Fe(III)-Aso4-SO4 Hydrothermal Systems

Precipitation and Characterization of Arsenate Phases from Ca(II)-Cu(II)- Fe(III)-AsO4-SO4 Hydrothermal Systems

Mario Alberto Gomez

Department of Mining and Materials Engineering

McGill University

Montreal, QC, Canada

October 2010

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Doctor of Philosophy

Abstract

The scope of this thesis is the study of three Fe(III)-As(V) hydrothermal systems. The

first one is the Fe(III)-AsO4-SO4 system and crystalline phases that are produced under high temperature (150-225°C); this was studied to clear up previous contradicting information on this system in relation to industrial arsenic products that are formed during the autoclave processing of arsenical sulphide gold feedstocks and asses their arsenic stability. The second system studied was Cu(II)-Fe(III)-AsO4-SO4 system at 150°C; this was investigated due to its relevance to industrial pressure leaching of copper concentrates. This system was studied in order to examine the possible effect of copper on the precipitation of scorodite. Finally, the structural and molecular examination of two

members of the Ca(II)-Fe(III)-AsO4 system, namely yukonite (synthetic and natural and arseniosiderite was undertaken due to their relatively unknown nature and the potential role play in controlling arsenic release in tailings. In the first case, three arsenate phases were found to be produced at various conditions explored here, these were: sulfate

containing-scorodite (Fe(AsO4)1-0.67x(SO4)x · 2H2O where 0.00≤x≤0.20), ferric arsenate

sub-hydrate (FAsH; Fe(AsO4)0.998(SO4)0.01 · 0.72H2O), and basic ferric arsenate sulfate

(BFAS; Fe(AsO4)1-x(SO4)x(OH)x · (1-x)H2O, where 0.3

i temperatures (< 150°C) before ultimately converting to the most stable phase (scorodite). These findings indicate that copper could coprecipitate at short retention times hence the importance of kinetics on the formation of scorodite. In the third system, the synthetic yukonite was found to be equivalent at the atomic, molecular and structural level to the natural yukonite. At the molecular level, arseniosiderite was found to have an H-bonding environment as in scorodite and exhibit extra protonated arsenate groups. In yukonite, in contrast, a wide diffuse H-bonding environment was observed with only arsenate groups present. At the electronic level yukonite and arseniosiderite were found to be identical, indicating that the local nature of the As, Fe and Ca atoms in these closely related but distinct phases is the same. Structural analysis of the materials showed that yukonite consists of nano and poorly crystalline domains while in the case of arseniosiderite micro-size single crystal domains exist.

ii Resume

La portée de cette thèse comprend l'étude de trois systèmes hydrothermiques Fe(III)-

As(V). Le premier système est composé de Fe(III)-AsO4-SO4 et les phases cristallines qui sont produites a haute température (150-225°C) ; celle-ci a été étudié pour élucider l'information précédente contradictoire sur ce système par rapport aux produits arsenicaux industriels qui sont formés pendant le traitement en autoclave de minerais aurifère arsenicaux sulfuré et de la stabilité des produits arsenicaux. Le deuxième

système étudié était de Cu(II)-Mg(II)-Fe(III)-AsO4-SO4 à 150°C. Ce système a été étudié due a son importance dan la lixiviation industrielle du cuivre et afin d’examiner l’effet du cuivre sur la précipitation de scorodite. Finalement, l'examen structural et moléculaire de

deux membres de la famille des systèmes Ca(II)-Fe(III)-AsO4 et yukonite (synthétique, naturel et arseniosiderite), ont étés observés en raison de leur nature relativement inconnue et de leur potentiel pour contrôler le dégagement d’arsenic dans les résidus. Dans le premier cas, trois phases d'arsenate se sont avérées a être produites à diverses

conditions; ceux-ci étaient : scorodite contenant de la sulfate (Fe(AsO4)1-0.67x(SO4)X ·

2H2O ou 0.00≤x≤0.20), de l’arsenate de fer(III) sub-hydraté ; FAsH (Fe

(AsO4)0.998(SO4)0.01 · 0.72H2O), et de la sulfate de fer d’arsenate basique ; BFAS

(Fe(AsO4)1 x(SO4)X(OH)X · (1-x)H2O, ou 0.3

iii forme de fer(III)-cuprique de sulfate d'arsenate basique dans ce système à des durées plus courtes (<1hr) et à de plus basses températures (< 150°C) ; avant de se transformer finalement en phase plus stable (scorodite). Les résultats de Therse indiquent que le cuivre pourrait co-précipiter à des durées de rétention plus courte donc l’importance kinétique pour la formation de scorodite. Dernièrement, le troisième système de yukonite synthétique s’est avéré équivalent au niveau atomique, moléculaire et structural au yukonite normal. Au niveau moléculaire, l'arseniosiderite s'est avéré a avoir un environnement de liaison-H comme dans le système de scorodite et comprends des groupes d’arsenate protonés. Dans le yukonite, on a observé un environnement étendu de liaison-H en présence seulement des groupes d'arséniate non-protonés. Les niveaux électroniques du yukonite et de l'arseniosiderite se sont avérés identique, indiquant que la nature locale des atomes de As, Fe et de Ca dans ces phases étroitement liées mais distinctes est identique. L'analyse structurale des matériaux a prouvé que le yukonite se compose de structures nano et médiocrement cristallines tandis que dans les systemes d'arseniosiderite des cristaux simple de taille micro existent.

iv Acknowledgments

I would like to first thank my official two supervisors: Professor George P. Demopoulos (Materials Engineering, McGill University) and Dr. Jeffrey N. Cutler (Canadian Light Source Inc) for their support, advice, patience and freedom to explore what I needed to in the pursuit of research. They offered me a chance to learn, study, use and combine two different disciplines together, i.e. the hydrometallurgical engineering field that produces the materials and the chemically based lab and synchrotron based analysis techniques that analyse the materials.

Support for the work described in this thesis was received via a NSERC strategic project grant co-sponsored by Areva Resources, Barrick Gold, Cameco, Hatch and Teck Metals for which the author is thankful.

In addition I would also like to thank some unofficial supervisors/colleagues that have been instrumental to guidance and direction of the author in the pursuit of research and learning. The first being Dr. L. Becze (Materials Engineering, McGill University), who reminded me and showed me how to work in a wet lab (particularly a hydrometallurgical lab) and was also always there to question the results gathered, something we had long discussions over in many projects but resulted in fruitful collaborations and numerous publications during the period of my time at McGill University. I would also like to thank Dr. Hassane Assaoudi (Chemistry Department, McGill University) who was the only person in Montréal when I came in 2007, that knew and could teach me how to do the correlation group analysis of vibrational spectra that also resulted in fruitful collaborations and publications together. I would also in this regard like to thank Dr. Samir Eluatik (Chemistry Department, University of Montréal) for showing me how to operate and provide the use of a state of the art Raman Microscope with 4 different wavelengths in addition to various vibrational tools.

I would also like to acknowledge the different technicians and people that assisted me in this research presented herein. 1st Slavek Poplawski for kindly doing XRD analysis for me when I did not have time. Xiu Dong Liu for taking of TEM pictures, Dr. J. F. Le

v Berre and Mert Celikin for taking SEM pictures. Ranjan Roy and Andrew Golsztajn for use of ICP-AES. Dr. J. Warner and Dr. N. Chen for help with making EXAFS samples and HXMA beam line use as well as Tom Regier, and Robert Blyth at the SGM line at the CLS. Dr. John Dutrizac and the late Dr. J. L Jambor are also kindly acknowledged for proving me with the Phase 3 and 4 materials, as well as useful discussion on the matter.

I would also like to acknowledge some people along the way in my time at McGill that helped and encouraged the learning-curiosity of the current topic or research area I was investigating at the time for the work presented in thesis. These are Prof. Robert Martin (Earth and Planetary Sciences) for introducing me to crystal chemistry of minerals; Prof. D. S. Bohle (Chemistry department) for showing me the basics of crystallography and for trying to get single crystal diffraction analysis of my compounds no matter how many times it did not work and Prof. Mark Sutton for letting me audit his solid state physics class, but also discussing some of my conflicts with some of the existing data of my research which I did not agree with .

I would also like to acknowledge my colleagues from the hydrometallurgical group (Kee Eun, Amandine, Levente, Guobin, Karl, Derek, Renaud, and Jean Christoph for french translation) and the materials engineering department have made my time at McGill University quite interesting and very different from the chemistry department from which I came from in Saskatchewan. Also Marina Lizon, thank you.

