsorption Versus Coprecipitation EXAFS of Sorbed Oxyanions

Ionic Strength Effects on Desorption Hydration Energies of the Oxyanion Versus the Types of Surface Complexes Stability of Oxyanion Substituted Ettringites

Oxyanion Interactions with Mineral Surfaces

Figure U . Experimental Design. 10 coprecipitation experiments, which provided information on oxyanion macroscopic behavior in the

presence of ettringite. The samples obtained in these reactions were probed with Infrared, Raman,

and Extended X-ray Absorption Fine Structure Spectroscopy (EXAFS) to speciate these oxyanions

in ettringte at different pH and ionic strengths. In a separate study, solvation of XO4 (X= As, P,

Se, Cr, Mo, and S) ions was studied with vibrational spectroscopy and this information was

combined with the data obtained from the ettringite studies to derive a hypothesis for oxyanion-

mineral surface interactions.

Organization

The dissertation is divided into six chapters (besides this). Chapter II provides information on ettringite solubility and stability. Experimental results on the geochemistry of the Ca-Al-S 0 4 -Hz0

system are also presented here. The thermodynamic constants used in this chapter are provided in

Appendix A. Results of ASO 4 and Cr 0 4 sorption in ettringite are presented in Chapters III & IV.

The arsenate sorption results are supplemented by the results of arsenate interactions with lime and gibbsite (Appendix B). These macroscopic experimental results were interpreted using the reactive

site density analysis (Appendix C) based on ettringite crystal structure. Chapters V and VI evaluate the microscopic details of ASO 4 and Cr 0 4 complexes in ettringite from the use of vibrational and EXAFS spectroscopies, respectively. The EXAFS data fits for each type of ASO 4 coordination are presented in Appendix D. Chapter VII summarizes the results and gives recommendations for future research. CHAPTER n

SOLUBILITY AND WEATHERING OF ETTRINGITE:

GEOCHEMISTRY OF THE Ca.Al SO4.H2O SYSTEM

INTRODUCTION

Fly ashes, FGD by-products and cement based waste solidification materials are alkaline, and primarily consist of ettringite, portlandite, gypsum, anhydrite, A1 hydroxides, quartz, dolomite and calcite minerals and minor concentrations of amorphous silica, clays and poorly crystalline Fe oxy hydroxides (MA'ITIGOD et ai, 1990; FOWLER et al., 1993). The aqueous chemistry of these materials is dominated by Al, Ca and SO 4 activities. This suggests that weathering and the behavior of major and trace elements in such environments can be understood from mineral-water equilibria in the Ca-Al-S0 4 -H2 0 system. The phase equilibria of this system at high pH (> 11.0) are well known (DAMIDOT et al., 1992; DAMIDOT and GLASSER, 1993); however, little is understood about the geochemical bdiavior at near neutral pH. The study of this system at low pH

(< 7.0) is complicated by the complex mineralogy of several basic aluminum sulfates

(NORDSTROM, 1982; WANG and HSU, 1994), Al hydroxide (HSU, 1989) and Ca-Al-hydroxy sulfate phases (this work),

Ettringite is a common phase in the Ca-Al-S 0 4 -H2 0 system above pH 10.0 and is reported to be abundant in alkaline materials (SOLEM and McCARTHY, 1992; HASSETT et al., 1991;

FOWLER et al., 1993; HILLS et al., 1993). Continuous weathering of alkaline materials would

11 12 result in lowering of the solution pH which may, in turn, dissolve ettringite and exhibit different phase equilibria (SOLEM and McCARTHY, 1990). Therefore, a study of the interactions among these mineral phases will be helpful in understanding the geochemistry of the Ca-Al-S 0 4 -H2 0 system at neutral pH. In this chapter, laboratory investigation results on: (1) ettringite solubility; and ( 2) the geochemistry of the Ca-Al-S 0 4 -H2 0 system in natural environmental conditions, are presented.

Previous Studies on Ettringite Solubility and Geochemistry o f the C a - A f System

Solubility and Standard State Free Enersy of Formation ofEttrimite

The congment dissolution of ettringite under alkaline conditions can be described as:

Ca« Alz (8 0 4 )3 (O H )u. 26HzO <=> 6 Ca^+ + 2 A1^^ + 3 8 0 / + 120H + 26HzO (1)

According to the Law of Mass Action,

(Ca^ + )^(AI^+)^(SO^-)^(OH')^(H O )^ equilibrium constant, K, = ------^ (Ettringite) or when (Ettringite) = (H 2 O) = 1

Ki = *^K.p = (Ca^*)W*)"(S04^f (0H-)“ where is the solubility product constant for ettringite dissolution, and the ions in curved brackets are activities. The equilibrium constant is related to the free energy change of the reaction

(AG“). According to Hess’s Law:

AG" = -RTln *“-K,p = AGp" - A G / where R is the Universal Gas Constant, T the absolute temperature and AGp" and A G / are the respective standard state Ace energies of products and reactants. In other words. 13 -RTln '“-Ksp = (6AG"ca+ 2AG” ai + 3AG°s04+ 12AG“oH +26AG°H jO) " AGg" (2)

From Eqn. 2, the standard state free energy of ettringite, AG e ° can be estimated using the standard state free energies of its ions and the solubility product constant '“’Kgp. However, these equations are valid, and the thermodynamic data can be derived from solubility measurements only when ettringite is in its standard state, dissolves congniently according to the reaction described above, and equilibrium is achieved during dissolution.

The first major investigation on ettringite solubility and its relation to other minerals in the Ca-

AI-SO4 -H2O system was conducted by JONES (1944), who reported that ettringite (Ca sulfoaluminate) dissolves incongruently to gypsum and Al hydroxide at low pH, and to portlandite at high pH. Using the data of JONES (1944) and the standard state free energies of formation of gypsum and hydrogamet, HAMPSON and BAILEY (1982) predicted that '^Kgp would vary between 10’^® and 10^® (with A 1(0 H)4‘replacing Al^^as the Al species in the dissolution products shown in Eqn. (1)). They attributed this '^Kgp variation to changes in ettringite crystal loigth and habit. A more complete ettringite solubility investigation was conducted by ATKINS et al. (1991).

These authors reported a ®^Kgp (according to Etpt 1) value of 10'"'^ which differs from the above because HAMPSON and BAILEY (1982) expressed ettringite dissolution in terms of A 1(0 H)4" species. On the basis of molar ratios of Ca/S04 and Ca/Al in ettringite and its saturated solution,

ATKINS et al. (1991) suggested that this mineral dissolves congniently above a pH of 10.7. The solubility studies of DAMIDOTet al. (1992) showed similar results and also suggest that *“"Kgp is invariant with changes in solution composition. Another number for '“'Kgp is available in the thermodynamic database of EQ3/6 (WOLERY and DAVELER, 1992) which differs from the 14 reported value of ATKINS et al. (1991) by 3 log units. Thus, the solubility of ettringite is uncertain with several published values present in the literature.

Geochemistry of the Ca-Al-SO^-H^O System at 298 K

Numerous investigators have examined the Ca-Al-S 0 4 -H2 0 system under alkaline pH conditions (JONES, 1944; ATKINS etal., 1991; HAVLICA and SAHU, 1992; DAMIDOT et al.,

1992; DAMIDOT and GLASSER, 1993), but little information is available for the pH range of 4 to 9. Since the geochemistry of Ca-S 0 4 *H 2 0 and AI-SO 4 -H2O systems is well understood, available literature on these two systems was gathered to interpret mineral interactions in the Ca-

AI-SO4-H2O system.

JONES (1944) studied the stable and metastable mineral phases of the Ca-Al-S 0 4 -H2 0 system and the effect of activities of alkalies such as Na and K on the stabilities of these phases in the alkaline pH range. He concluded that minerals that form in this system are solid solutions of ettringite and monosulfoaluminate (Ca4 Al2(S0 4 )(0 H) 12- 6H2O) and coexist with portlandite, gypsum, and gibbsite (Fig. 2.1). Ettringite was found to be the most stable phase in the presence of high concentrations of alkalies and sulfate (JONES, 1944; and DAMIDOT and GLASSER,

1993). In addition, temperature has a pronounced effect on the stability of minerals in this system

(DAMIDOT and GLASSER, 1992). For instance, monosulfoaluminate is more stable than ettringite as the temperature is increased from 298 to 358 K. NISHIKAWA et al., (1992) studied the effect of dissolved CO 2 on this system. They reported that ettringite, when exposed to 5%

CO2 at 95% relative humidity and pH greater than 11, decomposed first to vaterite (CaCOg) and 15

-2

Gypsum

-3 Ettringite cT

Al-Hydroxide Portlandite

-4

Hydrogarnet /

-5 11 12 13 pH

Figure 2.1. Ettringite stability in alkaline environments (HAMPSON and BAILEY, 1982). 16 then to aragonite (CaCOg). These transformations were foimd to be accelerated in dry conditions

(rate ~ 2X10'^ h ') or when water activity is low.

Relatively few minerals form in the Ca-SO^ ^ O system (DONER and LYNN, 1989).

Primarily, the mineralogy consists of gypsum (CaSO^. 2H2O), its analogues with different amounts of crystalline water such as bassanite (CaSO^. O.5H2O) and anhydrous phases including anhydrite (CaSO^) and its polymorphic forms a- and y-CaSO^. Of these phases, gypsum is commonly encountered in soils and weathered fly ashes (FOWLER et al., 1993). The AI-SO 4-

H2O system is more complicated and includes several Al hydroxide phases such as gibbsite (y-

A1(0 H)3), boehmite (y-AlOOH), and diaspore (a-AlOOH) found at near neutral pH (HSU, 1989), and basic Al-sulfate phases (NORDSTROM, 1982), such as basaluminite (ALt(OH)io(S 04).5H20), jurbanite (Al(OH)S04.5HzO) and alunogen (Al 2(S04)s.IVHiO), at acidic pHs (< 7.0).

Experimental studies on the chemistry, mineralogy and crystal structures of some basic Al-sulfates were conducted by BASSETT and GOODWIN (1949), HENRY and KING (1950),

JOHANSSON (1960,1962) and WANG and HSU (1994), and the available thermodynamic data on many of these phases are compiled by NORDSTROM (1982). Several of these minerals, with the exceptions of jurbanite, basaluminite (LONGMIRE et al., 1990) and alunogen, are not commonly identified in X-ray diffraction of natural weathered materials, such as acidic mine spoils. The chemistry of water in equilibrium with these materials is generally reported to be saturated with respect to jurbanite (NORDSTROM and BALL, 1986; SULLIVAN et ai, 1988) and basaluminite (SINGH and BRYDON, 1970; SJOSTROM, 1993). Even though jurbanite has been found in some of the precipitates, direct precipitation of basaluminite is still a subject of controversy. While SINGH and BRYDON (1969,1970) reported basaluminite formation when Ai sulfate solutions were allowed to react with bentonite, the experiments conducted by ADAMS and 17 HAJEK (1978) using mixtures of Ca( 0 H>2 with Al 2(S0 4 >3^ precipitated poorly crystalline compounds that have the same composition as basaluminite. However, aging of these solutions at

323 K resulted in the formation of crystalline basaluminite. In addition, water samples collected from several acid mine streams (NORDSTROM and BALL, 1986) show that the first Al hydrolysis reaction (around pH 4.9) has an impact on aqueous Al and SO 4 concentrations and equilibrium mineral phases. NORDSTROM and BALL (1986) suggest that this hydrolysis reaction initiates Al hydroxide precipitation and probably extends jurbanite stability to higher pH values. Nevertheless, the phases that control aqueous Al and SO 4 concentrations around pH 4.5 are still unresolved.

Previous studies on the geochemistry of Ca-S 0 4 -H2 0 and AI-SO 4-H2O systems provide information on the equilibria of constituent minerals and their probable pH-stability regimes. In the present study, this information is used to supplement the data obtained in the experiments described herein on the Ca-Al-S 0 4 -H2 0 system. In addition, this study used various microscopic tools to probe the existence of mineral phases that were identified by macroscopic thermodynamic modeling. As will be discussed, this study supports the existence of Al-hydroxy sulfate phases below neutral pH and shows that they can precipitate rapidly in natural systems and thus can have a potential influence on major and trace element dynamics.

EXPERIMENTAL MATERIALS AND METHODS Materials

Solutions. Glassware and Reaeents

Water used in the experiments was deionized and collected from a Bamstead NANOpure II system. The deionized water was boiled for an hour to drive off dissolved CO 2 and cooled in a 18 glove box filled with N 2 gas. The water was analyzed for S O /, Cl , Ca^*, and Na* before beginning the experiments, and none of these ions were found above detection limits of the instruments used in this study. Glassware was used for transferring deionized water or acidic solutions (pH < 8.0), and Teflon® and polypropylene bottles and centrifuge tubes for all high pH experiments (pH > 8.0). All labware was washed by the following protocol: 24 h soak in 2%

Micro Liquid Laboratory Cleaner (Cole-Parmer Instrument Company), 24 h soak in 10% HQ, wash and rinse with deionized water. Reagent grade CaO (Aldrich), Al2(SO4)3.16H20 (Baker), sucrose (Jenneile), NaOH (Jenneile), and HQ (EM Science) were used in the synthesis and dissolution experiments. The chemicals were used as supplied, and not purified further. Purity of the lime was tested for possible carbonate impurities with X-ray diffraction (XRD) and Fourier

Transform infrared spectroscopy (FTIR) and stored in a desiccator filled with N 2 gas. Reference solutions supplied by GPS Chemicals were used to calibrate the ion chromatograph (IC), atomic absorption spectrometer (AAS), and inductively coupled plasma spectrometer (ICP).

Synthesis and Characterization ofEttrimite

Natural ettringite is difficult to obtain in large quantities and is usually contaminated with foreign ions such as Si% Fe^*, K*, Na* and COs^ (MURDOCH and CHALMERS, 1958). To obtain pure material, ettringite was synthesized using the saccharate method as described by Me

MURDIE et al. (1986). This method is usefiil for preparing ettringite in large quantities because sucrose enhances Ca solubility. Ettringite was made by dissolving 13.63 g ^ 2(8 0 4 )3 in 40 mL water, and 13.4 g CaO in 890 mL of 10% sugar solution. These two solutions were mixed at room temperature in polypropylene bottles kept inside a glove box filled with N 2 gas. Nucléation was very rapid and a suspension appeared immediately after mixing. This suspension was 19 continuously stirred for 48 h and then the solids were separated by centrifugation. These solids were washed several times with aqueous NaOH at pH 12.0 to remove adsorbed sucrose and any gypsum that was precipitated during synthesis. The solids were allowed to dry in a desiccator tilled with N 2 gas.

The precipitated solids were examined by XRD, FTIR, scanning electron microscopy (SEM) and thermogravimetry (TG). XRD patterns of the precipitated solids were equivalent to the data reported in the International Center for Diffraction Data (ICDD) manual. This manual uses different references for synthetic ettringite XRD (card #: 31-251, 9-414, 37-1476 and 41-1451

(most recent)). While the d-spacings of the lines in these references match, their intensities are not identical. Similarly, the d-spacings of synthetic ettringite in these experiments also matched aU previously reported data but the intensities were similar to those reported in card # 37-1476

(McMURDBE et al., 1986). Infrared spectra of the precipitated solids were collected and compared with those reported by PÔLLMANN et al. (1989) and KUMARATHASAN et al.

(1991). Symmetry information obtained from vibrational modes of SO 4 and H 2O molecules in synthetic ettringite (for detailed discussion refer to Chapter V) agree with ettringite crystal structure refinements (MOORE and TAYLOR, 1970). Absence of vibrational frequencies of CO3 around ~ 2500-3000 and 750 cm'* (WHITE, 1974) indicated that the synthetic material was initially free from calcite. However, upon long exposure to the atmosphere (~ 60 days), a doublet appeared around 1450 (1480,1440) cm'* due to CO 2 adsorption. On the basis of XRD and FTIR data, the synthetic imterial was identified as crystalline ettringite. This synthetic material was also examined for its crystal composition by dissolving 0.1 g of oven-dried (383 K) material in 100 mL of 0.1 M HQ acid and analyzing for Al, Ca, and SO 4 . Obtained data were normalized to 6 moles of Ca per mole of ettringite. Loss of moisture (wt%) at 383 K (from TG analysis) was converted 20 to moles of H2O using the reported ettringite stoichiometric composition. This analysis showed that there were 26 moles of H 2O in the structure. However, this number may also include adsorbed and structural water. Thus, the absolute quantity of structural 2 HO remains uncertain.

Fortunately, this number has no effect on the solubility product analysis, as long as H 2O activity is unity.

Methods

Solubility Studies

The solubility product constant for ettringite was evaluated with synthetic ettringite and equilibrium was approached from both undersaturated and supersaturated solutions (i.e. common- ion effect). Ionic strength was not maintained by use of background electrolyte because of the possibility of ion substitution in ettringite (KUMARATHASAN et al., 1991; PÔLLMANN et al.,

1989). To achieve equilibrium from undersaturation, 0.25 g of ettringite was equilibrated with 25 mL of water in several polypropylene tubes inside a glove box filled with N 2 gas. The tubes were continuously agitated at 298 (± 1) K. After 0.02, 0.08,0.17, 0.25, 0.5, 1, 4, 24, 120 and 450 h, three replicate tubes were removed and centrifuged for 10 min at 7000 rpm. The supernatant pH was measured and the solutions were filtered through a 0.45 pm pore size Nucleopore polycarbonate membrane filter. The solids were air-dried, and studied using XRD and FTIR to characterize the type of dissolution (congruent or incongruent). The chemical composition of the filtered solutions and solids were determined as described previously. To approach equilibrium from supersaturation, ettringite saturated solutions were first prepared (on the basis of the above solubility data) by equilibrating synthetic ettringite in deionized water for a week. The solutions were separated from the solid phase and CaO and Al 2(S0 4 )3 were added. Ettringite precipitation 21 from these supersaturated solutions was induced by seeding with synthetic ettringite. The resulting solids were separated by filtration after 48 h, and chemical analyses were performed on the filtered solutions and solids. The effects of pH and suspension concentration on “‘‘Kgp were also evaluated.

Geochemical Studies on the Ca-Al-S0^-H20 System

Laboratory synthesized ettringite (0.25 g) was equilibrated with deionized water in polypropylene centrifuge tubes and the pH of the sample solutions was adjusted by adding 1.0 M

HCl in a glove box filled with N% gas. Equilibrium pH was adjusted between 4 and 9.5 in 0.5 pH units. The equilibrated solids were continuously agitated at 298 (± 1) K. Three vials for each pH were collected after 10, 120, and 240 days. The analyses of the filtered solids and solutions were performed as described previously.

Spéciation and Geochemical Modelim

Measured total ion concentrations from the solubility and weathering studies were speciated using the MINTEQA2 thermodynamic spéciation code (ALLISON et al., 1990). The Davies

Equation was used to estimate activity coefficients of the charged species. Saturation indices (SI = log [lAP/Ksp]) for the possible mineral phases were estimated using the computed activities of the ions and the reported solubility product data (^pendix A). Redox reactions were not considered in these computer simulations because the samples were well aerated. Reaction path calculations were performed using EQ3/6 (WOLERY and DAVELER, 1992) and the thermodynamic data compiled in ^pendix A. This program considers chemical reactions to be irreversible and represented by a sequence of partial equilibrium states. Each equilibrium step is reversible with 22 respect to the next, but overall irreversible to the initial state (HELGESON, 1968). For each infinitesimally small step in equilibrium state, the Law of Mass Action, mass balance, charge balance and non-ideality are solved as the reaction progresses towards the final state. As the reaction proceeds from the initial stage, changes in solution composition are represented by the reaction progress variable, %, which increases as equilibrium is approached. The reaction can only proceed along the considered path if free energy change of a reaction with respect to % is negative (dO/d^ < 0). Since the reaction rates are not available for several reactions, the total number of moles of the reactant consumed is used to establish relative rates in this model. The present data were simulated as a closed system, such that the precipitated solid phases during the reaction remained in the system and were allowed to change as the reaction progressed along the reaction path. Although real soil systems are open to the inputs of foreign ions that are not considered in the present system, mineralogically closed models offer heuristic value to studies of the Ca-Al-S 0 4 *H 2 0 system.

Instrumentation and Analysis

AU the reactants and precipitated solids were studied with XRD, SEM and FTIR. The XRD patterns were coUected using Cu-Ka radiation, and a Philips PW 1216/90 wide range goniometer fitted with a theta-compensating slit and a graphite monochromator. The instrument was calibrated using cholesterol for low 26 angles and Si powder (NIST SRM 640b) for high 20 angles. The samples were finely powdered, mounted tm a silicon holder, and the powder surface was smoothed with a glass slide prior to analysis. The diffraction scans ranged from 6 to 55 °26 with a step interval of 0.05 °26 and a counting time of 4 seconds per step. Sample d-spacings were compared with the reported data from the ICDD powder files. Diffuse reflectarKe FTIR spectra of the solids 23 were collected from 4000 to 400 cm'* with 2 cm * resolution on a Mattson Polaris FTIR. A JEOL

JSM-820 SEM was used to study the trace mineralogy. Energy Dispersive X-ray analysis (EDX) of the solid phases was also conducted on the same instrument and processed with software supplied by Oxford Inst. (Link Analytical eXL). Aqueous cations were analyzed with a Perkin-

Elmer 3030B AAS. Anion concentrations were measured on a Dionex 2(XX)i Basic

Chromatography Module (IC) using a Dimex-AS4A 4 mm separator column, and a Dionex-AG4A guard column. The chemical analysis obtained with the above techniques was also checked for accuracy with a Leeman PS2000 ICP spectrometer.

dH Measurement

An EA 920 Orion Expandable Ion Analyzer connected to a Ross pH electrode (8103) was used to measure solution pH. The pH meter was calibrated with pH 10.0 and 7.0 buffers supplied by VWR Scientific. The buffer solutions were stored in an incubator at 298 (± 1) K and transferred to a beaker before each electrode calibration. To minimize liquid-junction potential effects, samples were first centrifuged for 15 min at 7000 rpm and pH measurements were later made in particle free supernatants. The pH readings usually stabilized in 4 minutes, and the measurements were taken after tliat period. The sample and reference solution pH measurements were repeated to check for any drift in pH. 24 RESULTS AND DISCUSSION

Ettringite Solubility Product

The congruent dissolution of ettringite can be described according to Equation 1 on the basis of reported stoichiometry. Solubility data (Table 2.1) suggest that ettringite dissolution was fast

(Fig. 2.2) and that the reaction reached equilibrium within minutes. This rapid dissolution may be due to the fine grain size (2-5 pm in length) of the synthetic mineral. In contrast, the approach to equilibrium from supersaturated solutions was relatively slow and took about 24 h before equilibrium was reached. However, a white ettringite precipitate appeared immediately after the solutions containing CaO and Al 2(S04)3 were mixed, and the lAP was close to the "^Kgp after a few minutes of mixing. A slower rate of precipitation when the lAP is close to the '“"Ksp is observed for many mineral precipitation systems (STUMM, 1993; SPOSITO, 1994; OELKERS et al., 1994).

The Ca/Al and Ca/SO^ ratios for synthetic ettringite were 2.97 and 2.22, respectively (Table

2.2). These are in close agreement with the reported stoichiometry of 3.0 and 2.0 (based on unit cell composition). Little to no change was observed in the Ca/Al and Ca/SO^ ratios during ettringite dissolution (Table 2.2). In addition, similar molar ratios were observed for both the solid and solution phases. Taken as a whole, these data support congruent dissolution of ettringite as reported by ATKINS et ai. (1991). Further support for congruent dissolution was provided by

XRD and FTIR which showed no effect of dissolution on structural and spectral properties of ettringite, as long as pH was above 10.7. Thus, in this hyperalkaline regime (pH = 12.5 to 10.7), ettringite dissolution could be described by E g i 1. At pH values < 10.7, incongruent ettringite dissolution produced gypsum and amoiphous Al oxides, and will be discussed in the next section. Table 2.1. Solubility product constant (®^Ksp) measurement for ettringite. Under-Saturated Solutions*

Time pH Ionic Total Concentrations p(ion) p I A P A v g . p I A P P(IAP )3 Strength (mol.m’^ (h) (± 0 .1) mol.m^ Al Ca SO4 Af+ Ca** SO4* 0.02 11.0 6.72 0.78 2.21 1.26 24.45 2.86 3.13 111.4 111.8 112.9 0.08 11.1 7.95 0.87 2..35 1.24 24.79 2.84 3.15 110.9 110.8 112.3

0.17 11.0 7.35 0.90 2.12 1.12 24.38 2.87 3.18 111.5 111.6 113.0

0.25 11.2 7.35 0.77 2.13 1.11 25.25 2.87 3.18 110.9 111.0 112.3 0.5 11.2 8.27 0.77 2.11 1.08 25.25 2.88 3.19 110.9 111.0 112.3

1 11.3 8.58 0.66 2.01 0.97 25.72 2.70 3.24 109.7 111.0 111.1

4 11.1 7.48 0.73 2.05 1.02 24.87 2.88 3.22 111.5 111.4 112.9 24 11.0 7.43 0.72 2.00 1.08 24.48 2.89 3.19 111.9 112.1 113.4

120 11.0 8.41 0.75 2.12 1.05 24.68 2.71 3.05 110.8 111.9 112.3 450 11.0 8.06 0.79 2.08 0.92 24.68 2.72 3.10 111.0 111.0 112.5

* Equilibrium from undersaturated solutions: Synthetic ettringite was dissolved in deionized water, and samples were collected for différait reaction periods. These samples were analyzed in triplicate and the average -log (lAP) is given as avg. pIAP. For each reaction time, complete chonical analysis of one of the triplicates and its pIAP (according to Reaction. 1) are given. For the same samples, lAP was also calculated using the measured stoichiometry (according to Reaction. 3) and reported as p(IAP) 3. Suspension concentration was lO g L t Equilibrium from supersaturated solutions: Saturated solutions of ettringite were prepared initially by dissolving synthetic ettringite in deionized water, and ettringite was allowed to precipitate Aom these solutions by adding excess dissolved CaO. Activities of the ions were computed using MTNTEQA2 computer code and reported as p(ion).

(Table 2.1 continued on the next page) Table 2.1. (continued)

T i m e pH I o n i c Total C oncentrations p ( i o n ) p I A P A v g p I A P P(IAP )3 S t r e n g t h (mol.m’^ (h) (± 0 .1) moLm'^ Al C a S O 4 Af+ Ca** SO/ 24 12.4 8.97 0.02 14.7 0.28 31.71 2.26 3.73 107.4 108.0 108.1 24 12.2 9.46 0.11 13.7 0.23 30.16 2.24 4.13 107.7 107.7 108.5 48 11.5 8.47 0.17 3.73 0.29 27.21 2.63 3.83 111.5 112.0 112.6 48 11.4 7.98 0.17 3.03 0.45 26.72 2.71 3.61 111.7 111.7 113.0 Table 12. Chemical analysis of synthetic ettringite before (ETT 2), and after solubility measuronent (rest of the samples). All elonaits of the solid are reported as the number of moles per one mole of ettringite and are normalized to 6 moles of Ca.

Time Solid Equilibrium (h) Moles of Each Molar ratios Solution Component Molar ratios Ca A! SO4 OH Ca/Al Ca/S04 Ca/Al Ca/SÜ4

ETT 2 6 2.02 2.80 12.47 2.97 2.14 - - 0.17 6 2.00 2.72 12.57 3.0 2.21 2.35 1.90

4 6 2.04 2.67 12.80 2.94 2.25 2.79 2.01 24 6 2.01 2.70 12.64 2.98 2.22 2.80 1.85 120 6 2.01 2.67 12.69 2.98 2.25 2.82 2.02

450 6 2.03 2.81 12.47 2.96 2.14 2.62 2.25

720 6 2.02 2.70 12.67 2.97 2.22 2.84 2.04

y 28

1 2 0 - O Undersaturated Solutions • Supersaturated Solutions

116 -

112 - 0 ^ 0 0 O 8

108 -

104 -

100 L-i„ii L-i.i iiiiiiil ■ I I I I until I 1.11 I I iiiiui I I ■ ' ...... 0.01 0.1 1 10 100 1000 Time (h)

Figure 2.2. Ettringite solubility. 29 Effect of Stoichiometry on ‘"'K^p

The measured stoichiometries of synthetic ettringite (Table 2.3) were close to the ideal stoichiometry (as described in Eqn. 1), but with slight deviation in OH and SO4 composition. The normalization procedures that are employed in estimation of crystal stoichiometry can result in notable deviation in the '‘^Kgp values of minerals.

To analyze the variation in crystal composition with respect to changes in the reactant concentration, synthesis was varied by changing initial CaO and Al 2(S0 4 )3 concentrations in the presence or absence of sucrose. The final precipitates were analyzed for chemical composition and crystaUinity. These studies indicated that all of these procedures resulted in a final product that has the same composition after imrmalization to 6 moles of Ca. According to this measured stoichiometiy, the dissolution of ettringite can be expressed as

Ca« AI2.02 (8 0 4 )2.79 (OH)i2.4« .26 H2O <=> 6C a^ + 2.02A1** + 2.79504 ^ + 12.480H +

26H2O (3)

The solubility product constant for this reaction is also calculated with this estimated stoichiometry and reported in this dissertation as Kg (avg. pKg = 112.9). Thus, '“‘Kgp and Kg are ettringite solubility product constants estimated from ideal and measured stoichiometries respectively. As expected, the stoichiometry of the solids has a significant effect on the thermodynamic constant of ettringite Qog Kg versus log ^Kgp in Table 2.1). The solubility product estimated according to the measured stoichiometry (Kg) is about one log unit smaller than the average "^Kgp. However, in the case of ettringite these two values are within the standard error of the analysis. 30 Table 2.3. Chemical analysis of synthetic ettringites. Samples marked ETTl-2 were synthesized as reported by McMURDIE et al. (1986). Samples ETT3-4 and ETT5-6 were synthesized in a similar way but with excess CaO and Alz(S 04)3, respectively. Data are reported as number of moles of element per mole of ettringite and are nonnalized to 6 moles of Ca.

Sample me* niAi ltiS04 ntoH ETTl 6 2.03 2.76 12.57 ETT2 6 2.02 2.80 12.46 ETT3 6 2.02 2.79 12.48 ETT4 6 2.03 2.80 12.49 ETT5 6 2.05 2.73 12.69 ETT6 6 2.02 2.79 12.48

Table 2.4. Effect of pH and suspension concentration on ettringite solubility. The solubility product constant (*“'Ksp) was calculated on the basis of Reaction 1. Ion activities are reported as p(ion).

a. Effect of pH* pH p(ion) p*^Ksp

(± 0.1) Ca** 804*^ Al** 10.7 2.62 2.59 22.5 109.3 11.3 2.79 3.12 26.31 111.1 12.2 3.21 3.32 29.24 109.8 12.3 2.31 4.02 31.81 110.0

b. Effect of Suspension Concentration* Suspension pH p(ion) p"K sp density (g L ‘) (± 0.1) Ca* S O / Al** 4 11.4 2.90 3.14 25.80 110.1 10 11.1 2.84 3.06 24.49 110.3 12.1 11.0 2.78 3.00 25.30 112.4 20 10.8 2.85 2.92 23.44 111.0 30 10.5 2.66 2.84 22.14 110.8 Suspension concentration: 4 g L' 1. T = 298K. Equilibrium pH was allowed to change. T = 298 K. 31 Effect of dH on

The eflFect of pH on the solubility product constant was evaluated over the ettringite pH stability range. Dissolution experiments were conducted at pH 10.7, 11.3, 12.2, and 12.5 (Table

2.4a) while keeping the suspension concentration constant at 4 gL'\ Although the aqueous ion concentrations vary with pH (Table 2.4a), the estimated solubility product constant at all of these pH values was very similar, suggesting that changes in pH had little effect on *“‘Kgp. Similar observations were also repotted by ATKINS et al. (1991) and DAMIDOT et al. (1992).

Effect of Suspension Concentration on

Solubility product experiments are conducted in the laboratory at unusually small solid-to- solution ratios and this suppresses secondary mineral formation. Soils, on the contrary, exhibit very high solid-to-solution ratios. While the solid-to-solution ratio should not affect '“"Kgp, it may have a pronounced effect on dissolution rates and thus may control weathering reactions (CASEY et a l, 1993). Experiments were conducted to evaluate the variation of '“"Kgp with suspension concentrations of 4,10,12.1, 20, and 30 gL'*. The results show that, as the solid-to-solution ratio increased, pH dropped and aqueous Ca^*, Al^* and SO^^ concentrations increased (Table 2.4b).

However, the '"^Kgp was invariant at all examined suspension concentrations and equilibrium was achieved witliin minutes for all cases. At high suspension concentrations (>10 gL'‘), the Ca/Al, and Ca/S 0 4 molar ratios of the equilibrium solutions (2.3 and 1.7 respectively) were lower than those of the precipitate (2.97 and 2.22 respectively). This discrepancy may be attributed to the formation of surface polymers or a secondary precipitates (WOLLAST and CHOU, 1992; CASEY et a l, 1993) at high suspension concentrations. Thus, these metastable phases and ettringite may be at equilibrium with the solution phase. 32 Incongruent Dissolution of Ettringite and Geochemistry of the Ca-Al-S0 4 -H2 0 System

Ettringite incongruent dissolution was studied by reacting the synthetic solids with 0.01 M

HCl at initial pH values ranging from 4 to 10 and for reaction periods of up to 240 days. All dissolution experiments were performed in closed reactor vessels as described in the Materials and

Methods section.

Solution Chemistry

The chemical composition of the solutions reacted with ettringite were dependent on equilibrium pH, but no effect of reaction time was evident for the examined period of 10 to 240 days (Fig. 2.3). This suggests that equilibrium was achieved in < 10 days. The results show that there are two pH thresholds where aqueous ion concentrations changed dramatically (Fig. 2.3).

One is around pH 9.0 and the other at pH 4.5. As the pH was lowered from 10 and the first threshold was crossed, dissolved Ca increased from 13 mM to 90 mM (Fig 2.3a), SO 4 increased from 5.6 mM to 10 mM (Fig 2.3c), while Al decreased from 0.2 mM to 0.025 mM (Fig 2.3b). As the pH was lowered from the first threshold to the second threshold, the concentrations of aU the ions remained constant. At the second threshold, Ca decreased from 90 mM to 70 mM, whereas Al and SO 4 increased steeply. Concentrations and activities of the dominant chemical species were evaluated using the MINTEQA2 computer code. Activities of Ca^* and S 04^ showed the same behavior as their total concentrations with variation in pH (Fig. 2.4). However, the decrease in

Ca^* activity below pH 5 is noteworthy and will be addressed later in the geochemistry section.

Aluminum activity decreased with increasing pH and this behavior presiunably reflects hydrolysis.

In the experimental pH range, Al activity (Fig. 2.4d) exhibited several inflection points which correlate with the first, second and third Al hydrolysis constants (STUMM and MORGAN, 1982 ;

FAURE, 1989). 33 100

80 : [8 S 60 a i 40 & 20 □ D O 0 25 20 L a 15 1 10 Î 5 L A

L §3AQSlÛ0 £S> [A Û 0 O CHS3 O

16

12 i

□ 0 I * ‘ ■ I ■ ’ ‘ ‘ I ‘ ‘ ■ I ■ ‘ ■ I ■ ■ ■ ■ I ■ ■ ■ ■ I ■ ■ 3 4 5 6 7 8 9 10 11 pH

O 10 days O 120 days A 240 days

Figure 23. Ettringite dissolution in environmental pH conditions. Total measured concentrations are plotted. 34 - 1 .0 . a

-1.5 ^ gfCooBPGoea (0 oo p^8-i ; O Gypsum ^ S -2.0 □ Ettringite

-2.5

-3.0 I I I I I I I I I I I I I I I I I I I I Li .ii. I. 1,1 I I I j.

- 2.00 O Gypsum -2.25 □ Ettringite I o -2.50 CA I -2.75 Gypsum

-3.00 O * I I i I I i t t I I i I t I I I I I I I I t I I I I I I t I .,1 II 7 8 10 11 pH

Figure 2.4. Ettringite weathering in environmental pH range. Ion activities (in parenthesis) are plotted against pH, with possible mineral phases of the system. The solid symbols represent experimental results. Lines are drawn for different mineral phases by fixing the activities of all ions in a mineral, except the ion of interest. Open symbols represent the same but by considering the variability in activities of components. p(Ca^*) = 2.5, p(SO /) = 3.0, and p(Al^*) = 21.0 (for ettringite) and 6.5 (for jurbanite and basaluminite). Continued on the next page. 35 Figure 2.4. continued.

- 2 .0 0 Gypsum -2.25 Ettringite O O O Jurbanite Basaluminite I -2.50 -2.75

-3.00

3 4 5 6 78 9 10 11 pH

5

-10

-15

-20

3 4 65 7 8 9 10 11 pH 36 Saturation Index Calculations

Saturation indices [SI = log(IAP/Kgp)] were estimated using measured total ionic concentrations. The MINTEQA2 thermodynamic data base was modified with the data in y^pendix A, and SI calculations were performed using this computer code. Several AI hydroxide phases, such as gibbsite, bayerite, diaspore and boehmite, including amorphous forms, were found to be possible precipitates around pH 7. Unfortunately, thermodynamic modeling does not give information about the preference of one polymorph over any other during precipitation. In addition, because of its higher solubility and smaller grain size (LINDSEY, 1979; STUMM,

1993), amorphous AI hydroxide is preferred over the other crystalline phases in the early stages of precipitation. Therefore, only amorphous A1 hydroxide was considered in our simulations. In addition, several basic A1 sulfates and Ca sulfates (Appendix A) were also considered.

The SI calculations showed that only ettringite is stable above pH 10.7, and gypsum and amorphous A1 hydroxides precipitated between 10.5 and 4.5 (Table 2.5). In addition, basaluminite

(pH < 7.5) and jurbanite (pH < 5) were found to be the potential A1 sulfate phases below neutral pH. Basaluminite was always found to be supersaturated by several orders of magnitude over a pH range of 4 to 8 . Indeed, SI values for this mineral increased from 0.25 at pH 3.8 to a maximum of 7.0 at pH 6.0; and then decreased to 0.3 at pH 7.7. This kind of supersaturation is caused either by slow precipitation kinetics or precipitation of an amorphous precursor (SPOSITO,

1994).

Reaction Path Calculations

Reaction path calculations were made to understand mineral equilibria during ettringite weathering. In this study, computer simulations were carried out for the pH decrease (due to Table 2.5. Results of ettringite weathering experiments. 37

pH Equilibrium Total Identified Equilibrium Time Concentrations Solids (± 0 .1) (days) (mM) X = XRD Ca A1 SO4 S = Chemical Spéciation 3.8 10 71.28 23.07 16.03 jurbanitCs, basaluminiteg, 120 69.58 23.15 15.78 gypsumx,s 240 70.75 21.33 14.93 4.2 10 86.8 2.63 10.09 jurbanitCs, basaluminitCs, 120 85.65 3.22 10.09 gypsums, X 240 85.45 3.22 10 4.9 10 93.05 0.06 9.53 basaluminitCs, Al-hydroxidCs, 120 89.43 0.14 9.47 Gypsums, X 240 92.75 0.07 9.31 5.6 10 88.23 0.03 9.59 basaluminitCs, Al-hydioxidCs 240 90.28 0.03 9.47 Gypsums, X 6.0 10 90 0.03 9.77 basaluminitCs, Al-hydroxidCs 240 89.65 0.1 9.13 Gypsums, X 6.8 240 93.93 0.02 9.49 basaluminitCs, Al-hydioxidCs, Gypsums. X 7.7 240 87.3 0.03 9.47 basaluminitCs, Al-hydioxidCs, Gypsums, x 8.0 10 93.95 0.02 10.09 Al-hydroxidCs, Gypsums, x 120 87.4 0.02 9.63 240 89.83 0.11 9.66 8.5 120 73.98 0.06 9.63 Al-hydroxidCs, Gypsums, x 240 77.33 0.06 9.94 9.4 10 41.25 0.03 10.06 Al-hydroxidCs, Gypsum*, x, 120 43.05 0.22 10.31 EttringitCx 240 43.7 0.24 10.34 10.0 10 13.65 0.22 5.84 Al-hydioxidCs, Gypsum*, x. 240 13.4 0.34 5.59 EttringitCx 38 weathering) from ettringite equilibrium pH (~ 11.5) to a pH value of 4.0. During these simulations, minerals that have slower precipitation kinetics, and crystalline phases that succeed amorphous phases during precipitation (e.g. gibbsite versus amorphous A1 hydroxide) were suppressed. These calculations showed that ettringite dissolves incongruently to gypsum and A1 hydroxides below pH 10.7. As pH further dropped to 7.5, basaluminite appeared as a stable phase and persisted even at low pH. Amorphous Al-hydroxide phases disappeared at pH 4.5 and jurbanite formed at its expense as pH further decreased (Fig. 2.5). Similar observations on the mineralogy and activities of aqueous ions were also made from the actual laboratory weathering experiments described above. However, when pH was < 5 the predicted activities of Ca^^ and

304^ were much smaller (1 log unit) than the experimental values. Thus the stability of gypsum, jurbanite and basaluminite alone may not adequately explain the behavior of Ca^* and SO#^ in solution. This suggests that some other phase(s) not accounted for in the thermodynamic data base may also affect the activities of these two ions.

XRD and Microscopic Studies

XRD of the precipitated solids (Fig. 2.6) indicate the presence of ettringite above a pH of

10.7, gypsum and ettringite between pH 9 and 10, and only gypsum below pH 9. No other mineral phases were evident from XRD. The presence of ettringite below pH 10.7 is contradictory to the observations made from the solubility measuremmts and SI calculations. Although slower dissolution kinetics can cause this type of anomaly, this is not likely in the present system since the experimental data from 10 and 240 days gave similar results. This extended stability of ettringite into the lower pH range may be due to its incongruent dissolution to gypsum and A1 hydroxides.

The failure of thermodynamic models to predict the presence of ettringite below pH 10.7 is due to 39

Basaluminite

-10

-15 'Juibanite

Gibbsite

ingite

-25 2 3 4 5 6 7 8 9 10 11 12 pH Figure 2.5. Stability diagram for the Ca-Al-SO^-H^O system. Open circles are experimental data points and the solid line with arrow represents the reaction path predicted using EQ3/6. Estimated Al^* activity in the leachate of flyashes and acidic mine spoils are also plotted ( ■: SULLIVAN etal. (1988); ♦ : FRUCHTER etal. (1990); and ▼(flyashes) and A(FGD): MAITIGOD et al. (1990)). 40

5000

4000 I

t 3000 pH;9.3 f

2000

I pH:6.5 S

1000

pH:4.3

10 20 30 40 50

D egree 2 6

Figure 2.6. XRD patterns of ettringite weathering products. Peaks around 9.3 and 11.8 20 correspond to ettringite and gypsum, respectively. 41 the inappropriateness of using '“'Kgp determined from congruent dissolution to model incongruent processes.

Trace mineral identification was attempted with SEM, based on crystal morphology and energy dispersive X-ray elemental analysis (EDX). SEM revealed three morphologically distinct phases in the system (Fig. 2.7a). These are euhedral and anhedral crystals, and very fine grains.

The euhedral crystals are monoclinic, exhibit three sets of cleavage, and range in size from < 1 pm to as large as 60 pm. These crystals are the dominant phase in all the samples equilibrated below pH 9.0, and constitute > 90% of the sample. The anhedral crystals are found in samples of pH <

7.0, range in size from 20 to 130 pm, and have irregular, conchoidal-type fractures (Fig. 2.7b).

The fine-grained material is of submicron size and found in the pH range of 8 to 5.5. The abundance of this fine-grained fraction decreased dramatically below pH 5.0.

The euhedral crystals contain equiraolar Ca and S as determined by EDX (Fig. 2.8). The mineral morphology and compositional data indicate that these are gypsum crystals. In addition, some of these gypsum crystals have a high A1 content (0.2 to 0.8 (+ 0.03) atomic percentage).

Previous gypsum mineralogical studies have shown little variation in its chemical composition

(DEER et al., 1966; DONER and LYNN, 1989), and Al substitution has not been reported before.

These low Al concentrations may be due to adsorbed or fine nano-scale Al hydroxide coatings on gypsum. The anhedral crystals are identified as Al hydroxy sulfate phases. These grains have variable Al/S ratios (1.2 to 2.6) and contain no Ca. No apparent correlation was observed between this ratio and the grain size of these minerals. This ratio also agrees with the composition of known Al hydroxy sulfate phases, such as basaluminite. Compositionally similar grains were also observed in prismatic form but were few in number. WANG and HSU (1994) have precipitated morphologically similar, Al hydroxy sulfates by mixing NazS 04 with Al hydroxide solutions. -U. ■, 4 , .

X6,9b$ i}ü;w[

: : y ' . W (-V

«009 20KU X400 :)M'ND33 0010 101ÏU X4/500 jl-'iii H03

Figure 2.7. SEM micrographs of ettringite weathering products, al and a2 show gypsum and fine grain coatings of Al-hydroxides, b: Al-hydroxy sulfate, c; Ca-Al-hydroxy sulfate. 6 43

12000

10000

8000 a

(3 6000

4000

2000

13 4 Energy (k eV)

Figure 2.8. EDX analysis of different mineral phases (size < 50 pm) of ettringite reaction products in the pH range of 4.0-6.0. a: Al-hydroxy sulfate, b: Al-hydroxides, c & d: Ca-Al-hydroxy sulfates, and e: gypsum. 44 The fine grained material showed two distinctly different compositions, one (plate-shaped grains) purely Al and O (Fig. 2.7a), and tte other Ca, Al, S, and O (Fig. 2.7c). These two mineral phases are probably some form of Al hydroxide and Ca-Al-hydroxy sulfate, respectively. EDX can not distinguish between adsorbed and structural atoms. Thus, it is not possible to rule out the presence of adsorbed Ca on Al-hydroxy sulfate or adsorbed Al on gypsum. However, the high atomic percentages for Ca and Al (0.5-4.0) in these materials suggest that these ions were present in structural sites rather than as adsorbed species. To our knowledge, Ca-Al-S-(0H,0) phases have not been reported before from low pH materials, such as acidic soil horizons; however, specimens of a Ca-substituted alunite, minamiite (MAl3(S0 4 )2(0 H)g, where M = Na, K, Ca) have been identified from hydrotheimally altered quartz-bearing andésites (OSSAKA et al., 1982).

Geochemistry of the Ca-Al-SOd-HjQ System

Several mineral phases were plotted in an activity - pH diagram (Fig. 2.4) using the available thermodynamic data of minerals in the Ca-S04-H20 and AI-SO4-H2O systems (i^pendix A). The minerals considered are ettringite and gypsum in the Ca-S 0 4 -H2 0 system, and amorphous Al hydroxide, jurbanite and basaluminite in the AI-SO4-H2O system. Some newly identified minerals in the Al hydroxy sulfate system have variable composition when compared to the well known basic alumino sulfates (WANG and H SU , 1994), and the lack of thermodynamic data for them prevents me from representing their solubilities here. It should be noted that these morphologically different minerals are stoichiometrically close to basaluminite.

The two pH thresholds identified from solution chemistry of ettringite weathering correspond to major changes in mineral equilibria. At the first pH threshold (pH ~ 9.0), Ca^* and S 0 4 ^ activities (Fig. 2.4a, & 2.4b) are controlled by ettringite and gypsum, and Al^* by ettringite and Al hydroxide. The inflection in the activities of different ions at this pH correlates with the onset of 45 gypsum and Al-hydroxide precipitation. Although their precipitation begins at pH 10.7 (i.e. where ettringite starts to dissolve incongruently), the solution composition is controlled by these minerals only below pH 9.5. Between the first and second thresholds, gypsum controls Ca^* activity and Al hydroxide and Al hydroxy sulfates control Al^^ activity (Fig. 2.4d). Although S 0 4 ^‘ activity correlates to that of gypsum (solid line in Fig. 2.4b), calculations showed that basic Al sulfates also produce similar S 0 4 ^ activities (triangles in Fig. 2.4c), especially below pH 7.0. As noted above, SEM indicates the presence of different Al hydroxy sulfate phases below this pH. This suggests that 8 0 4 ^ activity is dominantly controlled by gypsum and Al hydroxy sulfate phases below pH 7.0 and only by gypsum in the pH range of 7 to 10. Thus, the geochemistry of Ca-S 0 4 -

HzO and AI-SO4-H2O systems may be simple in the pH range of 7 to 10 and behave independently of each other. Below this pH regime, changes in Ca activity can modify SO4 activity and thus can significantly influence Al activity, or vice versa.

The onset of Al hydoxy sulfate precipitation, at the expense of Al hydroxide, perturbed the

Ca-S0 4 -H2 0 system and this effect is distinct when pH ^ 5.0. The high solubility of the basic alumino sulfates gives rise to high Al** and 8 0 4 *" activities which, in turn, decreases Ca** activity in solutions in equilibrium with gypsum and Al hydroxy sulfates. This may explain the experimentally observed reduction in Ca** (Fig. 2.4a) activity and corresponding increase in 8 O4* activity below this pH. In addition, the observed Ca** and 8 0 4 *" activities are also in equilibrium with gypsum (circles in Fig. 2.4). It is not clear what is responsible for the decrease in Ca** activity below pH 5. Although the reaction path calculations predicted a decrease in Ca** activity, these values are much smaller than the experimental data. The model calculations predict the presence of gypsum, basaluminite, and jurbanite, but these minerals do not support the observed

Ca** activity. These large Ca activities may be caused by a soluble Ca phase that is not accounted for in the thermodynamic data base. Failure to attain equilibrium in the allotted reaction time could 46 also explain this data but the apparent invariance in solution chemistry with time makes this condition unlikely. The observed variation in ion activities may be caused by the Ca-Al-hydroxy sulfate solid phase tentatively identified by SEM.

Equilibrium geochemical modeling with MINTEQA2 showed that basaluminite was supersaturated by several orders of magnitude in the pH range of 4 to 8 . Aluminum activity in equilibrium with basaluminite would be smaller than is observed in these weathering experiments.

This may suggest that: (1) its precipitation kinetics are very slow and/or (2) other polymorphic or amorphous forms of basaluminite, for example, those identified in microscopic studies, may be controlling ion activities.

The reaction path calculations accurately predicted the experimentally observed Ca^*, Al^*, and S 0 4 ^ activities above pH 5.0 (Fig. 2.5). The observed similarities between the experimental values and field results of SULLIVAN et al. (1988) (for acid mine drainage), FRUCHTER et al.

(1990) (for fly ash leachates) and MATTIGOD et al. (1990) (for fly ash and FGD waste leachates) above pH 5 indicate the validity of the predicted equilibria. However, their marked disagreement at pH < 5.0 can be attributed to the formation of Ca-Al-hydroxy sulfates in the laboratory studies and the relatively Ca poor materials of the field sites.

Geochemistry o f the Ca-AUS 0 4 - H ^ System: Open to Kf, Si(OH) 4 , andCOj'

When the system is open to inputs of foreign ions such as Fe^\ Mg^\ K \ Si( 0 H)4°, and

COs^, the mineral equilibria of the Ca-Al-S 0 4 -H2 0 system will include different mineral phases.

To understand these effects, equilibrium spéciation modeling and reaction path calculations were conducted at different pHs and for a range of foreign ion concentrations. During these computations, Ca^*, Al^*, and S 04^ activities were fixed at the values estimated firom the weathering experiments. 47 If the Ca-Al-S 0 4 -H20 system is open to atmospheric CO 2, the equilibrium reactions exhibit carbonate mineral phases such as calcite and aragonite under alkaline pH conditions. However, the slow precipitation kinetics of calcite delays its formation relative to ettringite and gypsum.

Moreover, high S 04^ activity favors the formation of gypsum over calcite (DONER and LYNN,

1989). This has also been observed from field studies of ettringite-bearing, weathered, FGD wastes. In these alkaline materials, calcite did not appear imtil 240 days of weathering (FOWLER et al., 1993), whereas ettringite and gypsum precipitated within one week of their exposure to natural weathering conditions. Thus, carbonate equilibria has little influence on this system during the initial stages of weathering. However, elevations in CO 2 partial pressures, due to decomposition of organic matter beneath the surface layers of soils, may enhance the kinetics of calcite formation. Spéciation codes indicated that the presence of Mg^"^ formed dolomite

(CaMgfCOa):), magnesite (MgCOs). and huntite (Mg 3Ca(CC>3)4) in addition to calcite. Although huntite formation has been observed in weathered dolomite or magnesite bearing rocks (DEER et al., 1966), its rate of formation was not evaluated.

Thermodynamic spéciation calculations predicted that the addition of Fe^*, Mg^*, K \ and

Si(0 H)4° to the system formed Fe oxyhydroxides, minerals belonging to alunite-jarosite family, clays and leonhardite in addition to the other above mentioned mineralogy of the Ca-Al-S 0 4 -H2 0 system. Of the Si phases, zeolites (leonhardite), and some clays such as kaolinite and halloysite may be dominant minerals in the system. However, actual experiments are necessary to support these predicted mineral equilibria. 48 Conclusions

The results presented here demonstrate that ettringite dissolves congruently above pH 10.7 with a p"^Kgp of 111.6 (± 0.8) according to Equation 1. The measured crystal stoichiometry, after normalization to six moles of Ca, is different from that of the reported composition and it did not affect the log "^Kgp value significantly. In addition, changes in pH and suspension concentration had no effect on log *^Kgp. Ettringite can also exist below this pH but only in association with gypsum and Al hydroxide. Weathering at near neutral pH completely dissolved ettringite and precipitated Al hydroxides, Al hydroxy sulfates such as basaluminite and other unnamed minerals, gypsum, and unidentified Ca-Al-hydroxy sulfate phases. Although the geochemistry of the Ca-Al-

SO4-H2O system is simple above pH 7.0 and can be described as association of the Ca-S 0 4 -H2 0 and AI-SO4-H2O systems, it is complicated by the presence of several phases at near neutral pH.

The simplicity above pH 7.0 results from the absence of Al hydroxy sulfate phases. EQ3/6 calculations accurately predicted the weathering sequence of ettringite and relevant ion activities when the pH was > 5.0. These reaction-path simulations also suggested the existence of highly soluble Ca phases around pH 5.0 which were confirmed by the microscopic studies. Below neutral pH Ca activities can significantly influence Al activity or vice versa. Further study should explore the occurrence of these phases in field weathering environmoits and their potential to form in the presence of other ions not considered in the present study. CHAPTER n i

INTERACTIONS OF ASO4 WITH ETTRINGITE

INTRODUCTION

Arsenate (ASO4) is a common soil contaminant and has high mobility in alkaline environments. This bdiavior is due to repulsion of ASO 4 by negatively charged particle surfaces in soils and sediments (PARFITT, 1978; HINGSTON, 1981; STUMM, 1993). At near neutral pH, minerals such as goethite, schwertmannite and other Fe oxides (GUPTA and CHEN 1978;

LUMSDON et al., 1984; MERRILL et al., 1986; FULLER and DAVIS, 1994; BIGHAM et al.,

1994) were found to sorb significant quantities of ASO 4. In contrast, few mineral surfaces have been reported to sorb ASO 4 in alkaline environments. Studies on weathered alkaline flyashes, cements and alkaline soils have identified calcite (CaCOs), portlandite (CafOH):) and ettringite

(Ca6Al2(S0 4 )3(0 H)i2.2 6 H2 0 ) as stable mineral phases (GOLDBERG and GLAUBIG, 1988;

MATTIGOD et al., 1990; HASSETT et al., 1991; FOWLER et al., 1993; REARDON et al.,

1993; REDDY et al., 1994; van der HOEK et al., 1994) that can control trace element solubility at high pHs (> 8.0). Carbonation and induced calcite precipitatim from dolomite have been shown to decrease equilibrium ASO4 concentrations from arsenate-contaminated waste waters

(REARDON et al., 1993), but the responsible reaction mechanisms have not been ascertained

(GOLDBERG and GLAUBIG, 1988; HOEK et al., 1994). Recent weathering studies on

49 50

FGD wastes have shown that decreases in several oxyanion aqueous concentrations were concomitant with ettringite formation (HASSETT et ai, 1991; FOWLER et al., 1993). These preliminary studies showed that aqueous ASO 4 concentrations in equilibrium with ettringite are low and this mineral spears to have potential application for aqueous AsO# removal.

Extensive ASO4 substitution and solid-solution formation in ettringite is possible and is mainly attributed to the column and channel-like structure of ettringite (MOORE and TAYLOR, 1968)

(Chapter I). Arsenate substitution may take place by replacing solvated channel SO 4. While ASO 4 substitution in ettringite has been previously observed (KUMARATHASAN et al., 1990), the extent of substitution was thought to be limited by differences in ionic charge and size (when compared to SO 4).

Arsenate is present as As 04^ and HAs 0 4 ^ in high pH environments (DOVE and

RIMSTIDT, 1985). Interaction of these irais with calcareous minerals can result in the formation of surface complexes, solid solutions and/or precipitates. Despite extensive ASO 4 incorporation and solid solution formation in ettringite (KUMARATHASAN et al., 1990), precipitation of Ca arsenates in this system may not be ruled out. Clearly, ettringite has a high affinity for soiption of aqueous ASO4, but the total sorption capacity is not known. Furthermore, the exact molecular mechanisms responsible for ASO 4 sorption by ettringite have not been deteimined. This chapter addresses these deficiencies by 1) evaluating the efffciraicy of ettringite to remove aqueous ASO4,

2) examining the differences in ASO 4 sorption by adsorption and coprecipitatirai and 3) postulating the possible mechanisms of ASO 4 interactions with ettringite. 51 EXPERIMENTAL MATERIALS AND METHODS

Materials

Deionized water used in these studies was obtained from a Bamstead, NANOpure II system, boiled for 1 hr, followed by cooling under ascarite in a N% (g) - filled glove box to eliminate CO 2.

This procedure was repeated prior to each experiment. Acid-washed Teflon® and polypropylene bottles and centrifuge tubes were used in all experiments. Reagent grade CaO (Aldrich),

Al2(SO4>3.16H20 (Baker), sucrose (Jenneile), NaClO# (GFS), Na2S0 4 (Baker), Na2(HAs0 4 ). 7H2O

(Aldrich), NaOH (Jenneile) and HQ (EM Science) were used as supplied. Synthetic ettringite was used in all adsorption experiments, and the synthesis procedures were described in Chapter II.

SEM showed that the synthetic ettringite grains exhibit short prismatic habit, and are significantly smaller than the ettringites prepared in the absence of sucrose (FÔLLMANN et al., 1993; also see

Fig. 3.4).

Synthesis of Adsorbed and CovrecipitatedAsO^-Ettrineite

Arsenate sorbed ettringites (As 0 4 -ettringite) were synthesized and used to study the extent of

ASO4 desorption. Adsorbed As 0 4 -ettringite was prepared by reacting 5 g of synthetic ettringite with 14 mg of Na 2HAs0 4 .7 H2 0 in 1.25 L deionized CO 2 -free water. Solution pHs were maintained at 11.5 with 0.008 M NaOH to prevent gypsum (CaS 0 4 .2 H2 0 ) formation. The sample solutions were reacted in polypropylene bottles for 60 h and the final precipitates were separated using 0.45 pm polycarbonate filters. From the mass balance calculations, the solid phase ASO 4 concentration was estimated to be 8.3 mmol kg'\ Coprecipitated As 0 4 -ettringite was synthesized for two different ASO4 concentrations (0.18 and 0.30 mol kg'^) by simultaneous addition of

Na2 (HAs0 4 ).7H2 0 , CaO and Al 2 (S0 4 )3.16H2O. Lime addition was constant, and the concentrations of ASO4 and SO 4 were varied to achieve the desired solid phase ASO 4 52 concentrations. The equilibrium pH of the system was maintained around 11.9. The mixture was

allowed to equilibrate for 60 h at room temperature and then the precipitate was separated using

0.45 pm pore size polycarbonate filter. Sucrose was not used in this synthesis to minimize its complexation with As (V).

Methods

Arsenate adsorption, coprecipitation and desorption studies were conducted as discussed below. Whenever actual species or complexes are used in the discussion they will be represented with their respective charges.

Reactive Site Density Analysis

The number of reactive sites on ettringite external surfaces was estimated by using surface area of the bulk synthetic ettringite sample, ettringite crystal and unit cell; and the number of sites per unit cell (i^pendix C). During these calculations, ettringite was assumed to have prismatic shape, ‘c’ axes of the unit cells are oriented parallel to the lengths of the grain, and the sites parallel to the ( 0 0 1 ) plane were considered to be negligible (< 1 % of total sites) as compared to those exposed on other edges of the crystal. The surface area of synthetic ettringite was measured using a Micromeritics Flow Sorb I I 2300 BET surface area analyzer and single point calculations.

The instrument was calibrated with specimen kaolin (standard # 8570, surface area = 10.3 m^ g ') and alumina (standard # 8571, surface area = 153 m^ g"') supplied by National Bureau of

Standards. The average surface area of ettringite crystals was measured from SEM micrographs, using the Sigma Scan Image Analysis Software (Version 1.20.09). The unit cell surface area and the number of reactive surface sites were estimated from the crystaUographic data of MOORE and

TAYLOR (1968) (Chapter I). Details of these calculations are shown in Appendix C. 53 Adsorption Studies

Preliminary batch adsorption experiments showed that equilibrium between ettringite and

ASO4 was essentially instantaneous (apparent equilibrium was established before the samples could be filtered). Solution composition was invariant from < 1 min to 96 h, and, thus, a 24 h reaction time was adopted for all adsorption experiments for convenience. Generally, ion sorption experiments employ an “inert” background electrolyte to control ionic strength. This procedure was not adopted in any of the treatments in the current study to preclude incorporation of ions from the background electrolyte into the ettringte structure. Stock ASO4 solutions were prepared by dissolving Na 2HAs0 4 . 7HiO in COz-free deionized water. In all adsorption experiments, 0.1 g of synthetic ettringite was equilibrated with 25 mL of ASO 4 stock solution in polypropylene centrifuge tubes. All samples were reacted at 298 (± 1.0) K on an oscillating shaker. After 24 h, the samples were centrifuged, the supernatant pH was recorded, and the solutions were filtered through 0.25 pm polycarbonate membranes and saved for analysis of total dissolved Ca, Al, Na, ASO4 and SO4.

These adsorption experiments were conducted for a range of pH values (10.5-12.5), suspension densities (4,10,20,30 g L*) and ASO 4 concentrations (0.36 pM to 14.4 mM).

Coprécipitation Studies

In the coprecipitation experiments, ettringite was allowed to precipitate from mixed ASO 4 and

SO4 solutions. Initially, 100 mL aliquots of solutions containing dissolved Al2(SO4)3.16H20 and

Na2HAs0 4 .7 H2 0 were prepared to provide a range of ASO4/SO4 ratios. The source of Al was either Al2(SO4)3.16H20 or NaA 102. To each of these solutions, 0.13 g of CaO slurry was added and the resultant suspaisions were stirred for 24 h. The suspensions were then centrifuged, supernatant pH values were recorded and the solutions were analyzed for elemental composition as 54 described below. The suspension concentrations in these samples were approximately 1.2 g L *.

The solid phase ASO 4 concentration was estimated on the basis of mass balance calculations.

Desorption Studies

The ability of SO 4 to displace sorbed and coprecipitated ASO 4 from ettringite was examined in COz-free SO4 solutions with and without ionic strength control. In the latter case, 25 mL of 0-

20.83 mM SO 4 was reacted with coprecipitated As 0 4 -ettringite (0.1 g containing 0.30 mol ASO 4 kg ' ) for 0 to 336 h. pH of the system was also not controlled. In a separate set of experiments, 0.1 g of adsorbed (8.3 mmol ASO 4 kg ') and coprecipitated As 0 4 -ettringite (0.18 mol

ASO4 kg ') were reacted with 0-20.83 mM SO 4 for 24 h in the presence of sufficient quantities of

NaC104 to obtain ionic strengths of 0.1,0.3,0.5 and 1.0 moL L '. The samples were filtered after completion of the reaction and the resultant solids and solutions were analyzed as described below.

Instrumentation

Mineralogy of the solid samples was examined using XRD and SEM. Solution chemical analysis was performed with AAS, IC and ICP. Details of these instruments are described in

Chapter II and the analytical procedures were also similar except for the samples that contained

Q O 4 which were analyzed by a Perkin Elmer (Optima 3000) ICP for both cations and anions.

RESULTS AND DISCUSSION

Sorption is used in this paper to refer to both adsorption and coprecipitation reactions. The sorption experiments were conducted to simulate adsorption (presumably only surface coverage) and coprecipitation (ASO 4 incorporation into ettringite channels and solid solution formation) reactions. These laboratory studies are analogues of natural environments where ASO4 55 contaminated leachate interacts with existing ettringite of the waste matrix (adsorption) as compared to direct ettringite precipitation from ASO 4 containing leachate (coprecipitation). In principle, coprecipitation could include adsorption, structural substitution and precipitation of a different ASO 4 pdiase.

Site Density on Ettrineite Surfaces

Arsenate interactions in ettringite may result in surface complex formation or channel substitution, depending on the free energy change of the system. However, surface sites (A, B, C,

D, E, and F type sites. Fig. 1.2, Chapter I) may have more preference over the channel sites (G type sites. Chapter I) during adsorption, due to sluggish solid state diffusion Thus, the number of reactive surface sites per unit cell and the surface area of ettringite limit the extent of surface interactions. Site density calculations (Appendix C) indicate that the sorption maximum for synthetic ettringite is 0.11 mol kg '. When ASO 4 loadings exceed this maximum, sorbed icms may either substitute inside the channels (with enough reaction time) and/or break the ettringite structure and form Ca/Al ASO4 precipitate. In contrast, ASO 4 can substitute inside the channels more easily during coprecipitation, and ettringite structure may not be destroyed when the ASO 4 concentrations exceed 0.11 mol kg '.

ASO 4 Adsorption

Adsorption experiments were conducted at different initial aqueous ASO 4 concentrations (0.45 pM - 15.2 mM), pH values (10.5-12.5) and suspension concentrations (4, 10, 20, 30 g L ').

Despite the lack of an “inert” background electrolyte, the ionic strength of the experimental solutions in equilibrium with ettringite varied little (< 10 %) with ASO 4 loadings. During adsorption. A, B, C, D, E, and F type surface sites are more available as compared to ‘G’ type. 56

Adsorption at Low Concentrations (ASO4 <17 uM)

The efficiency of ASO 4 removal from solution was > 98 % (Fig. 3.1) when the initial dissolved ASO 4 concentrations were < 17 pM (Table 3.1). If ASO4 reactions with ettringite resulted in anion exchange of SO4, then migration of sorbed ASO4 into the channels should have occurred. This would result in displacement of SO4 from the solid and a corresponding increase in dissolved SO4. However, sorption of ASO4 did not result in detectable changes in aqueous SO4 concentrations. This suggests that channel anion exchange is not a dominant sorption mechanism for ASO4 in this concentration regime. As was observed for dissolved SO4, total dissolved Ca and

Al concentrations were also invariant with increases in solid-phase ASO 4 concentration. In addition, adsorbed ASO 4 concentrations were much smaller than the sorption maxima ( 0 .1 1 mol kg' b predicted from crystaUographic information (Appendix C). These results may imply a possible surface complexation reaction for ASO4 retention.

Adsorption at Hish Concentrations (AsOd >17 uM)

The efficiency of ASO 4 removal varied when the initial ASO 4 concentrations were in the range of 17 pM to 15.2 mM (Fig. 3.1). As the initial dissolved ASO 4 concentration increased from 17

|iM to 1 mM, the efficiency of ASO 4 removal ranged from 75 to 90 %. In the 1 to 10 mM concentration range, the efficiency was > 90%, and above this initial ASO 4 concentration range sorption efficiency decreased steeply. Presumably these changes in ettringite ASO 4 sorption reflect changes in reaction mechanisms. In contrast to the results observed at initial aqueous ASO 4 concentrations < 17 pM, the reaction of concentrated ASO 4 solutions (> 17 pM) with ettringite resulted in changes in dissolved SO4, Al, Ca and solution pH. As sorbed ASO4 increased, equilibrium pH decreased from 11.4 to 10.9. lUssolved Ca also showed a similar trend and 57

Table 3.1. Arsenate adsorption by ettringite at low initial ASO 4 concentrations®.

Sample [ASO4 ] (pM) Initial Final Mean S.D 1 46.0 9.01 0.5 2 17.5 1.37 0.23 3 8.57 0.14 0.01 4 6.85 0.09 0 .0 2 5 6.00 0.10 0.01

6 4.28 BDL -

7 2.56 BDL -

8 0.85 BDL -

9 0.45 BDL - 10 0 .0 BDL -

® Equilibrium Ca, Al and SO4 concentrations remained constant at 2.3 (± 0.1), 0.95 (± 0.02) and 1.28 (± 0.03) mM respectively. BDL Below detection limit. S.D Standard Deviation. 58

100 - o o oo (2) o

o O O H . 90 - O I O o f SO -

O pH not controlled 70 - ■ pH controlled at 12 ± 0.2 O

"I ■ ■ I ■ ' < il ■ ' ■ 0.001 0.01 0.1 1 10

Concentration of added AsO^ (mM)

Figure 3.1. Efficiency of AsO^ removal during adsorption. 59 decreased from 2.2 mM to 0.6 mM (Fig. 3.2). On the contrary, dissolved SO 4 and A1 increased from 1.2 to 9.0 and 0.8 to 4.6 mM respectively (Fig. 3.2). When adsorbed ASO 4 was < 0.15 mol kg ' (i.e. close to the maximum sorption capacity, 0.11 mol kg ') the moles of SO 4 released and

AsÛ4 sorbed were similar. In the concentration range 0.15 to 1.36 mol kg ' (ASO 4 sorption > maximum sorption capacity), the moles of sorbed ASO 4 were similar to the released A 1 and SO 4 combined. At high ASO 4 concentrations (> 1.36 mol kg '), the concentration of released ions were much smaller than the sorbed ASO 4. This behavior was not observed when the pH was maintained at 12 ± 0.2 (Fig. 3.2). Under these conditions, the concentrations of released SO 4 and Al, at all

ASO4 loadings, were several orders of magnitude less than the sorbed ASO 4.

Chemical spéciation and saturation index (SI) calculations were performed using

MINTEQA2 (ALLISON et ai, 1990). The results indicate that the dominant aqueous ASO 4 species in this system were HAs 04^ and As 0 4 ^ (DOVE and RIMSTIDT, 1985). This is because the equilibrium solution pH was controlled by ettringite, and was close to the third pK, of H 3ASO4

(~11.5). Saturation index calculations of the equilibrium solutions showed that gypsum

(CaS0 4 .2Hz0 ) was undersaturated, and gjauberite (Na2Ca(S0 4 )z) and gibbsite (A1(0H)3> were supersaturated. The available thermodynamic data on ASO 4 containing minerals is limited and Ksp values are listed ordy for Ca 3(As04)z.6 H2 0 and AS2O5. However, the experimental samples were undersaturated with respect to AS 2O5 by several orders of magnitude and close to the saturation for

Ca3(As0 4 )2.6 H2 0 . Although previous studies on Ca arsenate systems suggest that

Ca4(As0 4 )2(0 H)2.4 H2 0 can precipitate when ASO 4 containing solutions are added to CaO slurry, its solubility product constant has not been reported (REARDON et ai, 1993). However, my experiments on ASO 4 reaction with CaO at ambient temperature and pressures and very high initial 5 11.6 : 5 4 11.4 > O pH not fixed too : 5 11.2 - 0 0 0 ~ 0 ■ pH fixed at 12 + OJÎ so. : 0 0 : 0 11.0 0 r " 2 r 0 ^ : 0 10.8 1 : 10.6 r f ' - 3 _ 10.8 - - 0

L 0 7.5 4 - - Oo g - 0 B a 0 % 5.0 0 0 U 0 CO Q : oP s 2.5 0 0 iPiVl i - i 1 1 1 1 1 1 1 0.0* ■ ■■II'»I I I I I I 0 4 8 12 16 0 4 8 12 16 0 4 8 12 16 Initial ÂSO4 Concentration (mM) Initial AsO^ Concentrations (mM) Initial AsO^ Concentrations (mM)

Figure 32. Variation in concentrations of ions during AsO^ adsorption in ettringite. 61 As0 4 concentrations (> 72 mM) precipitated johnbaumite (Ca;(As 0 4 )3(0 H)). This mineral phase dissolves with a p’°*^Ksp of 39.60 (Appendix B).

In the current study, M1NTEQA2 calculations indicated that all ASO 4 adsorption samples were close to saturation with johnbaumite. However, this phase was not detected by XRD in any of the saturated and supersaturated samples (Fig. 3.3-1 b-e). In the absence of pH control, ettringite XRD peaks decreased in intensity (for e.g., the (100) peaks) with increases in sorbed

As0 4 (Fig. 3.3-la). Concomitantly, gypsum XRD peaks (7.63 and 4.28 Â) increased (Fig. 3.3-1 b

& c). When the sorbed ASO 4 reached 1.4 mol kg’*, ettringite peaks disappeared and only gypsum was evident in these samples (Fig. 3.3-1 c). At higher ASO 4 loadings, gypsum also completely disappeared and new solids precipitated. These latter samples showed some sharp XRD peaks with a broad uneven baseline, indicative of poorly crystalline material, which did not correspond to any solids reported in the ICDD manual (including johnbaumite). However, when solution pH was maintained at 12.0 (± 0.2), gypsum was absent in XRD patterns and ettringite persisted at all examined ASO 4 loadings (1.44 - 7.44 mM or solid phase concentrations of 0.34 - 1.66 mol kg’*)

(Fig. 3.3-1 d & e). XRD scans of these samples also exhibit a decrease in intensity for ettringite

(100) reflections, and poorly crystalline phases with increases in ASO 4 concentrations.

SEM indicated that the synthetic ettringite was fine grained and the maximum size (length) was approximately 2 pm (Fig. 3.4a). Gypsum crystals were identified in samples containing more than 0.1 mol kg’* ASO 4. At higher ASO 4 loadings, gypsum crystals exhibited dissolution features such as etch pits (Fig. 3.4b), and lost all of their clear euhedral geometry. The precipitates at the highest solid-phase ASO 4 concentrations were fine grained (< 1 pm) and difricult to image with

SEM. 1. A dsorption 2. C oprécipitation

I WAliJL/X^^Wvi I

I ÜH1

-1-Ji.^UL.i^J,. 10 20 30 40 50 10 20 30 40 50

D % r e e 2 8 D e g r e e 2 6

Figure 3.3. XRD of AsO# reacted ettringites. 3-1. Adsorption samples, a: pure ettringite. b & c: reacted with 2.16 and 5.4 mM of ASO 4 (no pH control), d & e: exposed to 2.88 and 7.2 mM of ASO 4 (pH=12 ± 0.2). Suspension density: 4 g L \ 3-2. Coprecipitation samples, a) pure ettringite. b, c & d: reacted with 0.36,1.42 and 2.52 mM of ASO 4. Suspension concentration: 1.2gL '\ pH: 11.9 + 0.3. Figure 3.4. SEM micrographs of ASO 4 reacted ettringites. a: synthetic ettringite, b: ASO 4 adsorbed ettringite (5.76 mM), c and d: coprecipitated As 0 4 -ettringite at initial ASO 4 concentrations of 0.144 and 1.44 mM. a 64 Effect of Suspension Concentration on AsOi Adsorption

Changes in the suspension concentration greatly affected ASO 4 sorption by ettringite.

Increases in suspension density resulted in enhanced ASO 4 sorption and aqueous Ca, Al and SO 4 concentrations (Fig. 3.5); and solution pH decreased by a log unit (from 11.4 to 10.5). These changes in Ca, Al and SO 4 were observed even in the absence of any added ASO 4, suggesting that this may have been due to ettringite dissolution. The solutions reacted at all suspension densities were close to ettringite saturation. Except for the lowest suspension concentration (4 gL"'), Ca/Al and Ca/S 0 4 ratios of the reacted solutions (approximately 2.3 and 1.70 respectively) were slightly smaller than those of the precipitate (2.97 and 2.22 respectively), indicating slight incongruency in the dissolution at high suspension concentrations. This may be attributed to the formation of some form of surface polymers or a secondary precipitate on ettringite surfaces. Similar results have been observed for the dissolution of feldspar (WOLLAST and CHOU, 1992), and olivine (CASEY etal., 1994 ).

The positive correlation of ASO 4 retention with suspension density is presumably due to ASO 4 interactions with ettringite surfaces. This may suggest that the surface area of the solids has a major influence on aqueous ASO4 in the examined concentration range. In addition, with increases in ASO4 uptake, aqueousSO4 concentrations also increased while Al and Ca remained constant at each suspension concentration. Despite the fact that the observed relation between suspension concentration and ASO4 soiption is indicative of surface interactions, the increase in aqueous SO4 with ASO 4 loadings suggests at least two types of ASO 4 interactions in ettringite: 1) surface reactions and SO4 release is to balance the excess negative charge from sorbed ASO4 and 2) SO4 release may be due to ASO4 channel substitution. 5 Suspension Concentration

. ^ j A A A (gL') 5 11.0 • 4 ^ 20 rÔ 3 . I f ■ ■ 10 ▼SO

10.5 F v * Z * 2 1 ■ • A - 4

0.12 -

3 * I I F _ . §■ 0.09 f m m m > f » . * * S . * rT 0.06 F o ' 2 » < 1 CO F i ♦ ♦ 0.03 » — Î’ 1 0.00 0 Ill'll J - l 1 1 1 1 1 1 1 1 1 1 1 1 BLx 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8

Initial AsO^ Concentration (mM) Initial AsO^ Concentration (mM) Initial AsO^ Concentrations (mM)

Figure 3.5. Effect of suspension density on equilibrium concentrations of ions.