Finally, I would like to thank my mom for her values, support and encouragement, without these I could not have completed this task nor have succeeded in my life. My math and chemistry undergraduate degrees I dedicated to my deceased father and grandfather for introducing me to these concepts at my infant stage. However, my PhD thesis which is the “mature” part of my academic work, I dedicate to my mother for all her efforts and struggles as a single mother raising a kid, living in three different countries (Guatemala, Mexico and Canada). I would also like to thank my family (in the USA and Guatemala) and my friends (Saskatoon and Montreal) for all their support and encouragement.

vi Contributions of Author and Co-Authors of the Published Work Presented in This Thesis

Gomez M. A., Becze L., Cutler J. N. and Demopoulos G. P. (2010a) On the hydrothermal reaction chemistry and characterization of ferric arsenate phases precipitated from

Fe2(SO4)3-As2O5-H2SO4 solutions. Hydrometallurgy. (Accepted January 2011)

Gomez M. A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G. P. (2010b) Vibrational

spectroscopy study of hydrothermally produced scorodite (FeAsO4·2H2O), ferric arsenate

sub-hydrate (FAsH; FeAsO4·0.75H2O) and basic ferric arsenate sulfate (BFAS;

Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O ). J. Raman Spectros. 41, 212-221.

Gomez M. A., Becze L., Celikin M. and Demopoulos G. P. (2010c) The effect of copper on the

formation of scorodite (FeAsO4·2H2O) from aqueous hydrothermal conditions: Evidence of a hydrated ferric cupric arsenate-sulfate short lived intermediate. Inorg. Chem. (In review)

Gomez M. A., Becze L., Blyth R.I.R., Cutler J. N. and Demopoulos G.P. (2010d) Molecular and structural investigation of yukonite (synthetic & natural) and its relation to arseniosiderite. Geochimica et Cosmochimica Acta. 74, 5835-5851

In all these works (except for yukonite), the author has produced the materials under investigation, performed all the solid and solution analysis, in addition to their detailed characterization and their arsenic stability measurements. Yukonite is the only instance in which the product was not synthesized by the current author; rather this was conducted by Dr. L. Becze who provided the synthetic material (natural samples were all collected by the current author), chemical and arsenic stability analysis. Dr. R. I. R. Bllyth kindly conducted some of the XANES measurements. In terms of analysis, the SEM and TEM images were taken by technicians (Xu Dong Li) in some cases and colleagues in others (J. F. Le Berre and M. Celikin); however, the interpretations, analysis and writing of this and all other information was conducted by the first author.

vii In addition to the normal supervisory role of the official supervisors (G. P. Demopoulos and J N. Cutler), guidance and direction was given to the current author by Dr. H. Assaaoudi and Dr. R.I.R Blyth on issues of vibrational spectroscopy and X-ray absorption spectroscopy but again all work was conducted by the current author.

All signatures and copyright forms have been given as separate files to McGill University and not included in this thesis. GPD______and JNC______MAG______LB______MC______RRB______HA______

viii Layout of Thesis “Prologue”

This thesis was prepared in a non-traditional fashion on the basis of manuscripts. The four original research papers have been slightly modified to fit the purposes of this thesis. The material in Chapter 3 was published first as a peer reviewed conference paper (Gomez et al., 2008) and later accepted to an international peer review journal (Gomez et al., 2010a). The content of Chapter 4 was published and is available to the public (Gomez 2010b) while that found in Chapter 5 (Gomez et al., 2010c) is under review. Finally, the content in Chapter 6 (Gomez et al., 2010d) has also been already published and is available to the public. In here we give a layout of the thesis and provide the connection and relations between each of the chapters.

The first chapter of the thesis gives a brief background on the overall problem which is investigated herein, namely arsenic and its form once the value metal has been extracted.

The second chapter of the thesis consists of a detailed literature review on the relevant chemical systems under interest here; this is done to provide the reader with the relevant background information and concepts of the work(s) that have been previously conducted and how these have been changed or improved with the research presented herein. In addition to this, in the literature review I have chosen to give a brief background information on the interaction of light (EMR) with matter as spectroscopic tools are intensly used in this work, in addition to a brief and simple treatment (explanation) of the phenomena that occur in Vibrational (IR and Raman) and Electronic (XANES) spectroscopy.

The third chapter of this thesis is really the core of the thesis. In here we show what controls (“experimental parameters”) the domain of formation for the phases formed in

the higher temperature (150-225°C) Fe(III)-AsO4-SO4 hydrothermal system and identify what these phases are at the elemental, electronic, molecular and structural level using a variety of lab and synchrotron based analysis techniques. The short and long term arsenic stability is then evaluated for the phases produced at various conditions to determine

ix which may be safe to use for the disposal of arsenic. The results obtained are then discussed in relevance to the previous works and thoughts on this system to further bring new insights into the complex chemistry that dominates. A detailed molecular account via vibrational spectroscopy of the phases produced in the previous chapter is then given in the fourth chapter. This was done to further advance the understanding of the molecular characteristics and properties of the ferric arsenate phases produced in the

Fe(III)-AsO4-SO4 system, demonstrate the value of the obtained spectroscopic data in characterizing the form(s) of arsenate in the ideal system but also in an industrial product (donated by Barrick Gold) which was produced under similar conditions.

Therefore once a good handle has been attained on the domain of formation and the types of phases that may be formed in the Fe(III)-AsO4-SO4 system, the effect of co-ion on the system is investigated in two parts (one at higher and and one at lower temperature):

The first part pertains to chapter five in which the effect of co-ions (Cu2+ and Mg2+) upon the formation of the phase(s) is studied at high temperatures (150 °C). Of particular interest here, is in what form does the arsenic phases occurs in; are the phases produced in presence of foreign cations related to the previous phases found (chapter 3 and 4) and is there any precious metal loss via copper or magnesium co-precipitates (either in an arsenate or sulfate form). After these results are evaluated, an extension to correlate the real products observed in the industry with the lab based phases is investigated via the analysis of an industrial product produced under CESL conditions (donated by Teck Metals). It should be noted here that only one temperature range was investigated in this portion since this is the temperature used industrially in the CESL process.

The second part in chapter six deals with the effect of Ca2+ upon the formation of the

arsenic phase(s) produced in the Fe(III)-AsO4-SO4 system at lower temperatures (95°C). This research was undertaken for three main reasons: (a) as a result of the evidence that

Ca(II)-Fe(III)-AsO4 phases (Yukonite and Arseniosiderite) were some of the main arsenic phases occurring in the Gold mining tailing operations found in Yukon territory as

x previously described by Paktunc et al., (2003 and 2004) as well as recent report of these phases found in Nova Scotia, Canada by Walker et al. (2005 and 2009) and those in the Mokrsko-west gold deposits, Czech Republic (Filippi et al., 2004; Filippi et al., 2007; Drahota et al., 2009; Drahota and Filippi 2009). As well as (b) parallel studies conducted

in our research group, in which the presence of Ca(II)-Fe(III)-AsO4 phases (Jia and Demopoulos, 2008; Bluteau et al., 2009) was detected which was derived from one of the phase (scorodite) which was produced in the third chapter and (c) evidence which also

showed that one of these synthetic Ca(II)-Fe(III)-AsO4 produced phases (yukonite) resulted in excellent arsenic retention properties (Becze and Demopoulos, 2007; Becze et al., 2010) under similar industrial conditions used in the Uranium industry disposal sites.

In spite of these studies and findings of these Ca(II)-Fe(III)-AsO4 phases their true nature and how these are related was still not yet well understood. Therefore, chapter six deals with the production of the synthetic yukonite and comparison with natural specimens from distinct locations around the world and how these compare at the elemental, electronic, molecular and structural level. Moreover, it is shown how yukonite is related to its almost identical relative (arseniosiderite-a natural sample) something which has been previously debated and in this study we give further insights into this system.

Lastly in chapter 7, overall conclusions of the studies on the various systems here are given, in addition to what new knowledge this work has provided for the relevant academic and industrial research fields and future directions of work in this area.

xi References

Becze L. and Demopoulos G.P. (2007) Hydrometallurgical synthesis, characterization

and stability of Ca-Fe-AsO4 Compounds. In Extraction and Processing Proceedings, B. Davis and M. Free, Eds., TMS, Warrendale, PA, pp. 11-17. Becze L., Gomez M. A., Petkov V., Cutler J. N. and Demopoulos G. P. (2010) The

Potential Arsenic Role of Ca-Fe(III)-AsO4 Compounds in Lime Neutralized Co- Precipitation Tailings, In proceedings of Uranium 2010 and 40th Annual Hydrometallurgy Meeting, Uranium Processing -Tailings Section, CIM-MET SOC, Saskatoon, SK, Canada,Vol II, 327-336. Bluteau M-C., Becze L. and Demopoulos G.P. (2009) The dissolution of scorodite in

gypsum-saturated waters: Evidence of Ca-Fe-AsO4 mineral formation and its impact on arsenic retention. Hydrometallurgy 97, 221-227. Drahota, P., Rohovec J., Filippi M., Mihaljevic M., Rychlovsky P., Cerveny V. and Pertold Z. (2009) Mineralogical and geochemical controls of arsenic speciation and mobility under different redox conditions in soli, sediment and water at the Mokrsko-West gold deposit, Czech Republic. Sci. Total. Environ. 407, 3372- 3384. Drahota, P. and Filippi M. (2009) Secondary arsenic minerals in the environment: A review. Environment International. 35, 1243-1255. Filippi M., Golias V., and Pertold Z. (2004) Arsenic in contaminated soils and anthropogenic deposits at the Mokrsko, Roudny, and Kasperske Hory gold deposits. Environ. Geol. 45, 716-730. Filippi M., Dousova B., and Machovic V. (2007) Mineralogical speciation of arsenic in soils above the Mokrsko-west gold deposits, Czech Republic. Geoderma. 139, 154-170. Gomez M. A., Becze L., Bluteau M. C., Le Berre J. F., Cutler J. N. and Demopoulos G.

xii Gomez M. A., Becze L., Cutler J. N. and Demopoulos G. P. (2010a) On the hydrothermal reaction chemistry and characterization of ferric arsenate phases precipitated from

Fe2(SO4)-As2O5-H2SO4 solutions. Hydrometallurgy. (Accepted) Gomez M. A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G. P. (2010b) Vibrational spectroscopy study of hydrothermally produced scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) and basic

ferric arsenate sulfate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O ). J. Raman Spectros. 41, 212-221. Gomez M. A., Becze L., Celikin M. and Demopoulos G. P. (2010c) The effect of copper

on the formation of scorodite (FeAsO4·2H2O) from aqueous hydrothermal conditions: Evidence of a hydrated ferric cupric arsenate-sulfate short lived intermediate. Inorg. Chem. (In review) Gomez M. A., Becze L., Blyth R.I.R., Cutler J. N. and Demopoulos G.P. (2010d) Molecular and structural investigation of yukonite (synthetic & natural) and its relation to arseniosiderite. Geochimica et Cosmochimica Acta. 74, 5835-5851 Paktunc D., Foster A. and Laflamme G. (2003) Speciation and characterization of arsenic in Ketza river mine tailings using X-ray absorption spectroscopy. Environ. Sci. Techn. 37, 2067-2074. Paktunc D., Foster A., Heald S. and Laflamme G. (2004) Speciation and characterization of arsenic in gold ores and cyanidation tailings using X-ray absorption spectroscopy. Geochim. Cosmochim. Acta 68, 969-983. Walker S. R., Jamieson H. E., Lanzirotti A., Andrade C. F. and Hall G. E. M. (2005) The speciation of arsenic in iron oxides in mine wastes from the giant gold mine, N.W.T.: application of synchrotron micro-XRD and micro-XANES at the grain scale. Can. Mineral. 43, 1205-1224 Walker S. R., Parsons M. B., Jamieson H. E. and Lanzirotti A. (2009) Arsenic mineralogy of near-surface tailings and soils: Influences on arsenic mobility and bioaccessibility in the Nova Scotia gold mining districts. Can. Mineral. 47, 533- 556.

xiii Table of Contents

Abstract...... i

Resume...... iii

Acknowledgments ...... v

Contributions of Author and Co-Authors of the Published Work Presented in

This Thesis...... vii

Layout of Thesis “Prologue”...... ix

References...... xii Table of Contents...... xiv

List of Figures...... xviii

List of Tables ...... xxv

List of Acronyms...... xxviii

1. Introduction...... 1

1.1 Background ...... 1 1.2 Research objectives and rationale...... 3 2. Literature Review ...... 4

2.1 Natural sources of arsenic and chemical properties...... 4 2.2 Previous and recent research conducted on the high temperature (150-225 °C) Fe

(III) - As (V) - SO4 system...... 6 2.3 Previous research on the effect of co-ions on scorodite precipitation ...... 17

2.4 Previous research conducted on the Ca (II) - Fe (III) - AsO4 system ...... 19 2.5 Interaction of Electromagnetic Radiation with Matter ...... 26 2.6 Theory of Vibrational (Infrared and Raman) Spectroscopies...... 28 2.7 Theory of Near Edge X-ray Absorption Fine Structure Spectroscopy (NEXAFS) ...... 32 2.8 References...... 37

xiv

3. The Hydrothermal Fe (III)-AsO4-SO4 system at 150-225°C...... 44

3.1 Abstract...... 44 3.2 Introduction...... 45 3.3 Experimental Methods ...... 47 3.4 Results and Discussions...... 49 3.4.1 Precipitation Chemistry ...... 50 3.4.1.1 Precipitation of Scorodite...... 50 3.4.1.2 Precipitation of Ferric Arsenate sub-Hydrate (FAsH)...... 55 3.4.1.3 Precipitation of Basic Ferric Arsenate Sulfate (BFAS)...... 57 3.4.1.4 Precipitation of arsenate-bearing Basic Ferric Sulfate...... 62 3.4.1.5 Precipitation Diagram...... 63 3.4.2 Characterization ...... 65 3.4.2.1 TGA, XRD, ATR-IR, Raman...... 65 3.4.2.2 Crystalline Particle Morphology ...... 82

3.4.1.3 Fe L2,3 NEXAFS ...... 84 3.4.3 Short and Long Term Leachability Response...... 88 3.4.3.1 Short term arsenic release ...... 88 3.4.3.2 Long term arsenic release...... 89 3.5 Summary and Conclusions ...... 92 3.6 References...... 93

4. Vibrational Spectroscopic study of hydrothermally produced Fe (III)-AsO4-SO4 phases ...... 98

4.1 Abstract...... 98 4.2 Introduction...... 99 4.3 Experimental Methods ...... 100 4.4 Results and Discussions...... 102

4.4.1 Scorodite (FeAsO4·2H2O)...... 103

4.4.1.1 Vibrations of OH units and H2O molecules...... 105 3- 4.4.1.2 Arsenate (AsO4 ) stretching and bending modes ...... 108

xv 4.4.2 Ferric Arsenate sub-Hydrate (FAsH; FeAsO4·0.75H2O)...... 109

4.4.2.1 Vibrations of OH units and H2O molecules...... 111 3- 4.4.2.2 Arsenate (AsO4 ) stretching and bending modes ...... 113

4.4.3 Basic Ferric Arsenate Sulfate (BFAS; Fe[AsO4]1-x(SO4)x(OH)x·wH2O]) ..... 114 4.4.3.1 Molecular solid solution behavior...... 117 4.4.4 Characterization of an Industrial Arsenate-Containing Residue...... 120 4.5 Summary and Conclusions ...... 123 4.6 References...... 124

5. The Hydrothermal Cu (II)-Fe(III)-AsO4-SO4 System at 150°C ...... 128

5.1 Abstract...... 128 5.2 Introduction...... 129 5.3 Experimental Methods ...... 131 5.4 Results and Discussions...... 133 5.4.1 Precipitation of Scorodite...... 133 5.4.1.1 Equilibrium results...... 133 5.4.1.2 Kinetic results ...... 135 5.4.1.3 Characterization of Scorodite...... 138

5.4.2 Formation of a Cu(II)-Fe(III)-AsO4-SO4 intermediate ...... 143 5.4.2.1 Solution Changes ...... 143

5.4.2.2 Characterization of Cu(II)-Fe(III)-AsO4-SO4 intermediate...... 145 5.4.3 Characterization of a copper pressure leach residue...... 156 5.5 Summary and Conclusions ...... 162 5.6 References...... 163

6. The Hydrothermal Ca (II)-Fe (III)-AsO4 System at 95°C ...... 168

6.1 Abstract...... 168 6.2 Introduction...... 169 6.3 Experimental Methods ...... 171 6.4 Results and Discussions...... 174 6.4.1 Chemical Composition...... 174 6.4.2 X-ray Diffraction and SEM analysis...... 177

xvi 6.4.3 ATR-IR Spectroscopy...... 179 6.4.4 TEM and Electron Diffraction ...... 186 6.4.5 Raman Spectroscopy...... 189 6.4.6 X-ray Absorption Analysis ...... 192 6.5 Summary and Conclusions ...... 198 6.6 References...... 199 7. Conclusions and Perspectives ...... 207