8i 66 Effect of dH on AsO^ Adsorption

Changes in pH can affect AsOa sorption since AsÛ 4^ protonates significantly below pH 11.5.

However, little to no effect of pH on AsO# sorption was observed for the samples with pH > 10.5

(Fig. 3.2). At pH 10.5, less AsO# sorption occurred as compared to high pH experiments. This likely resulted from ettringite dissolution since this mineral becomes markedly less stable at pH <

10.7. Solution pH did have a strong effect on aqueous Ca, Al and SO 4 across the entire range examined in this study. When pH was maintained at 12.0, ASO 4 sorption caused only small changes in dissolved Ca, Al and SO4 concentrations (Fig. 3.2).

ASO 4 Coprecipitation

ASO4 uptake in the coprecipitation experiments was several orders of magnitude greater than in the adsoiption experiments (Table 3.2), even at the lowest suspension concentration (1.2 gL ').

In addition, XRD showed that ettringite was the only stable crystalline phase at all but the highest

ASO4 concentrations. Ettringite was also stable when ASO 4 loadings exceeded the maximum adsorption capacity ( 0.11 mol kg ') estimated on the basis of the exposed ettringite surface area.

This suggests that ASO4 coprecipitation resulted in channel substitution. When essentially no SO4 was present during coprecipitation, a new mineral was precipitated. The stmcture, chemistry and

XRD of this new mineral phase were not reported in the ICDD powder files. Even though ettringite was a stable phase over a wide range of ASO 4 concantrations, a fiat baseline was not present in the ettringite XRD with higher solid phase ASO 4 concentrations (Fig. 3.3-2). This indicates the presence of a poorly crystalline to amorphous phase in addition to ettringite. SEM showed that the grain morphologies of the adsoiption and coprecipitation samples were different.

The grain size (length) of coprecipitated ettringite was approximately 6-10 pm in length as 67

Table 32. ASO4 coprecipitation in Ettrin^te. pH fixed at 11.9 + OJ Sample pH [ASO 4] (mM) [SO4] (mM) (± 0 .1) Initial Equilib. Initial Equilib. I 12.8 2.519 0.013 0 .0 0 0 .0 0 2 11.2 1.440 0.009 10.4 1.21 3 11.9 0.720 0.008 5.21 0.38 4 12.1 0.720 0.009 5.21 0.36 5 11.5 0.360 BDL 5.21 0.68 6 12.0 0.288 BDL 4.44 0.28 7 11.4 0.360 BDL 5.08 0.81 8 12.0 0.216 BDL 5.74 0.24 9 12.5 0.144 BDL 3.60 0.05 10 11.9 0 .0 0 0 0 .0 0 0 5.11 0.25 pH not controlled Sample pH [ASO4] (mM) [SO4] (mM) (± 0 .1) Initial Equilib. Initial Equilib. 1 12.7 2.52 0.025 0 .0 0 0 .0 0 2 12.7 1.44 0.026 1.56 0.07 3 12.6 0.72 0.026 2.60 0.11 4 1 2 .6 0.36 0.007 2.60 0.17 5 12.6 0.288 BDL 2.60 0 .1 0 6 12.6 0.216 BDL 3.65 0.09 7 12.6 0.360 BDL 3.65 0 .1 0 8 12.6 0.144 BDL 3.65 0.14 9 12.6 0.007 BDL 3.65 0.13 10 12.6 0.000 0.000 3.65 0.09

BDL: Below detection limit 68 compared to the 2 pm of adsorbed ettringites. However, with increases in AsO# loadings, the grain

length of coprecipitated samples decreased to approximately 2 pm (Fig. 3.4 c & d). Moreover, the

grain size variation was also large (Fig. 3.4d) at very high levels of ASO 4 incorporation (AsOa- 2

mol kg'^), possibly indicating a mixture of reaction products.

Coprecipitation experiments conducted at two pHs, 12.5 (+ 0.2) and 11.8 (+ 0.5), showed

that pH had no effect on either equilibrium ASO 4 concentrations (Table 3.2) or on the crystal

structures of the precipitates. This was in direct contrast to the pH-dependent ASO 4 adsorption onto ettringite.

ASO 4 Desorption

Desorption of ASO4 from As04-ettringite was evaluated on samples from the adsorption and

coprecipitation experiments. These studies were conducted to understand the rate and ionic

strength dependency of ASO4 desorption; SO4 was used as a displacing ion in all of these

experiments.

Kinetics of AsOdDesorption

Kinetic studies were conducted only on a coprecipitated sample since both adsorption and

coprecipitated samples exhibit similar behavior at the solid phase ASO 4 concentration of 0.30 mol

kg'* used. These experiments showed no evidence of ASO4 release with increases in SO 4 input and

this behavior did not change with time (Fig. 3.6). Calcium and Al concentrations also remained

constant with time and with increases in SO4 ad&tions. Although smaller pH changes (± 0.3) were

observed with time, they were within the pH measurement error. The desorption results suggest a

rapid equilibrium for ASO4-ettringite, and the sorbed ASO 4 is essentially not desorbable. 10.8

♦ i ♦ I o . 10-5

«5 10 Initial SO^ Concentration 10.2 (mM) 4 • No addition ▼ 10.42 ■ 2.08 ♦ 20.83 A 5.21

ô 2

g Ll-I I I I I I I I I I I I I I I I I t ■■■■ I ' 0 100 200 300 400 0 100 200 300 400 Time (h) Time (h)

Figure 3.6. Kinetics of AsO^ desorption from coprecipitated As-ettringite. Solid phase AsO^ concentration; 0.30 mol kg '.

$ 70 Ionic Streneth Dependence of Desorption

Desorption studies were conducted on both adsorbed (8.3 mmol kg ') and coprecipitated

(0.18 mol kg ') As 0 4 -ettringites, at different ionic strengths (0.1-1.0 mole L"' NaCHO#) and SO 4 loadings (0-20.8 mM). The rationale behind selecting these two samples was that ASO 4 interactions were expected to be different, i.e. dominantly surface interactions in the adsorbed sample as compared to channel substitutif in the coprecipitated sample, at these solid phase ASO 4 concentrations. During desorption, solutif chemistry of the adsorbed and coprecipitated samples was invariant with increases in ionic strength and SO4 loadings for all ions except ASO4. It was also found that final filtrate SO 4 concentrations were higher than added SO 4. This increase was approximately 1 mM for SO 4 loadings < 5.2 mM, and greater than 2 mM when SO 4 loadings were

> 10.4 mM. This SO 4 increase also did not correlate with increases in ionic strength. Since Ca and Al concentrations remained constant with increases in ionic strength and SO 4 loadings, the possibility of excessive solid dissolution and consequent increases in aqueous SO4 can be ruled out.

The experimental results on the adsorbed samples indicate that ASO 4 concentration remained constant with increases in ionic strength and SO4 concentrations (Fig. 3.7a). The measured ASO4 concentrations during desorption were similar to the equilibrium concentrations measured during adsorption experiments (Table 3.1). XRD and SEM indicated no changes in sample mineralogy.

On the other hand, ASO 4 desorption from the coprecipitated sample resulted in a release of As 0 4 to solution (Fig. 3.7b). This observed ASO4 release increased with increases in ionic strength and SO4 loadings, and reached a plateau at high ionic strengths (> 0.5 mole L‘‘). The released ASO 4 was less than 6% of the total ASO 4 in the solid. ASO 4 desorption in this experiment can not be attributed to dissolution of As0 4 -ettringite because dissolution would have resulted in increases in

Ca and Al in addition to ASO 4. When similar desorption studies were conducted on a concentrated coprecipitated As 0 4 -ettringite (0.30 mol kg '), no apparent release of ASO 4 was observed 0.025 0.04 - SO^mM • 0.5 ▼ 10.4 0.020 __

_ m 2.6 ♦ 20.8 0.03 A 5.2 0.015 — 1r

9 0.010 I 0.02

1.3 ▼ 10.4 0.005 5 2.6 4 20.8 0.01 5.2 0.000 ' ' ■ ■ I ■ ' ' ' I ■ ■ ■ ■ I ■ ■ I ■ 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Ionic Strength (mol L'b Ionic Strength (mol^ L'^)

Figure 3.7. Ionic strength dependant desorption of AsO.. a: adsorbed and b: coprecipitated ettringite. 72 (Fig. 3.6). At this solid phase ASO 4, the concentration of desorbed AsO# remained constant with increases in SO#. This variable behavior in desorption of solid phase AsO# is presumably due to the differences in the type of AsO# interactions with ettringite, i.e. OS versus IS complexation.

ASO4 Interactions with Ettringite

The results of the sorption experiments suggest that AsO# retention by ettringite could be due to three different mechanisms: 1) surface complexation, 2) channel substitution for SO# and 3) precipitation of a Ca arsenate or a Ca-Al-arsenate phase. These mechanisms may take place simultaneously and information from these macroscopic experiments is not sufficient to distinguish them clearly. However, the dominant mechanisms can be identified from the available solution chemistry, XRD and thermodynamic data.

If channel substitution is the dominant reaction during adsorption, then SO# concentrations will increase with increases in AsO# uptake. Depending on the released SO# concentrations, channel substitution may also accompany changes in Ca and/or Al concentrations to balance the excessive AsO# negative charge. Precipitation and surface complexation are unlikely to exhibit this kind of behavior. Disappearance of characteristic ettringite peaks (d spacing of 9.72 Â) and simultaneous development of new peaks and uneven base line in the XRD patterns are indicative of precipitation of new phases at high AsO# concentrations (> 1.0 mol kg '). The absence of these

XRD changes of samples at low AsO# additions (~ 0.01 mol kg ') may not completely rule out precipitation. However, precipitation is unlikely because: 1) AsO# reactions with lime (Appendix

B) and gibbsite (Appendix B; ANDERSON et a/., 1976) produced much higher equilibrium concentrations than were observed here, thus eliminating the possibility of direct interactions with

Ca or Al or formation of their precipitates; 2) precipitation of an AsO# solid phase would have resulted in uniform AsO#^ activity in reacted solutions; and 3) spéciation calculations indicated 73 very low As04 ^' activities in equilibrium with ettringite as compared to the known Ca arsenates such as johnbaumite or Ca3(As04)2 .6H2O (Table 3.3). In summary, surface complexation and channel substitution at low ASO4 loadings (< 0.1 mM), and precipitation in concentrated samples

(> 0.5 mM) are the likely mechanisms of ASO 4 sorption by ettringite.

During adsorption, ASO 4 may only form surface complexes at very low concentrations (ASO4

<17 pM). The dependence of sorption on suspension concentration (or surface area) is in support of this mechanism. In addition, XRD suggests that the ettringite structure was preserved, and no new mineral phases were observed. Changes in ionic strength did not affect ASO 4 desorption from these samples (Fig. 3.7a) which suggests that ASO 4 binds strongly to the surfaces. Since ettringite surfaces are negatively charged in the experimental pH range (CHEN and MEHTA, 1982), specific adsorption must occur through ligand exchange, i.e. by replacing Al/Ca coordinated surface OH, or Ca coordinated H2O. However, Pauling’s Electrostatic Valence Rule predicts that

Ca should have more affinity for ASO 4 (Chapter I)

Increases in ASO4 adsorption may result in both surface complexation and channel substitution (at AsÜ4 loadings ~ 0.1 mM). This is because the moles of released SO 4 and sorbed

ASO4 were similar, and ASO4 loadings were close to the saturation of available surface site density

(0.11 mol kg'*) in ettringite. However, channel sites become less accessible at high ASO 4 loadings due to steric and charge constraints on ASO 4 diffusion into the channels during adsorption. Thus, once the surface and channel sites adjacent to the crystal edges become occupied, it is likely that a change in the reaction mechanism occurs, reducing the efficiency of ASO 4 removal fiom solution.

This may correspond to the first observed decrease in ASO 4 removal efficiency (Fig. 3.1) at initial

ASO4 concentrations of approximately 0.1 mM. As the ASO 4 concentration increases above 0.1 mM, ettringite stability decreases and may finally result in the formation of a Ca arsenate precipitate. This is supported by the increase in aqueous phase Al and SO4, and decreases in Ca 74

Table 3 J. As04^' activity in equilibrium with ettringite, CasCAsO^): .6 H2O and johnbaumite.

Sample pH log(Ca'+) p (As04^)

ettringite* johnbaumite Ca3(As04>2 .6H2O 1 12.7 -2.49 <9.0*-5.0* 8.62 5.73

2 11.3 -2 .8 6 < 9.0*- 7.53 5.16 4.0* 3 10.9 -3.71 3.54 5.98 3.89

* AsO/'activities in equilibrium with ettringite are variable, and depend on the pH, initial ASO 4 and SO4; and suspension concentrations.

* As 0 4 ^' activity in equilibrium with ettringite when initial ASO 4 ~ 10 pM.

* As 0 4 ^ activity in equilibrium with ettringite when initial ASO 4 ~ 5 mM. 75 concentrations with increased ASO 4 loadings. XRD and SEM also support this hypothesis.

Precipitation may also be the reason for the increase in efficiency of ASO 4 removal (Fig. 3.1) in the concentration range of 1 to 10 mM. In this concentration range, the solutions may be saturated with respect to an amorphous phase suggested to be present from XRD. With further increase in

ASO4 loadings, available Ca decreases and this results in reduction of ASO 4 removal efficiency.

As expected, coprecipitation resulted in larger ASO 4 uptake than adsorption and this can be attributed to the greater accessibility of channel sites during coprecipitation. This accessibility of channel sites during coprecipitation makes channel substitution preferable over surface sorption because of the net positive charge of columns compared to the negative charge on the surfaces. If

ASO4 substitutes inside the channels, then it can be present: 1) as a completely solvated ion (or OS complex at G - type site. Fig. 1.2) and/or 2) as an IS complex by coordinating directly to column

Ca. HAYES et al., (1987) used the ionic strength dependence of oxyanion soiption to demarcate these two types of complexes (for e.g., in the case of SeOs and Se 0 4 sorption on Fe oxides). This interpretation was also supported by spectroscopic data (HAYES et al., 1987).

During the ionic strength dependent desorption of sorbed ASO 4 from arsenated ettringite, both

SO4 and CIO 4 can substitute insicte ettringite channels and replace ASO 4. However, it is likely that

SO4 played the dominant role in displacing ASO 4 because of its greater charge and smaller size. In the present study, ASO 4 desorption from a dilute, coprecipitated As 0 4 -ettringite (As 0 4 = 0.18 mol kg'*) was found to be dependent on ionic strength and SO 4 concentration. Sorbed ASO4 was not completely released and desorption reached a maximum at an ionic strength of 0.5 mole L *.

Desorption rate decreased with further increases in ionic straigth. These experiments suggest that

ASO4 is present as an OS complex inside the channels and only the channel sites close to the edges of the mineral are available for desorption, or a sluggish solid state diffiision of SO 4 which might resulted in only partial exchange of solid phase ASO 4. In contrast, desorption of ASO 4 from a 76 concentrated As 0 4 -ettringite (ASO 4 = 0.30 mol kg'*) resulted in no release of ASO4 into solution,

which suggests that ASO 4 binds specifically to the columns (without hydration) at high ASO 4

concentrations.

The relatively larger size and charge of ASO4 over SO4 restricts the substitution of ASO4

inside the channels without affecting ettringite structure. In addition, solid phase SO 4 may play a

dominant role in the behavior of ASO 4 and in stabilizing ettringite. When solid (Aase ASO 4 in

coprecipitated As 0 4 -ettringite is small, ettringite structure is well preserved by the high

concentrations of SO4 and allows ASO4 as an O S complex. However, at high incorporations of

ASO4, the charge and the size of ASO 4 may play a dominant role. With further increases in solid

phase ASO 4, it may produce IS complexes and form a new precipitate. During adsorption, ASO 4

interactions with ettringite may dominantly be at A, B, C, D, E, and/or F type - sites; and

coprecipitation includes ‘G’ - type sites in addition to these. The suggested mechanisms are however tentative and spectroscopic studies are necessary to further confirm them.

CONCLUSIONS

The results presented here demonstrate that ASO 4 can be readily sorbed by ettringite.

Arsenate solubility in equilibrium with ettringite is much lower than that of the other known Ca arsenate minerals in the experimental pH range (Table 3.3). The possible reaction mechanisms by which ettringite can retain ASO 4 are: 1) interactions with surface functional groups on the exterior and interior of the crystals (at A, B, C, D, E, and F - type sites), and 2) substitution into ettringite channels with displacement of SO4 through anion exchange (‘G’ - type sites). Predominance of one mechanism over the other likely depends on the type of sorption and solid phase SO4 concentrations. These mechanisms result in formation of O S and IS ASO4 complexes. Although

O S complexes are easily replaceable from mineral surfaces in general, ASO4 O S complexes in 77 coprecipitated ettringite are not exchangeable due to their presence inside the channels. At high solid phase ASO 4 concentrations (> 1.5 mol kg'*), ettringite breaks down and some form of Ca arsenate precipitates. Further studies are necessary to identify these phases and to understand the mineral-equilibria in this concentration regime. CHAPTER IV

INTERACTIONS OF Cr 0 4 WITH ETTRINGITE

INTRODUCTION

Chromium (Cr) occurs in natural environments as Cr (HI) and Cr (VI). Its presence in soil is mainly attributed to the weathering of chromite, or chromite containing basic and mafic rocks.

However, tanning, electroplating, painting, drug and dye manufacturing, petroleum refining, petrochemicals, steel and non ferrous metal industrial operations have resulted in elevated Cr concentrations around the world (ADRIANO, 1986). The carcinogenic effects of Cr (VI) have been well established and these health risks may be magnified by its h i^ solubility and mobility compared to Cr (III), and oxidation of Cr (III) in the presence of oxidants such as Mn oxides

(JOHNSON and XYLA, 1991). Reduction of Cr (VI) to the non-toxic Cr (III) in the presence of soil organic matter and Fe(II) minerals is possible, but Cr (VI) presence in natural waters indicates the kinetic limitations of these reactions (KENT et al., 1994; ANDERSON et al., 1994).

Cr(VI) exists as HCr0 4 and Cr 0 4 ^ (pKa = -0.2, 6.51) at neutral pH (SMITH & MARTELL,

1976; BROOKINS, 1988). Previous research has shown that these oxyanion species form OS complexes with mineral surfaces at near neutral pH, and protonation of CrO#^ has little effect on

Cr (VI) adsoiption and desorption kinetics on mineral surfaces (CHANG et al., 1994). In addition, Cr 0 4 (HCr0 4 and Cr 0 4 ^) soiption was observed to be affected by the presence of other

78 79 oxyanions such as SO4, ASO4 and PO4. Competitive adsorption of these ligands on the surfaces of

minerals reduced Cr04 sorption in soils and flyashes (APPLETON et al., 1989; TH EIS et al.,

1989; K ENT et al., 1994). C1O4 mobility is expected to be higher in alkaline environments

because of its low first pKa (HAYES, 1987). Although high concentrations of Cr04 have been

observed in alkaline flyashes and FGD wastes, its interactions with dominant and potential sorbent

minerals such as ettringite are not clearly understood. Oxyanion-substituted ettringite synthesis

experiments (HASSETT et al., 1990; KUMARATHASAN et a l, 1990; POLLM ANN et al.,

1993) showed that C1O4 can substitute for SO4 in ettringite channels. Solid solution formation

between SO4 and Cr04 end-member ettringites is limited by the synthesis procedure (presence or

absence of sucrose) and pH. Preferential ettringite formation over Cr04 end-member may result in

pure ettringite precipitation in association with Cr04-S04 solid solutions. High SO4 concentrations

are commonly observed in coal combusticm residues and alkaline waste materials and thus are

expected to have a strong influence on Cr04 uptake by ettringite. The main objectives of this

chapter are to: 1) estimate the extent of C1O4 sorption by ettringite during adsorption and

coprecipitation, and 2) evaluate the desorption of sorbed Cr04in the presence of high ionic strength

SO4 solutions.

EXPERIMENTAL MATERIALS AND METHODS

Materials

Synthesis ofSQj-Ettrimite

Sulfated ettringite was synthesized by mixing CaO and Al 2(S0 4 )) in stoichiometric proportions in

10 %, COa-free sugar solution. Details of the procedure and precipitate characterization are described in Chapter II. These synthetic materials were used in Cr 0 4 adsoiption experiments. 80 Synthesis of CrOd-Ettrineite

A synthetic solid solution of C 1O4 and SO 4 ettringite was used in the desorption experiments.

The solid phase Cr 0 4 concentration was 16.7 mmol kg'* and the molar ratio of C 1O 4 to SO4 was

approximately 0.02/2.98. This sample was prepared by mixing 6 g of CaO with 4 L of COz-free deionized water containing 0.65 mM of K 2C1O4 and 3.13 mM of Al 2(S0 4 )3. Sucrose was not used

in this synthesis because its presence induced a miscibility gap in Cr 0 4 ettringite solid solution

(PÔLLMANN et al., 1993). After 48 h, the precipitated solids were separated and air-dried in a desiccator. The XRD of these solids were similar to that of the S 0 4 -end member ettringite. SEM indicated that the grain lengths were approximately 12 pm. This larger grain size as compared to synthetic ettringite (2 pm) was also observed in Cr 0 4 -free ettringite synthesis and was attributed to the absence of sucrose. Similar behavior was also observed in previous sucrose-free ettringite synthesis procedures (POLLMANN etal., 1993).

Methods

CrOd Sorption Studies

Sorption experiments were conducted to simulate adsoiption and coprecipitation. Preliminary adsorption studies showed that Cr 0 4 uptake by ettringite was instantaneous and solution composition was invariant from < 1 min to 96 h of observation. So for convenience, a 24 h reaction time was used for all soiption studies. Adsoiption experiments were conducted by reacting 0.1 g of synthetic ettringite with 25 mL of C1O4 solutions (0.05-7.94 mM) in polypropylene centrifuge tubes (suspensim density 4 g L'*), For coprecipitation, ettringite was precipitated from Cr 0 4 -containing solutions by mixing CaO with different ratios of K 2Cr0 4 and

Al2(SÛ4)3 solutions (suspension density 1.2 g L'*). The samples were continuously agitated on a 81 reciprocating shaker at 298 (± 1) K. After reaction, the samples were centrifuged at 7000 rpm for

15 min. After measuring supernatant pH, the samples were filtered with 0.2 |mi pore size

polycarbonate filters and analyzed as discussed below. Solids were studied with XRD and SEM to

identify the mineral fraction. Solid phase C1O4 was estimated from mass balance calculations.

Cr04 Desorption Studies

Ionic strength-dependent C1O4 desorption from ettringite was studied for coprecipitated ettringite. The replacing ion was SO4. The details of ionic strength adjustments and other experimental procedures are discussed in Chapter III.

Anahses

The concentrations of Ca and Cr in the reacted solutions were analyzed with a Perkin Elmer

303GB AAS and the concentrations of all other ions were determined with ICP. Instrumental description is provided in Chapter II.

RESULTS AND DISCUSSION

The soiption and coprecipitation experiments were meant to mimic actual waste environments where adsoiption on ettringite surfaces is presumably a dominant process when CrÛ 4 containing wastewaters interact with already existing ettringite. In contrast, direct precipitation of ettringite from Cr0 4 containing leachate (coprecipitation) may result in formation of a solid solution of

Cr0 4 -S0 4 ettringite, C1O4 and SO4 end-member ettringites, and/or a Cr 0 4 precipitate. Although adsorption can exhibit similar phenomena, replacement of channel SO4 by Cr0 4 is limited by the rate of C1O4 diffusion into the crystal lattice. 82 Cr04 Adsorption

Adsorption experiments were conducted at a constant suspension density (4 gL'') and for a range of C1O4 concaitrations (0.086-8.62 mM). Despite the lack of pH and ionic strength control, these samples showed little variation for these two parameters (Table 4.1). With increases in Cr 0 4 uptake, the concentrations of released SO4 increased while Ca and Al remained constant in the reacted solutions. The molar concentratirai of released SO4 was lower than the sorbed Cr 0 4 (Fig.

4.1). If C1O4 replaces SO4 from the channel sites, then the increased aqueous SO4 should have been equal to the sorbed C1O4. In addition, displacement of channel SO4 started even before aU the surface sites were saturated (0.11 mol kg''). If we assume that the released SO4 is equal to the chaimel substituted C1O4, then the surface reacted C1O4 increased with increases in C1O4 uptake.

MINTEQA2 spéciation indicated that Cr 0 4 ^ was the dominant aqueous Cr0 4 species in the experimental pH range. Reduction of existing Cr (VI) to Cr (III) or any other oxidation states is not expected, since there is no reductant in the syston. Saturation index (SI = log (lAP/Ksp)) calculations conducted with the same computer code gave no indication of any Cr 0 4 solids in the system. The SI calculations for C1O4 are limited by the lack of free energies of formation of several Ca and Al chromate phases. The thermodynamic database of this program only lists a solubility product constant tor CaCr0 4 , which was found to be undersaturated in Cr 0 4 reacted solutions in equilibrium with ettringite.

XRD of Cr0 4 -sorbed ettringite samples neither showed new peaks corresponding to other mineral phases nor changes in ettringite peak intensities. This suggests the absence of new minerals in the reacted solids with increases in Cr 0 4 loadings. As shown by SEM, sorbed Cr 0 4 did not affect ettringite crystal size and overall grain morphology (Fig 4.2a, 4.2b). These experimental results suggest that sorbed C 1O4 interacted with ettringite external surfaces and also replaced channel SO4. Table 4.1. Chromate adsoiption in ettringite®.

Initial Cr0 4 Equilibrium Concoitrations (mM)

Concentration pH Ca Cr04 SO4 (mM) (± 0 .1) Mean S.D Mean S.D Mean S.D G 11.4 2.46 0.24 0 0 1.43 0 .0 2 0.086 11.2 2.32 0.23 0.054 0.003 1.45 0.03

0.34 11.2 2 .2 2 0.19 0.331 0 .0 1 2 1.43 0.01

0.69 11.2 1.99 0.06 0.653 0 .0 1 2 1.36 0.01

0 .8 6 11.3 2.2 0 0.19 0.835 0.011 1.34 0 .0 2 2.59 11.2 2.30 0.27 2.46 0.03 1.52 0.06

4.31 11.2 2.33 0.07 4.01 0.1 1.51 0.01 6.03 11.3 2.57 0.31 5.71 0 .0 2 1.64 0 .0 2 7.76 11.2 3.00 0 7.3 0.08 1.68 0.04

8.62 11.3 2.76 0 .2 2 7.94 0.09 1.69 0.01

@ pH was iMt controlled, and the suspension concentration was maintained at 4 g L \ Concentration of Al remained constant at 1.11 ±0.1 mM with increases in Cr 0 4 loadings.

S.D Standard deviation.

82 84

1.0

0.8

0.6

0.2

0.01 0.0 0.2 0.4 0.6 0.8 1.0

A dsorbed CrO^ (mM )

Figure 4.1. CrO^ adsoiption in ettringite. 0

Figure 42. SEM micrographs of C 1O4 reacted ettringites. a: synthetic ettringite, b: Cr 0 4 adsorbed ettringite (8.62 mM), c and d: coprecipitated Cr 0 4 -ettringite at initial C 1O4 concentrations of 1.29 and 2.15 mM. % 86

Cr0 4 Coprecipitation

Little change in pH (12.4 ± 0.1) or in dissolved Ca or A1 concentrations occurred as CiOa concentrations increased during ettringite coprecipitation. Cr 0 4 uptake was also much larger during coprecipitation than adsoiptioa The coprecipitation experiments, conducted at variable

SO4 concentrations while keeping Cr 0 4 constant, showed that Cr 0 4 sorption decreased with increases in S04 /Cr0 4 ratio. This suggests that SO4 is preferred to Cr 0 4 for the channel sites

(compare samples C0P4 and C0P5 in Table 4.2). Although Ca chromate precipitates were not observed in M1NTEQA2 SI calculations, microscopic studies indicated heterogeneity in the grain size distributioa SEM of the samples indicated that the grain length of the coprecipitated samples was 10-12 pm when the molar ratio of Cr 04/S0 4 in reacted solution was < 0.3 (Fig. 4.2c). With increases in C1O4 concentration, the crystal size decreased and the heterogeneity in grain size distribution increased (Fig. 4.2d).

XRD showed no new phases when C1O4 concentrations were less than 1.29 mM (Fig. 4.3).

However, at high concentrations the (100) peaks of ettringite (at a d-spacing of 9.72 Â) decreased in intensity and a new peak developed at lower 20 (10.33 A). These changes persisted at all high solid phase C1O4 concentrations (> 0.63 mol kg'') and the observed XRD patterns do not match any of the reported C1O4 phases in the ICDD manual. Other known Ca or A1 chromate compounds

(for e.g. chromatite and its hydrated phases and polymorphs, hydrated A1 chromate, etc.) do not exhibit strong reflections at 10.33 A These XRD patterns are also significantly different from that of Cr0 4 ettringite reported by PÔLLMANN et al. (1993). However, their synthesis used an equilibrium time of 90 days as compared to 24 h in my study. The larger size of Cr 0 4 over SO4

(KRESTOV, 1991) may not cause this increase in ettringite d-spacings, since the difference between the two ions is much lower than the observed increase (0.61 A). Although probable 87

Table 4.2. Chromate coprecipitation in ettringite®.

Sample pH Ca Cr04 SO 4 (mM) (± 0.04 mM) (± 0.04 mM) (±0.1) Concn. Initial Equilib. Initial Equilib. Copl 12.3 13.1 3.35 2.62 0.0 0.0 Cop2 12.3 14.5 2.15 1.48 0.52 0.13 Cop3 12.3 14.6 1.73 1.23 1.55 0.15 Cop4 12.4 17.7 1.29 0.98 1.55 0.26 CopS 12.4 18.2 1.29 1.15 2.08 0.33 Cop6 12.5 18 0.86 0.87 2.60 0.23 Cop7 12.4 16.6 0.65 0.63 3.13 0.14 CopS 12.4 15.4 0.463* 0.448* 3.51 0.29 Cop9 12.4 13.8 0.411* 0.408* 3.96 0.30 CoplO 12.4 14.7 0.0 0.0 2.74 0.29

® pH was not controlled. Suspension concentration was maintained at 1.2 g L \ Concentration of A1 was below < 0.07 mM. Ca, CrO^ and SO4 in the table represent total measured concentrations.

* The measurement error was less than 0.2 pM. 1000 \- k J L

800 k I I 600 I.

400

200

i— I. I I 10 20 30 40 50

D e g r e e 2 8

Figure 4.3. X-ray diHraction patterns of coprecipitated Cr 0 4 - ettringite. The solid phase Cr0 4 concentrations increase in the order: a

the gallery width, this would have resulted in much larger d-spacing than observed, even if we

assume linear bonding for Cr-O-Cr in CrzO?^ and its attachment parallel to the columns.

Interestingly, when Cr 0 4 -containing ettringites were exposed to the atmosphere for a long time (~20 days), the XRD patterns of the samples showed stronger reflections at 29.4 °20, indicating the probable presence of calcite. FTIR also showed similar results, and the CO3 concentrations increased with increases in solid phase C 1O4 concentration. This may be due to some catalytic effect of sorbed Cr 0 4 on carbonation reactions. Mechanistic information for this behavior is difficult to predict from the available data and more experiments are necessary to understand the carbonation reactions of ettringite in the presence of C 1O4.

C 1O 4 Desorption

Desorption experiments on coprecipitated Cr 0 4 ettringite showed no C1O4 release upon exposure to increases in ionic strength (0 .1 -1 .0 mole L'*) and aqueous SO4 concentrations (0 -2 0 .8 3 mM) (Fig. 4.4). Ca and A1 concentrations were not affected by changes in ionic strength or SO4.

Chromate concentrations also did not change with ionic strength and SO 4, which shows the inability of SO4 to interact with and displace C1O4 from ettringite. XRD and SEM also indicated no changes in the crystallinity and morphologies of ettringite during C1O4 desorption.

Cr04 Interactions with Ettringite

In contrast to the ASO 4 system (Chapter 111), C1O4 uptake by ettringite was retarded by aqueous SO 4 during adsorption and coprecipitation. This may be due to the high selectivity of channel sites for SO4 over C1O4. In addition, coprécipitation resulted in more Cr 0 4 uptake than 90

0.050 -

0.045 -m * e -▼ A • T -m ■ ^ 0.040 m m $

§ -• - # 1 = 0.1 mol, L'^ - ■ I = 0.3 mol, L* 0.035 A 1 = 0.5 mol, L* - ^ 1 =1.0 mol, L'*

0.030 —1—t—1 1___ 1.1 1___ 1___ L , .1, 1___ 1___ L_l_ 10 15 20 25 [SOJ (mM)

Figure 4.4. CrO^ desorption from coprecipitated CrO.-ettiingite. 91 adsorption, which indicates that C1O4 migration into the channels and SO4 replacement is the controlling step in Cr 0 4 adsorption.

The dominant sorption mechanisms for Cr 0 4 adsorption can be attributed to two processes; 1) adsorption on ettringite external surfaces, and 2) replacement of channel SO4. This reasoning is based on the adsorption data which shows that the moles of sorbed C1O4 is larger than the moles of released SO 4, and the difference between the two may represent the concentration of surface C1O4 species. In addition, channel substitution began before all the surface sites were saturated (0.11 mol kg'^), which may indicate that only certain surface sites (A, B, etc.) are reactive for Cr 0 4 and their availability iiKreases with increases in C1O4 activity. Externally sorbed Cr 0 4 may be complexing either with surfacial Ca or A1 polyhedra, and spectroscopic measurements are necessary to identify the actual mechanisms and the type of complexing surface sites.

Coprecipitation resulted in higher C1O4 uptake than adsorption. Although SO4 is preferred to C1O4 by the channels, the identified high C1O4 sorption during coprecipitation is due to the rapid precipitation and growth of ettringite crystals. Sirtce both of these ions have the same charge and similar radii (thermochemical radii of SO4 and Cr0 4 are 230 and 240 pm respectively), substitution of these ions should not affect ettringite structure. The observed increase in the d-spacings (~ 0.6

Â) o f ( 1(X)) ettringite reflections at very h i^ solid phase Cr04/S04 ratios may not correspond to any probable configuration of C1O4 or Cr^O? species, and this may reflect the formation of a new solid phase. In addition, SEM showed a heterogeneous grain size distribution at high solid phase

C1O4/SO4 ratios, which indicates the poisoning effect of C1O4 on ettringite growth. This may further support the above hypothesis. Absence of correlation between C1O4 desorption and the ionic strength of the medium indicates IS nature for Cr0 4 in ettringite. However, spectroscopic studies are necessary to better understand the C1O4 complexes on ettringite surfaces and in the channels. CHAPTERV

VIBRATIONAL SPECTROSCOPY OF OXYANIONS IN ETTRINGITE

INTRODUCTION

Sorption studies of AsO# and C1O4 on ettringite have suggested that these oxyanions can substitute inside the channels, form complexes on external surfaces; or form precipitates with ettringite constituents (Chapters III & IV). The mechanistic information extracted from the macroscopic investigations in the previous chapters is tentative, however, and specific mechanisms

(e.g. IS or OS complexation) can only be ascertained using spectroscopic methods. Among the different spectroscopic tools available, vibrational spectroscopy has been used extensively in surface science because it is sensitive to small changes in local molecular environments.

Available literature on the theory and application of this technique in probing molecular environments is large (e.g. HERZBERG, 1945; SZYMANSKI, 1967; FARMER, 1974; Me

MILLAN, 1985; NAKAMOTO, 1986; JOHNSTON, 1990; URBAN, 1993; FERRARO and

NAKAMOTO, 1994; and others). This technique has been applied successfully to study the sorption of inorganics (RUSSELL et al, 1974; RUSSELL, 1979; PARFITT et a l, 1975;

HARRISON and BERKHEISER, 1982; TWU and DUTTA, 1989; TEJEDOR and ANDERSON,

1986 and 1990; WERSIN e ta l, 1994) and organics (JOHNSTON et al., 1991; TEJEDOR et al,

1990 and 1992; BIBER and STUMM, 1994; HUG et al, 1994) on mineral surfaces and to the

92 93 Study of glasses or poorly crystalline materials (CAMBIER, 1986; AMONETTE and RAJ, 1990;

GENGEgra/., 1995).

Oxyanions such as S04^, As04^' and Cr 0 4 ^ are tetrahedral in alkaline solutions

(NAKAMOTO, 1986) and exhibit four fundamental vibrations in the infrared region: Ai

(symmetric stretching, Vi), E (symmetric bending, Vz), and 2F (antisymmetric stretching and bending, V 3 and V4 respectively). Protonation (pH effects), close proximity of another anion

(concentration effects) and complexation of these oxyanions with mineral surfaces (or a cation) affect the force fields of oxyanion tetrahedra and lower their symmetry from Td to either Csv, Czv or Cl (Fig. 5.1). The IR and Raman selection rules for appearance of bands are distinctly different

(COTTON, 1971; GRAYBEAL, 1988; NAKAMOTO, 1986). This reduction in symmetry splits the peaks of doubly and triply degenerate vibrations, and latent IR bands will appear in the vibrational spectrum. The amount of splitting and the number of Vz, V 3 and V4 peaks, and shifts in

Vi vibrational mode, reflect the structure of a complex, and the probable complexing ion or molecule.

The molecular symmetry of oxyanions is lower when adsorbed on mineral surfaces or incorporated in a crystal structure than when present as uncomplexed, aqueous molecules. Thus, the vibrational spectra of adsorbed oxyanions are far more difficult to interpret because of the presence of more than one type of symmetry, combinations of low frequency lattice modes with vibrations of the molecule of interest, and ion-ion coupling which can produce more bands than would be predicted from symmetry interpretation (HEZEL and ROSS, 1968; BEECH and

LINCOLN, 1971; URBAN, 1993). Site (WINSTON and HALFORD, 1949) and factor group analysis (BHAGAVANTAM and VENKATARAIDU, 1969) and correlation method (FATELEY et al., 1971) are commonly used to analyze complete vibrational spectra of crystals. 94

Td

C3v

C2v

0 #

I8 Cl

Wavenumber (cm"1 )

Figure 5.1. Arsenate in different symmetries and a cartoon of the IR spectra of vg vibrations 95 The main objectives of this chapter are to evaluate: 1) the molecular environments of sorbed

ASO4 and Cr 0 4 in ettringite and their variation with the mode of sorption (adsorption versus coprecipitation); 2 ) applications of semi empirical molecular orbital calculations to identify modifications in vibrational spectra with changes in sorbate symmetry and complexing ion; and 3) the variation in mineral surface interactions of different oxyanions in relation to their hydration eneigies.

DATA COLLECTION AND ANALYSIS

Diffuse reflectance IR spectra of powdered samples were collected on a Mattson Polaris

Fourier Transform Infrared Spectrometer (FTIR) equipped with a broad-band mercury-cadmium- telluride (MCT) detector, a KCl beam splitter and a Harrick Praying-Mantis diffuse reflectance cell. Scans of the solids were collected fran 4(XX) to 400 cm ' with 1 cm ' resolution, as an average of 200 scans. All samples were diluted (2 % sample) with IR-grade KBr and background subtractions were made for the diluent. Aqueous samples were probed using an Attenuated Total

Reflectance (ATR) cell with a premounted ZnSe reflection crystal. To minimize oxyanion interaction with the ZnSe crystal, the spectra were collected immediately after the solutions were transferred to the ATR cell. However, the spectra were invariant from < 1 min to 60 min.

Background subtractions were made with deionized water.

Raman laser excitation was provided by a Ti:sapphire laser (coherent) pumped by an Ar laser

(coherent 90) operating at 784 nm and filtered by a holographic band pass filter (Kaiser) which delivered about 50 mW at the sample. A single stage spectrograph (Instruments SA, HR 640) with a 300 mm ' grating coupled to a CCD detector (photometric CH260 camera head/EEV05-10CCD with 296 X 1152 pixels, 0,66 cm X 2.59 cm active area) was preceded by a holographic band reject filter (Kaiser) in 180° backscattering geometry. The scans were collected from 2000 to 300 96 cm'\ Depending on the solid phase sorbate concentration and background fluorescence, the scan

time ranged from less than a minute to a maximum of 15 minutes. The scans were collected at low

energy laser output to avoid local heating in the sample.

The collected FTIR and Raman spectra were later processed for peak deconvolution and

curve fitting with the Grams 2.02 software package (Galactic Industries Corp.). When

superimposed peaks were noticed (from the peak asymmetry and appearance of shoulders), second

order derivatives were used to identify approximate peak positions and the number of overlapping

peaks, and curve fitting was employed to obtain information on peak intensities and positions. The

composite peaks were fit with Cauchy-Lorentzian profiles, using nonlinear least squares refinement

procedures based on a finite difference Levenberg-Marquardt algorithm (Fig. 5.2). The details of

these curve fitting procedures for FTIR spectra were described by GILLETTE et al. (1982) and

MADDAMS (1980). The peak positions obtained from these two methods were close and (fiffered

by < 10 cm'^ in all samples. However, when peaks did not overlap, their positions were identified

with more accuracy, for e.g., in the case of SO4 symmetric stretch (± 2.0 cm'*). Site symmetry and

correlation analysis for oxyanion coordination in ettringite and Ca arsenate model compounds were

carried out as reported by FATELEY et al. (1971) and are presented in Table 5.1,

Semi empirical molecular orbital calculations were conducted using the PM3 method

(STEWART, 1989a & b) which is based chi the NDDO (Neglect of Diatomic Differential Overlap)

approximation. These computations were carried out with the Hyperchem computational

chemistry software (Autodesk, Inc.). The vibrational calculations were performed on molecules

after they were optimized for geometry. PM3 was used for both the geometric and vibrational computations. Geometry optimization was conducted with a Restricted Hartree Fock (RHF) set up, and the termination condition was set with Polak and Ribiere approximation. The convergence limit was usually ~ 0.01 kcal (molA) '*. The bond lengths and angles of geometrically optimized Table 5.1. Site symmetty analysis of AsO^ coordination in different minerals. OS: Outer-s^ere and IS: inner-sphere complexes.

Normal Vibrations

Symmetry '^2 '^3 '^4

Td E Fa Fa C3V E Aj+E Aj+E C2v A. Aj+Aj Aj+Bj+Bj Aj+Bj+Bj Cl A 2A 3A 3A

Mineral / Species Space Group Site Symmetry c, Haidingerite Pcnb (d \)

Rauenthalite P-1 q

Sodium Arsenate P2j/n (C^2h) c,

Ettringite P 3 1 c ( c \) OS - AsO/ c. C, IS - AsO/

AsO/'(aq.) Td/C,

5 0.60

0.25 0.45

I I 0.20 •Ê 0.30 I 0.15 0.15

0.00 4000 3000 2000 1000 1000 900 800 700 Wavenumber (cm^) Wavenumber (cm ')

0.25 8 8 I I 0.20 g I ■§ 0.15

0.10 1000 900 800 700 1000 900 800 700 Wavenumber (cm ') Wavenumber (cm ')

F igure 5.2. FTIR spectral analysis, a. collected data, b. As -0 stretching band, c. second derivative of b and d. curve fit.

VO 00 99 molecules (e.g. PO4) are in good agreement with the experimentally measured values (e.g.

CRUICKSHANK and ROBINSON, 1966; CRUICKSHANK, 1961).

RESULTS AND DISCUSSION

Vibrational spectroscopy of ASO4 and C1O4 interactions in ettringite are presented in three major sections. In the first part, vibrational spectra of ettringite and the coordination environment of constituent molecules are discussed. The latter two sections mainly focus on ASO4 and C1O4 interactions with ettringite.

COORDINATION ENVIRONMENT OF Ca, AI, SO4 , OH AND H2O IN ETTRINGITE

The ettringite structure consists of columns and channels (MOORE and TAYLOR, 1968)

(Figure 1.1, Chapter I). The column ligands are OH and H 2O, and AI and Ca polyhedra are linked through OH. H 2O molecules are connected to only Ca and also participate in H-bonding with channel SO 4. Oxyanion soiption either inside the channels or on ettringite external surfaces may displace or strongly interact with Ca and AI polyhedra ligands and thus, can perturb AI and Ca coordination. This would result in changes in vibrational bands of ligands, and Ca-0 and Al-0 with increases in solid phase oxyanion concentration. Although metal-0 bond vibrations give clues to the coordination environment, they oftai lack sensitivity as compared to the ligand vibrations, such as S-0 (SO4) and 0-H (OH or H 2O) vibrations. The latter type of vibrations are focused on in this dissertation to understand oxyanion complexation in ettringite.

Previous IR studies on ettringite (PÔLLMANN et al,. 1989; KUMARATHASAN et al,

1990) have recorded bands at 3638 and 3420; 1670 and 1450; 1114 and 616; and 854 cm ' which 100 were assigned to OH stretching, OH bendng, S-0, and Al-OH vibrational modes respectively.

This band assignment was incomplete and in some cases was inaccurate (for CO3). A complete ettringite vibrational spectrum analysis and the molecular environment of constituent species are presented below.

OH Related Bands

Water exhibits three bands due to OH stretching and bending at 3445 (antisymmetric stretching), 3220 (symmetric stretching) and 1630 cm ' (bending) in IR (AINES and ROSSMANN,

1984). In ettringite, several types of OH exist, such as the free channel OH, surface exposed OH,

OH linked to Ca and AI polyhedra, OH of channel H 2O and OH of H 2O connected to Ca. This variety of OH bonding environments results in a broad IR absorption band around 3300 and 1640 cm ', and all these OH’s are not clearly distinguishable.

The distinguishable peaks related to OH stretching vibrations are free OH (3744 cm '), Al-

OH (3611 cm ') and Ca-OHz (3562,3356 cm '). These band assignments were made after making comparisons with specimen gibbsite and portlandite. Other peaks related to OH bending vibrations are 1645, 916 and 541, 643 and 317 cm ' which correspond to OH in H 2O, Al-OH, and Ca-OH respectively. These band assignments are in agreement with the published data (BISHOP et ai,

1994; POLLMANN etal., 1989; AINES andROSSMAN 1984; FARMER, 1974).

Water in ettringite is only bonded to Ca [Ca(0H>4(H20)4] and behaves similarly to free OH since the average bond distances of Ca-OH (MOORE and TAYLOR, 1970) are not significantly different from that of Ca-OHz (avg. 2.45 and 2.55 (± 0.05) Â, respectively). However, the OH stretching frequencies are expected to be slightly smaller since H 2O participates in H bonding with

SO4. BISHOP et al. (1994) showed that solvated Ca in clay interlayers produced OH stretching bands at 3540 and 3365 cm '. Thus, the strong peak at 3562 cm ' in ettringite IR spectra is 101 assigned to asymmetric stretching of OH in H 2O bonded to Ca, and related symmetric stretching bands are at 3356 cm'*. These bands are expected to change whai different oxyanions are substituted for SO4. Inner sphere complexation of oxyanions inside ettringite channels may take place by replacing structural HzO and, as a result, the intensity of this band should decrease.

Although IR intensity of a vibrational mode of a molecule mainly depends on ( ^ p/

1989).

ATR-FTIR spectra of aqueous SO4 exhibit a broad band at the 3600-3100 cm'* which resemble the broad OH band of H 2O in ettringite (Fig. 5.3). This suggests that H 2O in ettringite forms H bonds with SO4 in ettringite channels. Similarly, ASO 4 and Cr 0 4 in ettringite may produce new bands that correspond to OH stretching vibrations of H 2O in their solvation shell. In addition

OH bending modes are also expected to display similar behavior, for e.g. OH bending modes of

H2O in the solvation shell of SO4 versus column H2O in ettringite are 1655 and 1645 cm * respectively. The ettringite OH bending vibrations are somewhat smaller since H 2O bonds with Ca and participates in H bonding with SO4.

SO 4 Vibrations

The ettringite crystal structure refinement data indicate that channel SO 4 is completely solvated and is expected to be present as an undistorted tetrahedron (MOORE and TAYLOR,

1970). However, the S-O bond lengths were variable (1.31-1.56 Â) which made MOORE and

TAYLOR (1970) doubt the atomic coordinates. Site symmetry analysis (WINSTON and

HALFORD, 1949) of SO 4 in ettringite suggests that its local symmetry is Ci (Table 5.1). SO 4 ions 102

1.6 1.4 I

g R 1 oC <5 I 0-8 .9 m I 0.6 I g I 0.4 W 00 < 0.2

0.0 2000 1500 1000 500 Wavenumber (cm'^) 0.4

0.3 VO I

0.1

0.0 3800 3600 3400 3200 3000 2800 2600 2400 Wavenumber (cm'^)

Figure 5.3. Vibrational spectra of ettringite and aqueous SO^. 1. ATR-FTIR spectra of aqueous SO^ (a), and FTIR (b) and Raman (c) of ettringite in 2000-250 cm'* range. 2. ATR-FTIR spectra of aqueous SO^ (a) and FTIR spectra of ettringite in 3800-2400 cm'* range. 103 in this symmetry with the observed variation in S-0 bond distances should result in removal of degeneracy and exhibit aU nine fundamental vibrations in Raman and IR.

The IR measurements indicate that SO4 asymmetric stretching frequencies (V3) exhibit splitting (1192, 1131, 1090 (?) cm'*), and the center of mass for this band shifted to higher wavenumbers (~ 1136 cm'*) when compared to aqueous SO4 (1098 cm'*) (Table 5.2). The SO4 symmetric stretching frequency (vi) was identified (989 cm'*) with low intensity in ettringite IR, but S - 0 V2 and V4 vibrations are difficult to distinguish because of their overlap with OH bending modes. However, Raman spectra exhibit distinct bands corresponding to V 4 (669, 627, and 606 cm'*) and V 2 (489 and 451 cm'*). Raman spectra also exhibit splitting in V 3 (center of maxima ~

1131 cm'*), but it is difficult to extract the peak positions since this band is at the base of a strong

CO3 band. A strong band at 989 cm * in Raman corresponds to Vi and this data agree with the IR.

Strong SO4 interaction with other ions should result in removal of degeneracy and shifts in the Vi.

However, splitting of V 3 and V4 vibrations and the absence of shifts in Vj of SO4 in ettringite as compared to aqueous SO4 make it difficult to interpret the SO4 local coordination in ettringite.

Strong complexation of one of the O of SO4 results in an increase in the S-0 bond length and thus, the S -O -X (X = complexing ion) symmetric stretch should be smaller than the symmetric stretch of uncomplexed S-0 and degeneracy on V 3 and V4 vibrations should be lifted. Sulfate complexes in anhydrite (CaS 0 4 ) (FARMER, 1974), bassanite (CaSO 4.0 .5 H2O) (FARMER, 1974), and aqueous SO4 complexes of Co (NAKAMOTO, 1986) produced Vi bands (S -O X ) at higher wavenumbers as compared to uncomplexed SO4. In contrast, heavy metal (Sr, Ba, and Pb) SO4 minerals produced Vi (S -O X ) vibrations at lower wavenumbers (FARMER, 1974), Although these compounds exhibit similar splitting in V 3 and V4 bands, the shifts in the Vi band are noteworthy.

Semi empirical calculations (PM3) could not be conducted for these complexes because of the 104

Table 5.2. SO4 Normal modes in ettringite and related systems.

Species/Mineral Vi Vi V3 V4 cm' SO4* 983 (R) 450 (R) 1105 (R,IR) 611(R,IR) S 0 4 (a q .) 987 (w ) — 1098 (br) gypsum (C 2) 1004 476 (br) 1170,1146, 670,625,602,586 1119,1102 Ettringite * 989 489, 1192,1131(s), 669,627,606 451(sh) 1090

* collected Aom NAKAMOTO (1986). * Peak information is from both Raman and IR. R: Raman, IR: Infrared, s: strong, w: weak, sh: shoulder, br: broad. 105 absence of parameters for Ca and other metal atoms, and thus, the exact nature of these changes

are uncertain in the case of SO4. Because ettringite doesn’t exhibit any shifts in the Vi peak, SO4

may be described as a regular tetrahedron. This, however, contradicts the crystal structure

refinement data of MOORE and TAYLOR (1968). The inconsistency between IR and structure

refinement data may be ascribed to the combined effects of cation and coordinating H 2O on the Vj

band position, which may be difficult to interpret with available data in the case of SO4.

CO, Contamination in Ettrinsite

Synthetic ettringite samples were expected to be free from CO 3 contamination since the

mineral was prepared in COz-fiee deionized water. However, the presence of CO 3 V3 stretching

vibrations around 1450 cm'* indicate the presence of adsorbed CO 3. PÔLLMANN et al. (1989)

did not identify this band (although their spectra shows the presence of a strong peak) and

KUMARATHASAN et al. (1990) attributed this to OH bending. Despite the use of C 0 2 -ffee

deionized water and N 2 (g) filled glove boxes in the ettringite synthesis, the spectroscopic results

indicated that it is difficult to avoid CO 3 contamination in the final product, because these alkaline materials are highly reactive towards CO 2. This may also indicate that the identified CO 3 was

incorporated into the structure during ER measurements, or present as adsorbed species in the reagents such as CaO and NaOH used in ettringite synthesis. In the present study, the peaks

around 1450 cm'* were identified as CO 3 after comparing with a set of standards (calcite and goethite-adsorbed CO 3 (CARLSON and SCHWERTMANN, 1990)). In addition to the CO 3 V3 stretching frequencies, several other peaks corresponding to CO 3 were identified in both Raman andlR. 106 Summary

The spectral information together with the structural data provide useful clues to the local bonding environments of different polyhedra and functional groups. When ettringite is reacted with

ASO4 or Cr0 4 , these ions interact with the surface or penetrate into the channels, and changes in the vibrational modes of OH, H2O, SO4 and the incoiporated oxyanions are anticipated.

ASO4 COMPLEXATION IN ETTRINGITE

Vibrational spectroscopic studies of AsÜ 4 reacted goethite (LUMSDON et al., 1984) and freshly prepared Fe oxides (HARRISON and BERKHEISER, 1982) have shown that ASO 4 forms binuclear complexes through the rq)lacement of ‘A’-type OH’s (A - type OH for Fe oxides or clays refers to the surface OH connected to a single cation; SPOSITO, 1989. This A-type OH has no relation to the A -type site in ettringite.). Spectral information, such as V 3 band splitting, appearance of Raman modes in IR, and disappearance of peaks corresponding to A-type OH

(connected to a single cation) stretching frequencies were used by these researchers to determine the structure of adsorbed ASO4. On the contrary, the IR studies of KUMARATHASAN et al.

(1990) showed a single peak for As-0 V 3 vibrations (873 cm'^) in ASO 4 ettringite. Due to the preservation of degeneracy for tl» As-0 V 3 vibrations, it can be argued that ASO 4 retains its tetrahedral symmetry and probably forms OS complexes in ettringite. However, other IR studies on Ca arsenates (FARMER, 1974; MOENKE, 1962) have shown that As-0 V 3 vibrations are split, indicating a lower symmetry than Td for ASO 4 in these chemically analogous materials. Although these studies do assist in band identihcatim, they have little influence on the understanding of

ASO4 complexes in ettringite. This is mainly because ASO 4 in Ci and Czv symmetries produce the 107 same number of bands and the amount of splitting varies with the type of complexing metal and the number of attached metal atoms. To evaluate ASO4 bonding in ettringite, a detailed investigation was conducted by 1) theoretically evaluating the vibrational frequencies of ASO 4 and its aquo complexes and 2) studying ER spectra of aqueous arsenates and several known Ca arsenate compounds with differing ASO 4 coordination environments. This information was later employed in interpretation of ASO4 molecular structures in ettringite.

Semi Empirical Molecular Orbital Calculations o f ASO4 Species

A NEK) (Neglect of Differential Overlap) based PM3 method was used in this study to calculate vibrational frequencies. This method considers overlap density between two orbitals centered on the same atom interacting with the overlap density between two orbitals also centered on a single atom. Previous semi empirical calculations on smaller molecules have shown that PM3 produced results close to the experimental values when compared with other available methods

(SEEGER et al„ 1991; STEWART, 1989 a & b). The calculated vibrations are usually at a higher energy level than the experimental values and this error is due to the neglect of correlated motion of the electrons and approximation of the potential as a harmonic function.

The semi empirical calculations were performed by optimizing the geometry and also by constraining ASO 4 polyhedra bond lengths and angles (from crystalline Ca arsenates). Although the predicted vibrational frequencies from these two methods do not differ greatly, bond length- constrained simulations (in the case of ASO4 in haidingerite and rauenthalite) produced values closer to the experimental values for both V 3 and Vi vibrations (< 60 and < 40 cm '\ higher respectively). Because the bond parameters are not available for all ASO 4 complexes of interest in this study, geometry optimized ASO 4 species were used to compute vibrational spectra. In addition, a small molecular cluster, instead of a single molecule (used in this study), may predict 108 the experimental data more accurately, but such a level of complexity was beyond the

computational resources available in this study.

In the current study, the computations were conducted for protonated forms of As 0 4 ^‘, both

gaseous and aqueous (H bonding of four H2O to the four O of AsO#); and its Al, Mg, Zn and Cd

complexes. In addition, only stretching vibrations are considered in this study, since their

interference with other molecular vibratims are minimal. The PM3 method in Hyperchem 2.02

lacks parameters for Ca which prevented the study of Ca arsenate complexes. However, a working

hypothesis was developed on the basis of computations for protonated and metal As 0 4 ^' complexes

which is later used in interpretation of spectral information Aom arsenated ettringites.

Protonation atid Solvation Effects on As-0 Vibrations

Semi empirical calculations were made on gaseous As 0 4 ^', HAs0 4 ^,and H 2ASO4 ; and their

partly (H bonded to one or two H 2O) and completely solvated molecules (H bonded to four H 2O)

(Fig. 5.4, Fig. 5.5, Fig. 5.6). These simulations may represent ASO 4 in different solvation states

within ettringite. Calculated As-0 fundamental vibrations and BR. vectors of different As 0 4 ^

species are close to the experimental data. The differences in the experimental and calculated

frequencies (~ 1(X) and < 20 cm ' higher for calculated As-0 V 3 and Vi stretches) are mainly due to

the assumptions involved in the calculations.

The calculated As-0 bond lengths are longer in gaseous ASO 4 (1.741 A ) than solvated species

(1.726 A ). The same behavior was also observed in the case of experimentally verified protonated

sulfates, for e.g., S-OH bond lengths of gaseous (1.57 A ) and solvated (1.528 A ) H2SO4 species

(STRUCTURE REPORTS, 1991; MOODENBAUGH et al., 1983). The calculations indicated

that protonation and solvation of gaseous ASO 4 will decrease As-OX (X = H or H bonded H 2O)

bond lengths in the order HAs 0 4 ^ > H2ASO4 » As0 4 ^ (H2O) ~ As0 4 ^ (H20)2 (Table 5.3). Table 53. Tbeoieticai bond lengths and stretching vibrations for different ASO 4 species (based on semi empirical calculations).

Molecule Bond Lengths in AsO^ (A) As-O Stretching Vibrations (cm'*) As-O As-0-(X) Vi’ V3 AsQ*^ (g) 1.74 — 773 824

HAsO/(g) 1.68 1.88 637 905,823 H2As04'(g) 1.64 1.81 744 (sy), 733 1022,957,880 As04^(H20) 1.73 1.76 806 (sy), 791 946,844

As04^(H20)2 1.72 1.75 836(sy), 821 923,863,836

As04^(H20)4 1.73 (1.68) — 849 (837) 878 (878) HAS04^(H20)4 1.68 1.87 697,684(sy) 999,944,895 H2AS0 4 (1)20)4 1.63 1.81 781(sy), 758 1041,937, 805 (762) A1-0 2 -ASQ2 1.61 1.90 601 1028,1022, 899

A1-Q2-A sQ2(H20)4 1.62 1.88 (avg.) 637 (sy) 1037,1008 Mg-Q2-As02(H2O)4 1.65 1.80 794 (sy), 696 952,923, 875 Cd-Q 2-As02(H20)4 1.66 1.79,1.78 (avg.) 826 (sy), 754 976,910 &l"0z"As02(H2O)4 1.64 1.79,1.78 756 1004,928

* Symmetric stretching (sy) of As-O-X bonds. Asymmetric stretching bands arc also reported in the same column. Niunbers in parenthesis are experimental data. t 110

Fa s 4

0

% «% •••V

*4

r .

Figure 5.4. AsO/ and its vibrational modes in Td symmetry, a & b: asymmetric and symmetric stretching, and c&d: asymmetric and symmetric bending vibrations respectively. I l l

V--

Figure. 5 J . Symmetric stretching of As-0 vibrations in (AsÜ4^')(H20)4. a 112

0 v b

0 >

Figure 5.6. Symmetric stretching of As-O-X (X = H* or H^O) vibrations in a: HA sOa^, b: HgAsO/ and c: hydrogen bonded AsO/". 113 However, the As-O (uncomplexed O) bond lengths of the above species exhibit tlie reverse trend.

Complete solvation of the protonated species did not affect the above trend, although the bond lengths were slightly smaller in solvated species. Asymmetric stretching (V3) of As-O vibrations of protonated and solvated molecules exhibit splitting according to their symmetry.

The symmetric stretching of As-OX (X = H or H bonded H 2O) vibrations decreases (HAsOa^'

< H2ASO4 « AsO/ (H2O) < As04^' (H20)2> with increases in As-OX bond lengths (Table 5.3).

Similarly, symmetric stretching of As-0 (uncomplexed O) vibrations increases in the order

HAs0 4 ^ < As04^ (H20>2 < As04 ^' (H2O) « H2ASO4 . This discrepancy in the observed trend is mainly due to the strong effect of addition of another proton to HAs 0 4 ^ species on As-0

(uiqrrotonated O) bond lengths. Similar behavior was also observed in experimentally obtained bond lengths and vibrational frequencies of P-OH in aqueous HP 0 4 ^ and H 2PO4 solutions

(CRUICKSHANK and ROBINSON, 1966; CRUICKSHANK, 1961).

The OH stretching frequencies of As-OH and AS-OH 2 may also be significant in identifying the coordination of As. However, these vibrations are modified by H-bonding of OH and H 2O with other atoms in the ettringite structure (such as SO4). Thus, the OH frequencies are affected by the other nearest neighbors around ASO 4 which makes it difficult to differentiate the solvation and protonation of ASO4 species in the solids. However, OH vibrations are useM in identifying oxyanion-H20 interactions (discussed in the next section).

Thus, solvation and protonation of As 0 4 ^ in solids (either partial or complete) can be identified from the location of the Vi band in the vibrational spectrum. In summary, protonation resulted in relatively small As-OH vibrations (~ 750 cm ') as compared to H-bonded ASO 4 (~ 810 cm '). Although similar information can be obtained from V 3 vibrations, the large deviation from the experimental data (~ 12 %) makes it difficult to compare with the experimental values. 114 Metal Complexation with A sO /'

The effect of a complexing metal on As-0 vibrations was studied by selecting solvated bidentate mononuclear metal arsenate complexes (e.g. Fig. 5.7). In this study, Al, Mg, Cd and Zn

ASO4 complexes were studied since only these metals were parameterized in PM3. The computations showed that the estimated metal-As bond distances are close to the measured values in crystalline solids. For example Al-As bond distances are around 2.9 (±0.1) Â, depending on O-

Al-0 and As-O-Al angles, as compared to the measured values of As-Al in crystalline AlAsO#. Of all the examined metals, Al was most effective in distorting the ASO 4 tetrahedron. Accordingly, the

As-OX Vi band (X = Al, Cd, Mg, Zn) shifted to 637 cm ' (1037 cm ' for AsOz) as compared to

794,756 and 826 (754 as.) cm ' in the case of Mg, Zn and Cd, respectively. This behavior may be due to inductive effects of different cations and their relative masses.

Thus, the calculated As-0 vibrational modes agree with experimental values (LUMSDON, et al., 1984; FARMER, 1974; Fig. 5.8). This strongly supports the validity of assigned relative shifts in the Vi band for different coordinations. In summary, the ASO 4 tetrahedron was distorted more by protonation than by solvation. Except for Al, complexation of As 0 4 ^ with metals shifted

Vi band to lower wavenumbers in the range of 800-750 cm '. Al produced a strong shift in the Vi band (637 cm ') which apparently is due to its high electronegativity.

ASO 4 Spéciation in Water

Acidity constants of H3ASO4 (pKa = 2.24,6.86,11.49) suggest that it dissociates to H2ASO4', and HAs 0 4 ^ at neutral pH and As04^ at alkaline pH. The symmetries of these four aqueous

species with decreasing proton concentration are Csv, C2 ,, Csv and Td. Collected ATR spectra of different H 3ASO4 species are in agreement with the symmetry assignments of ASO 4, with the 115

Figure 5.7. Symmetric stretching of As-O-Cd vibrations in bidentate mononuclear CdAs 0 4 (HzO)4. 116

500

1000

1500 lî 2000 il 2500 3000

3500

4000 4000 3500 3000 2500 2000 1500 1000 500 Experimental Results Wavenumber (cm')

Figure 5.8. Comparison of PM3 computed results with the experimental data for solvated and protonated AsO^ species. 117 exception of HAs 0 4^ (Table 5.4). The IR peak for HAsO#^ was very sharp, without any shoulders, and the maximum shifted to lower wavenumbers (859 cm'*) (Fig. 5.9) as compared to the V3 of AsO/' (878 cm'*). This suggests a relatively weaker bond for protonated As-0 in

HAs0 4 ^ as compared to As 0 4^. The presence of a single sharp peak at lower wavenumbers may be due to rapid exchange of a H* with terminal O of AsÛ 4 at time scales smaller than the IR sensitivity ('-10*'* s'*). However, this hypothesis is tentative and more studies are required to understand these spectra.

The ATR spectra of ASO 4 at pH 5.7 (dominant aqueous ASO4 species at this pH is H 2ASO4') showed three distinct peaks corresponding to a split in the V 3 band at high wavenumbers (Table

5.4) and a low frequency As-OH band (Fig. 5.9). These peak positions are in agreement with the predicted frequencies of protonated As 04^' from first principles.

IR studies on different aqueous Co ASO4 complexes showed splitting in the V 3 vibrations and a shift in the Vi band (As-O-Co) to lower wavenumbers (BEECH and LINCOLN, 1971).

Although this Vi band was at a lower wavenumber than that of uncomplexed ASO 4 (837 cm'*), it shifted to higher wavenumbers with decreases in symmetry and, overall the band was within the

800-750 cm * range. These data also support the theoretical predictions of wavenumber increase with decrease in the symmetry of protonated and solvated species.

The ATR-FTIR spectra of aqueous As04^' also produced broad peaks in the OH stretching region (~ 3400 cm'*) which are due to OH stretching of H 2O in the ASO 4 hydration shell (Fig.

5.10); and asymmetric and symmetric OH stretching bands are at 3290 and 3012 cm * respectively. To understand the nature of solvation around ASO 4, different Na salts of SO4, PO4,

C1O4, M0 O4 and Se 0 4 were examined. Differences in the charge of the central atom (As, Cr, P, etc.) have different contributions to H bonding with H 2O in the first few layers of the solvation 118

Table 5.4. ATR-FTIR spectra of ASO4 aqueous species.

pH Species Medium Symmetry As-O V3 cm'* > 12.0 A s O /- Water Td 878* 9.8 HAs 0 4 ^ Water Csv 859 5.7 H2ASO4 Water, SO4 C2V 909, 876, 860 (762) 5.7 H2ASO4 Water, NO3 C2V 909, 876, 858 (766) 5.7 H2ASO4 Water, Q C2V 909, 876, 864 * Reported by NAKAMOTO (1986). Number in parentheses represent Vi modes.

Table 5.5. FTIR spectra of As-0 vibrations in different crystalline arsenates.

Mineral Vi V3 cm* As0 4 (aq) 837 878* Sodium arsenate 736(w),702(s) 982(sh), 885(s), 818 Phaimacolite 786,735(s) 978,907, 891,850, 831,813 Haidingeiite 804,737(s) 983, 894,852, 832, 812 Sainfeldite 804,786,712(s) 996, 932,900,866, 840 Rauenthalite 790,736(s), 991(sh), 925, 877, 863, 834, 815 721(sh) * Reported by NAKAMOTO (1986) sh: shoulder, w: weak, s: strong 0.10 g

0.08

I I ^ 0.04 S

0.02

0.00

950 900 850 800 750 700

Wavenumber (cm'^)

Figure 5.9. ATR-FTIR spectra of a) HAsO/ and b) HzASO^ 120 shell. This results in shifts of OH stretching and Ooxymion—HOH (H bond) vibrational frequencies of oxyanion IR spectra (Fig. 5.10) (KRESTOV et al., 1994). Ooxyuiion-.HOH vibrational frequencies were not explored in this study because of their low intensity and occurrence in far-lR region (~ 300 cm '), which was beyond the instrumentation range used in the study. Similarly, when H 2O in solids H bond to oxyanions, the OH stretching should reflect oxyanion-HzO interactions in solids, e.g. hydrated Ca and Na arsenates (discussed in the next section).

Structure ofAs04 in Crystalline Ca Arsenates

IR spectra of Ca arsenates rauenthalite (Ca 3(As04 )z. IOH2O), pharmacolite (CaHAs 0 4 .2H2 0 ), haidingeiite (CaHAs 0 4 .2H2 0 ) and sainfeldite (Cas(HAs 0 4 )2(As0 4 )2.4 H2 0 ) were collected and their As-0 vibrational frequencies were canpared with ASO 4 local symmetries reported from single crystal refinement work (CALLERI and FERRARIS 1967; FERRARIS and ABBONA 1972;

CA Tn and IVALDI1983; and FERRARIS et ai, 1971).

The site symmetry of ASO 4 in all of these minerals is Q but they differ in the number of Ca atoms they are bonded to and the intensity of H bonding with H 2O. In all of these minerals, ASO 4 binds with more than one Ca in either bidentate mononuclear and/or bidentate binuclear fashion. It is not surprising to see severe distortion in the ASO 4 polyhedron and elimination of the degeneracy for V3 vibrations (Table 5.5). In haidingeiite (and also in pharmacolite and sainfeldite) only one type of As coordination is present (Fig. 5.11). The appearance of more than three V 3 vibrational peaks represent lattice mode interactions with the fundamental vibrations and/or probably surface species which are in different coordination than the lattice ions. In the case of diffuse reflectance

IR spectra the latter effect may be prominent, since IR absorption is primarily due to interaction with surface species. Rauenthalite has two types of ASO 4 species: one forms bidentate binuclear 121

0.16

en

0.14 AsO,

m 0.12 en

0.10 PO o v

00 00 C r O en 0.06 I MoO 0.04 leO

0.02

0.00 I so

- 0.02 3800 3600 3400 3200 3000 2800 2600

Wavenumber (cm‘^)

Figure 5.10. OH stretching of HjO in oxyanion solvation shells. 0.300

g 0.225 //

0.150

0.075

1050 900 750 600 Wavenumber (cm )

Figure 5.11. FTIR spectra of As-O vibrations and the local symmetry of arsenate tetrahedron in haidingeiite. 123 and the other multidentate complexes (Fig. 5.12); and the V 3 vibrations exhibit a corresponding split in the peaks. It should be noted that the As-OH symmetric stretching frequency in all of these minerals decreased to lower wavenumbers.

ASO4 in Na arsenate (Na2HAs0 4 .7 H2 0 ) is bonded to a and exhibits no direct interaction with Na but participates in H bonding with H 2O in the Na coordination sphere (FERRARIS and

CHIARI, 1970a) (Fig. 5.13). Site symmetry analysis, however, suggests a Ci symmetry for ASO 4 in this mineral. IR spectral analysis indicates that the ASO 4 €3, symmetry is preserved and the appearance of shoulders around the main band indicate ASO 4 site symmetry effects in this crystal.

It should be noted that the fundamental modes may also combine with lattice modes and may cause the observed low intensity bands in the IR spectra.

Crystal structure refinements of the examined minerals suggests that ASO 4 symmetry is more disturbed by its bonding to a H* than by the number of bonded Ca atoms. For instance, As-0 bond distances of Asl and As2 polyhedra in rauenthalite differ by only 0.02 Â (1.66 and 1.68 Â respectively). In haidingerite, the As-OH bond is 1.73 Â long as compared to 1.66 Â for As-0

(Ca), and there is an intermediate As-0 distance of 1.69 Â for H-bonded terminal O. Similarly, in

Na arsenate, the bond lengths for As-OH and As-0 (H bonded) were 1.74 and 1.68 Â, respectively. These bond lengths give clues to the variation in As-0 Vi vibrational frequencies for these minerals (inverse relation of Vi with As-OX bond length). The relatively high wavenumber for this vibrational mode represents bonding to Ca, where as a lower number (around 730 cm ') represents ASO4 protonation. Theoretical predictions for the shifts in the As-OX (X = H, Al, Mg,

Cd, Zn, H 2O ) Vi band also showed similar results for protonated, solvated and metal complexed

ASO4. Similar arguments can be made for the ASO 4 Vi vibrations in ettringite and, thus, its chemical environment can be evaluated. Aluminum arsenates were not available to extract As 1 0.15

0.10 I 0

As 2 $ 0.05

0.00 1050 900 750 600 Wavenumber (cm )

Figure 5.12. FTIR spectra of As-O vibrations and the local symmetry of AsO. tetrahedra in rauenthalite. 0.4

0.3

0.2 I 0.1

0.0 1050 900 750 600

Wavenumber (cm'^)

Figure 5.13. FTIR spectra of As-O vibrations and the local symmetry of AsO. tetrahedron in Na arsenate. W: water. 126 information on Al-O-As bond vibrations. However, as discussed above, semi empirical molecular

orbital calculations showed that the Vj band would shift to very low wavenumbers (~ 630 cm'*).

ASO4 Spéciation in Ettringite

OH Stretchine and Bendine Modes

Sorption of ASO4 by ettringite perturbed the OH stretching frequencies, and this perturbation

was dependent on the mode of sorption (adsorption versus coprecipitation) and pH. Ettringite

exhibits peaks at 3648, 3611, 3562, 3356 and 3222 cm'*, and, with increases in sample ASO 4

concentration (adsorption and coprécipitation), the OH stretching maxima shifted to around 33(X)

cm * (Fig. 5.14). This new band corresponds to H 2O participating in H bonding with ASO 4, which

is in agreement with OH stretching of H 2O in the ASO 4 solvation shell (3350 and 3011 cm'*) (Fig

5.10).

In adsorption samples, increased ASO 4 loadings caused decreases in peak intensity near 3562

cm * (Ca- 0 H2 vibrations) and a shift in the peak maximum to 33CX) cm * (H 2O H bonded to ASO 4).

This decrease in peak intensity corresponds to the removal of Ca-bonded H 2O. The structural

analysis of ettringite indicates that such interactions with Ca may take place at B', ‘C’, and ‘E’

type sites (Fig. 1.2). In the absence of pH control and at 6-20 % solid phase ASO 4 concentration,

new sharp peaks developed at 3550 and 3398 cm'*. The presence of these peaks in the vibrational

spectra indicate the presence of gypsum (OH vibrations of H 2O). Corresponding changes in OH

bending were also observed. XRD analysis of the same samples also indicated gypsum formation

at these solid phase ASO 4 concentrations (Fig. 3.3). When pH was maintained at 11.8, these changes were absent and only the peak maximum at 3562 cm * decreased (Fig 5.14).

The OH stretching frequencies of coprecipitated ASO 4 ettringites were not affected by pH

(10.5-12.5). With increases in solid phase ASO 4 concentration, new, shaip peaks developed at 0.35 0.35

0.30 0.30

0.25 0.25 P s 0-20 I 0.20 a 0.15 I1 0.15

0.10 0.10

0.05 0.05

0.00 0.00 3500 3000 2500 2000 30003500 2500 2000

Wavenumber (cm ) Wavenumber (cm'^)

Figure 5.14. OH stretching in adsorbed AsO^-ettringite. 1. No pH control. Solid phase AsO^ concentrations are a: 0, b: 2, c: 4,d: 11 and e: 30%. 2. pH: 11.8. AsO^ concentrations of the samples are a: 0, b: 4, c: 11, d: 17 and e: 20 %. 128 3671, 3618 and 3542 cm'* (Fig. 5.15). In addition, a broad band developed around 3023 cm'*, corresponding to H 2O in the ASO 4 solvation shell (Fig 5.10). OH bending vibrations in coprecipitated samples produced a peak at 1645 cm * and did not show significant deviation from ettringite values. This was due to the close proximity of OH bending peaks for H2O in the SO4

(1655 cm'*) and ASO 4 (1645 cm'*) solvation shells. The absence of peak intensity changes at 3562 cm * may indicate that coprecipitated ASO 4 dominantly substitutes at G-type sites in ettringite.

SO 4 Vibrations

S0 4 in ettringite exhibited a split V)band (maxima at 1138 cm'* with two shoulders), where as the Vi (989 cm'*) remained unchanged as compared to aqueous SO4. With increases in solid phase

ASO4 concentration, changes in the S - 0 V3 vibrations were large, especially in adsorbed samples with no pH control. The S-0 V 3 vibrations of As0 4 -reacted ettringite did not show any deviation from those of pure ettringite until the solid phase ASO 4 concentration reached 5 % (Fig. 5.16).

Above this concentration, the S - 0 V3 vibrations showed clearer splitting, and sharp peaks developed at 1156 and 1138 cm *. These peaks correspond to SO4 S-0 vibrations in gypsum and correlate with the observations made from the OH stretching frequencies at this concentration range. Symmetric S-0 stretching vibratiœis increased to 1(X)1 cm'*, indicating changes in SO4 coordination. Sulfate antisymmetric baiding also followed the trend of the V 3 vibrations (Fig.