7.1 Conclusions and Contributions to Original Knowledge ...... 207 7.2 Future Directions ...... 210 8. Appendix...... 212

8.1 Additional material from chapters ...... 212 8.1.1 (Chapter 3)...... 212 8.1.2 (Chapter 4)...... 214 8.1.3 (Chapter 5)...... 217 8.1.4 (Chapter 6)...... 222 8.2 Additional details on methods used ...... 224 8.3 Complete list of publications, conference proceedings and projects that have been produced during the coarse of PhD studies...... 229

xvii List of Figures

Chapter 2

Figure 1. Phase diagram for the Fe (III)-AsO4-SO4 system. (Swash and Monhemius, 1994)…………………………………………………………………………………..pg 6

Figure 2. Arsenic ppt curves for the 150 and 190 °C tests (heating solution from ambient temperature) (Swash 1996)………………………………………………………...…..pg 8

Figure 3. Arsenic ppt curves for the 150 and 190 °C tests (injecting iron solutions at elevated temperatures) (Swash, 1996)….………………………………………….....pg 10

Figure 4. Arsenic removal from solution by precipitated hematite at 190 °C (Swash, 1996)……………………………………………………………………………….....pg 11

Figure 5. Arsenic precipitation curves at (a)150 °C and (b)190 °C (Monhemius and Swash, 1999)………………………………………………………………….………pg 12

Figure 6. Iron precipitation curves at 150 °C via (a) heating solution from ambient temperature (Swash and Monhemius, 1996) and (b) injecting iron solutions at the target temperature (Monhemius and Swash, 1999)……………………………...………….pg 13

Figure 7. Basic Ferric Sulfate MDO’s polytpes (left) orthombic and (right) monoclinic. The octahedra are the iron units and tetrahedra the sulfate. (Ventruti et al., 2005)……………………………………………………………………………….....pg 15

Figure 8. Yukonite mieral from Tagish Lake, Canada ……………………..…….....pg 21

III -12 Figure 9. Basic structural units of mitridatite: (left) sheets of [Fe 9O6(AsO4)9] and (right) Ca2O10(H2O)2 dimers. (Moore and Ito, 1977a) …………………………….....pg 24

Figure 10. Wavelength and Frequency representation of Electromagnetic Radiation……………………………………………………………………..…...…..pg 26

Figure 11. Energy level schematic of the vibrational transition processes that are observed in vibrational spectroscopy (Coates, 1998)………………………...……....pg 31

Figure 12. Energy level schematic of the electronic transition processes that are observed in (a) X-ray photoelectron spectroscopy and (b) X-ray Absorption Spectroscopy (Nilsson 2002)……………………………………………………………………………...…..pg 32

Figure 13. Energy level schematic of the electronic transition processes that are observed during a Near Edge X-ray Absorption Fine Structure experiment. (Watts et al., 2006)……………………………………………………………………………...…..pg 33

xviii Figure 14. The various regions of an X-ray Absorption Spectrum. (George, 1998; George 2006)……………………………………………………………………………...…..pg 34

Figure 15. Scheme of interaction of a photoelectron with the atoms of the nearest environment (A is the atom absorbing an X-ray photon and B is a neighbouring atom): (a) the energy-level diagram of electrons in a crystal lattice at different excitation energies corresponding to single-scattering (EXAFS) and multiple scattering (XANES) processes. (b) The emerging wave corresponding to a free electron and the interference between the emerging and scattered waves as well as (c) the energy dependence of X-ray absorption in the absence of scattering from neighbouring atoms and in the presence of scattering.(Aksenov et al., 2006)………………….……………………………...…..pg 35

Chapter 3

Figure 1. (a) Plot of Fe/As molar ratio in initial solution against molar ratio of precipitated As/Fe molar ration in the case of scorodite precipitation (data from Table 1), (b) plot of Fe/As molar ratio in initial solution against molar ratio of precipitated As/Fe in the case of FAsH precipitation (data from Table 3) and (c) plot of (Fe/As) solid against (Fe/As)ppt for BFAS precipitation (data from Table 5)…………………………..pg 52-54

Figure 2. The Gomez - Becze - Demopoulos (“GBD”) Phase Diagram of the arsenate phases found in the Fe (III) - AsO4 - SO4 system……………………………….…....pg 63

Figure 3. TGA (left) and XRD (right) analysis of Scorodite, Ferric Arsenate sub-hydrates (FAsH), Basic Ferric Arsenate Sulfate (BFAS).………………………..………...…..pg 65

Figure 4. XRD patterns obtained for experimental scorodite and reference JCPD file (005-0216)………………………………………………………………………...…..pg 66

Figure 5. (a) XRD of FAsH, Phase 4 and reference Type 1……………………...… pg 67 (b) Raman of FAsH and Phase 4…………………………………………………...... pg 67 (c) ATR-IR of FAsH and Phase 4…………………………………….……………....pg 68

Figure 6. (a) XRD of BFAS, arsenate containing-BFAS, Phase 3, reference Type 2 and BFS……………………………………………………………………………………pg73 68 (b) Raman of BFAS and Phase 3……………………………………………………..pg73 (c) ATR-IR of BFAS and Phase 3……...... pg 74

Figure 7. (a) XRD of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3)…..pg 76 (b) Raman of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3)…………....pg 76 (c) ATR-IR of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3)………...... pg 77

Figure 8. (a) XRD of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3)…....pg78 (b) Raman of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3)…...……….pg 79 (c) ATR-IR of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3)…………..pg 79

xix

Figure 9. (a) XRD of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3)………pg80 (b) Raman of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3)…………….…pg81 (c) ATR-IR of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3)………….…..pg 81

Figure 10. FEG-SEM images of Scorodite (Sc-4) at (a) 5x and (b) 20x magnification ……………………………………………………………………………………...... pg 82

Figure 11. FEG-SEM images of FAsH (Fs-3) at (a) 5x and (b) 20x magnification ……………………………………………………………………………………...... pg 83

Figure 12. FEG-SEM images of BFAS (Ba-5) at (a) 5x (b) 10x showing recrystallization of the rounded particle to the monoclinic-orthorhombic like crystal and (c) 20x magnification showing the full monoclinic-orthorhombic like crystal ………….pg 83

Figure 13. Fe L-edge XANES of Scorodite, FAsH, BFAS, α-Fe2O3 and FeSO4·7H2O………………………………………………………………………...... pg 85

Figure 14. TCLP-like sequential test on Scorodite, FAsH and BFAS………………....pg 88

Figure 15. Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 3……………………………….………………pg90 Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 5…………………………………………………………pg90 Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 7.5…………...……..…………………………………...pg 91

Chapter 4

Figure 1. ATR-IR (top) and Raman (bottom) spectra of synthetic scorodite……....pg 105

III Figure 2. Molecular cluster of isostructural orthorhombic M AsO4·2H2O derivatives of Scorodite. The dashed lines indicate the hydrogen bonds that occur between tetrahedral groups and that of the metal water molecules. Numbers besides the bonds are the bond lengths (Å) ………………………………………………………………………….pg 106

Figure 3. ATR-IR (top) and Raman (bottom) spectra of FAsH …………………....pg 110

Figure 4. Molecular cluster of isostructural orthorhombic FeAsO4·3/4H2O derivatives of Scorodite. The dashed lines indicate the hydrogen bonds that occur between tetrahedral groups and that of the metal water molecules……………………………………...pg 111

Figure 5. ATR-IR (top) and Raman (bottom) spectra of BFAS…………………....pg 115

Figure 6. Raman spectra of three BFAS samples with various arsenate and sulfate contents. All spectra have been normalized to one for comparison………………...pg 117

xx Figure 7. ATR-IR spectra of BFAS phases with various arsenate and sulfate contents. The spectra above (samples 1-3) correspond to BFAS phases with higher arsenate content, and the dashed spectra below (samples 4-6) for solids with higher sulfate content…………………………………………………………………………..…...pg 118

Figure 8. The ATR-IR spectra (arsenate and sulfate υ3 antisymmetric stretching region) of an arsenate containing residue and of the different synthetic high temperature Fe(III) - AsO4 -SO4 phases. The spectra have been set to a relative scale for easier comparison………………………………………………………………..…………pg 121

Figure 9. XRD spectra of the scorodite, FAsH, BFAS and of an arsenate containing industrial residue………………………………………………………………….....pg 122

Chapter 5

Figure 1. Concentration profiles as a function of reaction time for the Case 1 (a), Case 2 (b), Case 3 (c) and Case 5 (d). The reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached………………………………………….……………………..pg 136-137

Figure 2. XRD of the solids produced at 150 °C once the target temperature was reached at (left) 2.5 hrs and (right) at the end of the reaction period (12.5 hrs). All were found to be scorodite …………………………………………………..……………………..pg 140

Figure 3. ATR-IR spectra of the solids produced at 150 °C once the target temperature was reached at (left) 2.5 hrs and (right) at the end of the reaction period (12.5 hrs). All were found to be scorodite …………………………………..….………………….pg 140

Figure 4. SEM micrographs of scorodite particles of Case 1(a,f), Case 2 (b,g), Case 3 (c,h), Case 4 (d,i) and Case 5 (e,j) at 2.5hrs (top) and 12.5 hrs (bottom) respectively. The scale of the images is 5 μm using x10K magnification. The reaction time of 2.5 hrs corresponds to the time the target temperature of 150 °C was reached and 12.5 hrs at the end of the reaction. The inset in Figures b and i, display a graphic description of what an ideal orthorhombic-dipyramidal crystal looks like superficially ……………….…..pg 141

Figure 5. Concentration profiles as a function of reaction time for Case 4 including the heat up period from 25 °C to 150 °C; the reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached ………………………………………………..pg 143

Figure 6. Photographs of products formed at (a) 30 min, 90 °C; (b) 40 min, 101 °C ; (c) 60 min, 125 °C and (d) 96 min, 135 °C during the heat up period of Case 4 test…..pg 146

Figure 7. ATR-IR spectra of intermediate formed during the heat up period from 25 to 135 °C of Case 4 test. All Spectra have been offset vertically for easier comparison. Reference spectra for scorodite and BFAS prepared via hydrothermal synthesis at McGill

xxi (Gomez et al., 2008; Gomez et al., 2010c) are also shown for easier comparison ………………………………………………………………………………...……..pg 147

Figure 8. Raman spectra of intermediate formed during the heat up period from 25 to 135 °C of Case 4 test. All Spectra have been offset vertically for easier comparison. Reference spectra for scorodite and BFAS prepared via hydrothermal synthesis at McGill (Gomez et al., 2008; Gomez et al., 2010c) are also shown for easier comparison ……………………………………………………………………………………....pg 149

Figure 9. Polarized microscopy images of the intermediate gelatinous product obtained at 40 min and 101 °C. The scale on the images is 100 μm …………………………...pg 150

Figure 10. Selected area XRD of the intermediate gel-type of product produced at 40 minutes and 101 °C in two distinct spots on the sample as well as the powder XRD pattern of Case 4 at 2.5 hrs (150 °C) and that of a reference scorodite prepared by hydrothermal synthesis. (Gomez et al., 2008; Gomez et al., 2010c) ……………………………………………………………………………………..pg 151

Figure 11. TEM, EDS and SAED of the gel material produced at 40 min and 101 °C. The SAED was taken in different locations on the sample showing both an amorphous and a polycrystalline type of material ……………………………………………....pg 152 . Figure 12. TEM of the semi-crystalline component of the gel product (40 min and 101 °C) showing the formation of ordered atomic planes from atoms in the unordered phase to the lattice fringes. All lattice fringes that were observed were measured to be 3.27 Å, which are close to the lattice spacing observed in BFAS/Type2/Phase 3 (3.25Å).( Swash, 1996; Dutrizac and Jambor, 2007; Gomez et al., 2010c) The corresponding lattice spacings in scorodite are usually 3.17 and 4.45 Å. (Swash, 1996; Dutrizac and Jambor, 2007 Gomez et al., 2010c) ………………………………………………….....pg 153-154

Figure 13. Picture of the as-received CESL pressure leaching arsenic-containing residue (left) and the same residue after the extraction of elemental sulfur with toluene (right)………………………………………………………………………………..pg 156

Figure 14. XRD (a), Raman (b) and ATR-IR (c) of the CESL industrial pressure leaching arsenic-containing residue before and after the extraction of elemental sulfur with toluene. Reference XRD, Raman and ATR-IR (hematite spectra) from Sigma-Aldrich are also shown for comparison. All samples have been offset vertically for easier comparison……………………………………………………………………..pg 157-158

Figure 15. Raman spectra (a) of reagent grade (Sigma-Aldrich) elemental sulfur and that of the extracted sulfur obtained from the Teck residue. Figure (b) shows the ATR-IR spectra of arsenate region before and after elemental sulfur (S°) extraction…………………………………………………………………………...pg 160

xxii Figure 16. ATR-IR spectra of the ν3(AsO4) mode, for the CESL industrial pressure leaching arsenic-containing residue compared to other relevant high temperature ferric arsenate phases.(Gomez et al., 2010a)……………………………………………....pg 161

Chapter 6

Figure 1. X-ray diffraction patterns of the synthetic and natural (Grottal della Monaca and Tagish Lake) yukonites and arseniosiderite (Romanech). The symbol Δ in the Grotta della Monaca sample (left) indicates the presence of scorodite impurity…………..pg 177

Figure 2. SEM images of synthetic and natural yukonites [(a) Synthetic (Syn-1); (b) Tagish Lake; (c) Grotta della Monaca] and arseniosiderite [(d) Romanech]. The scale on the images is 5 μm… …………………………………...pg 178

Figure 3. ATR-FTIR (left) and Micro-Raman (right) spectra of synthetic (Syn-1) and natural (Grotta della Monaca and Tagish Lake) yukonite and arseniosiderite (Romanech)………………………………………………………………………….pg 180

Figure 4. ATR-IR of arseniosiderite (Romanech) (left) in the arsenate and water stretch region and (right) the hydroxyl and/or water region at 25, 450 and 650 °C………...pg 183

Figure 5. ATR-IR of yukonite (Tagish Lake), (left) in the arsenate and water stretch region and (right) the hydroxyl and/or water region at 25, 450 and 650 °C………..pg 184

Figure 6. TEM and SAED images of natural yukonite (Tagish Lake) (a, b, e, f), and arseniosiderite (Romanech) (c, d)……………………………………………….…..pg 186

Figure 7. Micro-Raman spectra of natural yukonite (Tagish Lake) and arseniosiderite (Romanech) showing the arsenate modes. (On the bottom spectrum, the FWHM is also shown for the ν2 arsenate band)……………………………………………………..pg 190

Figure 8. As K-edge XANES (left) and EXAFS (right) spectra of synthetic (Syn-1) and natural (Tagish Lake) yukonite, arseniosiderite (Romanech), reagent grade As2O3 and As2O5………………………………………………………………………………..pg 193

Figure 9. Fe L edge (left) and Ca Ledge (right) XANES spectra of synthetic (Syn-1) and natural (Grotta della Monaca and Tagish Lake) yukonite, arseniosiderite (Romanech), reagent grade hematite and calcium hydroxide………………………………..pg 194-195

xxiii Appendix

Chapter 5

Figure S1. Raman spectra of the solids produced at 150 °C once the target temperature was reached at (left) 2.5 hrs and (right) at the end of the reaction period (12.5 hrs). All were found to be Scorodite. (Gomez et al., 2010a)…………………………………pg 217

Figure S2. SEM micrographs of scorodite particles of Case 1(a,f), Case 2 (b,h), Case 2’ (c,i), Case 3 (d,j), Case 4 (e,k) and Case 5 (f,l). All figures represent the particles at 2.5hrs (top) and 12.5 hrs (bottom) respectively, with the exception of figure for Case 2’ which is for experimental conditions of Fe/As 1 a 24 hr time and 160 °C and the inset display a graphic description of what ideal orthorhombic-dipyramidal crystal looks like. The scale of all the images is 5 μm using an x10K magnification; except for Case 2’ (c) and Case2’(i) which were taken x5K and x10 K (the scale on the images is 10 μm and 5 μm) respectively. The reaction time of 2.5 hrs corresponds to the time the target temperature of 150 °C was reached and 12.5 hrs at the end of the reaction………...pg 218

Figure S3. TEM and EDX of the semi-crystalline component of the gel product (this is a distinct particle then that shown in Fig 12). Lattice pacing was found to be 3.27 Å……………………………………………………………………………………..pg 219

Figure S4. Full scale TEM images of the gel product of three distinct particles showing the lattice fringes and the formation of the lattice fringes from disordered atoms to atomic planes and finally lattice fringes………………………………………………….....pg 219

Figure S5. (a) TEM images of the gel formed showing different orientation of lattice fringes (highlighted with red lines). The inset shows the SEAD of the polycrystalline component. Figure (b) displays a porous structure at the nanometer level of the gel material. Figure (c) and (d) are optical (visible light) microscopy images taken of the gel product in two different locations of the sample again depicting macroscopically the porous nature………………………………………………………………………...pg 220