5.17). These changes in S - 0 vibrations with increases in solid phase ASO 4 do not correspond to changes in ettringite SO4 polyhedra, but to precipitation of a new phase, gypsum. Spectral interpretation was easier in these samples since gypsum was identified by XRD. All these changes were not observed in the adsorbed samples when pH was maintained at 11.8, indicating that 129

0.7

00 0.6 m

0.5

■e I 0.3

0.2

0.1 I

0.0 3500 3000 2500 Wavenumber (cm'^)

Figure 5.15. OH stretching in coprecipitated AsO^-ettringite (pH 12.4). Solid phase AsO^ concentrations are; a: 0, b: 4, c: 9, d: 20 and e: 30 %. Experiments at pH 11.8 also produced similar IR spectra. 1.0 00 ON II 0.8 s s ^ CO < 0.6 VO cn 00 0.4 — b

o' 0.2 VO I 9 i 0.0 1800 1600 1400 1200 1000 800

Wavenumber (cm *)

Figure 5.16. Fl'iK spectra of adsorbed AsO^ ettringite (1800-720 cm‘‘). Solid phase ASO 4 concentrations are: a: 0, b: 2, c: 4, d: 11 and e: 30 %. 0.9 0.7 0.8 0.6 f 0.7 0.5 i 0.6

0.5

? 0.4 « 0.3 i < 0.3 0.2 VO 0.2 0.1 0.1

0.0 0.0 750 700 650 600 550 500 450 400 750 700 650 600 550 500 450 400 Wavenumber (cm‘^) Wavenumber (cm^)

Figure 5.17. vibrations of SO_^ in ettringite. 1. AsO^ coprecipitation in ettringite. Solid AsO^ concentration are a: 0, b; 4, c: 9, d: 20 and e: 30 %. 2. AsO^ adsorption in ettringite. pH: not controlled. AsO^= a: 0, b: 2, c: 4, d: 11 and e: 30 %. 132 splitting of S-0 vibrations in the above samples was caused by gypsum formation and not by

interactions of adsorbed ASO 4 with ettringite SO 4 polyhedra.

In coprecipitated samples, the S-0 V 3 and V4 vibrations were only slightly disturbed by ASO 4

incorporation into ettringite, suggesting no changes in the local symmetry of channel SO 4 (Fig.

5.17,5.18,5.19). This behavior was insensitive to the pH of the reacting solutions.

A s04 Vibrations

The As-O vibrations were different in adsorption versus coprecipitation samples (Fig 5.16,

5.18, 5.19). The samples exhibited splitting in the As-0 V 3 vibrations, which overlapped each other, and the Vi band. Some of the CO 3 and Al-OH bending vibrations appear at 941, 917, 870,

and 845 cm'* as small sharp peaks, which were easily separated fiom the As-0 vibrations.

When pH was not controlled during adsorption, V 3 splitting remained the same (947,900, 842 cm'*) and peak intensities increased with increases in ASO 4 concentrations. Although splitting of the V3 vibrations indicated reduction in ASO 4 tetrahedral symmetry, it provided little information

about its bonding environment. Adsorption experiments conducted at controlled pH (11.8) showed

the same behavior as above for the As-0 V 3 vibrations (949, 891, 843 cm'*). This splitting in V 3 is

indicative of either Cz, or Ci ASO 4 symmetry.

In the absence of pH control, symmetric As-0 stretching modes showed two bands at 807 and

787 cm * in the ASO 4 ettringite (ASO 4 ~ 2 %). With increases in ASO 4 concentration, the Vi band

at 807 cm * remained stable as the other decreased in intensity, and a new shoulder developed at

736 cm * when the concentration reached 20 % by weight (equilibrium pH 10.9). After comparison with Ca arsenate model compounds and semi empirical calculations, the Vi band around 807 cm * may be attributed to ASO4 bonded to Ca and H 2O. The bands at approximately 736 cm * are likely 1.6

1.4

1.2 $

1.0

I 0.8 S A < 0.6 i 0.4

5 0.2 ss

0.0 *n I ^ 00 1800 1600 1400 1200 1000 800 Wavenumber (cm'^)

Figure 5.18. FTIR spectra ofcoprecipitatedAsO^-ettringite (1800-680 cm ‘) (pH: 12.4). Solid phase AsO^ concentrations are: a: 0, b: 4, c: 9, d: 20 and e; 30 %. 1.6

(S 1.4 -®

ÿ 1.0 ■e s 0.8 < 0.6

0.2 I 0.0 1800 1600 1400 1200 1000 800 Wavenumber (cm'*)

Figure 5.19. FITR spectra of coprecipitated AsO^-ettringite (1800-680 cm ’) (pH 11.8). Solid phase AsO^ concentrations are: a: 0, b: 4, c: 9, d: 20 and e: 30 %. 135 due to protonated species. Shifts in the band positions (787 cm ') likely resulted from changes in the overall ASO 4 bonding environment (includes H-bonding). The equilibrium pH of these samples decreased from 11.4 (2 % solid phase ASO 4) to 10.9 (20 % solid phase ASO 4) with increases in

ASO4 loadings (Chapter III). The ratio of the aqueous concentrations of HAs 0 4 ^/ As0 4 ^‘ increases with decreasing pH (pKa = 11.49) and at pH 10.9 HAs 0 4 ^ is the dominant ASO 4 species. Thus, the spectroscopic results indicate the direct sorption of HAs 0 4 ^ into ettringite or precipitation of Ca arsenate. As expected, this band was not observed in adsorbed samples at pH 11.8 and coprecipitated samples at pH 12.5 and 11.9 (discussed below).

Coprecipitated ASO 4 ettringites showed complex IR patterns as compared to the adsorbed samples (Table 5.6). Coprecipitation experiments were conducted at pH 12.5 and 11.8 (pH has no effect on ASO4 uptake). The IR spectra of these samples indicate that high SO4 concentrations in the solid had a major influence on ASO 4 symmetry (Table 5.6). At pH 12.5, the V 3 band exhibited several peaks which may correspond to different symmetries for ASO 4 polyhedra or to the interactions of fundamental vibrations with low frequency lattice modes. However, at pH 11.8 the number of bands for As-0 V3 vibrations were relatively small, which may suggest a high symmetry for ASO4. In addition, when the sorbed ASO 4 was <4% , there was very little splitting in V3, which is indicative of highly ordered symmetry for sorbed ASO 4, i.e. Td or C 3V. As-0 (bonded) stretching in coprecipitated samples showed two peaks (overlap with V 3 vibrations) around 800 and 760 cm '.

The position of symmetric stretching bands at these wavenumbers indicate that ASO 4 mainly interacts with Ca in the columns and also participates in H bonding with H 2O molecules. These samples have relatively high solid phase SO4 concentrations as compared to the above and may indicate that high SO4/ASO4 in ettringite stabilizes ettringite columns and allow ASO4 in relatively high symmetry. 136

Table 5.6. FITR spectra of As-O vibrations in coprecipitated As 0 4 -ettringite.

AS0 4 pH Vi V3 (%) cm' 4 12.4 801,776,756 989,945,910, 880(s), 863 (s) 9 12.4 803,766 989,946,912(s), 863(s), 829 20 12.4 799,750 948,895, 880(s), 862(s), 843, 834(s) 1.4 11.8 — 861(sh), 847(s) 4 11.8 795,758(w) 915 (br), 830 9 11.8 758 937,915, 885, 865,835 20 11.8 752 931,900, 827 30* 12.5 793,769 957,935, 883 (s), 852

* SO4/ASO4 ratio in the reacting solutions was zero, s; strong, w; weak, sh: shoulder, br: broad. 137 As0 4 interactions with A1 polyhedra are difficult to predict since both As-0-AI and SO 4 V3 vibrations occur around 600 cm'\ Since no major peak was observed at this wavenumber with increases in ASO4 concentration, it is unlikely that ASO4 interacts with A1 in ettringite. This eliminates the possibility of ASO 4 coordination with ettringite at A, D, and F type sites in the examined concentration range (0.01 - 2.0 mol kg'^). In addition, the predicted Czv or Ci symmetry for ASO4 is only possible when it interacts with ettringite at ‘B’, ‘C’ and ‘E’ - type sites. Channel substitution (substitution for SO4 at ‘G’ type sites) may not be a dominant mechanism during adsorption since this should not result in decreases in intensity of Ca-0H2 OH stretching band. In contrast, the ‘G’-type sites are dominant for substitution during coprecipitation, since such intensity changes (OH stretching of Ca-OH%) were not observed with increases in solid phase

ASO4. However, ASO4 interactions at B, C, and E- type sites in coprecipitated samples can not be completely ruled out on the basis of available data.

Summary ofAsOd Coordination in Ettrlneite

In summary, IR and Raman studies indicate that:

1). ASO4 forms IS complexes in all adsorption samples, either as bidentate mononuclear or bidentate binuclear complexes. However, it is difficult to estimate their relative proportions with the available information. The dominant ASO 4 adsorption sites in ettringite are B, C and E types.

2). Coprecipitated samples exhibit similar ASO 4 bonding environments. However, ASO 4 also forms OS complexes at very low solid phase ASO 4/ SO4 ratios. ASO4 interactions at ‘G’ type sites are dominant in coprecipitated samples.

3). ASO4 coordinates to Ca rather than to Al. This is supported by decreases in OH stretching frequencies (corresponding to Ca-OHz) with increases in ASO 4 concentration, and the aidaient absence of very low frequency Al-O-As Vi bands. 138 Cr04 COMPLEXATION IN ETTRINGITE

Chromate can interact with ettringite to form surface complexes, polymerize, or precipitate

Ca or Ca-Al chromâtes. Preliminary IR studies of chromated ettringite and its solid solution with

SO4 were examined by KUMARATHASAN et al. ( 1990 ) (Table 5.7). This study suggests that

C1O4V3 vibrations do not show any splitting which further indicates that its tetrahedral symmetry is preserved upon incorporation into ettringite. Similar behavior was also observed for IR-active

V3 and V4 SO4 vibrations. According to these IR results, it can be argued that SO4 and Cr 0 4 are present as O S complexes in ettringite. Although they reported shifts in OH stretching fi-equencies, these authors have not correlated these with any possible ettringite structural changes.

Previous vibrational spectroscopic studies of Ca chromâtes (EI-RAFEI, 1981 ; DOYLE and

EDDY, 1967 ) showed that C1O4 is unperturbed by Ca, and the V3 frequency remains at 880 cm*' without any splitting. Their site symmetry analysis suggests a Td structure for C1O4 in CaCr0 4 and is also in support of their vibrational spectroscopic data. Substitution of larger divalent cations such as Ba^* for Ca^* perturbed Cr-0 bonds and the V 3 vibrations showed clear splitting, which were consistent with Cs site symmetry of C1O4 in BaCr0 4 . MILLER et al. ( 1971 ) studied the IR spectra of KBr-doped Ca and Cr 0 4 , and their results showed several peaks aroimd 900 cm"' which they attributed to the formation of different cationic chromate complexes such as comer-

(mono or binuclear) and edge-sharing complexes. Thus, previous studies indicate that Ca may not strongly perturb Cr-0 bonds (except the high temperature anhydrous salts) and the C1O4 tetrahedral symmetry is well preserved, in the case of CaCr 0 4 . To my knowledge, no vibrational spectroscopic studies have been conducted on Al chromâtes.

The available information on C1O4 Va and V4 band identification is limited and is also very difficult to interpret because of the close occurrence and overlap of these bands in the vibrational 139

Table 5.7. IR spectra of CrO^-ettringite (KUMARATHASAN et al. 1990).

Sample OH-stretching OH-bending SO4 Cr04 cm'*

(C r04)2 3485 , 3417,3238 1635 886 (S 04)2.7(C iO 4)i 3 3632, 3425,3240 1637 1116 886 (SO 4)2j(C lO 4)0.6 3640,3430 1637 1116,( 610) 878 (S 04) j.3 3638,3420 1670 1114,( 616)

The numbers in parenthesis are V4 vibrations of SO4

Table 5.8. Vibrational snectra of C 1O4 in NaÆiOd.

Na%Cr0 4 Vi V2 V3 V4 cm* Cr0 4 ^ (IR) aq. 880 N.A Ct0 4 ^* 846 349 890 378 IR 858-851 N.A 946 , 890,879 N.A Raman 851 351 933 , 900 , 873 408 , 392,361

* Reported by NAKAMOTO (1986). N. A: not analyzed since these bands are beyond the range of InstrumenL 140 spectra. Raman studies by WEINSTOCK et al. (1973) have identified these bands for aqueous

Cr0 4 ^ species, but no reports are available for solids. Additionally, crystal structure refinements are not available for hydrated Ca and Al chromate salts. Thus, it is difficult to evaluate Ca and Al effects on C1O4 local symmetry. Since this information is crucial in assigning vibrational modes and extracting structural information, solution and solid state IR and Raman measurements were conducted for Na chromate (NazCr 0 4 ) and dichromate (Na 2Cr207.2 H2 0 ). In these solids C1O4 bonds directly to Na (similar to IS complex) (NIMMO, 1981) and also dimerizes (in

Na2Cr2 0 ?.2 H2 0 ). This information was later used for band identification and interpretation of sorbed Cr 0 4 symmetry in ettringite.

CrOs Spéciation in Water. Na^CrOd and Na'>Cr707.2Hi0

Chromate forms dominantly Cr 0 4 ^ species in alkaline solutions (pKa = -0.2, 6.51 ) and dimerizes at acidic pH to Cr 2 0 ?^. ATR - FTIR measurements of Na and K chromate aqueous solutions produced a single peak at 880 cm ' (V3) (Fig. 5.20). With dimerization, the symmetry of the molecule changes from Td to Csv, C 2V. C, or C% (depending on Cr-O-Cr angle), the V 2, V3 and V4 vibrations split accordingly, and latent IR bands appear in the spectrum. The collected ATR -

FTIR spectra of aqueous Cr2 0 ? exhibit four distinct peaks at 950 (st), 901 (w), 881(s) and 772 (w) with a small shoulder at 931 cm ' (Fig. 5.20). On the basis of these vibrations, Cr 2 0 ?^ symmetry can be assigned to C 3v. The peak at 772 cm"' corresponds to Cr-O-Cr asymmetric stretching and other peaks at 950,901 and 931 cm'* correspond to the remaining C1O3 system. Since Cr207^ is always in equilibrium with monomeric species, the small peak at 881 cm * is assigned to asymmetric stretching of Cr-0 in C1O4. 141

0.6

CrO,

0.4 I •£ I 0.2

0.0

1100 1000 900 800 700 600 Wavenumber (cm'^)

Figure 5.20. ATR-Fl'lR spectra of aqueous CrO/ and CrzCh . 142 IR and Raman spectra of solid phase CrO/ and CrzO?^ V 3 vibrations exhibit splitting with

several new peaks as compared to the aqueous spectra of the same species (Fig. 5.20, Fig. 5.21).

Oxygen of Cr 0 4 in Na2CrÜ4 (s) are bonded to octahedral Na (NIMMO. 1981). This distorts the

C1O4 polyhedra with Cr-0 bond distances of 1.635 to 1.656 Â , which resulted in splitting of the V 3

vibrations (Table 5.8). Similar studies on NazCr 207.2H2 0 produced several sharp ER peaks, which

suggests that C1O4 symmetry is less than Td. The experimental results and theoretical calculations

of BROWN and ROSS (1972) suggest C^v symmetry for Cr2 0 ?.

Although C1O4 bonds to Na in Na 2Cr0 4 , the Cr-0 Vi band (851 cm ') did not show much

variation when compared to aqueous species (846 cm '; NAKAMOTO, 1986). However, strong

bond formation in Cr 2 0 ? produced Cr-O-Cr asymmetric stretch at lower wavenumbers and the Vi

band of the remaining C1O3 system shifted to higher wavenumbers because of shortening of these

three Cr-0 bonds (1.62 as compared to 1.78 Â in Cr-O-Cr, PANAGIOTOPOULAS and

BROWN, 1972). In summary, Cr-0 asymmetric stretching in Na 2Cr0 4 and Na 2Cra0 7 .2 H2 0

exhibit splitting, which indicate a lower site symmetry for C1O4 in these solids (Fig. 5.21) than

aquo C1O4 and the symnietric stretching (Cr-OX, X = Na) was not affected by C1O4 complexation

with Na. These spectral changes with variation in bonding were used to predict C1O4 coordination

in ettringite.

Cr04 Spéciation in Ettringite

Vibrational spectroscopic studies on Cr 0 4 -sorbed ettringites showed that the adsorbed

samples exhibited little variation in IR spectra as compared to coprecipitated samples. This may be attributed to the relatively low solid jAiase Ct 0 4 concentrations of the adsorbed samples and 143

00

1.5

en I 00 g 1 . 0 I

S

0.0 1000 800 600 400

W avenum ber (cm ‘^)

Figure 5.21. Vibrational spectra of CrO/ and Cr^O?^' in solids. FTIR spectra of a. NaaCrO»,

and b. Na^Cr^O^.ZH^O; c. Raman spectra ofNa^CrO^. 144 insensitivity of IR to such low concentrations (0.08 mol kg '). Thus, most of the current discussion is limited to the coprecipitated and concentrated (> 0.1 mol kg ') adsorbed CrOA-ettringites.

OH Stretching and Bendins Vibrations

Sulfate ettringite exhibits OH stretching bands at 3611, 3562, 3356, and 3222 cm ' and a small shoulder at 3648 cm '. With increases in coprecipitated CrO^, the peak at 3611 cm ' shifted to 3641 cm ', and new peaks developed at 3518, 3481, and 3190 cm ' (Fig. 5.22) Interestingly, IR measurements on solvated C1O4 (aqueous CrÛA) also exhibit asymmetric and symmetric OH stretching frequencies around 3508, 3348 and 3069 cm ' (Fig. 5.10), which suggests that at least some of the H 2O in ettringite is solvating or forming H bonds with CiOa. Coprecipitated ettringite samples with high CiOa and SOa concentrations showed features that are combinations of the two.

Assuming CiOa is present inside the ettringite channels and interacts with Ca polyhedra, shifts in the Al-OH stretching frequency from 3611 in ettringite to 3641 cm ' in CiOa ettringite may be attributed to the relatively high electronegativity of S over Cr.

Water in C1O4 and SOa solvation shells exhibited bending vibrations at 1638 and 1655 cm ', respectively. As the solid phase CiOa concentration increased, the bending vibrations decreased from 1650 to 1630 cm ', which offers similar chemical informaticm as that of stretching modes

(Fig. 5.23). Due to the close proximity of the strong V 3 peak of SOa to that of Al-OH bending, it is difficult to examine Al-OH vibrations with changes in sorbate concentration. In addition, Ca-OH stretching and bending modes (far-IR region) showed little variation.

SO 4 Vibrations

Sulfate vibrations showed several changes with increases in solid phase CiOa concentration in coprecipitated ettringite. At low soibate concentrations, SOa exhibited clear V 3 splitting and the 145

0.8 3518 3481

m rf)

0.6

0.2

I - cn 0.0 3800 3600 3400 3200 3000 2800 2600 2400 Wavenumber (cm^)

Figure 5.22. OH stretching bands of coprecipitated CrO^ ettringite. Solid phase CrO^

concentrations are: a: 0, b: 0.5, c: 4, d: 14, e: 18 and f: 20 %. 146

0.30

0.25

1629 (?)

o S

s

0.10

§ 0.05 _b

0.00 1750 1700 1650 1600 1550 1500 Wavenumber (cm^)

Figure 5.23. OH bending vibrations of coprecipitated CrO^-ettringite. Solid phase CrO^

concentrations are: a: 0, b: 4, c: 14, d: 18 and e: 20 %. 147 peaks were at 1190, 1158, 1106 cm ' (Fig. 5.24). Although V 4 should exhibit similar features in

IR, the presence of several bands around 600 cm ' complicated band identification (Fig. 5.25).

However, bands at 680, 642 and 609 cm ' were considered to be the result of degeneracy removal

in V4 vibrations. These band assignments were made after comparing with the IR spectra of gypsum SO4 vibrations. The S - 0 symmetric stretching modes were close to 990 cm ' and exhibited little variation with increases in sorbed Cr 0 4 . Dimerization or strong complexation of SO 4 would have increased its frequency (S - 0 in SO3) to higher wavenumbers (TWU and DUTTA, 1989), which is apparently not observed in these samples. Raman spectra also showed similar results, i.e. a strong Vi band at 989 cm"', a split V 3 (1138 and 1116 cm ') and broad V 2 and V4 bands centered at

455 and 617 cm ', respectively. This indicates that SO4 in ettringite is not greatly affected by C1O4 except at veiy high Cr 0 4 /S0 4 molar ratios.

Cr04Vibrations

IR and Raman showed distinct changes in all Cr 0 4 fundamental vibrations with increases in solid phase Cr 0 4 concentration. Aqueous Cr0 4 species produced a strong V 3 peak at 880 cm ' in

IR. However, the IR of Cr 0 4 in coprecipitated ettringite showed strong peaks at 933,915, and 789 with shoulders at 777 (± 3.0) cm ' (Fig. 5.26). This suggests lowering of Cr 0 4 tetrahedral symmetry in coprecipitated C1O4 ettringites. With increases in solid phase C1O4 concentration, the peak at 933 cm ' increased in intensity with no changes in other peak positions. This reduction of

Cr0 4 symmetry should result in the appearance of a latent Vi band in IR. Its absence in the IR spectra of these samples may be due to overlapping with a strong CO3 peak at 876 cm''. Since the

IR spectrum was collected only for the mid-lR region, V; and V 4 C1O4 bands were not reported in this study. However, this problem was alleviated by using Raman for this region. 148

0.8

0.6 o\ o I 0.4 I o

0.2

0.0

1300 1200 1100 1000 900 Wavenumber (cm'^)

Figure 5.24. S-0 stretching vibrations in coprecipitated CrO^-ettringite. Solid phase CrO^ concentrations are: a: 0, b: 4, c: 14, d; 18 and e; 20 %. 149

0.4 S

I 2 0.2

2

S

0.0 700 650 600 550 500 450 400

Wavenumber (cm‘^)

Figure 5.25. S-0 bending vibrations (as.) in CrO^-coprecipitated ettringite. Solid phase CrO^

concentrations are: a: 0, b: 4, c: 14, d: 18 and e: 20 %. 150

0.3

0.2 I I <

0.1

800 0.0 1000 900 800 700 Wavenumber (cm-i)

Figure 5.26. CrO^ coprecipitation in ettringite. Solid phase CrO^ concentrations are, a: 0

b: 4, c: 14, d: 18 and e: 20 %. 151 Raman spectra were collected for both adsorbed and coprecipitated Cr 0 4 -ettringite samples

(Fig. 5.27). Scans with longer duration were used for adsorbed samples, and were probed with a low intensity laser beam to avoid local heating effects. Sample fluorescence restricted the maximum scan time to about 15 minutes. Thus, peak intensities in the Raman spectra do not relate to the species concentration, since the samples have variable scan times.

Raman spectra of CrO# adsorbed ettringites showed distinct peaks at 933, 902, 883 cm'^ which correspond to V 3 splitting, and a peak at 861 cm'* indicates the Vi vibrational mode. With increases in solid phase C1O4 concentration, several new peaks at 903, 855 and 780 cm'* appeared in addition to the above mentioned peaks. This suggests that C1O4 is present in more than one coordination environment with increases in concentration. Longer sample scans produced very large V2 and V4 SO4 bands and these peaks masked the other Cr-0 vibrational modes at low wavenumbers.

Coprecipitated Cr 0 4 -ettringite samples showed strong Vi peaks at 860 and 855 cm * with several peaks. The small Vi shift of these samples to high wavenumbers (846 cm * for aqueous

Cr0 4 ) may indicate that Cr 0 4 binds to Ca. Similar shifts were also observed for S-0 vibrations in gypsum and anhydrite. Although SO 4 peaks masked several C1O4 bands in dilute samples, the V 2 and V4 peaks were observed at 418, 407, 375, 364 and 356 cm * in concentrated samples. The strong peak at 356 cm * was tentatively identified as V 4 on the basis of its intensity (WEINSTOCK et al., 1973). Thus, the complimentary information obtained from Raman clearly suggests that the

C1O4 symmetry in chromated ettringites is lower than Td. Identification of C 1O4 reactive sites in ettringite (A, B, C, etc.) are difficult, since Ca and Al chromate vibrational spectra are not available for comparison. 152

18000

16000

14000

g 12000 a I* 10000

I 00 s 8000 N î ON I 6000 1 4000 n OO 00 2000 vo i 00

950 900 850 800 750 Raman Shift (cm'^)

Figure 5,27. Raman spectra of Cr-0 vibrations of CrO^ in adsorbed (b & c) and coprecipitated (d, e, & f) CrO^-ettringite. Solid phase CrO^ concentrations are: a: 0, b: 0.5, c: 4. d: 14, e: 18 and f: 20 %. 153 Carbonate IR peaks (va stretch) were relatively undisturbed by increases in solid phase CrO^ concentration, suggesting that sorbed Cr 0 4 did not affect adsorbed CO3. Cr0 4 sorption also has no effect on Ca-0 vibrations in the far-IR region (~ 320 cm ') which indicates little variation in coordination of Ca polyhedra.

Summary of C 1O 4 Coordination in Ettringite

On the basis of these vibrational spectroscopic studies, the following conclusions can be made on Cr0 4 complexation in ettringite:

1) Cr0 4 forms IS complexes in adsorption and coprecipitation samples, and reduction in Cr 0 4

Td symmetry may be attributed to its conq>lexation (IS) with ettringite Ca or Al polyhedra. Sorbed

Cr0 4 forms H bonds with ettringite H 2O molecules.

2) Precipitation of a Ca chromate in the system can be ruled out since these phases exhibit OS

C1O4 complexes.

3) Cr0 4 may not polymerize in this system since the Cr-0 V 3 band does not exhibit extensive splitting due to the formation of Cr-O-Cr bonds. CHAPTER VI

EXAFS OF ASO4 AND CrO^ ETTRINGITE

INTRODUCTION

Macroscopic sorption and desorption experiments indicated that ASO 4 and Cr 0 4 can interact with ettringite surfaces and substitute inside channels, and gave clues for the existence of inner- and outer-sphere complexes (Chapters HI & FV). Convincing evidence for these oxyanion complexes was later provided by vibrational spectroscopy (Chapter V). However, uncertainties still exist regarding the nature of these oxyanion interactions. Specifically more information is needed to determine the reasons for breakdown of the ettringite structure during oxyanion adsorption as compared to coprecipitation, and to evaluate oxyanion surface polymerization (in the case of Cr0 4 >. Similar questions have been answered with EXAFS (Extended X-ray Absorption

Fine Structure) spectroscopy by numerous investigators.

The theory, limitations, and applications of EXAFS to earth materials have been well documented (e.g. ASHLEY and DONIACH, 1975; LEE and PENDRY, 1975; STERN et al.,

1975; CRAMER and HODGSON, 1979; EISENBERGER and LENGELER, 1980; TEO 1986

WAYCHUNAS et al., 1986; KONINGSBERGER and PRINS, 1988; BROWN et al., 1988

CHARLET and MANCEAU, 1992; MANCEAU et al., 1992; WAYCHUNAS et al., 1993

O’DAY et al., 1994; ZABINSKY et al., (in print); REHR et al., 1994). Thus they wiU not be

154 155 presented here. The reader is referred to one of the citations listed above for detailed background information.

EXAFS spectra produced above the absorption edge of an element due to backscattering from neighboring atoms can be represented by:

AT xik) = (Nj/R/) Fj(ifc) S,(A) exp(-2o/ 1^) exp[-2Rj/A(ifc)] sinI2kRj +

For backscattering from a particular coordination shell, all the above terms can be combined and represented as,

A" X(A) = A " S Aj(A) srn[Y(j(A)] (2)

In Equation 2, Aj(k) represents the total backscattering amplitude and sinl'PyfA:)] represents the total phase for a pair, i,j. These phase and amplitude functions relate to a set of bond distances

(Rj) and number of scatterers (Nj) respectively. In routine EXAFS analysis, application of this equation to the study of unknown compounds and extraction of Rj and Nj depends on how well the phase and amplitude functions can be obtained from reference compounds or theoretical calculations (WAYCHUNAS et al., 1986; O’DAY et al., 1994). In addition, study of oxyanion 156 complexation with weak backscatterers, such as Al and Ca, in ettringite depends on the accuracy of

(f.

EXAFS studies of Cd and selenite sorption on Al oxides exhibited very little backscattering from the neighbors immediately beyond the first shell oxygens (PAPELIS et a i, 1995), despite their IS complexation with Al oxide surfaces. These authors attributed this behavior to low backscattering from Al; however, they did not make any comparisons with structurally known Al- selenito complexes or minerals. Similar studies conducted on selenate substitution in calcite

(REEDER et al., 1994) showed significant backscattering from Ca, and obtained (f were in the range of 0.005-0.01 Similar high (f for Ca backscatterers in hedenbergite and andradite restricts the accurate evaluation of Nj at higher Rj for Ca (O’DAY et al., 1994). Ettringite column ions (Ca, Al) are poor backscatterers and the EXAFS parameters (such as (f, Nj and Rj) obtained in the data fits for these backscatterers are crucial to identify the sorbed oxyanion molecular structures. Thus, the EXAFS of oxyanion sorbed ettringites were interpreted with caution, and were supplemented by the symmetry information obtained fi-om vibrational spectroscopy. The main objectives of this study are to evaluate: 1) the sorbed ASO 4 and Cr 0 4 molecular environment and the type of reactive sites in ettringite; 2) the effect of sorption mechanisms (adsorption and coprecipitation) on oxyanion solid phase spéciation; and 3) the concentration limit where ettringite structure breaks.

MATERIALS AND METHODS Materials

EXAFS analysis was conducted on ASO4 and Cr 0 4 reacted ettringites obtained in previous experiments (Chapters III and IV). Modd AsÛ 4 compounds were obtained from the Smithsonian

National Museum of Natural History, Washington D C. The samples were haidingerite (# 157 B14402) from Schneeberg, Germany; rauenthalite (# 119358) and sainfeldite (# 140952) from

Gottes Mine, France; and phaimacolite (# R9079) from Rauenthal, Alsace, France.

EXAFS Data Collection

X-ray absorption spectra were collected o t the wiggler beamlines IV-1 and IV-3 at the

Stanford Synchrotron Radiation Laboratoiy (SSRL) for As and Cr K-edges at ambient and liquid

He temperatures (~ 10 K, As only). An unfocused beam and Si (111) monochromator crystal were used in this study. During data collection, higher order harmonics were suppressed by detuning the monochromator until flux from the incident beam was reduced to ~ 40-50 %. Both fluorescence and absorption spectra were collected simultaneously from 200 eV below the absorption edge to

1000 eV above the absorption edge. Ar and N 2 -filled ionization chambers were used to measure incident and trairsmitted beam intensities. A Lytle-type detector (LYTLE et al., 1984) with Ge filter for As, and V filter for Cr were used to collect fluorescent radiation. In the case of Cr, the beam path inside the sample hutch was enclosed in a polyethylene bag and continuously fed with

He to suppress beam absorption by air. The synchrotron electron beam aiergy was 3.0 GeV and the beam current ranged from 20 to 90 mA. The incident X-ray beam energy was calibrated using

GaAs for As and Cr (III) metal foil for Cr. Energy resolution was better than 4 eV at the As and 2 eV at the Cr K-edges.

Powdered samples were mounted either between mylar windows or two thin kapton tape sheets which were supported by plastic slide holders. Concentrated samples were diluted with boron nitride, while low concentration samples were run without dilution. The number of scans

(individual spectra collected) varied from sample to sample depending on the absorber concentration. There were usually 6 to 25 scans, with the duration of each scan approximately 1.2 h (0.5 h in the case of model compounds and highly concentrated samples). All measurements at 158 10 K were performed with an Oxford Instruments CF1208 liquid He cryostat loaned by the SSRL

Biotechnology Group. The temperature inside the He cryostat sample compartment was maintained at 10 (± 3) K.

EXAFS Data Analysis

Data were analyzed with the EXAFSPAK software package (GEORGE and PICKERING,

1993) which is based on curved-wave backscattering formalism and the single scattering EXAFS approximation. The collected sample scans were averaged and processed after calibration for energy and edge position. If any of the scans were found noisy (due to changes in the incident beam, etc.), they were not considered further. EXAFS analysis of model compounds and arsenated ettringite samples showed that the transmission and fluorescence spectra gave similar information; however, fluorescence data were used for dilute samples due to better signal-to-noise ratio, and transmission data for concentrated samples (e.g. model compounds) to avoid any self absorption amplitude effects for fluorescent radiation (TEO, 1986; WAYCHUNAS et ai, 1993).

The first stage of EXAFS data analysis was background subtraction (Fig, 6.1). This was carried out by using a polynomial pre-edge fit below the absorption edge, and a cubic spline fit with 3 or 4 spline segments above the edge. The spectra were normalized with a Victoreen function using Victoreen coefficients calculated in the EXAFSPAK. EXAFS were extracted after the spectra were converted from a photon energy function, E (eV), to a photoelectron wave vector, k (‘‘). During the conversion, Eo (energy at wiiich /: = 0) for As and Cr were fixed at 11880 and

6002 eV respectively. The spectra were weighted by to compensate for damping of oscillations at high k. The ^-weighted EXAFS spectra were then Fourier transformed using a Gaussian-limited window to produce radial structure functions (RSF). The RSF peaks correspond to central absorber and neighboring backscatterer (at distance, Rj) pair correlations. Each peak in the RSF, 0.09 1.5 0.06

I 0.03 I I 0.00 0.5 -0.03

-0.06 12000 12300 12600 4 6 8 10 12 Energy (eV) k(A-*) 4 0.008 ■ë 3 I 2 0.000

1

-0.008 0 0 1 2 3 4 5 6 4 6 8 10 12 R + A(A) k (A ')

Figure 6.1. EXAFS data reduction. Solid line represents experimental data and dotted line shows the fit a:Absoiption edge, b: Raw EXAFS, c: FT of the EXAFS, and d: Fourier filtered second shell and its fit.

% 160 when filtered and backtransformed into k space, has frequency and amplitude representing the type and number of backscatterers at a distance Rj from the absorber. If the RSF showed any peaks at very small distances (in R-space), or if the peaks were not well resolved, Fourier transforms were collected with a different Gaussian window or the background subtraction procedures were iterated until the best results were obtained.

The conspicuous peaks in the RSF represent backscattering from neighboring atoms around the central absorber. For analysis, these peaks were first backtransformed into k space and the resulting ‘filtered’ EXAFS fit separately. Using the obtained fit parameters for these shells, the complete RSF (0.2 5.0 Â in R+A) was also backtransformed and analyzed as described below. A non linear least squares method (Marquardt algorithm) was used to fit the Fourier filtered peaks with phase shifts and amplitude functions calculated from FEFF6 (theoretical) or extracted from model compounds or structurally similar materials, as described later in this section. In this fitting procedure, the number of backscattering atoms (Nj), their distance from the absorber (Rj), and the

Debye-Waller parameter ((f) were usually considered as adjustable parameters. Although a shift in threshold energy (A Eo) was treated in a similar fashion, its value obtained in the first shell fitting was later used as a fixed parameter for the other shells. To confirm the values obtained in the above exercise, the curve fitting is carried out following the protocol:

a) fit the first shell, extract AEo and collect residua,

b) fit the second and higher order shells by fixing A Eo to that of the first shell and floating all other variables, and

c) fit the complete back - Fourier transform with the values obtained from the component shells. 161 The number of variables used in the fitting process were always smaller than the maximum number given by the Nyquist equation, , where AR and AJfe are the window widths in R and k space ji of the Fourier filtered data.

Theoretical phase and amplitude fiinctions for O, As, Ca, and A1 were calculated from first principles using the FEFF 6 code (REHR and ALBERS, 1990) and the published AIASO 4

(GIFFON et al., 1983); haidingerite, CaHAs 0 4 .H2 0 (FERRARIS et al., 1972); and weilite,

CaHAs0 4 (FERRARIS and CHIARI, 1970b) crystal structure refinement data. These phase and amplitude functions were used to fit the EXAFS spectra of structurally known compounds such as

Na2HAS0 4 .7 H2 0 (FERRARIS and CHIARI, 1970a), rauenthahte, Ca3(AsO4)2.10H2O (CATTI and IVALDI, 1983), pharmacolite, CaHAs 0 4 .2 H2 0 (FERRARIS, 1969); and haidingerite

(FERRARIS et al., 1972). Relative errors in Rj and Nj and obtained from model compound

EXAFS fits were evaluated for their variation with distance and backscattering atoms. The errors in bond distances and coordination numbers were of the order of 0.01 Â and <10 % for As-0, respectively; and 0.02-0.1 Â and 25-35 % for As-Ca, respectively. Goodness-of-fit (F) is measured by taking the sum of squares of the differences between experimental and calculated values of the backtransforms (GEORGE and PICKERING, 1993). The data fits are shown for each backscatterer type and model compounds in i^pendix D. Cr samples were analyzed in a similar way; however, the absence of good model compounds restricted the analysis. Nevertheless, the RSF gave significant information on CrÛ 4 spéciation in ettringite. Amplitudes due to multiple scattering paths were observed to be poor in these samples due to the presence of weak backscatterers and low stmctural symmetry and thus were not considered during EXAFS fitting. 162 RESULTS AND DISCUSSION

EXAFS results of oxyanion coordination in ettringite are discussed under three sections.

Possible A sO a and C1O4 coordination sites in ettringite are presented with bond distances between

As/Cr and other ions in ettringite. As and Cr K-edge EXAFS are later discussed separately.

STRUCTURE AND OXYANION COORDINATION IN ETTRINGITE

A brief introduction to the ettringite crystal structure and its possible oxyanion coordination sites are presented in Chapter I (Fig. 1.2). However, a detailed discussion of the same with

As04/Cr0 4 near neighbors (once adsorbed onto ettringite) and their approximate bond distances are presented here for each site (Table 6.1, Fig. 6.2). As shown earlier, the ettringite structure consists of columns and channels. The columns (parallel to the ‘c’ axis) are made up of A1 octahedra (Al(OH)g) sandwiched between three Ca antiprisms (Ca(H20)4(0H)4) on either side and are linked through OH (Fig. 6.2). In ettriiigite, SO4 occupy the channels and balance the column cationic charge.

Arsenate and Cr 0 4 sorption by ettringite may result in formation of solvated OS; and comer-, edge- or face-sharing IS complexes with column A1 and/or Ca polyhedra. The sizes of ASO 4 and

C1O4 polyhedra are very similar (As-0 = 1.68 Â, Cr-0 = 1.64 Â; Oak-Oas = 2.75 Â, Ocr-Ocr =

2.63 Â) and thus, their ability to interact at Ca/Al polyWra edges should be similar. However, the differences in charge and reactivity of these oxyanions may restrict them to only certain sites.

The size of reacting polyhedra (Al/Ca) edges and faces should be similar to those of AsOa (or

C1O4) if edge- or face-sharing complexes are to form. Edge sharing complexes are common in minerals. However, face sharing ASO 4 complexes with A1 (e.g. in the case of Fe octahedra,

WAYCHUNAS et al., 1993) or Ca polyhedra (studies of various Ca-arsenate model compounds Table 6.1. Oxyanion reactive sites in ettringite.