Figure S6. Photograph of intermediate gel products after 7 days exposed to air.….pg 220

Figure S7. XRD of the Teck residue before (a-c) and after elemental sulfur (S°) extraction (d-f). Figures (a,d), (b,e) and (c,f) show the JCPDS match for hematite, elemental sulfur and scorodite. In all cases the confidence in the matches for the JCPDS database was ≤ 45%………………………………………………………………....pg 221

xxiv Chapter 6

Figure S1. X-ray analysis of natural and synthetic (Syn-1) yukonites at different X-ray wavelengths: (a) Powder x-ray diffraction using lab based (λ=1.54Ǻ) and synchrotron based radiation (λ=0.458Ǻ) and (b) lab based (λ=0.71Ǻ) single crystal diffraction on the Tagish Lake sample……………………………………………………………..…..pg 222

Figure S2. Micro-Raman spectra of Tagish Lake Yukonite, Romanech Arseniosiderite and relevant synthetic compounds (scorodite, ferric arsenate, arsenate-ferrihydrite, and ferrihydrite) prepared by synthetic methods developed at McGill (Singhania et al., 2005, 2006; Jia and Demopoulos, 2008; Gomez et al., 2009) that can be found in mine tailings disposal ponds……………………………………………………………………....pg 223

List of Tables

Chapter 2

Table 1. Arsenic mineral groups and examples of common minerals. (Smedley and Kinniburgh, 2002)……………………………………………………………….……..pg 4

Table 2. Typical arsenic concentrations in sulfide and oxide rock-forming minerals (Smedley and Kinniburgh, 2002)…………………………………………….….……..pg 5

Chapter 3

Table 1. Scorodite experimental and precipitation data…………………….……...... pg 50

Table 2. Chemical composition data for scorodite………………………….………..pg 51

Table 3. Ferric Arsenate sub-Hydrate experimental and precipitation data……...…..pg 56

Table 4. Ferric Arsenate sub-Hydrate chemical composition data………….……….pg 56

Table 5. Basic Ferric Arsenate Sulfate experimental and precipitation data………...pg 58

Table 6. Basic Ferric Arsenate Sulfate chemical composition data………….……....pg 59

Table 7. Empirical and theoretical formulae of the various BFAS products ……...... pg 60

Table 8. XRD of experimental FAsH compared to the XRD data of FeAsO4·3/4H2O (Jakeman et al.,1991)…………………………………………………………….…...pg 69

Table 9. Experimental FAsH and reference FeAsO4·3/4H2O lattice parameter data (Jakeman et al.,1991)…………………………………………………..…..…….…...pg 69

xxv Table 10. Comparison between XRD patterns obtained from the experimental FAsH product, the structure simulated with CaRIne and the reference FeAsO4·3/4H2O material (Jakeman et al.,1991)……………………………………………………..…….…….pg 70

Table 11. XRD data of experimental BFAS, Phase 3 (Dutrizac and Jambor, 2007), Arsenate-containing BFS and reference Type 2 (Swash and Monhemius, 1994)…....pg 72

Chapter 4

Table 1. Raman and IR Spectral assignment for Scorodite, FAsH and BFAS……..pg 104

Table 2. IR spectral assignments for BFAS samples with various sulfate and arsenate content…………….………………………………………………………………....pg 119

Chapter 5

Table 1. Initial solution compositions and precipitation efficiencies at the end of the reaction period (12.5 hrs).………………………………………………..…….…....pg 131

Table 2. Composition of precipitated solids at the end of the reaction period (12.5 hrs)……………….………………………………………………………...………..pg 134

Table 3. Analysis of solid/gel and solution during the heat up period and beyond of Case 4 experiment……………………………………………………………...……...…..pg 144

Chapter 6

Table 1. Initial solution compositions and precipitation efficiencies at the end of the reaction period (12.5 hrs)……………….…………………………………………...pg 175

Table 2. ATR-IR and Micro-Raman band assignment for yukonite and arseniosiderite……………………………………………………………….……....pg 181

Table 3. Hydrogen bond lengths calculated using the regression functions from Libowitzky (1999) and the ATR-IR hydroxyl stretching frequencies observed in this study…………………………………………………………………………...... …..pg 182

Table 4. Peak positions for the Ca-Ledge XANES analysis of the Ca-Fe-AsO4 phases (eV)……….……………………………………………………………………...….pg 197 .

xxvi Appendix

Chapter 3

Table S1. Experimental (this work) and reference Scorodite lattice parameter data…………………………………………………………………………………..pg 212

Table S2. Comparison between XRD patterns obtained from the experimental Scorodite product, the structure simulated with CaRIne and the reference data………………pg 213

Chapter 4

Table S1. Solid composition analyses for the various phases characterized ……....pg 214

Table S2. Spectral predictions and correlation schemes for the internal modes of H2O in Scorodite …………………………………………………………...…….…………pg 214

3- Table S3. Spectral predictions and correlation schemes for the internal modes of AsO4 in Scorodite………………………………………………………………………….pg 215

Table S4. Spectral predictions and correlation schemes for the internal modes of H2O in Ferric Arsenate sub-Hydrate………………………………………………………...pg 215

3- Table S5. Spectral predictions and correlation schemes for the internal modes of AsO4 in Ferric Arsenate sub-Hydrate……………………………………………………...pg 216

Chapter 5

Table S1. XRD (λ = 1.78897 Å) of semi-crystalline component of the green gel formed in Case 4 (40 minutes and 101 °C) shown in Figure 10…………………………….pg 217

xxvii List of Acronyms

ATR. Attenuated Total Reflection

BFAS. Basic Ferric Arsenate Sulfate

BFS. Basic Ferric Sulfate

CESL. Cominco Engineering Services Ltd

CLS. Canadian Light Source

EMR. Electromagnetic Radiation

EDX. Energy Dispersive X-ray

ED. Electron diffraction

EXAFS. Extended X-ray Fine Structure Spectroscopy eV. Electron Volt

FAsH. Ferric Arsenate sub-Hydrate

HXMA. Hard X-ray MicroAnalysis

H-bond. Hydrogen bond

IR. Infrared

NEXAFS. Near Edge X-ray Fine Structure Spectroscopy

PPT. Precipitation

SGM. Spherical Grating Monochromator

SEM. Scanning Electron Microscopy

TCLP. Toxicity Characterization Leachability Procedure

TEM. Transmission Electron Microscopy

XANES. X-ray Absoprtion Near Edge Spectroscopy

XRD. X-ray Diffraction

XPS. X-ray Photoelectron Spectroscopy

xxviii 1. Introduction

1.1 Background

Arsenic and its adverse effects on the environment and society impose a serious risk as it can be observed in cases such as India, Bangladesh, Vietnam, Nepal, and Cambodia. These problems arise from geochemical processes but man-made activities like mining and processing of minerals may further contribute to this risk. In both cases, problems and environmental disasters or accidents may be traced at least in part to the lack of chemical understanding of the arsenic solubility from natural or man-made solids. Much suffering has been endured from an economical and environmental point of view but more importantly from a sociological and human perspective, where villages in developing countries still must deal with and rely on ground water that is contaminated with arsenic. Although these issues have been mainly centered in developing countries, others such as US, Canada and Europe are not free from these threats, especially with the increasing demand (as world population increases) for recovery of valuable metals (Copper, Gold, Cobalt, Nickel and Uranium to name a few) from arsenic-containing ores as the processing of the latter generates arsenic-containing solid wastes that are disposed in tailings management facilities.

The Fe (III)-AsO4 waste solids produced from the hydrometallurgical processing of mineral feedstocks can be classified into two groups depending on their degree of crystallinity. Thus at ambient temperature typically, poorly crystalline Fe(III)-As(V) solids(Krause and Ettel, 1989; Langmuir et al., 1999; Jia and Demopoulos, 2005a; Jia and Demopoulos, 2005b) are produced by co-precipitation, which according to recent studies appear to consist of ferric arsenate and arsenate-adsorbed onto ferrihydrite. (Langmuir et al., 1999; Jia and Demopoulos, 2005a; Jia and Demopoulos, 2005b) These co-precipitates are produced from high Fe (III) to As(V) molar ratio solution (typically >3) by lime neutralization (De Klerk 2008). Today this method is still considered to be the most suitable method to treat low arsenic containing process effluent solutions.

In the case though of arsenic-rich and deficient in iron solutions an alternative to co-

precipitation is to produce crystalline scorodite (FeAsO4 · 2H2O). This can be done for example at elevated temperatures, near the boiling point of water (80-95 °C) and under controlled supersaturated conditions.(Singhania et al., 2005; Singhania et al., 2006; Demopoulos et al., 1995; Demopoulos, 2005; Fujita et al., 2008) Crystalline scorodite is

at least 100 times less soluble than its amorphous counterpart (FeAsO4 · xH2O[am]) (Krause and Ettel, 1989; Langmuir et al., 2006; Bluteau and Demopoulos, 2007; Le Berre et al., 2007) and given its high arsenic content (in comparison to the conventional Fe (III)-As(V) co-precipitates) is advocated currently for the fixation of arsenic-rich wastes. (Filippou and Demopoulos 1997; Demopoulos, 2003; Fujita et al., 2008) At even higher temperatures (>100 °C, the hydrothermal precipitation range) during autoclave processing of copper (Berezowsky et al., 1999) and gold (Dymov et al., 2004) sulphide feedstocks, other crystalline phases than scorodite are reported to form (Swash and Monhemius, 1994; Dutrizac and Jambor, 2007). These high-temperature ferric arsenate phases constitute indeed the major part of the subject matter of this thesis as their true identity, conditions/mechanisms of formation, and long term stability have not been unequivocally established or determined.

2 1.2 Research objectives and rationale

Given the importance of processing of arsenic bearing mineral feedstocks and the associated environmental concerns with the disposal of the produced (in-situ) arsenical waste solids, the present research project which this thesis is based on, seeks to advance our understanding of the chemistry of these solids. In particular this study is divided into

three industrially relevant systems: (1) the pure Fe(III)-AsO4-SO4 system related to the

Copper and Gold industry (2) the Cu(II)- Fe(III)-AsO4-SO4 system related to the Copper and (3) the Ca(II)-Fe(III)-AsO4-SO4 system related to the Gold and Uranium industry.

This research aims at answering the following questions: (a) what are the phases formed (at the elemental, electronic, molecular and structural level) under hydrothermal 2+ 2+ 2+ conditions from X(II)-Fe(III)-AsO4-SO4 solutions (X = Ca ,Cu or Mg ), (b) what are the process parameters (experimental conditions) favouring the formation of these phases, and (c) what is the short and long term arsenic stability of these phases.

To this end, a systematic series of hydrothermal experiments was carried out under a variety of conditions; followed by comprehensive characterization using a variety of techniques and an arsenic stability evaluation of the produced phases. In addition to investigating the pure systems, products produced under real industrial conditions were obtained and analyzed from our industrial partners (Barrick Gold and Teck Metals) to correlate our laboratory finding with that produced industrially.

3 2. Literature Review

2.1 Natural sources of arsenic and chemical properties

Arsenic exists as a major constituent in more then 200 minerals (Table 1); these minerals can occur as primary minerals (e.g. FeAsS) or their alteration products (e.g.

FeAsO4 · 2H2O).

Table 1. Arsenic mineral groups and examples of common minerals.

(Smedley and Kinniburgh, 2002)

These minerals are not common in natural environments with their greatest concentrations occurring in mineralised areas in close association with transition metals as well as Au, Sb, W, Mo, Ag, Pb and Cd. The most abundant arsenic mineral is arsenopyrite (FeAsS) which forms under high temperature conditions in the earth’s crust; similar occurrence is observed for realgar (AsS) and orpiment (As2S3). In addition arsenic is found in nature in varying concentrations (minor components) in other common rock-forming minerals such as silicates, carbonates, sulfates, phosphates, sulphides and oxides. Only the latter two will be discussed due to their relevance to industrial processes (Table 2). (Smedley and Kinniburgh, 2002)

4 Table 2. Typical arsenic concentrations in sulfide and oxide rock-forming minerals

(Smedley and Kinniburgh, 2002)

The incorporation of arsenic in sulfides is due to the fact that the chemistry of arsenic follows closely that of sulfur; pyrite and marcasite contain the greatest concentration of arsenic among the sulfide minerals. Generally arsenic acts as a substituent for sulfur, phosphate or vanadium in elemental isovalent solid substitution or through molecular substitution for groups such as phosphates, vanadates and sulfates. Significant high arsenic concentrations are found in many oxide minerals and hydrous metal oxides, either as sorbed species or as part of the mineral structure. Arsenic concentrations in iron oxides can reach significant values (Table 2) when they form as oxidation products of primary iron sulfide-arsenide minerals. Hydrous iron oxides have a high capacity for surface arsenic retention; it has been shown, through the use of extended X-ray absorption fine structure spectroscopy (EXAFS) and vibrational (IR and Raman) spectroscopy that in the case of these hydrated iron oxides, bidentate-binuclear bonding occurs between the arsenate species and the hydrated iron oxide surfaces.(Waychunas et al., 1993;Waychunas et al., 1995 ; Myneni et al., 1998)

5 2.2 Previous and recent research conducted on the high temperature (150-225 °C) Fe (III) - As (V) - SO4 system

Swash and Monhemius were the first to report on the synthesis and characterization

of Fe(III)-AsO4-SO4 compounds under hydrothermal autoclave conditions from the natural pH of the solutions (pH < 1).(Swash and Monhemius 1994; Swash 1996; Swash and Monhemius, 1998; Monhemius and Swash, 1999) For their first series of experiments (“test tube scale”), the solids were synthesized in an autoclave for 24 hrs at pH<1 from starting solutions having different Fe/As ratios which were placed in a glass test tube and heated in an autoclave at the temperature (150 to 225 °C) of interest.

Figure 1. Phase diagram for the Fe (III)-AsO4-SO4 system.

(Swash and Monhemius, 1994)

Initially, solutions of Fe(NO3)3.9H2O, Li2SO4.H2O and As2O5 were mixed in selected

proportions. Subsequently, solutions of Fe2(SO4)3.xH2O and As2O5 were instead used in order to avoid potential interference from lithium and nitrate ions. The latter solutions were also closer to real hydrometallurgical solutions. According to their work four

distinct crystalline phases were found to form: scorodite, FeAsO4 · 2H2O; basic iron

sulfate, FeOHSO4; Type 1, Fe2(HAsO4)3 · zH2O for z<4; and Type 2,

Fe4(AsO4)3(OH)x(SO4)y for x+2y=3. The formation of these phases was correlated to temperature (150 to 225 °C); Fe(III)/As(V) ratio (9:1 – 1:1 and 1:0) at a fixed retention time of 24 hours as shown by Figure 1 above. In all their experiments starting solutions

6 were prepared with variable Fe(III) concentration (introduced as 21-10.5g/L Fe3+) and 5+ fixed As(V) concentration (introduced as 13g/L As ) without the addition of H2SO4 to the starting solutions. Of these phases the ones labeled as Type 1 and Type 2 was not reported by any other investigators in previous reports. Characterization of these phases was mainly done via XRD and chemical analysis; although some IR was done no discussion of the results were reported and the poor quality of the data collected made it impossible to extract any useful information. Of the two phases, Type 2 ( < 0.34 mg/L As) was the only one found to meet the TCLP leachability criterion of 5 mg/L As (released after 24 hours of equilibration at pH 5) exhibiting similar behavior with that of scorodite (< 0.8 mg/L As). It must be however noted here that the TCLP leachability criterion has now been revised to 1 mg/L of As (Le Berre et al., 2007). In the dissertation thesis by Peter M. Swash (1996) more details on the experimental methods for the system discussed above (pH of starting solutions < 1) may be found. Test tube scale type of experiments were also conducted for the same system but this time the initial solution pH was adjusted to 5 prior to heating, this was done to investigate the type of solids, which may grow from gelatinous ferrihydrite sludges. In this case LiOH (aq) was used to increase the pH rather then NaOH or Ca(OH)2 as the use of the latter two would cause the precipitation of other types of compounds such as sodium jarosite

(KFe3(SO4)2(OH)6) which would complicate the identification and solubility testing of the solids. In this case ferric nitrate and lithium sulfate were used. In these experiments, the Type 2 structure was not observed and the Type 1 phase was found as a minor component. Instead a third unknown phase labeled “Type 3” which had the form

Fe2(AsO4)0.33(6-2x)(HAsO4)x where Fe(III)/As(V)~1:1 and x < 1 was observed to form from arsenic rich starting solution (Fe(III)As(V)<2) and temperatures > 125 °C. When iron levels increased, ferrihydrite (Michel et al., 2007) and hematite became more plentiful with minor amounts of Type 3. Using ferric sulfate as a source of iron and sulfate produced a similar set of phases as that with ferric nitrate and lithium sulfate. At lower temperatures (< 125°C) amorphous and poorly crystalline arsenical ferrihydrite (2- line ferrihydrite-(Michel et al., 2007) formed, while at higher temperatures and iron-rich solutions, hematite and more crystalline ferrihydrite (6 line; Michel et al., 2007) were formed. The Type 3 phase was only observed when Fe(III)/As(V) ratio was one and

7 225°C. At this temperature when the iron content was increased, arsenic did not precipitate along with it or as a separate phase but rather reported as adsorbed onto ferrihydrite (6 line) and/or hematite. The mechanism of formation of the solids is very different at natural (pH <1) and higher pH. In the case of pH 5 tests, solid formation (crystallization) occurs via a gelatinous hydrated sulfate deficient iron-arsenate-hydroxide mixture while at pH <1 (natural pH of solutions) direct nucleation and growth out of the solution occurs.