Type of Site Description Oxyanion Location in Ettringite Structure Âs-Ca/Âl Bond Distances Site ‘A’ Comer sharing with Ca or A1 Al: Crystal edges parallel to the (001) plane. 3.6 for linear As-O-Al, 3.1 polÿhedra. for a tilt of 55” (ZAs-O-Al = 125”). Ca: On all edges * Ca-OH sites parallel to the (001) plane. 4.2 - 4.0 for linear As-O-Ca 3.8 - 3.6 for ZAs-O-Ca = 125” * Ca-OHz sites on edges other than (001). 4.4 - 4.1 for linear As-O-Ca 4.0 - 3.8 for ZAs-O-Ca = 125”

‘B ’ Comer sharing, bidentate Al: Can not form. binuclear complexes. Ca: * Ca-OH sites parallel to the (001) plane (at ‘C’ -3 .8 type site). * Ca-OHz sites on edges other than ( 00 1 ). -4 .0

® C1O4 also produces similar cationic distances since Cr-0 bond lengths are close to those of As-0.

* At ‘C’ type site, ASO 4 may also fonn tridentate complex with three Ca polyhedra (Ca-OH) parallel to the (001) plane. However the OH-OH distances are about 2.46 and 2.67 Â, which arc slightly smaller than the edges of ASO 4 (~ 2.75 Â).

Table 6.1. Continued on the next page.

a Table 6.1. Continued.

Type of Site Description Oj^anion Location in Ettringite Structure As-Ca/AI Bond Distances Site (Â) ‘C’ Corner sharing, monodentate Al: Can not form. binuclear and bidentate binuclear (discussed under ‘B’ Ca: Ca-OH sites parallel to the (001) plane. ~ 3.8 for linear As-O-Ca type site).* arrangement

‘D’ Edge sharing with Al polyhedra Al-OH sites parallel to the (001) plane. -2 .4

‘E’ Edge sharing with Ca Ca-OHz sites on edges other than (001). 3.3 - 3.1 polyhedra Ca-OH sites on edges parallel to the (001) (at ‘C’ type site).

‘F’ Edge or comer sharing with Al At the broken edges parallel to the (001) plane. A l-A s: 2.4(edgesharing) and comer sharing with Ca -3.2

Ca - As : 4.0

‘G’ Outer-spheie inside the channels.

& 165

yVater

Hydroxyl

Calcium

Aluminum

Figure 6.1. Structure of an ettringite column. Alphabetic labels indicate different sites that may interact with oxyanions. 166 discussed later in this section) have not been observed which may be due to the very close approach of cations, and thus, are not considered any further.

Comer sharing ASO 4 complexes can form on ettringite external surfaces with Al and Ca polyhedra by replacing OH or H 2O without any restrictions of polyhedral dimensions (‘A’-type sites; Fig. 1.2, 6.1). In this case, the distances between the cations (As and Ca/Al) depend on the

As-O-X (X = Ca/Al) bond angle. Arsenic-Ca distances of 4.4 to 4.0 Â are expected for As-O-Ca moieties in linear arrangemaits. This corresponds to Ca-OH^/OH bond lengths of 2.75 to 2.36 Â respectively. However, when As-O-Ca is not linear, then As-Ca bond lengths are expected to be lower depending on As-O-Ca angje (e.g. 3.75 Â for ZAs-O-Ca ~ 125°). Similar interactions with

Al can produce As-Al bond lengths of 3.5 to 3.1 Â [for linear and a tilt of 55° (from linear As-O-

Al configuration), respectively; e.g. 3.15 Â in AIASO 4, GIFFON et al., 1983]. However, as shown by the crystal structure, Al is exposed only on the edges parallel to the (001) plane (Fig.

6.2). Since Al polyhedra are surrounded by Ca polyhedra within the columns, and also due to steric constraints, ASO4 complexation with Al parallel to the columns may not be possible.

However, similar interactions with Ca may take place either parallel to (001) surfaces or to the columns.

Formation of bidentate binuclear complexes (sharing of two Ca/Al polyhedra with two comers of a single ASO 4 polyhedron, B-type sites) produce smaller As-X bond lengths. Al polyhedra can not form these kinds of complexes, since two consecutive Al polyhedra within a column are separated by Ca polyhedra (closest approach of OH is ~ 4.3 A); and are farther apart for a linkage betweai Al of neighboring columns (~ 9.0 A). A similar situation exists for inter­ column linkage in the case of Ca (sq>arated by ~ 4.5 A). However, bidentate binuclear complexes are possible for intra-column Ca polyhedra by replacing channel H 2O. The closest approach of two H2O of neighboring Ca polyhedra are 2.82 and 3.14 A. A complex with the latter one may be 167 difïïcult (0 -0 in ASO4 is 2.60-2.85 Â) since it creates extensive bond stretching in both As and Ca polyhedra. Arsenate interactions at these B-type sites may produce As-Ca bond lengths of 4.0 Â

(assuming no tilt along the two bonded oxygens), since Ca-OHz bond lengths are greater than 2.60

Â. In Ca-arsenate model compounds used in this study (discussed later), bidentate binuclear complexes produced As-Ca distances of 3.5 to 3.6 Â. These small distances when compared to those at B-type sites in ettringite are attributed to the relatively short Ca-0 distances of model compounds, for e.g., 2.33 Â in the case of rauenthalite.

Arsenate interactions with intra-column OH are difficult because of steric constraints; however, surface OH can exhibit this kind of complex. When these complexes are formed with Ca polyhedra at the column broken edges (parallel to ((X)l)), ASO 4 exhibits bidentate mononuclear symmetry with one Ca (As-Ca ~ 3.2 Â) and bidentate binuclear symmetry with the remaining two

Ca atoms (As-Ca ~ 3.7-4.0 Â) (‘C’-type sites in Fig. 6.2). ASO 4 monodentate binuclear complexes are also possible at this site when one O of an ASO 4 unit connects to an exposed column OH (As-

(Ha ~ 3.8 Â or less depending on angle of tilt).

In the case of edge sharing complexes, Al and Ca polyhedra edges should be similar in length to those of ASO 4. In Ca arsenates, ASO4 edges are typically in the range of 2.60 - 2.85 Â, and edge sharing ASO 4 with Ca are dominantly 2.70 (± 0.05) Â long. The edge length in Al polyhedra are in the range of 2.45 to 2.80 Â, close to those of ASO 4, which makes it possible to form edge-sharing complexes with ettringite surface aluminals (As-Al ~ 2.4 Â) (‘D’-type sites; Fig. 1.2, Fig. 6.2).

However, these edge sharing complexes results in very close approach of cations (~ 2.4 Â); and due to charge repulsion between the cations these kinds of complexes are highly unlikely to form.

Even if the complexes are formed, the charge repulsion may increase the separation between the cations. This is also observed in semi empirical calculations of edge sharing Al arsenate complexes. These theoretical calculations produced an As-Al bond distances of 2.9 Â, and similar 168 As-Al distances may be observed for edge sharing complexes at ‘D’-type sites. The Ca polyhedra edges, however, are slightly longer (2.91-3.20 Â) than those of ASO 4. Thus, the shortest Ca polyhedra edges may form bidentate complexes, but, this may result in Ca-OHz and As-O bond distortion (As-Ca ~ 3.2-S.3 Â, since Ca-OHz ~ 2.60-2.75 Â) (‘E’-type sites. Fig. 1.2, Fig. 6.2).

As discussed above, Ca polyhedra exposed on the surface (with terminal OH) can exhibit bidentate complexes since OH-OH distances are in the range of 2.60-2.85 Â (‘C’-type, Fig. 1.2, Fig. 6.2).

In addition, ASO 4 may also form edge-sharing complexes with Al polyhedra and by connecting to two Ca at the column broken edges (‘F’-type sites). This produces As-Ca and As-Al bond distances of 3.7 and 2.4 Â respectively.

The O S ASO4 complexes may form inside the channels by replacing SO4 (‘G’-type sites; Fig.

1.2). In this case, As-Ca/Al distances are > 4 Â. In summary, As-Ca bond distances of > 4.0, 3.5-

4.0 and 3.2 Â correspond to comer sharing (linear arrangement) at A and B-type sites, bidentate binuclear and tilted monodentate binuclear comer sharing complexes at B and C-type sites, and bidentate mononuclear complexes at E-type sites respectively. Similarly, As-Al bond distances of

3.1-3.6 and 2.3 Â represent corner sharing at A-type sites and edge sharing complexes at D and F- type sites respectively. In the case of Cr 0 4 complexes, Cr-Ca and Cr-Al bond lengths are expected to be smaller than the above mentioned distances by about 0.1 Â (different tilt angles for C1O4 may have different lengths). Thus, the structural analysis indicate that it is difficult to distinguish comer sharing monodentate complexes and bidentate binuclear complexes in the case of Ca in ettringite.

The reactive sites and the As-Ca/Al bond lengths presented in this section correspond to oxyanion adsorption on surfaces or channel substitution in ettringite. On the other hand, precipitation of new solids may exhibit different coordination environments, such as As-As 169 correlations (< 4.5 Â), which are discussed under Ca arsenate model compounds. These differences may help to distinguish adsorption versus precipitation reactions.

A s - EXAFS OF ASO4-ETTRINGITE

EXAFS ofCa Arsenate Model Compounds

Arsenic EXAFS of the model ASO 4 compounds were in excellent agreement with their reported crystal structures (Table 6.2). Fourier transforms of these model compounds showed small amplitudes for second shell Ca. Similar behavior was also reported for Ca-containing garnet and pyroxene EXAFS (O’DAY et al., 1994), and this also resulted in relatively high ( f for similar low atomic number backscatterers. The XAS spectra collected at 10 K for haidingerite did not reduce the ( f of the fits significantly compared to the results for 298 K, suggesting that the disorder is mainly static rather than thermal (Table 6.3). The model compound fits were good for

As-0 and As-Ca bond lengths < 3.5 Â; ard these varied by < 0.01 Â for As-0 and < 0.03 Â for

As-Ca. At longer distances, the error in estimation of bond distances increased to 0.1 Â and difficulties were experienced in resolving Ca at 3.41 and 3.78 Â (e.g. in the case of pharmacolite).

The values for As-0 were in the range of 0.001-0.(X)3 and for As-Ca they were in the range of 0.006-0.011 Since the amplitudes from single scattering were small, multiple scattering amplitudes are expected to be even smaller, and, hence were not considered in these sample data fits (Fig. 6.3).

Arsenate exists in two different types of polyhedra in equal density in rauenthalite; one (Asl) forms bidentate binuclear complexes with two Ca, and the other (As2) forms bidentate mononuclear, bidentate binuclear and monodentate complexes (Table 6.2 and also see Chapter V). Table 6.2. Comparison of crystal refinement and EXAFS fits of Na and Ca arsenates.

Mineral Crystal Structure Fits of EXAFS Data Refinement

Backscattering N R(Â) N R(Â) o* AEo(eV) F atom (P) Sodium Arsenate As-O 1 1.74 1 1.74 0.0021 -2.10 54 NazHAs04.7H20 3 1.67 3 1.68 0.0023

Rauenthalite Fluorescence Data Ca3(AsÜ4)2.10H2G As-0 4 1.68 4 1.69 0.0016 -3.92 66* As-Ca 1 3.21 1 3.22 0.0094 2 .1* As-Ca 3 3.67 3 3.64 0.0097 As-As 0.5 4.06 0.5 4.09 0.0057 As-As 0.5 4.53 0.5 4.51 0.0062 Transmission Data

4 1.69 0.0011 -3.36 90* 1 3.25 0.0067 2.4* 3 3.65 0.0084 0.5 4.12 0.0066 0.5 4.42 0.0051

Pharmacolite As-0 4 1.68 4 1.68 0.0014 -7.4 67* CaHAs04.2H20 As-Ca 1 3.20 1 3.22 0.0063 1.5* As-Ca 1 3.41 1 3.38 0.004 As-Ca 2 3.78 2 3.56 0.0106 As-As 2 3.89 2 3.83 0.017 Goodness-of-fit for the complete FT. * Goodness-of-fit for all shells combined excepting the first As-0 shell. Table 6.3. Comparison of haidingerite EXAFS fits collected at 298 and 10 K.

Mineral Crystal Structure Fits of EXAFS Data Refinement (XRD) Backscattering N R(A) N R(Â) AEo(eV) F atom (A') Haidingerite Temperature 298 K CaHAs0 4 . HzO As-0 3 1.653 4* 1.68 0.0039 -8.0 72* As-OH 1 1.77 (avg. 1.683) 4.7* As-Ca; 1 3.25 1 3.32 0.0055 As-Ca% 2 3.60 2 3.50 0.0062 As-Caz 2 3.73 2 3.67 0.0047 As-As 2 4.31 2 4.22 0.0171

Temperature 10 K 4* 1.69 0.0017 -0.45 35* (avg. 1.683) 4.1* 1 3.32 0.0109 2 3.52 0 .0 0 2 2 3.69 0.004 2 4.25 0.019 * Goodness-of-fit for the complete FT. * Goodness-of-fit for all shells combined excepting the first As-0 shell. 172

7 As-0

6

5 I 4

3 I As-As & Na 2 As-Ca

1 As-Ca

0 0 1 2 3 4 56 7 R + A(A)

Figure 6.3. Fourier transforms of a: haidingerite, b: rauenthalite, and c: Na^HAsO^.VH^O. 173 The As2-Ca distances for these complexes are 3.21, 3.63 and 3.75 Â respectively and Asl-Ca bond distances also follow the same trend (3.63 and 3.71 A). Average As-Ca distances of 3.21 and 3.67 A were used to fit the EXAFS spectra since it was difficult to resolve atoms separated by

< 0 .1 A. The data fits suggest that the values (0.005-0.01 A^) are reasonable and not unusually large. A good agreement between the transmission and fluorescence data supports the consistency of these analyses.

Pharmacolite has only one type of As, but it forms all the complexes that rauenthalite exhibits, and the As-Ca bond distances in these two minerals are comparable. However, in pharmacolite one of the As-Ca bidentate mononuclear complexes has an As-Ca bond distance of

3.41 A, which is due to binding of the longest Ca - 0 (2,96 A) with a protonated terminal O of

HASÜ4. Ettringite has no such long Ca-0 bonds to exhibit extended bidentate complexes with

ASO4. Haidingerite also exhibits As-Ca distances of 3.59-3.61 A, but these correspond to bidentate binuclear complexes; in addition, the protonated terminal ASO 4 O do not participate in bonding and none of the Ca-0 bonds are longer than 2.75 A. In contrast to these multidentate

ASO4 coordination environments in Ca arsenates, ASO4 is completely solvated in Na 2HAs0 4 .7 H2 0 , and its RSF showed no distinct peaks around 2.5-4.0 A; however, backscattering from Na and As atoms can be seen above this distance (Fig, 6.3).

Studies of these model compourxls indicate possible coordination environments and their bond distances, and Debye-Waller parameters for second shell Ca and As atoms. Although Al arsenate model compounds were not studied, Hyperchem simulations (Chapter V) showed that As-Al distances are expected to be around 2 .8 A (bidentate mononuclear) and 3.1 A (monodentate or bidentate binuclear complexes). 174 As0 4 Spéciation in Ettringite

The spectra collected both at 298 (concentrated samples) and 10 K (dilute samples) were used in the study, since these samples exhibit no variation in values excepting for significant improvement in signal-to-noise ratio (e.g. in the case of haidingerite. Table 6.3). The results of adsorption, coprecipitation and desorption are presented and discussed separately for different shells (Tables 6.4 & 6.5).

Collected X-ray absorption spectra showed beat patterns in raw EXAFS (Fig. 6.4) indicating a possible second neighbor either as a bidentate and/or monodentate complex, excepting for the dilute coprecipitated samples (Fig 6.4-2b) (details are presented later in this section). A typical

RSF of ASO4ettringite EXAFS exhibits three predominant peaks, at 1.35, 2.8 and 3.4 Â (Fig. 6.5).

The strong peak at ~ 1.35 Â (uncorrected for phase) corresponds to backscattering from first shell

O around As. Fourier filtered and phase corrected data indicate that this peak represents 4 (± 0.3)

O at an average distance of 1.685 (± 0.005) Â. A second weaker peak at 2.8 Â in the RSF represents backscattering from Ca or Al bonded as edge or comer sharing complexes respectively.

The third peak at 3.4 Â corresponds to Ca coordinated to ASO 4 in monodentate mononuclear/binuclear and/or bidentate binuclear symmetry. ASO4 ettringite RSF often showed a broad peak around 3.4 Â representing significant backscattering from neighboring, closely spaced

As/Ca. Goodness of data fits (F) for the entire backtransform mainly depended on how well the first shell was fitted, and the addition of second and higher order shells produced only a slightly improved overall fit. ASO 4 complexation in ettringite is expected to exhibit different species as discussed above, and EXAFS data analysis of these samples should indicate the average number and type of these species. However, information obtained from second shell coordination, such as Table 6.4. EXAFS analysis of ASO4 - adsorbed ettringite.

Solid Phase ASO4 pH Backscattering N R(Â) 0^ AE«(eV) F Concentration atom (A') mol kg ' 3.24 X 10^ 11.4 As-0 4* 1.69 0 .0 0 0 2 0.14 78* As-Al; 1.28 3.21 0.0046 2.4* As-Caz 1.4 3.57 0.008 As-Caa 0.7 4.42 0.01 6.48 X 1 0 " 11.4 As-0 4* 1.68 0.001 -3.25 94* As-Ali 0.71 3.17 0.0006 4.0* As-Ca^ 1.78 3.59 0.012 As-Cas 0.9 4.20 0.0077 0.289 11.8 As-0 4* 1.69 0.001 -4.22 66* As-Al; 0.5 2.99 0.0023 0.5* As-Caz 0.75 3.18 0.007 As-Caz 1.08 3.53 0.0089 As-Caz 2* 3.73 0.008 (0.89) (3.73) (0.003) 0.648 11.8 As-0 4* 1.69 0.0011 -3.4 54* As-Ca; 0.80 3.20 0.0052 1.1* As-Caz 0.5 3.47 0.0076

As-Cai 2 * 3.70 0.0093 As-As 0.51 4.46 0.0068 * Coordination number fixed either because floated value is close to the fixed number, or is relatively small when compared to model compounds. * Goodness-of-fit for the complete FT. * Goodness-of-fit for dl shells combined excepting the first As-0 shell. Numbers in parenthesis represent the values obtained by floating everything, but this shell is fixed at a higher coordination number to get reasonable «f that is compatible with the models. Subscripts 1,2, and 3 represent first, second and third shells for Al/Ca neighbors. „ DÎ Table. 6.4 Continued

Solid Phase AsO. pH Backscattering N R(Â) é' AEo(eV) F Concentration atom (A') mol kg’* 1.44 11.8 As-0 4* 1.69 0.0017 -3.16 69* As-Cai 1.0* 3.20 0.0049 4.4* (0.53) (3.19) (0.0013) As-Cai 2* 3.70 0.0092 (0.98) (3.68) 0.0055 As-As 1 4.51 0.0067

2.30 10.9 As-0 4* 1.69 0.0021 -3.63 44* As-Cai 0.98 3.25 0.0161 0.2* As-Caz 3.04 3.61 0.0193 As-Caz 1.12 4.51 0.007 As-As 1.01 4.95 0.0073

S\ Table 6^. EXAFS analysis of ASO4 - coprecipitated ettringite.

Solid Phase AsO< pH Backscattering N R(Â) 0 ^ AEo(eV) F Concentration atom (A') mol 5.75 X 10'^ 12.4 As-0 4* 1.69 0.0006 -4.88 110* As-Cas 1.27 4.07 0.0074 4.9* 0.144 12.4 As-0 4* 1.69 0.0009 -4.99 46* As-Caj 1* 3.61 0.0118 0.68* As-Caz I* 3.90 0.0076 As-Ca$ 1* 4.09 0.0084 0.216 11.9 As-0 4* 1.68 0.0009 -2.50 72* As-Ca; I* 3.18 0.05 7.5* (0.09) (3.18) (0.0032) As-Ca% 1* 3.60 0.011 (0.15) (3.60) (0.004) 0.648 12.4 As-0 4* 1.69 0.0013 -3.52 56* As-Ca% 0.87 3.55 0.0097 0.4* As-Ca% 1.14 3.72 0.0073 As-Caj 1* 4.09 0.0086 (0.61) (4.05) (0.005) As-Al (also fits) 0.93 4.22 0.010 1.44 12.4 As-0 4* 1.69 0.0012 -4.3 92* As-Caz 1* 3.21 0.0054 8.8* As-Caz 0.71 3.66 0.003 As-Cag 1.12 5.09 0.0034

Table 6.5. Continued on the next page. Table 6.5. Continued,

Solid Phase AsO^ pH Backscattering N R(A) AEo(eV) F Concentration atom (A*) mol kg’* 2.16 12.4 As-0 4*(4.63) 1.69 0.0014 -3.0 81* As-Caz 1* 3.21 0.0076 1.5* (0.36) (3.21) (0.0005) As-Caz 1.2 3.70 0.0085 As-Caz 0.97 5.09 0.002

* Coordination number was fixed. * Goodness-of-fit for complete FT. * Goodness-of-fit for all shells combined excepting the first As-0 shell. Numbers in parenthesis represent the values obtained by floating everything, but diis shell is fixed at a higher coordination number to get reasonable drat is compatible with the models. Subscripts 1,2, and 3 represent Ca/Al coordination in different shells.

00 11900 11950 12000 12050 12100 12150 11900 11950 12000 12050 12100 12150 Energy (eV) Energy (eV)

Figure 6.4. Raw EXAFS of AsO^ adsoibed (1) and coprecipitated (2) ettringite. Solid phase AsO^ concentrations are: a: 3.24 XIO'^, b: 57.5 X10 \c : 0.288, d: 0.648, and e: 1.44 mol kg'*. Arrow marks show different frequency components in EXAFS that correspond to backscatterers at different distances. I

»

R + A (A)

Figure 6.5 Fourier Transform of coprecipitated arsenate ettringite. § 181 the type of backscatterers and their bond lengths from the absorber, are helpful in the evaluation of solid phase AsO# spéciation.

ASO 4 Adsorption in Ettringite

Macroscopic sorption experiments, SEM and XRD showed that the probable sorption mechanisms are surface interactions at low coverages, and channel substitution and precipitation of a Ca arsenate phase with further increases in solid phase A sOa concentration (ASO4 >0.11 mol kg'*). EXAFS results indicated several changes in ettringite ASO 4 complexation and these are discussed by shell below (Fig 6 .6).

As-0

Arsenic is expected to be in tetrahedral coordination in ettringite since ASO 4 was used in the reacting solutions. The As-0 bond distances would be similar for all four O unless they bind to H or strongly complex with ettringite surfaces. In the latter case, different bond distances may not be resolved by EXAFS if they are close (< 0.1 Â), and instead variation in (f may be useful to predict bond length variation. WAYCHUNAS et al. (1993) have observed uniform oP for AsÛ 4 sorbed to different Fe oxides, and on this basis they have suggested uniform As-0 bonds in the ASO 4 polyhedron.

In all adsorption samples, As-0 coordination was fixed at 4 (when floated, the fits produced 4

± 0.3), and the corresponding bond distances from fitting were 1.685 (± 0.005) Â (Table 6.4). The samples exhibit an increase in oP from 0 .0 0 0 2 to 0.0 0 2 1 with increases in sorbate concentration. The smallest ( f values were obtained for the dilute samples studied at 10 K. The observed increase in (fis not related to equilibrium pH; however, a positive correlation with 182

6

5

I 4

3 I

2

1

0 0 1 2 3 56 74 R + A(A)

Figure 6.6. RSFs of adsorbed AsO^-ettringite EXAFS spectra. Arsenate concentrations are: a: 0.14, b: 0.28, c: 0.65, and d: 1.44 mol kg '. Arrows labled 1 and 2 indicate peaks corresponding to back scatterers at different distances in the second sheU. 183 increased heterogeneity in the AsOa complexes can be established, as shown by appearance of more

Ca at different distances from the absorber. Vibrational spectroscopic results of these samples

gave evidence for protonation of sorbed MO 4 , and increased complexity in ASO 4 bonding environment (Chapter V) with increases in solid phase ASO 4 concentration, values of Ca arsenate model compound fits were 0.0014 - 0.(X)39 (Table 6.2, & 6.3), and these relatively high values may represent disorder and variation in As-0 bonds as they coordinate differently with

Ca.

As-AUCa

Major changes in ASO 4 coordination with A1 or Ca polyhedra were observed as solid phase

ASO4 concentrations increased. In general these changes were manifested by differences in the type of solid phase complexation and the complexing cation. These changes are discussed according to distance away from tlie absorber, i.e., 3.2 (first shell), 3.5-3.75 (second shell), and > 4,0 Â (third shell) (Table 6.4).

The presence of A1 and Ca in the first shell corresponds to comer and edge sharing complexes respectively. At very low sorbate concentrations, the cation at 3.2 Â was identified as Al. Comer- sharing ASO 4 polyhedra with a small tilt in As-O-Al bonds (~ 35-55° from linear arrangement) can produce this bond length. The fits completely failed when a small component of Ca was added at this distance (both in the presence and absence of Al), suggesting that there are no major bidentate

Ca complexes. With increases in sorbate concentration, the number of Al atoms decreased and no

Al was observed in the fits when solid phase ASO 4 concentrations reached ~ 0.6 mol kg '. In the latter case, the second neighbor was observed to be Ca with coordination numbers close to 1.0

(Table 6.4). Debye-Waller parameters were similar to the values obtained for model compounds, indicating that the obtained coordination numbers were realistic. No distinct peaks were observed 184 around 2.3 Â in the RSF that corresponded to Al bidentate mononuclear A sOa complexes.

However, such complexation with Al may not be ruled out. This is because the side lobes of the largest peak at 1.35 A (corresponding to backscattering from O) produce several small peaks at shorter distances in R-space, and thus, it is difficult to separate edge sharing Al backscattering contributions from these.

Studies of Ca arsenate model compounds indicate that peaks in the range of 3.6-3.75 Â

(corrected for phase) in the RSFs represent comer sharing (bidentate binuclear or monodentate)

As0 4 complexes with Ca (second shell). These distances are too long for an Al arsenate complex.

ASO4 ettringite samples showed very high Nj and smaller Rj (3.57 Â) at very low sorbate concentrations. With increases in concentration, this component decreased and a new component appeared at 3.7 Â. These complexes at low R values may represent bridging (B-type sites) or monodentate binuclear complexes (C-type sites); however, the Ca atoms around 3.7 Â may represent either bridging (B-type site) or monodentate (A, C type) ASO 4 complexes. However, it is difficult to distinguish among these with available data. At very high ASO 4 concentrations, the component at 3.6 Â showed very high Nj (~ 3.0) and (f (0.0193 Â^). This may be due to overlapping peaks at 3.6 and 3.7 Â and the inability of fitting to resolve these two peaks.

Ca atoms at approximately 4.0 Â (third shell) may correspond to comer sharing with ASO 4

(close to a linear As-O-Ca arrangement), but very long As-Ca bond lengths (> 4.2 Â) represent either adjacent Ca polyhedra not directly participating in bonding or comer sharing with the longest

Ca-OHz moieties in linear arrangement (> 2.60 Â). It appears that As-Ca bond lengths of 4.0 Â are highly unlikely for comer sharing poljhedra, since they are not observed in any of the Ca arsenate compounds examined in this stu^. However, long Ca-OH/OHz bond lengths in ettringite

(~ 0.5 Â longer than the average Ca-OH/OHz in models) may produce such complexes with As-Ca bond distances ~ 4.2 Â. Due to weak backscattering from Ca and As, it is difficult to distinguish 185 these atoms > 4.5 Â from the absorber (As). The presence or absence of Ca, Al and As at this distance was identified on the basis of the N j, (f and F values. Unrealistic N j, very higli or negative ( f values and poor fits (high F) for a particular backscatterer indicate its probable absence. Arsenate adsorbed ettringite showed Ca atoms at 4.2 (± 0.2) Â, and their presence decreased with increases in sorbate concentration. This decrease is also concomitant with disappearance of Al at 3.2 Â. This component reappeared when adsorbate concentrations were > 2 mol kg'*, but was probably due to a different coordination environment in a new solid (XRD indicated that ettringite completely disappeared at this sorbate concentration). In some of these samples (AsOa >1.0 mol kg'*) Al was also observed at 4.2 (± 0.2) Â but with very small Nj (<

0 .2 ), and thus is not considered as a major component.

As-As

Arsenic as a next neighbor was not identified until solid phase ASO 4 concentrations reached

0.6 mol kg'*. At this solid phase concentration, ettringite reactive surface sites were saturated

(As0 4 ~ 0.1 mol kg'*) and XRD indicated an amoiphous phase in the sample. As-As distances in these samples were usually around 4.5 Â, except at very high ASO 4 loadings where the As-As distances were 4.95 Â; and Nj also increased with increases in concentration. Debye-Waller parameters were 0.007 when Nj and R were floated (Table 6.4). These values are smaller than those obtained in the model compound fits (>0.01) for As backscatterers. If one considers that the disorder in Ca arsenates and ASO 4 ettringites are similar, then the obtained coordination numbers in the above fits for this shell of arsenated ettringites almost doubled There was little to no difference in based on data collected at 298 and 10 K. Evidently the ASO 4 ettringite samples 186 had a high static disorder. Thus assigning a specific coordination for As at these distances may be erroneous, and must be approached with caution.

As04in CoprecipUated Ettringite

During coprecipitation, the probability of AsO# substituting inside the channels is large

(Chapter III). Thus the channel ‘G’ and ‘B’ type sites (with restrictions on ‘E’-type as discussed before) are more available for ASO4 interactions (Fig. 1.2, Fig. 6.2), which may result in OS complexes inside the channels and comer sharing with Ca polyhedra respectively. In addition, ettringite crystal edges (external surfaces) are also available for reaction with ASO 4 in the same way as in the adsorption experiments. EXAFS spectra from the coprecipitated samples showed completely different types of surface complexes (Fig. 6.7) at low sorbate concentrations than were observed in the adsorption samples. However, the differences in ASO 4 coordination between coprecipitation and adsorption treatments disappeared at solid phase ASO 4 concentrations > 1.4 mol kg'\

As-0

First shell As coordination with O and their bond distances are 4.0 (± 0.2) and 1.685 (±

0.005) Â respectively. The ( f values were very small in all cases and close to 0.001 (± 0.0(X)4)

When compared to the concentrated ASO 4 adsorbed samples, these values are relatively small, which may indicate that the disorder in As-0 is small and probably all As-O distances are close to the average value obtained. 187

8

6 I

4 •I

2

0 0 1 2 3 4 5 6 7 R + A(A)

Figure 6.7. RSFs of coprecipitated AsO^-ettringite EXAFS. Solid phase AsO^ concentrations are a: 3.24X10'^, b: 7.2X10'^, c: 0.072 and d:1.44 mol kg'\ Arrows labled 1 and 2 indicate peaks corresponding to backscatterers at difierent distances in the second shell. 188 As-AUCa

Next neighbor Ca or Al coordination in coprecipitated samples is very different from that of adsorbed As 0 4 -ettringite (Fig 6.6 & 6.7). In dilute coprecipitated samples, essentially no neighbors were observed in the range of 2.5 - 3.7 Â (for the data collected at 298 and 10 K). At very low solid phase ASO 4 concentrations (5.75 mmol kg *), only Ca are the closest neighbors at

4.1 Â, probably due to ASO4 comer sharing with Ca (linear arrangemoit). Arsenate location inside ettringite channels is difficult to establish since comer sharing complexes are formed both at external surfaces and in channels. However, these interactions may happen inside the channels, because of their absence in adsorption samples at low sorbate concentrations. When the sorbate concentrations reached 0.058 mol kg *, several Ca atoms were observed above 3.6 Â (3.61, 3.90,

4.09 Â) which indicate comer sharing ASO 4 complexes in different As-0-Ca arrangement (with different angles at O) than discussed above. These distances correspond to ASO 4 interactions with

B-type sites in bidentate binuclear fashion with two Ca atoms (at 3.61, and 3.90 Â). The other Ca at 4.09 Â was likely linked in monodentate coordination with adjacent As polyhedra. This environment persisted until the solid phase ASO 4 concentrations reached 0.65 mol kg'*.

Interestingly, samples of coprecipitation experiments conducted at high SO4 concentrations

(ASO4/SO4 ratio close to 0.2), and relatively low pH (11.8) produced no significant backscattering from next nearest neighbors. However, two low amplitude peaks at 2.8 and 3.2 Â (in R space) were observed in the RSF, which were difficult to fit with reasonable Nj and

Â. This may indicate possible OS ASO4, with a small quantity as IS complexes with Ca (both comer and edge sharing). At high ASO4/SO4 ratios (> 0.28), Ca were observed at 3.55, 3.72 and 189 4.10 Â (Table 6.5), suggesting a similar coordination environment as described above. These may indicate different types of comer sharing ASO 4 complexes with Ca polyhedra (IS).

With further increases in solid phase ASO 4 at pH 11.8 and 12.4, strong peaks appeared at 3.2

Â, and the neighbors were identified as Ca. This indicates that Ca is in bidentate mononuclear configuration with ASO 4 polyhedra and the probable sites are either ‘E’ and/or ‘C’ type. However,

‘C’ type sites may be preferred because the OH-OH distance (of Ca polyhedra) is close to the

As0 4 edge size. Preference of C-type sites by ASO 4 poisons ettringite growth, since these sites are only exposed at the column broken edges. This may also explain the ettringite grain size decrease

(from 10 pm to < 1 pm) with increases in ASO 4 in coprecipitated samples. In addition, peaks corresponding to other Ca neighbors appeared above 3.5 Â, supporting the presence of ASO 4 at C- type site (Fig. 1.2, Fig. 6.2).

AslAs

In all dilute samples, second neighbor As was not observed in the range of 3.7 -4.5 Â.

However, at very high solid phase ASO 4 concentrations. As was identified as one of the neighbors around 4.2 Â with relatively small Nj.

Desorption ofAs04from Arsenated Ettringite

Desorption experiments were conducted on adsorption samples that exhibited comer sharing

Al and Ca arsenate complexes (dominantly on the surface); and on coprecipitated samples with OS

ASO4 complexes (Chapter III). The macroscopic results in^cated that there was no apparent ASO 4 release when ASO4 adsorbed ettringites were exposed to high ionic strength SO4 solutions. The collected EXAFS of the adsorbed samples before and after desorption studies are very similar (Fig.

6 .8 ) indicating that the sotted ASO 4 is present as an IS complex. In contrast, there was some ASO 4 190

7

6

5

î 4 I 3 2

1

0 0 1 2 3 4 5 6 7 R + A(A)

Figure 6 .8 . Desoiption of AsO^ from adsorbed AsO^-ettringite. Solid phase AsO^ concentration is 8.3 mmol kg ’, a; I = 0.1 mol^L’’, SO 4 = 2.08 mM; b: I = 1.0 mol,L'\ S04 = 2.08 mM; and c: I = 0.1 mol^L’, SO 4 = 20.83 mM. 191 release from coprecipitated samples with increases in the ionic strength of the equilibrium

solutions. EXAFS analysis of these samples showed no changes in solid phase spéciation (Fig.

6.9) with changes in ionic strength of the contacting solutions. This indicates that OS complexes

still persist in coprecipitated ASO 4 ettringite even after treating with high ionic strength solutions.

Thus, these studies suggest that OS ASO 4 is present inside the channels ("O' type sites), and are not

completely replaceable due to slow ion diffusion into the channels.

ASO4 Interactions in Ettringite

The results of EXAFS analyses clearly indicate that ASO 4 forms IS complexes with ettringite during adsorption and coprecipitation (Fig. 6.6 & 6.7), except at highly sulfated and relatively low pH coprecipitated samples. As discussed previously these complexes are different in adsorption and coprecipitation samples.

In dilute adsorbed As 0 4 -ettringites (ASO 4 = 3.24 mmol kg '), nearest cations are Al with Nj and R j of 1.28 and 3.21 Â respectively. This indicates that ASO 4 formed comer sharing complexes with Al polyhedra. Structural analysis of ettringite columns suggest that these interactions are possible only on the broken edges or external surfaces (‘A’-type sites). Since Al coordination is greater than one, it appears that it may fonn bidentate binuclear complexes. However, considering the ettringite column structure, it is possible to rule out this kind of complex. The Ca at 4.42 Â corresponds to the second neighbor nearer to this Al-interacted ASO 4. At the same solid phase concentrations, ASO4 also forms comer sharing complexes with Ca polyhedra (N j = 1.4, R = 3.57).

On the basis of these bond lengths, the possible reactive sites at this concentration are ‘C’ (without having bidentate mononuclear for one of the Ca, i.e. monodentate binuclear complexes), ‘B’ or ‘F’

-type. However, the probable site may be tentatively identified as ‘F’, since Al at 3.2 Â has Nj > 1, and no Al neighbors are around 4.5 Â (so ‘B’ can be ruled out). This situation persists when 8

I_ 6 1 I

r2

-10 0 4 6 8 10 12 0 1 2 3 4 6 75 t(A") R + A(A)

Figure 6.9. Desoiption of AsQ, from coprecipitated AsO^-ettringite. 1) Â:-weighted EXAFS, and 2). their RSFs. Solid phase AsO^ = 0.18 mol kg '. a; unreacted sample; b; 1 = 0.1 mol^L"', SO^=2.6 mM; c: 1= 1.0 mol^L ', SO^ = 2.6 mM; d; I = 0.1 mol^L"', SO. = 20.8 mM; and e: I = 1.0 mol L ', SO. = 20.8 mM. 193 As0 4 concentrations are increased to 6.48 mmol kg'*. However, at these solid phase ASO 4 concentrations, coordination to Ca at about 3.57 Â increased to almost 2.0; while Nj for Al at about 3.21 Â was < 1. EXAFS gives information on the average bonding environment, and thus, these observed changes in coordination may explain the increased binding of ASO 4 to Ca polyhedra.

When solid phase ASO 4 concentrations reached 0.3 mol kg'*, ASO 4 edge sharing with Ca polyhedra was observed in addition to the previously discussed coordination. Considering these results, it can be hypothesized that ASO 4 interacts at C, B and B type sites, which also agree with the presence of other neighboring Ca atoms. Macroscopic sorption experiments showed that aU the surface sites were saturated (maximum ~ 0.1 mol kg'*) at these solid phase concentrations. Thus, formation of an edge sharing complex with Ca may indicate the beginning of ettringite breakdown and precipitation of a Ca arsenate phase.

With further increases in solid-phase ASO 4 concentration (> 0.3 mol kg'*), bidentate mononuclear complexes with Ca persisted, and bidentate binuclear complexes with Ca increased

(N j > 2). In addition, next neighbor As atoms appeared at > 4.5 Â, indicating Ca arsenate rather than substitution in ettringite. This is because, at these solid phase ASO 4 concentrations, ASO4 in the solid has exceeded the available reactive site density, and XRD indicated amorphous phases in the samples. In addition, adsorption experiments showed that S O 4 was released at a different stoichiometry than sorbed ASO 4, and aqueous Al also increased, indicating the instability of ettringite at high ASO 4 concentrations. Similar behavior was also observed in high pH adsorption samples (pH = 11.8) (Table 6.4). It is difficult to predict solid phase ASO 4 spéciation for these samples since the structure or fraction of the newly precipitated pliase is not known, and the

EXAFS spectra is an average of ASO 4 coordination in the precipitate and in ettringite. The 194 destruction of ettringite may be attributed to local high AsOa concentrations close to the surfaces, which resulted in increased bidentate complex formation (Ca at 3.2 Â).

In contrast to As adsorption, coprecipitation allowed direct formation of comer sharing AsO^ complexes with Ca and no interactions with Al were observed, which clearly indicates ASO 4 preference for Ca. The As-Ca bond distance at low solid phase ASO 4 concentrations is 4.07 Â, which suggests almost linear As-O-Ca bonds. These complexes have to be inside the channels or at the channel edges, since no other neighbors were identified after Ca and such complexes were not identified in adsorption samples. With increases in ASO 4 concentrations to 0.058 mol kg ', backscattering from Ca at 3.6 (N j ~ 1) and 4.0 (N j ~ 2) Â was identified, which indicates bidaitate binuclear configuration. This may be possible for ASO 4 complexation at ‘B’ or ‘C’ (monodentate binuclear with a Ca) type sites. With further uKreases in solid phase ASO 4 concentration, no changes were observed in ASO 4 complexes, except for the appearance of another Ca at 3.7 Â, which may also participate either in comer-sharing monodentate or bidentate binuclear configurations. When essentially no SO 4 was present during the coprecipitation studies, As-Ca at

3.2 Â (ASO4 edge sharing with Ca) and As-As interactions above > 4.0 Â appeared, indicating the formation of a new precipitate. However, when pH was maintained during coprecipitation at pH

11.8 in the presence of high aqueous SO 4 concentrations, no neighbors were observed at low solid phase ASO 4 concentrations. This suggests that ASO 4 is present as an OS complex in addition to minor concentrations of edge and comer sharing complexes with Ca polyhedra, when ASO 4/SO4 ~

0.2

Summary

On the basis of observed EXAFS, ASO 4 interactions with ettringite surfaces can be summarized as follows: 195 1) AsO# forms only comer sMnng complexes with Al/ Ca at low solid phase concentrations irrespective of the sorption mechanism (adsorption or coprecipitation). Appearance of only Ca as second neighbors in coprecipitation samples may suggest that AsOa has stronger affinity for Ca and channel sites than for Al in this pH range.

2) Adsorption resulted in high ASO 4 concentrations on ettringite surfaces. Once the surface sites were saturated, the strong affinity of ASO 4 for Ca destabilized ettringite and precipitated Ca arsenates (as is observed from As-As and As-Ca correlations).

3) Easy accessibility of ASO4 to the channels during coprecipitation resulted in a variety of complexes, and stabilized the ettringite structure up to very high ASO 4 loadings.

Cr - EXAFS OF CrOi ETTRINGITE

XAS studies of Cr0 4 interactions with mineral surfaces showed that this oxyanion forms OS complexes with magnetite (PETERSON et al., 1994); and IS complexes in the case of redox sensitive transformations on Mn oxides (CHARLET and MANCEAU, 1992). Macroscopic experimental data is also supportive of both types of complexes (EARY and RAJ, 1987;

JOHNSON and XYLA, 1991). In the case of C 1O 4 in ettringite, ionic strength-independent desorption is in support of IS complexatiwi (Chapter IV). Vibrational spectroscopic studies indicated similar results but the nature of these IS complexes is not clearly understood. Although

XAS studies provided more information, a thorough analysis of this system, as has been presented for ASO4, is unfortunately not possible because of a lack of structurally well characterized model compounds, and strong absorption of X-ray radiation by air and the ettringite matrix in the energy range of study. 196 Adsorption experiments resulted in very little C 1O 4 uptake as compared to coprecipitation

(Chapter IV). Because of the above mentioned reasons, there was not a strong enough signal at the

absorption edge to analyze the Cr-EXAFS of adsorption samples, and dilute coprecipitation

samples. Thus, the results presented here are only for concentrated coprecipitated ettringite

samples.

The size of Cr 0 4 tetrahedron (Cr-0 ~ 1.64 Â) is very similar to that of ASO 4 (As-O ~ 1.68

Â), and thus the bond distances between Cr and Ca, and Cr and Al are also expected to be very similar to those of As. A typical RSF of C1O 4 EXAFS (Fig. 6.10) showed essentially three peaks at 1.28, 2.8, and 3.3 Â (not corrected for phase). The first peak (strongest) at 1.28 Â may corresponds to back scattering from Cr 0 4 O atoms. The other weaker peaks at 2.8 and 3.3 Â represent comer sharing of Cr 0 4 with Al or another CrÛ 4 polyhedra (dimer, CrzO?), or edge sharing with Ca polyhedra; and Cr 0 4 comer sharing complexes with Ca polyhedra respectively.

Theoretical phase and amplitude functions for Cr-0, Cr-Ca, and Cr-Cr were extracted using a

CaCr0 4 .3H2 0 cluster (BARS et al., 1977). Due to the absence of crystal refinement data on Al chromâtes, Fe chromate was tried by substituting Al for Fe. Due to presence of probable error in atom coordinates, Fe chromate could not be used. This limited the current discussion to backscattering from O, Ca and Cr atoms only.

Cr^O

The coordination of O around Cr was close to 4 (error less than 10 %) with little variation in

Cr-0 bond distances (< 0.01 Â) (Table 6.6). For the model compound Na 2Cr0 4 , the Cr-0 bond length from EXAFS was 1.656 Â which is longer than the bond lengths measured from XRD

(1.649 A). In the case of C1O 4 in ettringite, the average Cr-0 bond lengths were 1.645 A, close to the value measured for Na2Cr0 4 . The goodness-of-fits were poor (F > 200) when this shell was Table 6 .6 . EXAFS analysis of Cr0 4 - coprecipitated ettringite.

Solid Phase CrO^ Concentration pH Backscattering N R(A) o* AEo(eV) F mol kg'* atom NazCrO^ --- Cr-0 4* 1.656 0.0017 0.20 207* 1.649* 0.043 12.5 Cr-0 4.10 1.636 0.0013 8.55 777* Cr-Ca 1.06 3.24 0.0053 18 (Cr-Cr) (1.11) (3.18) 0.0067 26 0.60 12.5 Cr-O 3.73 1.648 0.0 8.1 327* Cr-Ca 1* 3.26 0.0097 0.64 3.24 0.0058 33 (Cr-Cr) (0.98) (3.19) (0.009) 28 1.55 12.5 Cr-0 4* 1.652 0.00012 4.0 203* Cr-Ca 1.677 3.27 -0.0161 10 (Cr-Cr) (0.62) (3.18) (0.0018) 24 * Coordination number fixed either because floated value is close to the fixed number, or is relatively small when compared to model compounds. * Goodness-of-fit for the complete FT. * Obtained from XRD data. Numbers in parenthesis are other good fits. 198

I

S

0.0 3.0 4.51.5 6.0 R + A(A)

Figure 6.10. Coprecipitation of C1O4 in ettringite. Solid phase CrO^ concentrations are: a; 0.4 %, b: 0.5 %, c: 7 %, d: 18%, and e: NagCrO^. 199 fitted with one type of O. However, when a second shell was added, the fit significantly improved

(F ~ 65) and the Cr-O distances of this shell differ from the other by < 0.02 A. Since the Debye-

Waller parameter was small (< 0 for one of the O) it is unlikely that two different Cr-0 bond

lengths actually existed in these samples..

Cr-AllCalCr

Identification of the backscatterer that produced a peak around 2.8 Â (not corrected for phase) was difficult because of closeness of Ca and Cr atomic numbers. However, linear phase analysis of this peak (EXAFSPAK) showed that the backscatterer was Ca for aU Cr 0 4 -ettringite

samples examined. On the basis of these preliminary results, the peak around 2.8 Â was Fourier filtered into k-space and then fit with both Ca and Cr atoms. The Cr-Ca bond lengths were 3.25 (±

0.02) Â and those of Cr-Cr were 3.18 (± 0.01) Â; the fits were poor for either of these atoms,

suggesting that there may have been backscattering due to second neighbor Al. It is surprising to

see longer Cr-Ca distance than As-Ca (discussed in previous section), when Cr-0 < As-O, and this can not be explained with the available data. Debye-Waller parameter values for Ca and Cr were close, except for concentrated Cr 0 4 -ettringite, which showed large negative values for Ca and very small values for Cr.

Another distinct peak around 3.2 A (after phase correction ~ 3.68 A) in the RSF (Fig. 6.10 &

6.11) probably corresponds to Ca coordinated to C 1O4 as comer-sharing polyhedra, either as bidentate binuclear or monodentate complexes. Since the validity of the ( f values in these samples could not be confirmed with any model compound, Nj values obtained in these fits remain luiceitain. 200

7

6

5 I 4

3

A 2

1

0 0 1 2 3 4 5 6 7 R + A(A)

Figure. 6.11. CrO^ desoiption from coprecipitated CrQ,-ettringite. a: unreacted ettringite. b: I = 0.1 molgL'\ SO^ = 20.83 mM; and c; I = 1.0 m ol,L\ SO^ = 2.06 mM. 201 Cr-Cr

No Cr-Cr interactions were observed below 4.0 Â other than the possible presence of Cr at

3.2 Â discussed above.

Desorption of Cr04From CrO^-ettringite

EXAFS analysis of CrOfettringite samples at different ionic strengths were not analyzed for backscatterers, because of arguable information on Nj and atom type. However, a comparison of the RSFs indicated that ionic strength and S O 4 concentration increases did not produce any major variation on overall Cr0 4 coordination (Fig. 6.11).

C1O 4 Interactions in Ettringite

Macroscopic desorption and vibrational spectroscopic results indicated that C1O 4 forms IS complexes with ettringite. Although the types of C1O 4 complexes are still unidentified from

EXAFS, these studies strongly indicated the presence of neighbors close (3.2 Â) to Cr. These backscattering atoms were tentatively identified as Ca, rather than Cr. The presence of Ca at 3.25 and 3.68 Â indicate edge and comer sharing C 1O 4 complexes with ettringite. The presence of Cr at

3.2 Â may also have been due to dimer (Cr^O?) formation. This would have resulted in the increase of one of the Cr-0 distances by > 0.1 Â and should be reflected in multiple shell fits for

Cr-0 at 1.65 Â, and their ( f values. In addition, due to the Csv symmetry of CtiOi, the peak at 2.8

 (not corrected for phase) should have more amplitude due to multiple scattering within Cr 2 0 ?.

Absence of this behavior ruled out the presence of Cr neighbors. However, the presence of Al neighbors is still uncertain and further studies are required to distinguish them. CHAPTER Vn

CONCLUSIONS

The current studies on arsenate and chromate sorption in ettringite have provided new insights into oxyanion-mineral surface interactions. The salient points of these results are presented here with recommendations for future research.

Ettringite Solubility, Weathering and Geochemistry o f the Ca-Al-S04-H20 System

1) Ettringite dissolves congruently above pH 10.7 with a p*^Ksp of 111.6 (± 0.8). The solubility product of this mineral was not affected by changes in pH or suspension density. Below this pH, ettringite dissolves irKongruently to gypsum and Al hydroxides but controls Ca^\ Al% and S 0 4 ^' activities up to pH 9.5.

2) Ettringite weathering at near neutral pH showed several interesting aspects on the geochemistry of minerals in the Ca-Al-SO^HzO system. These studies indicate that,

a. The geochemistry of Ca-Al-SO^HzO is simple and behaves as an association of Ca-

SO4-H2O and Al- SO 4-H2O above neutral pH. However, formation of Al-hydroxy sulfates has an influence over the Ca-S0 4 -H2 0 system or vice versa under acidic pH conditions.

b. Basaluminite, jurbanite or other stoichiometrically similar Al-hydroxy sulfates precipitate rapidly at acidic pHs and thus significantly influence Al activities in natural waters.

202 203 c. An identified low pH, amorphous, Ca-AI-hydroxy sulfate phase is tentatively assigned

for the observed decreases in Ca^* activity below pH 5.0. This variation in Ca^* activity could not

be explained by other known phases in the Ca-Al-S 0 4 -H2 0 system.

Arsenate and Chromate Interactions with Ettringite

1) Ettringite has higher selectivity for A SO 4 than C1O4. ASO4 sorption was affected by mode of

sorption (higher sorption during coprecipitation than adsoiption), pH (sorption decreased with

decreasing pH), and suspension concentration (sorption increased with increasing suspension

density). Sulfate concentrations do not have a major effect on ASO 4 sorption. On the other hand,

C 1O 4 sorption decreased with increases in S O 4 concentration of the reacting solutions.

2) Arsenate Sorption: Initial AsO# concentrations and the mode of sorption (coprecipitation

versus adsorption) control ASO4 spéciation in ettringite. Macroscopic and spectroscopic studies

indicated similar mechanisms for ASO 4 sorption (Table 7.1). Lack of sufficient solubility data on

Ca, and Al arsenates restricts the application of tl%rmodynamic spéciation calculations to identify the mineral phases that control ASO 4 activity in the Ca/Al-AsOA-HzO system.

a. During adsorption and at low sorbate concentrations (< 10 mmol kg '), AsOa primarily

forms comer-sharing complexes with Al and Ca polyhedra exposed at ettringite surfaces. In this concentration range, sorption is > 99 %. These surface arsenato complexes are not easily displaced by increases in aqueous SO 4 concentration or ionic strength. With increases in initial

A SO 4 concentrations, ASO 4 substitutes for channel S O 4 and also forms bidentate mononuclear complexes (> 0.11 mol kg '). With fiirther increases in concentration, ettringite breaks apart and Table 7.1. Summary o f ASO 4 spéciation in ettringite.

Adsorption o f ASO4

Initial AsOa Sorption Scanning Electron IR +Raman EXAFS Concentrations Efficiency X-Ray Diffraction Microscopy (mM) (%) 0.017 >99, can no changes ~ 2 pm ettringite not detectable corner sharing with Al, and Ca not be grains desorbed

1.44 98 ettringite ( 100) gypsum + ettringite* IS with Ca, structural H 2O comer sharing with Al decreased reflections decreased, forms H bonds with ASO 4 relatively, and Ca increased, peaks indicative of first appearance of edge sharing gypsum appeared* ASO4 complexes with Ca.

2.88 94 ettringite ( 100) gypsum + ettringite* IS, coordinates with Ca, comer-sharing with Ca, reflections decreased*, structural water forms edge-sharing with Ca, gypsum peaks present hydrogoi bonds with ASO 4 As-As correlations

5.76 92 ettringite completely gypsum with IS, Samples with low pH comer-sharing with Ca, disappeared * corroded surfaces indicate the presence of edge-sharing with Ca, protonated ASO 4 As-As correlations

15.2 73 ettringite and gypsum fine grained, size < 1 IS, Samples with low pH comer-sharing with Ca, absent; presence of pm indicate the presence of edge-sharing with Ca, poorly crystalline protonated ASO 4 As-As correlations material " When pH was maintained at 11. 8 , gypsum was absent in the precipitates. * Ettringite was present when pH was maintained at 11.8, however, the base line of XRD patterns was not smooth.

Table 7.1 Continued on the next page. g Table 7.1. Continued.

Coprecvitation of ASO4

Initial AsOa Sorption Scanning Electron IR +Raman EXAFS Concentrations Efficiency X-Ray Diffraction Microscopy (mM) (%) 0.007 >99 no changes homogeneous, long not detectable corner sharing with Ca, linear ettringite grains As-O-Ca arrangement

0.144 >99 no changes no changes IS with Ca corner sharing with Ca, linear As-O-Ca arrangement + bridging complexes

0.216 >99, no changes no changes ASO4 in high symmetry, either not much backscattering from pH =12.0, partly Csv or Td, OS ASO4 near neighbors, OS or highSÜ4 exchange­ disordered ASO 4 able

0.65 >99 no changes ettringite grain size IS with Ca, As 0 4 not corner sharing with (]a, linear decreased, protonated As-O-Ca arrangement + heterogeneous grain bridging complexes size-distribution

1.44 >99 indicates presence of heterogeneous grain IS with Ca, As 0 4 not comer-sharing with Ca, poorly crystalline sizes protonated. edge-sharing with Ca material + ettringite

IS; inner-spheie complexes OS: Outer-spheie complexes

a 206 precipitates some form of unidentified Ca arsenate. These studies indicate that precipitation of this new phase started wlien the sorbate concentrations were close to 0.11 mol kg'\

b. Coprecipitation also resulted in the formation of comer-sharing complexes, but dominantly with the Ca polyhedra and no interactions with Al were observed at low sorbate concentrations (< 60 mmol kg"‘). With further increases in sorbate concoitrations, ASO4 forms bidentate mononuclear complexes with Ca polyhedra. In contrast to the adsorption studies, ASO 4 coprecipitation in ettringite did not affect its structure up to very high ASO 4 loadings (~ 1.5 mol kg '). This is due to the better accessibility of channel sites during coprecipitation. High S O 4 concentrations in ettringite allowed the formation of O S A SO 4 inside ettringite channels. All of these O S complexes, however, are not replaceable by any exchangeable ions because of their sluggish diffusion into the channels.

c. Ettringite structure was more stable during coprecipitation and was not affected by high

ASO4 loadings (ASO 4 > the surface coverages, 0.11 mol kg ') as compared to adsorption samples.

This may be due to the low As activity in equilibrium with coprecipitated ASO 4 ettringite (or a solid solution of A SO 4 - S O 4 ettringite) than any other Ca arsenate phases. During adsorption,

ASO4 channel substitution and solid solution formation is restricted by solid state diffusion, which finally resulted in only surface coverages and breakdown of ettringite structure.

d. Combined application of EXAFS and vibrational spectroscopic tools helped in complete evaluation of ASO4 macroscopic behavior in ettringite, which otherwise would not have been possible. Stretching vibrations of As-O-X (X = metal, H*, H 2O) were helpful to identify the

As coordination environment in ettringite, especially to distinguish ASO 4 protonation and H bonding in solids. 207 3) Chromate Sorption: Chromate forms IS complexes with ettringite in the examined concentration range of 0.04 - 1.55 mol kg *. Sorbed Cr 0 4 does not polymerize on ettringite surfaces but forms both edge- and comer-sharing complexes with Ca (and probably Al in case of comer-sharing complexes). Although surface complexation is the dominant mechanism for Cr 0 4 uptake at low sorbate concentrations (< 0 .0 2 mol kg'*) during adsoiption, channel substitution accompanies this process at higher concentrations (> 0.03 mol kg'*). Sorbed C 1O4 may not be exchanged by increases in SO 4 and also is independent of ionic strength, further indicating the formation of inner-sphere C 1O4 complexes.

Recommendations for Future Research

Based on the results Aom these experiments, the following research is recommended to improve understanding of geochemistry of the Ca-Al-S 0 4 -H2 0 systan, and oxyanion interactions at mineral-water interfaces:

1) Identification of Ca-Al-hydroxy sulfates in acidic soils, and their structural and solubility characterization will help to understand the Ca and Al activities in acidic waters. In addition, structural studies (e.g. transmission electrwi microscopy) of identified Al-hydroxy sulfates are necessary to understand the mineralogy and geochemistry of basic Al-sulfates.

2) Synthesis and solubility studies of different Ca and Al arsenates and chromâtes will help to evaluate the solid phases that control ASO 4 and C1O4 activities in waste materials. 208 3) The stabilities of other minerals of the ettringite group should be evaluated to help identify the most insoluble ettringite phase which miÿit then be used to treat oxyanion contaminated wastes at neutral pH.

4) Similar studies should be conducted for sorption of other oxyamons by ettringite to evaluate sorption efficiency and their stabilities after sorption.

5) A detailed spectroscopic investigation of oxyanion interactions in different solid phases, and theoretical calculations on metal-oxyanion interactions from first principles will help us to understand the oxyanion interactions at mineral-water interface. Vibrational spectroscopic techniques such as Sum Frequency Generation (SPG) may aid in probing water in oxyanion solvation shells, which can further supplement the information obtained from above studies.

Environmental Applications

1) The high stability and sorption capacity of ettringite can be used to treat oxyanion contaminated alkaline waste materials. Very low activities of As in equilibrium with ettringite (< 50 p.g mL'‘) and its rapid sorption kinetics makes it a good engineering material in As-waste treatment processes.

2) The observed relationship between the symmetric stretching ftequency of an adsorbed oxyanion and the type of complexing surface-ion may aid solid ;Aase oxyanion spéciation. APPENDIX A

SOLUBILITY PRODUCTS OF MINERALS

Mineral logKsp Source* Alunogen -7.0 1 Basaluminite (Crystalline) -117.7 1 Basaluminite (Amorphous) -116.0 1 Calcite -8.48 2 Ettringite - 111.6 3 Gibbsite -33.9 1 Gypsum -4.58 2 Johnbaumite -39.6 3 Poitlandite 2 2 .6 4 Ca3(As0 4 )2.6H2 0 -22.3 4 AS2O5 -6.69 4

1 NORDSTROM, 1982. 2 DREVER, 1988. 3 This Study. 4 MINTEQA2 Database. * Free energies of constituent ions of these minerals were obtained from FAURE (1989).

209 APPENDIX B

ARSENATE INTERACTIONS WITH CaO AND GIBBSITE

INTRODUCTION

Arsenate reactions with ettringite result in the formation of surface complexes, solid solutions and/or Ca or A1 arsenate precipitates. The currently available thermodynamic data is limited for

Ca or A1 metal arsenates, which is crucial in differentiating different mechanisms of AsO# sorption in ettringite. In addition, no spectroscopic data is available for AsOa interactions with Ca or A1 mineral phases or their precipitates. If precipitation of Ca and A1 arsenates is the dominant mechanism for the reduction of observed aqueous ASO 4 concentrations in equilibrium with ettringite, separate experiments with CaO and gibbsite should simulate this behavior. This chapter supplements this information to the dissertation.

MATERIALS AND METHODS

Adsorption experiments were conducted for CaO and gibbsite by reacting separately 0.1 g of reagent grade CaO, or natural gibbsite, with ASO 4 (0-3.6 mM). Natural gibbsite was collected from an oxisol of Parana, Brazil, well characterized for its PZC, etc. by PEIXOTO (1994). The suspension density was maintained at 4.0 g L"' in all of these experiments. The samples were allowed to react for 24 h, after which they were centrifuged and the supernatant pH was recorded.

210 211 These samples were filtered with 0.2 pm polycabonate filter and the solutions were analyzed with

Perkin Elmer, Optima 3000 Inductively Coupled Plasma spectrometer. The solid reaction products

were later studied with vibrational spectroscopy (FTIR and Raman), Extended X-ray Absorption

Fine Structure Spectroscopy (EXAFS), Scanning Electron Microscopy (SEM) and X-Ray

Diffraction (XRD). Data collection and analysis for these spectroscopic techniques are discussed

in Chapters V and VI; and the details of analytical instruments and reagents are described in

Chapters II and in.

In a separate experiment, CaO was reacted with aqueous ASO 4 (71.4 mM) in an attempt to

precipitate a Ca arsenate mineral phase. The reacted solids were characterized for their crystal

structure (XRD) and arsenate coordination (FTIR and EXAFS). Solubility studies were conducted

on this mineral by reacting it with deionized water in the presence and absence of an “inert”

electrolyte (perchlorate). The experimental procedures are similar to those for ettringite (Chapter

II).

RESULTS AND DISCUSSION

ASO4 Interactions with CaO

Lime readily formed poitlandite when reacted with water, which may indicate that ASO 4

actually interacted with poitlandite during ASO 4 sorption. For the remainder of this section, poitlandite is used to refer to the CaO phase.

A5 O4 Adsoivtion

Arsenate interactions with poitlandite (ASO 4 < 3.6 mM) resulted in a decrease in aqueous Ca

concentrations (Fig. B.l). In all samples examined, ASO 4 was below detection (< 1 pM) and the 212

a

I I 0.0064 A A

0.004 - A : A 0.002 18.0 0.15 - 15.0 12.0 □ u 9.0?- □ 0.4% I 0.3 a 0.05

(2* 0.2 0.1

12.8

S. oo 12.6(V

l i 111111111 I 11,1.1 I I 0.0 1.5 3.0 70.0 71.0 72.0 0 1 2 3 4

Initial AsO^ Concentration (mM ) Initial AsO^ concentrations (mM )

Figure B.I. Soiption of AsO^ on portlandite (a), and gibbsite (b). 213 equilibrium pH remained constant (~ 12.6). XRD of the reacted solids showed that the dominant

solid phase was portlandite with traces of calcite. With increases in sorbate concentration, calcite

peaks decreased in intensity and disappeared completely (> 0.2 mol kg'*). No new minerals were

observed by XRD. The peak widths corresponding to portlandite increased with increases in

sorbed AsO» (Fig. B.2). However, at very high initial AsOa concentrations (~ 70 mM), XRD of

the solid precipitate was different from portlandite. These new solids were identified as johnbaumite after comparing with the model spectra in the ICDD powder files. SEM showed that portlandite was fine grained (< 1 pm), and upon reaction with ASO 4 solutions, portlandite crystals did not exhibit any recognizable changes in morphology. Saturation indices could not be calculated because ASO4 concentrations were close to detection limits of the instruments used in this study.

Vibrational Spectroscopic Studies ofAsOt in Portlandite

Raman and IR studies of As0 4 -reacted CaO showed distinct splitting in the peaks corresponding to As-0 V 3 vibrations (Fig. B.3). Since these peaks were overlapped by sharp peaks of CO3 at 871 and 859 cm'*, accurate As-0 peak positions were obtained from peak-deconvolution procedures (Chapter V). The peaks related to As-0 vibrations are at 885, 879, 837, 812 (broad) cm'* for samples with low sorbate concentratiwis. The positions of these peaks remained stable with further increases in ASO 4 concentration; however, new shoulders developed at 905 (± 5.0) cm * and 795 (± 10.0) cm *. These results indicate that Vi vibrations are close to 800 cm *, which may suggest that ASO 4 is linked to Ca and is not protonated in this sample (Chapter V). Hydroxyl stretching frequencies of Ca-OH (3642 cm'*) in the reacted solids decreased with increases in solid phase ASO 4 concentrations. The Ca-0 vibrational modes (659, and 527 cm'*) however, did not 214

•I I i a

calcite

10 20 30 40 50 Degree 26

Figure B.2. XRD of AsO^ reacted CaO. Hie reacted concentrations decrease in the order d > c> b > a. a: portlandite and d: johnbaumite. 215

00 0.8

00

00

0.6

ON 00 en

: 0.2 m

“ s 0.0 1000 950 900 850 800 750 700 Wavenumber (cm^)

Figure B J. FTIR spectra of AsO^ reacted CaO. AsO^ concentrations decrease in the order d > c> b. a: unreacted CaO, and e; johnbaumite. 216 show major variation. These results, combined with XRD, indicate that ASO 4 fonns IS complexes

with Ca by replacing the surface OH of portlandite.

EXAFS ofAsOd in Portlandite

One of the As 0 4 -reacted CaO samples was studied with XAS (solid phase ASO 4 = 0.172 mol

kg'b and the obtained EXAFS were fit with theoretically-calculated phase and amplitude functions

of O, Ca and As backscatterers (Table B. 1). A typical RSF of this sample has four distinct peaks

at 1.35,2.8, 3.2, and 4.5 Â (uncorrected for phase). The fits for the Fourier filtered first peak gave

a coordination of 3.73 for first shell O at 1.688 Â. The values obtained for these As-O fits were

close to 0.0005 Â^, close to the As-0 fits of model compounds (Table 6.2). This low Debye-

Waller parameter value may indicate that all As-0 bond lengths are similar. The next major peak

at 2.88 Â (not corrected for phase) in Fourier transforms corresponded to the second neighbor Ca

(first shell Ca). However, these fits for Ca produced very small (f values (~ 0) when coordination

number (Nj) of Ca was floated (Nj ~ 0.35). Since the model compounds showed relatively high (f

values (~ 0.005-0.01 Chapter VI), Nj is expected to be larger than the obtained value. On the

basis of this preliminary data fitting, Nj for Ca was fixed at 1.0 and 2.0. The final fits were found

to be good for both Nj values (Table B.l). This leads to uncertainty in the value of Nj for Ca

atoms at this distance. Calcium atoms were also identified in the second and third shells at 3.55

and 3.70 Â, with a smaller (f value (~0.(X)4 A^) for the latter (corresponding to a peak at 3.2 A in

RSF). This indicates that the coordination for Ca at 3.70 A is expected to be larger than is given

in Table B.l. Arsenic as a backscatterer was also identified at 4.94 A These bond distances for

different backscatterers may indicate that ASO 4 forms bidentate mononuclear (Ca at 3.22 A); and bidentate binuclear and/or monodentate complexes (Ca at 3.55 and 3.70 A) with Ca polyhedra. Table B .l. EXAFS analysis of ASO4 reacted portlandite, gibbsite and johnbaumite.

Solid Phase ASO4 pH Backscattering N R(A) o^(Â^> AEo(eV) F Concentration atom mmol kg'^ 0.172 12.6 As-O 4* 1.69 0.0005 -3.69 57* (portlandite) As-Cai 1* 3.22 0.0024 2 .0 * 2* 3.25 0.0086 (0.35) (3.22) (-0 .0002) As-Caz 0.97 3.55 0.0093 As-Caz 1.14 3.70 0.0036 As-As 0.74 4.94 0.0051 johnbaumite 12.7 As-0 4* 1.69 0.0014 -2.33 49* As-Cai 1.09 3.23 0.0117 1.1* As-Caz 1.91 3.73 0.0128 As-As 1.06 4.23 0.0057 0.5 (gibbsite) 8.6 As-O 2* 1.65 0.0029 0.03 59* As-OH 2.7 1.728 0.0001 1.3* (4.66) (1.705) 0.0024 As-Ali 1* 2.52 0.0130 As-Alz 2 * 3.20 0.0075 * Coordination number is fixed. * Goodness-of-fit for complete FT. * Goodness-of-fit for all shells combined excepting the first As-0 shell. Numbers in parenthesis represent the values obtained by floating everything, but this shell is fixed at a higher coordination number to get reasonable oP that is compatible with the models.

3 218 However, it is cüfficult to ascertain whether johnbaumite actually precipitated at these low solid

phase ASO 4 concentrations.

Characterization o f Johnbaumite

When CaO reacted with 71.4 mM Na 2HAs0 4 .7 H%0 for 24 h in deionized water (suspension

density: 4 g L"'), at ambient temperature, and in the presoice of atmospheric CO 2; johnbaumite,

Ca;(As0 4 )3(0 H), an arsenate analogue of apatite precipitated. These precipitated solids were fine

grained (< 1 pm) and their XRD patterns were very similar to johnbaumite, reported by DUNN et al., (1980) (Fig. B.4). All of these reflections were similar to those of the reported natural mineral,

in their intensities and d-spacings, except for a reflection at 45.5 "29. This may be attributed to the presence of some other arsenate phase in addition to johnbaumite. On the basis of XRD, this

sample is tentatively identified as johnbaumite, or its poorly crystalline phase. Precipitation of johnbaumite in this system is surprising since previous Ca arsenate synthesis procedures of

FERRARIS (refer to Ca arsenate model cwnpounds in Chapter VI) and WARREN (personnel communication and REARDON et al., 1993) produced hydrated minerals such as sainfeldite, pharmacolite, and Ca 4(As04 )2(0 H) 2.4 H2O. The details of their synthesis procedures, however, were not available for comparison.

Chemical composition of the solutions in contact with johnbaumite precipitate showed 10.69 mM ASO4 and very low concentrations of Ca (< 0.05 mM). On the basis of obtained chemical analysis, XRD and vibrational data from As 0 4 -reacted portlandite (discussed above), the following reactions can be ascribed for the formation of johnbaumite:

3HAsO/ <=>3AsÛ4^- + 3H+ (1) SCaO + SHiO*» 5Ca(OH)2 (2) 5Ca(OH)2 + 10 H+ C» 5Ca^+ + lOHjO (3) 200

175

150

125

Si 100

10 15 20 25 30 35 40 45 50 55

Degree 26

Figure B.4. Comparison of XRD patterns of published (a) and synthetic johnbaumite (b). 220 5Ca^+ + 3AsO/ + HjO <=> Cas(As0 4 )3(OH) + H+ (4) Combining 1,3, and 4, 3HAsO/ +5Ca(OH)2+ 6H+ Cas(As0 4 )3(0 H) + 9H:0 (5)

These reactions indicate an increase in pH as johnbaumite precipitates. Small increases in solution pH (12.59 to 12.72 in Fig. B.l) and disappearance of intense IR OH stretching band at 3642 cm * support the above mechanism.

Johnbaumite Solubility

Johnbaumite dissolved rapidly and equilibrium was established within minutes (Table B.2).

The aqueous Ca/AsÜ4 ratios are slightly smaller than their stoichiometry in the solid, which may be caused by incongruency in the dissolution reaction or protonation of AsO#^ below pH 11.

Changes in pH should magnify the incongruency, which was not observed. Ionic strength (0.116 mole L'*) also did not affect the solubility product of johnbaumite, On the basis of these studies, a value can be estimated at 39.60. Data from supersaturation could not be used to predict ^"*^K«p since the measured Ca concentrations were close to detection limits (Fig. B.l).

Metastable dissolution products similar to CaHP0 4 , or octa-calcium phosphate, as has been observed in the dissolution of hydroxyapatite, may also be possible in the dissolution of this mineral.

CoordinationofAsOd in Johnbaumite

Vibrational spectroscopic studies of this mineral produced a sharp OH peak at 3562 cm * which corresponds to sample OH. A small and broad band around 3058 cm * (Fig. B.5) corresponds to crystalline HzO, and its low intensity indicates its relatively low concentration in the precipitated solids. The position of this band indicates that As 0 4 H bonds with HzO, since OH Table B^. Solubility of johnbaumite.

Ionic Total Time pH Strength Concentrations p(Activities of (h) (molcL‘) (mM) Ions) POAP) Ca ASO4 Ca^* A s O t 10 9.9 0.116 0.28 0.523 4.01 5.46 40.52 10 10.1 * 0.13 0.52 3.96 4.92 38.51 24 10.1 0.116 0.28 0.525 4.02 5.33 40.44 24 10.0 0.116 0.35 0.67 3.92 5.24 39.28 24 9.9 * 0.18 0.578 3.82 5.05 38.38 24 8.8 * 0.63 0.758 3.32 6.03 39.89

* Ionic strength of the medium was not controlled, however, it was 1.5 X 10"^ (± 0.1) moU L \ 0.6 As-O V,

0.5

0.4

? 0.3 < § 00

0.2 I 0.1

0.0 4000 3500 3000 2500 2000 1500 1000 500

Wavenumber (cm )

Figure B.5. FTIR spectra of johnbaumite.

g 223 frequencies are similar to those of H2O in ASO4 solvation shells (Chapter V). The OH holding for

H%0 is observed at 1661 cm'*. The presence of HjO in these samples may indicate the probable presoice of another hydrated Ca arsenate impurity or poorly crystalline johnbaumite. Arsenate exhibits V 3 splitting, and removal of degeneracy produced peaks at 940, 877, 859, 834 and 824 cm'*. Symmetric stretching vibrations of As-O-X (X = Ca, H, or H 2O) were observed at 798 and

736 cm'*, which may correspond to ASO 4 bonding with Ca; or relatively large numbers of Ca (as in rauenthalite. Chapter V) / H 2O, respectively. As-0 asymmetric bending vibrations are difficult to interpret since they produced a broad band at 470 cm * in IR. Ca-OH librational vibrations at 659 cm * in unreacted portlandite shifted to 630 cm * in johnbaumite, indicating strong interactions between Ca and ASO4 polyhedra.

EXAFS analysis of johnbaumite (Table B.l) showed four distinct peaks in its RSF. The peaks at 1.35 Â (uncorrected for phase) corresponds to the first shell O, 2.8 to Ca, 3.2 to Ca, and

3.85 to As. The Fourier filtered first peak fits weU with 4 O in the first shell around As at an average distance of 1.688 Â. Data fits for this shell produced a relatively large ( f (0.0015 Â^) when compared to ASO 4 in ettringte or portlandite. This may indicate non-uniform As-0 bond distances in ASO4 polyhedra. Fourier filtered data for the peaks at 2.8, 3.2, and 3.85 Â were fit well with 1.08 Ca, 1.91 Ca and 1.06 As, and the values were in the range observed for Ca arsenate model compounds (Table B.l). The RSF of johnbaumite also exhibited strong peaks at distances > 5 Â, which produced good fits with both Ca and As. Due to the ambiguity in the actual coordination and the identity of the backscatterer, this information is not presented here.

In summary, ASO4 polyhedra participate in extensive bonding and are distorted, as is evidenced by vibrational spectra and of the As-0 shell. In johnbaumite, ASO 4 forms both bidentate mononuclear, bidentate binuclear and/or mononuclear complexes. 224 As0 4 Interactions with Gibbsite

Arsenate removal by gibbsite was below 60 % in the concentration range evaluated (0-3.6 mM). Solution pH increased from 6.5 to 8 .6 with increases in ASO 4; whereas A1 concentrations remained stable until 3.0 mM, after which dissolved A1 increased with further increases in ASO 4

(Table B.l). The observed increases in solution pH were a manifestation of aqueous spéciation of added ASO 4, and ligand exchange with surface OH in monodentate, and bidentate fashion.

Although previous studies on ASO 4 sorption indicated that it fonns IS complexes with gibbsite

(ANDERSON et al., 1976), details of these interactions were not evaluated.

Infrared spectroscopic studies did not exhibit detectable changes in OH frequencies, which may be due to the very little ASO 4 sorption and strong bands of Al-OH in the region of As-0 V 3 vibrations. However, a small peak devel(^)ed at 880 cm'* at the base of a strong Al-OH band.

Although this peak increased in its intensity, it did not exhibit any shift with increases in solid phase ASO 4 concentration This may suggest that ASO 4 complexation with gibbsite was indifferent to increases in ASO4. The addition of Raman spectroscopic information and IR studies on deuterated samples may help to identify the precise nature of these surface complexes.

XAS studies were conducted on one sample of ASO 4 adsorbed gibbsite (solid phase ASO 4 =

0.5 mol kg'*). RSF of the extracted EXAFS indicate two distinct peaks, one at 1.35 and the other at 2.8 Â (uncorrected for phase). The first peak corresponds to O around As, which upon frtting with theoretical phase and amplitude functions showed 4.6 O at 1.705 Â with a high of 0.0024

(Table B.l). The bond distances and value may indicate that this As-0 shell may be comprised of two different parts, one with As-0 and the other As-O-X (X= Al, or H). When fitted for these differences in As-0 distances, the quality of the fit improved and this shell was resolved into two As-0 distances at 1.65, and the other two at 1.728 A. Longer As-0 bond distances may 225 correspond to protonated or bonded (with Al-po!yhedra) O. Another peak at 2.8 Â in the Fourier transform likely corresponds to As-Al distances of 2.5 and 3.2 Â, consistent with bidentate mononuclear, and bidentate bridging, or monodentate complexes with Al polyhedra.

Summary

AsOi Interactions with Lime

1) The spectroscopic studies indicate that AsO# forms IS complexes with portlandite by exchanging with surface OH. Uptake of ASO 4 by portlandite is essentially due to surface reactions rather than precipitation of johnbaumite at low concentrations (ASO 4 < 3.6 mM), because of absence of As-As correlations at 4.2 Â (EXAFS analysis). This is also supported by the absence of XRD and IR peaks corresponding to johnbaumite.

2) Johnbaumite may precipitate in alkaline calcareous environments, when lAP of solutions reaches 39.60.

A s04 Interactions with Gibbsite

1) There is very little ASO 4 uptake by gibbsite and changes in solution pH during adsorption is indicative of ligand exchange.

2) Sorbed ASO4 forms both bidentate mononuclear, and comer sharing (bidentate bridging and/or mononuclear) complexes. On the basis of available information, it is difficult to predict the precise geometry of corner sharing complexes. APPENDIX C

ETTRINGITE SURFACE SITE DENSITY CALCULATIONS

The reactive surface site density analysis was calculated by using the crystal structure refinement data of MOORE and TAYLOR (1970) and the measured surface area of synthetic ettringite (7.9 m V ’> Chapter III). For these calculations, it is important to know the exposed faces and number of sites on each of these faces. The surfaces of ettringite predominantly exhibit Ca-

OH2 functional groups as compared to the Ca-OH or Al-OH, since the c-axis is parallel to grain lengths (length ~ 2 -1 0 pm, width 0 .2 pm) and the latter two types are exposed only parallel to the

((X)l) plane (Fig 1.1,6.2). In addition, ‘A’ - type sites (mainly Ca-OHz) are dominant as compared to any other type (Chapter VI) per unit cell (UC) (A-type: 8 /UC, B-type: 1/UC, E-type: 2/UC).

The following procedure was used to estimate ettringite surface site density.

Surface area of ettringite = 7.9 m^g '

Area of unit cell (a.c or b.c) = 1.206 X 10^ pm^ (along C axis)

Number of unit cells per kg = (7.9*1000)/(1.206*10'‘®) =6.55*10^*

Considering one mole has 6.02*10“ atoms (Avogadro’s Nvunber), and all 8 Ca-OH: sites in the unit cell are reactive, then there will be 6.55*10^**8 As atoms per a kg of ettringite. This means 0.87 mol kg'* of As. By adding other sites parallel to the (001) plane (‘D’ and ‘C’ type)

(note that other sites parallel to the c-axis, such as ‘B’, ‘E’ and ‘F’ type sites are already included in the above calculation, since all exposed Ca-OHz are considered) the maximum site density is

226 227 estimated at 0.11 mol kg'*. However, all 8 Ca-OH; sites of a unit ceU are not reactive and the realistic number would be close to this predicted maximum. APPENDIX D

EXAFS DATA FITS

The EXAFS data fits given in this section supplement the arsenate sorption data presented in

Chapter VI. The following data fits are for Ca arsenate model compounds, and different backscatterers with different geometry in ettringite. The actual EXAFS fitting parameters are given in Chapter VI (Table 6.2,6.4,6.5).

228 229

-1.5

13

Figure D .I. Rauenthalite. EXAFS fits for second and higher shells (top), and for complete RSF (bottom). Fitted data (transmission data) is given in Table 6.2. 0.2 M

-0.4

- 0.6

- 0.6

Figure D ^. EXAFS fit for second and higher shells in pharmacolite. Fitted data is given in Table 6.2.

O 231

Tl

1 0.5

M /

1

r '

! h

M i

-10 12

Figure D J. EXAFS fits for the second and third shells (top), and for complete RSF (bottom) of ASO4 adsorbed ettringite. ASO4 = 3.24 mmol kg '. Fitted data (fluorescence data) is given in Table 6.4. 232

0.5

- 0.5

- 1.5 - 12

T2

-10 .i,f

Fleure D.4. EXAFS fits for the second and third shells (top), and for complete RSF (bottom) of ASÛ4 adsoited ettringite. ASO4 = 6.48 mmol kg‘\ Fitted data (fluorescence data) is given in Table 6.4. 233

-1

Jâ

- 0.5

a -2

-10 13 0

FipiirpDj;. EXAFS fits for the second and third shells (top), and for complete RSF (bottom) of ASO4 adsorbed ettringite. ASO4 = 0.29 mol kg'\ Fitted data (fluorescence data) is given in Table 6.4. ■0.5

-1.5

Figure D.6. EXAFS fit for second and higher shells of ASO4 adsorbed ettringite. ASO4 = 0.65 mol kg'\ Fitted data is given in Table 6.4.

g M M c X

-0 2

-0.4 -

12

Figure D.7. EXAFS fit for second and higher shells in AsO^ coprecipitated ettringite. ASO4 =1.44 mol kg'*. Fitted data is given in Table 6.5. m

-10

1 k ) R+À (A)

Figure D.8 . EXAFS fit for complete RSF of ASO4 coprecipitated ettringite. ASO4 =0.144 mol kg *. Fitted data is given in Table 6.5. Example of an outer-sphere ASO 4 complex in ettringite.

5 o\ -0.5 - 12 I 1 k (fi *) R+i (fi)

Figure D.9. EXAFS fit for second shell of AsO^ coprecipitated ettringite. ASO4 = 5.75 mmol kg'*. Fitted data is given in Table 6.5.

« LIST OF REFERENCES

ADAMS R, and HAJEK B.F. (1978) Effects of solution sulphate, hydroxide, and potassium concentrations on the crystallization of alunite, basaluminite, and gibbsite from dilute aluminum solutions.Soil Set. 126,169-173.

ADRIANO D C. (1986) Trace Elements in the Terrestrial Environment. SçûngQr-Weûa.'g NY.

AINES R.D., ROSSMAN G.R. (1984) Water in Minerals ? A peak in the Infrared. J. Geophys. Res. 89,4059-4071.

ALLISON J.D., BROWN D.S., and NOVO-GRADAC K.J. (1990) MINTEQA2/PRODEFA2, a geochemical assessment model for envircHunental systems: Version 3.0 user’s Manual. Environ. Res. Lab., USEPA, Athens, G A.

AMONETTE I.E., and RAI D. (1990) Identification of noncrystalline (Fe,Cr)(0H)3 by infrared spectroscopy. Clays and Clay Minerals 38,129-136.

ANDERSON M.A., FERGUSON J.F., and GAVIS J. (1976) Arsenate adsorption on amorphous aluminum hydroxide. J. Colloid Interface Sci. 54,391-399.

ANDERSON L.D., KENT D.B., and DAVIS J.A. (1994) Batch experiments characterizing the reduction of Cr(VI) using suboxic material from a mildly reducing sand and gravel aquifer. Env. Sci. Tech. 28,178-185.

APPLETON A.R., PAPELIS C., and LECKIE J O. (1989) Adsorptive removal of trace elements from coal fly-ash wastewaters onto iron oxyhydroxide. In 43rd Purdue Ind. Waste Conf. Proc., Lewis Pub. MI. 161-172.

ASHLEY C.A., and DONIACH S. (1975) Theory of extended X-ray absorption edge fine structure (EXAFS) in crystalline solids.Phys. Rev. 11,1279-1288.

ATKINS M., MACPHEE D., KINDNESS A., and GLASSER F.P. (1991) Solubility properties of ternary and quaternary compounds in the Ca 0 -Al2 0 3 -S0 3 -H2 0 system. Cem. Conor. Res. 21,991-998.

BARAN E.J., and AYMONINO P.J. (1968) The infrared spectra of barium tetraoxometalates. Spectrochim. Acta 24A, 291-295.

BASSETT H., and GOODWIN T.H. (1949) The basic aluminum sulphates. /. Chem. Soc., 2239- 2279.

238 239 BEECH T A., and LINCOLN S.F. (1971) Arsenato Complexes of Cobalt (III). Aust. J. Chem. 24, 1065-1070.

BHAGAVANTAM S., and VENKATARAIDU T. (1969) Theory of groups and its application to physical problems. Academic Press, NY. 279 pp.

BIBER M.V., and STUMM W. (1994) An in-situ ATR-FTIR study: The surface coordination of salicylic acid on aluminum and iron (III) oxides. Environ. Sci. Technol. 28,763-768.

BIGHAM J.M., CARLSON L., and MURAD E. (1994) Schwertmannite, a new iron oxyhydroxysulfate from Pyhasalmi, Finland, and other localities. Miner. Mag. 58, 641- 648.

BISHOP J.L., PIETERS C.M., and EDWARDS J O. (1994) Infrared spectroscopic analysis on the naturae of water in montmorillonite. Clays and Clay Minerals 42, 702-716.

BROOKINS D.G. (19Ü) Eh-pH Diagrams for Geochemistry. Springer-Verlag, 175 pp.

BROWN G.E. Jr., CALAS G., WAYCHUNAS G.A, and PETIAU J. (1988) X-ray absorption spectroscopy: implications in mineralogy and geochemistry. In Spectroscopic Methods in Mineralogy and Geology. Eds: F.C. Hawthrone. Rev. Mineralogy 18, Mineral. Soc. Am. Washington DC.

BROWN R.G., and ROSS S.D. (1972) The vibrational spectra of some condensed tetrahedral anions Spectrochim. Acta 2%k, 1263-1274.

BUTLER W.M., ANGELL C.L., McALLISTER W., and RISEN Jr. W. (1977) Far-Infrared study of cation motion in dry and solvated mono- and divalent cation containing zeolites X and Y. J. Phys. Chem. 22,2061-2068.

CALLERIM., and FERRARIS G. (1967) Struttura della haidingerite: CaH(As04).H20. Period. Miner. 36,1-23.

CAMBIER P. (1986) Infrared study of goethites varying crystallinity and particle size: I. Interpretation of OH and lattice vibration frequencies. Clay Miner. 21,191-200.

CARLSON L., and SCHWERTMANN U. (1990) The effect of CO 2 and oxidation rate of the formation of goethite versus lepidochrocite from and Fe(II) system at pH 6 and 7. Clay Minerals 25,65-71.

CASEY W.H., BANFDELD J.F., WESTRICH H R., and McLAUGHLIN L. (1993) What do dissolution experiments tell us about natural weathering? Chem. Geol. 105,1-15.

CATTI M., and IVALDI G. (1983) On the topotactic ddiydration Ca 3(As0 4 )2.11H2 0 (phaunouxite)- Ca3(AsO4)2.10H2O (rauenthalite) and the structures of both minerals. Acta Cryst. B39,4-10. 240 CHANG K.-S., LIN C.-R, LEE D.-Y., LO SHANG. -L., and YASUNAGA T. (1994) Kinetics of Cr(ni) adsorption/desorption at the yA1203/water interface by the pressure-jump technique./. Colloid Intetface Sci. 165,169-176.

CHARLET L., and MANCEAU A. A. (1992) X-ray absorption spectroscopic study of the sorption of Cr(m) at the oxide-water interface. II. Adsorption, coprecipitation, and surface precipitation on hydrous ferric oxide. /. Colloid Interface Sci. 148,443-459.

CHEN S.-S. and MEHTA P.K. (1982) Zeta potential and surface area measurements on ettringite. Cera. Concr. Res. 12, 257-259.

COTTON RA. (1971) Chemical Applications of Group Theory. Wiley Eastern Ltd., 386 pp.

CRAMER S.R, and HODGSON K.O. (1979) X-ray absorption spectroscopy: a new structural method and its applications to bioinorganic chemistry. Prog. Inorg. Chem. 25, Ed: S.J. Lippard, 1-40.

CRUICKSHANK D.W.J. (1961) the role of 3d-orbitals in rt-bonds between a) silicon, phosphorus, sulphur, or chlorine and b) oxygen or nitrogen. J. Chem. Soc., 5486-5504.

CRUICKSHANK D.W.J., and ROBINSON E.A. (1966) Bonding in orthophosphates and orthosulfates. Spectrochim. Acta 22,555-563.

DAMIDOT D., and GLASSER F.P. (1993) Thermodynamic investigation of the Ca 0 -Al2 0 3 - CaS0 4 -H2 0 at 25 “C and the influence of Na20. Cem. Concr. Res. 23,221-238.

DAMIDOT D., and GLASSER F.P. (1992) Thermodynamic investigation of the Ca0-Al20g- CaS0 4 -H2 0 at 50 °C and 85 °C. Cem. Concr. Res. 22,1179-1191. DAMIDOT D., ATKINS M., KINDNESS A., and GLASSER RP. (1992) Sulphate attack on concrete: limits of the Aft stability domain. Cem. Concr. Res. 22,229-234.

DEER W.A., HOWIE R.A., and ZUSSMAN J. (1966) An Introduction to the Rock-Forming Minerals, 528 pp, Longman Group Ltd. England.

DONER H E., and LYNN W.C. (1989) Carbonate, halide, sulfate and sutflde minerals. In Minerals in Soil Environments (eds. J.B. Dixon and S.B.Weed), SSSA Book Series: 1, 279-324.

DOVE P., RIMSTIDT (1985) The solubility and stability of scorodite. Am. Miner. 70, 838-844.

DOYLE W.P., and EDDY P. (1967) Infrared spectra of chromâtes (V) of calcium, strontium and barium. Spectrochim. Acta 23A, 1903-1907.

DREVER J.I. (1988) The Geochemistry of Natural Waters. Prentice Hall NY. 437 pp. 241 DUNN P.J., PEACOR D R., and NEWBERRY N. (1980), Johnbaumlte, a new member of the apatite group from Franklin, New Jersey. Am. Mineral. 65,1143-1145.

DUNN P.J., PEACOR D.R., LEAVENS P.B., and BAUM J.L. (1983) Charlesite, a new mineral of the ettringite group, from Franklin, New Jersey. Am. Mineral. 68,1033-1037.

EARY L.E., and RAI E. (1987) Kinetics of chromium (HI) oxidation to Cr(VI) by reaction with manganese dioxide. Env. Sci. Tech. 21,1187-1193.

EISENBERGER P., and LENGELER B. (1980) Extended X-ray absorption fine structure determinations of coordination numbers: Limitations. Phys. Rev. 22,3551-3561.

El-RAFEI E.A. (1981) Valency of chromium in the equilibrium ternary phase belonging to the noncondensed system CaO-CrzOs-Oz. Inorg. Chem. 20,222-227.

FADINI A., and SCHNEPEL F.-M. (1989) Vibrational Spectroscopy. Ellis Horwood Ltd., 203 pp.

FARMER V.C. (1974) The infrared Spectra of Minerals. Miner. Soc., 537 pp.

FATELEY W.G., McDEVITT N.T., and BENTLEY F.F.(1971) Infrared and Raman selection rules for lattice vibrations: The Correlation Method. Appl. Spectr. 25,155-173.

FAURE G. (1989) Principles and Applications of Inorganic Geochemistry. Mcmillan Publishing CO., New York, 626pp.

FERRARIS G. (1969) The crystal structure of pharmacolite CaHAs 0 4 .2 Hz0 . Acta Cryst. B25, 1544-1550.

FERRARIS G., and CHIARI G. (1970a) The crystal structure of NazHAs 0 4 .7 Hz0 . Acta Cryst. B26,1574-1583.

FERRARIS G., and CHIARI G. (1970b) The crystal structure of CaHAs 0 4 (Weilite). Acta Cryst. 826,403-410.

FERRARIS G., JONES D.W., and YERKESS J. (1971) A neutron diffraction study of the crystal stmcture of sodium arsenate heptahydrate NazHAs 0 4 .7 Hz0 . Acta Cryst. 827,354-360.

FERRARIS G., and ABBONA F. (1972) The crystal structure of Ca 5(HAS0 4 )2(AS0 4 )z.4 Hz0 (sainfeldite). Bull. Soc.fr. Mineral. Cristallogr 95,33-41.

FERRARIS G., JONES D.W., and YERKESS J. (1972), Neutron and X-ray refinement of the crystal structure of CaHAs04.H20 (haidingerite). Acta Cryst. 828,209-214.

FERRARO G., and NAKAMOTO K. (1994) Introductory Raman Spectroscopy. Academic Press Inc. NY. 370 pp. 242 FOWLER R.K., TRAINA S.J., BIGHAM J.M., SOTO U.I. (1993) Solution chemistiy and mineralogy of CiMn Coal Technology By-products in a long term equilibration study. Ag. Abst., 30.

FRUCHTER J.S., RAI D., and ZACHARA J.M. (1990) Identification of solubility-controlling solid frfiases in a laige fly ash field lysimeter. Environ. Sci. Tech. 24,1173-1179.

FULLER C.C., and DAVIS J.A. (1994) Evaluation of the processes controlling dissolved arsenic in Whitewood Creek, South Dakota. In Origin, Transport, and fate of arsenic- contaminated alluvial sediments in the Cheyenne Rive system. South Dakota: U.S. Geological Survey toxic substances in surface waters and sediments program Ed. K.E. Goddard. U.S.G.S. Water Supply Paper 2385.

GENGE M.J., and JONES A.P., and PRICE G.D. (1995) An infrared and Raman study of carbonate glasses: Implications for the structure of carbonatite magmas. Geochim. Cosmochim. Acta 59,927-937.

GEORGE G.N., and PICKERING l.J. (1993) EXAFSPAK: A suit of computer programs for analysis of X-ray absorption spectra. Documentation. Stanford Synchrotron Radiation Laboratory, Stanford, Ca.

GILLETTE P.C., LANDO J.B., and KOENIG J.L. (1982) Band shape analysis of Fourier Transform infrared spectra. Appl. Spectr. 36,401-404.

GOLDBERG S. and GLAUBIG R.A. (1988) Anion sorption on a calcareous, montmorillonitic soil-As. Soil Sci. Soc. Am. J. 52,1297-1300.

GRAYBEAL J.D. (1988) Molecular Spectroscopy. Me Graw-Hill NY., 732 pp.

GUPTA S.K., and CHEN K.Y. (1978) Arsenic removal by adsorption. Journal WPCF 50, 493- 506.

HAMPSON C.J., and BAILEY J.E. (1982) On the structure of some precipitated calcium alumino-sulphate hydrates, J. Mater. Sci. 17. 3341-3346.

HARRISON J.B., and BERKHEISER V. (1982) Anion interactions with freshly prepared hydrous iron oxides. Clays and Clay Minerals 30,97-102.

HASSETT D.J., HASSETT D.F.P., andMcCARTHY G.J. (1991) Ettringite formation in coal ash as a mechanism for stabilization of hazardous trace elements. In 9th Intern. Coal Ash Symp., 31:1-31:17.

HASSETT D.J., MCCARTHY G.J., KUMARATHASAN P., and PFLUGHOEFT-HASSETT (1990) Synthesis and characterization of selenate and sulfate-selenate ettringite structure phases. Mat. Res, Bull. 25,1347-1354.

HAVLICA J., and SAHU S. (1992) Mechanism of ettringite and monosulphate formation. Cem. Concr. Res. 22, 671-677. 243 HAYES K.F. (1987) Equilibrium, Spectroscopic, and Kinetic Studies of Ion Adsorption at the Oxide!Aqueous Intetface. Ph. D. Dissertation, Stanford University.

HAYES K.F., ROE A.L., BROWN G.E. Jr., HODGSON K.O., LECKIE J.O., and PARKS G.A. (1987) In situ X-ray absorption study of surface complexes: selenium oxyanions on a- FeOOH. Science 238,783-786.

HELGESON H.C. (1968) Evaluation of irreversible reactions in geochemical processes involving minerals and aqueous solutions-I. Thermodynamic relations. Geochim. Cosmochim. Acta 32,853-877.

HENRY J.L., and KING G.E. (1950) Phase rule investigation of the system AI 2O3 -SO3 -H2 O at 60 °C, basic region, J. Amer. Chem. Soc. 72,1282-1286.

HERZBERG G. (1945) Molecular Spectra and Molecular Structure. Vol II. Infrared and Raman Spectra o f Polyatomic Molecules. Van Nostrand, NY.

HESS Jr. B.A., SCHAAD L.J., and ZAHRADNÎK P.C (1986) Ab initio calculations of vibrational spectra and their use in the idaitification of unusual molecules. Chem. Rev. 8 6 , 709-730,

HEZEL A , and ROSS S.D. (1968) Forbidden transitions in the infra-red spectra of tetrahedral anions-FV. The vibrational spectra (4(X)0-4(X) cm-1) of some cobalt (HI) sulphato- and phosphato-complexes. Spectrochim. Acta 24A, 985-992.

HILLS C. D., SOLLARS C.J., AND PERRY R. (1993) Ordinary portland cement based solidification of toxic wastes: The role of OPC reviewed, Cem. Concr. res. 23,196-212.

KINGSTON F.J., ATKINSON R.J., POSNER AM., and QURK J.P. (1967) Specific adsorption of anions. Nature 215,1459-146.

KINGSTON F.J. (1981) A review of anitm adsoiption. In Adsorption of inorganics at solid liquid interfaces, Eds. M.A. Anderson and A.J. Rubin. Ann Arbor Pub. Co. 357 pp.

HOEK E.E. van der, BONOUVRIE P.A, COMANS R.N.J. (1994) Sorption of As and Se on mineral components of fly ash: relevance for leaching processes. App. Geochemistry 9, 403-412.

HSU, FA HO (1989) Aluminum Oxides and Oxyhydroxides. In Minerals in Soil Environments (eds. J.B. Dixon and S.B.Weed), SSSA Book Series: 1,331-371.

HUG S., and SULZBERGER B. (1994) In situ FTIR studies of interactions at hydrous metal oxide surfaces. Env. Chem. Div., ACS National Meet San Diego, CA, 471.

JONES F.E. (1944) The quaternary system Ca 0 -Al2 0 3 -CaS0 4 -H2 0 at 25 °C: Equilibria with crystalline AI2 O3 .3 H2O. alumina gel, and solid solution, J. Phys. Chem. 48,311-378. 244

JOHANSON G. (1960) The crystal structures of some basic aluminum salts;Acta Che. Scand. 14, 771-773.

JOHANSON G. (1962) The crystal structures of {Al2(0H)2(H20)g}(S04)2.2H20 and {AÎ2(0H)2 (H2 0 )g(Se 0 4 )2 . 2H2 O: Acta Che. Scand. 14,771-773.

JOHNSTON C.T. (1990) Fourier Transform infrared and Raman spectroscopy. In: Instrumental Surface Analysis of Geologic Materials, Ed. Perry, D.L., VCH NY., 121-155.

JOHNSTON C.T., TIPTON T., STONE D.A., ERICKSON C., and TRABUE S.L. (1991) Chemisoiption of p-dimethoxybenzene on Cu-montmorillonite. Lagmuir 7, 289-296.

JOHNSON C.A., and XYLA A.G. (1991) The oxidation of Cr(III) to Cr(VI) on the surface of Manganite (MnOOH). Geochim. Cosmochim. Acta. 55, 2861-2866.

KENT D.B., DAVIS J.A., ANDERSON L C D., REA B.A., and WAITE T.D. (1994) Transport of chromium and selenium in the suboxic zone of a shallow aquifer: Influence of redox and adsoiption reactions. Wat. Resour. Res. 30 ,1099-11 14.

KONINGSBERGER D C., and PRINS R. (1988) X-ray absorption: principles, applications, techniques of EXAFS, SEXAFS and XANES. John Wiley & Sons, 673 pp.

KRESTOV G.A. (1991) Thermodynamics of Solvation; Solution and Dissolution; Ions and Solvents; Structure and Energetics. Ellis Horwood Pub., NY.282 pp.

KRESTOV G.A., NOVOSYOLOV N.P., PERELYGIN, KOLKER A.M., SAFONOVA L.P., OVCHINNIKOVA V.D., and TROSTIN V.N. (1994) Ionic Solvation. EUis Horwood, NY. 347 pp.

KUMARATHASAN P., McCARTHY G.J., HASSETT D.J., and HASSETT D.F.P. (1990) Oxyanion substituted ettringites: synthesis and characterization and their potential role in immobilization of As, B, Cr, Se and V. In Fly Ash and Coal Conversion By-Products; Characterization, Utilization and Disposal VI (eds. R.L. Day and F.P. Glasser) Mat. Res. Soc. Symp. Proc. 178, 83-103.

LEE P.A., and PENDRY J.B. (1975) Theory of the extended x-ray absorption fine structure. Phys. Rev. 11, 2795-2811.

LINDSEY W.L. (1979) Chemical Equilibria in Soils. John Wiley & Sons, New Yoik. 449 p.

LONGMIRE P., BROOKINS D.G., and THOMSON B.M. (1990) Hydrogeochemical interactions and evolution of acidic solutions in soil. In Chemical Modeling o f Aqueous Systems II (Eds. D C. Melchoir and R.L. Bassett), ACS Symposium Series 416,154-168. 245 LUMSDON D.G., FRASER A.R., RUSSELL J.D., and LIVESEY N T. (1984) New Infrared band assignments for the arsenate ion adsorbed on synthetic goethite (a-FeOOH), J. Soil Sci. 35, 381-386.

LYTLE F.W., GREGOR R.B., SANDSTROM D R., MARQUES E.C., WONG J., SPIRO C.L., HUFFMAN G.P., and HUGGINS F.E. (1984) Measurement of soft X-ray absorption spectra with a fluorescent ion chamber detector. Nucl. Instr. Methods 226,542-548.

MADDOMS W.F. (1980) The scope and limitations of curve fitting. Appl. Spectr. 34,245-267.

MANCEAU A. A., CHARLET L., BOISSET M.C., DIDIER B., and SPADINI L. (1992) Sorption and spéciation of heavy metals on Fe and Mn oxides. From microscopic to macroscopic. App. Clay Sci. 7,201-223.

MANCEAU A. A., and CHARLET L. (1992) X-ray absorption spectroscopic study of Cr(III) at the oxide/water interface. I Molecular mechanism of Cr(IlI) oxidation on Mn oxides. J. Coll. Intetf. Sci. 148,425-442.

MATTIGOD, S.V., RAI D., EARY L.E., and AINSWORTH C.C. (1990) Geochemical factors controlling the mobilization of inorganic constituents from fossil fuel combustion residues: I. Review of the major elements. J. Envirorn. Qual. 19:187-201

McMURDIE H.F., MORRIS M.C., EVANS E.H., PARETZKIN B., WONG-NG W., and ZHANG Y. (1986) Calcium Aluminum Sulfate Hydroxide Hydrate, CagAl 2 (S0 4 )3 (0 H)i2 . 26H2O. Powder Diff. 4, 337.

McMILLAN P. (1985) Vibrational spectroscopy in the minerals. In: Microscopic to Macroscopic- Atomic Environments to Mineral Thermodynamics. Reviews in Mineral. 14, (Eds. Kieffer S.W., and Navrotsky A.), 9-63.

MERRILL D.T., MANZOINE M.A., PETERSON J.J., PARKER D.S., CHOW W., and HOBBS A.O. (1986) Field evaluation of arsenic and selenium rmnoval by iron coprecipitation. Journal WPCF SS, 18-26.

MILLER P.J., CESSAC G.L., and KHANNA R.K. (1971) Infrared spectra of KBr crystals doped with Cr 0 4 ^ and some dipositive ions, Spectrochim. Acta 27A, 2019-2026.

MOENKE H. (1962) Mineralspektren I, Akademie-Veiiag, Berlin.

MOODENBAUGH A.R., HARTT J.B., and HURST J.. (1983) Neutron Diffraction study of polycrystalline H2SO4 and HzSe 0 4 . Phys. Rev. B 28,3501-3505.

MOORE A.E., and TAYLOR H.F.W. (1968) Crystal structure of ettringite. Nature 218, 1048- 1049.

MOORE A.E., and TAYLOR H.F.W. (1970) Crystal structure of ettringite, Acta Cryst. B26,386. 246 MURDOCH J. and CHALMERS (1958) Geol. Soc. Am. Bull. 6 9 . 1620.

NAKAMOTO K. (1986) Infrared and Raman Spectra of Inorganic and Coordination Compounds. John Wiley & Sons NY., 483 pp.

NIMMO J.K. (1981) Sodium chromate (II) at 296 K (Neutron). Acta Cryst. B37,431-433.

NISHIKAWA T., SUZUKI K., ITO S., SATO K., and TAKEBE T. (1992) Decomposition of synthetic ettringite by carix)nation. Cem. Concr. Res. 22, 6-14.

NORDSTROM D.K. (1982) The effect of sulfate on aluminum concentrations in natural waters; some stability relations in the system AI 2O3 -SO3 -H2O at 298 K. Geochim. Cosmochim. Acta 46,681-692.

NORDSTROM D.K., and BALL J.W. (1986) The geochemical behavior of aluminum in acidified surface waters. Sci. 232,54-56.

O’DAY P.A., REHR J.L, ZABINSKY S.L, and BROWN, Jr. G.E. (1994) Extended X-ray Absorption Fine Structure (EXAFS) analysis of disorder and multiple-scattering in complex crystaline solids. J. Am. Che. Soc. 116,2938-2949.

OELKERS E.H., SCHOTT L, and DEVIDAL J.-L. (1994) The effect of Aluminum, pH, and chemical affinity on the rates of alumino silicate dissolution reactions. Geochim Cosmochim. Acta 58,2011-2024.

OSSAKA L, HIRABAYASHI JUN-ICHI, OKADA K., KOBAYASHI R., and HAYASHI T. (1982) Crystal structure of minamiite, a new mineral of the alunite group. Am. Mineral. 67,114-119.

PAPELIS C., BROWN G.E. Jr., PARKS G.A., LECKIE JO (1995) X-ray absorption spectroscopic studies of cadmium and selenite adsorption on aluminum oxides. Langmuir 11,2041-2048.

PANAGIOTOPOULAS N. CH., and BROWN I D. (1972) The crystal structure of p-NazCrzOy. Acta Cryst. B28,1352-1357.

PARFITT R.L., RUSSELL J.D., and FARMER V.C. (1975) Confirmation of the surface structures of goethite (a-FeOOH) and phosphated goethite by infrared spectroscopy. J. Chem. Soc. Faraday 1 72,1082-1087.

PARFITT R.L. (1978) Anion adsoiption by soil and soil minerals. Adv. Agron. 30,1-50.

PAULING L. (1960) The Nature of the Chemical Bond. Cornell Univ. Press, Ithaca NY.

PEACOR D R., DUNN P.J., and DUGGAN M.J.(1983) Sturmanite, a ferric iron, boron analogue of ettringite. Can. Mineral. 21,705-709. 247 PEIXOTO R.T.G. (1995) Surface chemical properties of oxisols derived from Parana Basalt (Brazil). Ph. D. Dissertation, The Ohio State University.

PETERSON M.L., BROWN G.E. Jr., and PARKS G.A. (1994) Direct evidence for heterogeneous redox at the aqueous chromium/magnetite interface. Presented at the 21st annual SSRL User’s Meeting, Stanford, CA.

POLLMANN H., KUZEL H.-J., and WENDA R. (1989) Compounds with ettringite structure. N. Jb. Miner. Abh. 160,133-158.

POLLMANN H., AUER St., KUZEL H.-J., and WENDA R. (1993) Solid solution of ettringites. Part II: Incorporation of B (O H )4 - and C 1O 4- in 3CaO.Al20) 3CaS04.32H20. Cem. Concr. Res. 23,422-430.

PLUMMER L.N., PARKHURST D.L., and THORSTENSON D C. (1983) Development of reaction models for ground-water systems. Geochim. Cosmochim. Acta 47,665-686.

REARDON E.J., WARREN C.J., and HOBBS M.Y. (1993) Reduction of trace element concentrations in alkaline waste porewaters by dedolomitization. Environ. Sci. Tech. 27, 310-315.

REDDY K.J., GLOSS S.P. WANG L. (1994) Reaction of CO 2 with alkaline solid wastes to reduce contaminant mobility. Wat. Res. 28,1377-1382.

REEDER R.J., LAMBLE G.M., LEE J.F., STAUDT W.J. (1994) Mechanism of Se 0 4 ^ substitution in calcite: An XAFS study. Geochim. Cosmochim. Acta 58,5639-5646.

REHR J.J., and ALBERS R.C. (1990) Phys. Rev. B41,8139.

REHR J.J., BOOTH C.H., BRHXiES P., and ZABINSKY S.L (1994) X-ray absorption fine structure in embedded atoms. Phys. Rev. 49,12347-12350.

RUSSELL J.D., PARFITT R.L., FRASER A.R., and FARMER V.C. (1974) Surface structures of gibbsite goethite and phosf^ated goethite. Nature 248,220-221.

RUSSELL J.D. (1979) Infrared spectroscopy of ferrihydrite: evidence for the presence of structural hydroxyl groups. Clay Minerals 14,109-114.

SEEGER D M.. KORZENIEWSKI C., and KOWALCHYK W. (1991) Evaluation of vibrational force fields derived by using semiempirical and abinitio methods. J. Phys. Chem. 95, 6871-6879.

SMITH R., and MARTELL A.E. (1976) Critical Stability Constants. V. 4. Inorganic Complexes. Plenum Press NY, 257 pp. 248 SOLEMN J.K., and McCARTHY GJ. (1992) Hydration reactions and ettringite formation in selected cementitious coal conversion by-products. In Mat. Res. Symp. Proc. Series 245, 71-79.

SPOSITO G. (1989) The Chemistry of Soils. Oxford University Press, New York, 277 pp.

SPOSITO G. (1994) Chemical Equilibria and Kinetics in Soils. Oxford Univ. Press NY, 268 pp.

STERN E.A., SAYERS D.E., AND LYTLE F.W. (1975) Extended X-ray absorption fine structure technique. HI. determination of physical parameters. Phys. Rev. 11,4836-4846.

STEWART J.J.P. (1989a) Optimization of parameters for semiempirical methods; I. Method, J. Comp. Chem. 10,209-220.

STEWART J.J.P. (1989b) Optimization of parameters for semiempirical methods: II. Applications, 7. Comp. Chem. 10,221-264.

STRUCTURE REPORTS, 1994

STUMM W. (1993) Chemistry of the Solid-Water Interface, Wiley-Interscience Pub., New York, 428 pp.

STUMM W., and MORGAN J.J. (1982) Aquatic Chemistry, 2nd ed, Wiley Interscience, New York.

SINGH S.S., and BRYDON J.E., (1969) Solubility of basic aluminum sulphates at equilibrium in solution and in the presence of montmorUlonit Soil Sci. 107,12-17.

SJOSTROM j. (1993) Combined studies of soil and ground water chemistry, southwest Sweden, focusing on A1 and SO 4 . Appl. Geochim. suppl. issue 2,281-283.

SULLIVAN P.J., YELTON J.L., and REDDY K.J. (1988) Solubility relationships of aluminum and iron minerals associated with acid mine drainage. Environ. Geol. Water Sci. 11, 283- 287.

SZYMANSKI H A. (1967) Raman Spectroscopy: Theory atuiPractice, V0I.I. John Wiley, NY.

TAYLOR H.F.W. (1973) Crystal structures of some double hydroxide minerals. Mineral. Mag. 39, 377-389.

TEJEDOR-TEJEDOR M., YOST E C., and ANDERSON M.A. (1990) Characterization of benzoic and phenolic complexes at the goethit^aqueous solution interface using cylindrical intenal reflection Fourier Transform Infrared Spectroscopy. Part 1. Methodology. Lanmuir 6,979-987. 249 TEJEDOR-TEJEDOR M., YOST EC., and ANDERSON M.A. (1992) Characterization of benzoic and phenolic complexes at the goethite/aqueous solution interface using cylindrical internal reflection Fourier Transform Infrared Spectroscopy. 2. Bonding Structures. Lanmuir 8,525-533.

TEO B. -K. (1986) EXAFS: Basic Principles and Data Analysis. Springer-Verlag NY., 346 pp.

THEIS L.T., RIPP J.A,, and VILLAUME J.F. (1989) Physical and chemical characteristics of unsaturated pore water and leachate at a dry fly ash disposal site. In 43rd Purdue Ind. Waste Conf. Proc., Lewis Pub. MI. 161-172.

TOSSEL J.A., and VAUGHAN D.J. (1992) Theoretical Geochemistry: Application of Quatum Mechanics in the Earth and Mineral Sciences. Oxford Pub. NY., 514 pp.

TWU J., and DUTTA P.K. (1989) Structure and reactivity of oxovanadate anions in layered lithium aluminate materials.J. Phys. Chem. 93,7863-7868.

URBAN M.W. (1993) Vibrational Spectroscopy of Molecules and Macromolecules on Surfaces. John Wiley & Sons Inc. NY., 384 r).

WANG, WEI-ZI and HSU, PA HO (1994) The nature of polynuclear OH-Al complexes in laboratory-hydrolyzed and commercial hydroxyaluminum solutions. Clays & Clay Minerals 42, 356-368.

WAYCHUNAS G.A., BROWN G.E. Jr., APTED M.J. (1986) X-ray K-edge absorption spectra of Fe minerals and model compounds: H. EXAFS. Phys. Chem. Minerals 13,31-47.

WAYCHUNAS G.A., REA B.A., FULLER C.C., and DAVIS J.A. (1993) Surface chemistry of ferrihydrite: Part 1. EXAFS studies of the geometry of coprecipitated and adsorbed arsenate. Geochim. Cosmochim. Acta 57,2251-2269.

WEINSTOCK N., SCHULZE H., and MULLER A. (1973) Assignment of Vz (E) and V 4 (Fz) of tetrahedral species by the calculation of the relative Raman intensities: The vibrational spectra of V043-, CrOA M oO/, WOA Mn 0 4 , TcO/, Re0 4 , RUO4 and OSO4. /. Chem. Phys. 59,5063-5066.

WERSIN P., HOCHELLA Jr. M.F., PERSSON P., REDDEN G., LECKIE J.O., HARRIS D.W. (1994) Interaction between aqueous uranium (VI) and sulfide minerals: spectroscopic evidence for sorption and reduction. Geochim. Cosmochim Acta 58, 2829-2843.

WHITE W.B. (1970) The Carbonate Minerals. In The infrared Spectra of Minerals (Ed. V.C. Farmer), 227-284.

WINSTON H. and HALFORD R.S. (1949)7. Chem. Phys. 17,607. 250 WOLERY T.J., and DEVELER S.A. (1992) EQ6, a computer program for reaction path modeling of aqueous geochemical systems: theoretical manual, user’s guide and related documentation. Version 7.0. Lawrence Livermore Laboratory, UCRL-MA-110662 PTN. 339 p.

WOLLAST R., and CHOU L. (1992) Surface reactions during the early stages of weathering of albite. Geochim. Cosmochim. Acta 56,3113-3122.

ZABINSKY S.I., REHR J.J., ANKUDINOV A., ALBERS R.C., and ELLER M.J. (to be published) Multiple scattering calculations of X-ray absoiption spectra.