Large scale hydrothermal precipitation tests involving the Fe(III)-AsO4-SO4 system were also conducted using a 4L autoclave to validate the findings of the “test tube” types of experiments described above and to get kinetic data. (Swash 1996) In these large scale precipitation experiments, two types of tests were conducted, the first was a “batch type” of experiment (similar to the “test tube” experiments) where the desired amount of ferric sulfate and arsenic pentoxide (Fe(III)/As(V) ratio 1:1, 1.5:1 and 2:1) were placed in the container vessel and then heated from room temperature to the desired temperature (150 and 190 °C) for 10 hrs while taking samples, i.e. these tests involved a heat-up period. The second type of tests (“at temperature”) involved heating the arsenic (or iron) solution to the desired temperature (150 and 190 °C) and then injecting the desired amount of iron (or arsenic) and taking samples for a period of 10 hours. The same Fe(III)/As(V) ratios were used (namely, 1:1, 1.5:1 and 2:1) for the “batch type” and at “temperature experiments”.

Figure 2. Arsenic ppt curves for the 150 and 190 °C tests (heating solution from

ambient temperature). (Swash 1996)

8 For the batch type of experiments involving heat-up (Fig 2; Swash, 1996) at 150 °C the removal of arsenic from solution (i.e. precipitation of arsenic) was much slower at all Fe(III)/As(V) ratios than that at higher temperature (190 °C). Increasing the iron content in the solution (Fe(III)/As(V) ratio 1:1 vs. 2:1) showed slower rates and this was reasoned to arise from slower rates of nucleation and mineral growth. But in general after 10 hrs, ≥ 95 % of the arsenic had been removed from solution and precipitated as scorodite

(FeAsO4·2H2O) when the Fe(III)/As(V) ratio ≤ 1.5. This is in contrast to the higher temperature (190 °C) experiments (Fig 2; Swash, 1996) where the arsenic removal from solution (i.e. precipitation of arsenic) was complete after 1.5 hrs at all Fe(III)/As(V) ratios

investigated. In this case the phase formed was the Type 2 (Fe4(AsO4)3(OH)x(SO4)y) but only when the Fe(III)/As(V) ratio was ≥ 1.5. The iron removal from solution (Swash, 1996) for the batch type of experiments at 150 and 190 °C was greatly influenced by both temperature and Fe(III)/As(V) ratio.

In terms of kinetics at 150 °C, iron precipitation were greatly slowed as the amount of iron was increased in the reaction solution with equilibrium iron removal values of 50 %, 66% and 100 % (corresponding to experimental Fe(III)/As(V) ratios 2:1, 1.5:1 and 1:1) achieved after 10 hrs. This was justified on the basis of unspecified chemical interference upon nucleation and growth of the scorodite particles. Iron precipitation kinetics at 190 °C again were greatly slowed by the increase of iron content in the initial reaction solutions, where now experimental Fe(III)/As(V) ratios of 1.5:1 and 2:1 had iron removal of ~ 70%, while at Fe(III)/As(V) of 1:1, ~ 90% of the iron was removed at the end of the reaction time (10 hrs). In this case the products formed were identified as an amorphous iron phase and scorodite. Oddly enough, Swash (1996) stated in his thesis that iron recoveries were very high at higher temperatures (190 °C) and this was due to co- precipitation of amorphous iron compound with scorodite.

Figure 3. Arsenic ppt curves for the 150 and 190 °C tests (injecting iron solutions at

elevated temperatures). (Swash, 1996)

In the case of the injection type of experiments (“at temperature”) that involved heating the arsenic solution to the desired temperature and then adding the required amount of iron, the precipitation of arsenic and iron occurred within the first 25 minutes(Fig 3; Swash, 1996). Fe(III)/As(V) solutions with composition of 1:1 were the fastest in terms of kinetics at 150 °C, while those at 190 °C exhibited no real change in terms of the fastest or slowest. The product obtained at 190 °C was largely scorodite (after 5 minutes) but a large portion of the produced solids remained as an unidentified amorphous phase. At longer times (they did not specify how long) either scorodite or Type 2 was formed. The phases precipitated at lower temperature (150 °C) were reported to be totally amorphous (at least under XRD analysis).

Figure 4. Arsenic removal from solution by precipitated hematite at 190 °C.

(Swash, 1996)

In the case of the injection type of experiments that involved heating the ferric sulfate solution to temperature (only 190 °C was used) and then injecting arsenic, it was found iron to have precipitated during the heat up (2 hours) in the form of hydronium jarosite

[(H3O)Fe3(OH)6(SO4)2] and hematite (α-Fe2O3). Swash commented that when arsenic was present in the solution from the beginning of the test, no jarosite phase was formed due to blockage of the nucleation and growth of this phase at low pH. Despite the precipitation of iron during the heat up period arsenic still precipitated upon injection at high temperature. The amounts of arsenic precipitated were ~ 80 % and 100 % (Swash, 1996) when the Fe(III)/As(V) ratio was 1.5:1 or 2:1 and 1:1 respectively. The removal of arsenic from solution took place via reaction of hematite with dissolved arsenate to produce scorodite as observed via XRD. Oddly enough the same author (Swash, 1996) reported that the produced solids were red in color suggesting the presence of unreacted hematite but yet still referred to the product as scorodite.

Figure 5. Arsenic precipitation curves at (a) 150 °C and (b) 190 °C .

(Monhemius and Swash, 1999)

In (1999) Monhemius and Swash published a study on the removal and stabilization of arsenic from spent copper refining electrolyte solutions via hydrothermal processing, in which the effects of copper and acid (which will be discussed more in detail in section 2.3) on arsenic precipitation were studied. More importantly and relevant to this section is the kinetic effect of the Fe(III)/As(V) ratio and temperature on the precipitation of arsenic and iron via injection type of experiments (“at temperature”); in this paper the authors reported that the arsenic solution was first heated to the desired temperature (150 and 190 °C) and then the required amount of iron was injected. At 150 °C the rate of arsenic precipitation (Figure 5; Monhemius and Swash, 1999) decreased with increasing Fe(III)/As(V) ratio and an induction period of 40 minutes was observed before arsenic started precipitating. At 190 °C they reported that the Fe(III)/As(V) ratio affected the arsenic precipitation in a similar but less marked way. In all cases it was reported that the precipitation of arsenic was ≥ 95 % and the phase formed was scorodite. At this point it is noted by the present author (M.A. Gomez) that the data included in the 1999 article by Monhemius and Swash (1999), specifically Figure 5a and 5b (150 and 190 °C) is in fact not for the injection experiments (“at temperature”) but rather for the batch type experiments involving heating up as discovered in the thesis of P. M. Swash (Figure 2a and 2b; Swash, 1996). Somehow this important oversight on the part of Swash and Monhemius went unnoticed in spite of the 16 citations their paper has received.

Finally, Monhemius and Swash (1999) reported that the precipitation of iron was similarly affected by temperature and Fe(III)/As(V) ratio and stated that at 150 °C (Monhemius and Swash, 1999), the iron removal values were 100, 66 and 50 % at experimental Fe(III)/As(V) ratios of 1:1, 1.5:1 an 2:1, indicating that part of the iron combined with arsenate to form scorodite while the rest remained in solution. Furthermore the number for iron equilibrium removal values at 150 °C given by Monhemius and Swash (1999) was exactly the same numbers given for the batch runs by P.M. Monhemius (1996) as can be observed from Figure 6a and 6b. Therefore the data presented for both arsenic and iron removal in the 1999 paper of Swash and Monhemius refer to the batch with heat up precipitation tests and not the injection “at temperature” tests as wrongly labeled. In 2001 the hydrothermal precipitation of Fe(III)-arsenates at high temperatures (210 to 280 °C) was also studied by Mambote et al. (2001), who in their study used ferric (III) sulfate and arsenic (III) trioxide sources along with hydrogen peroxide as an oxidant. The choice of arsenic in a trivalent form used in their studies was based on a suggestion (from P.M. Swash) that in many hydrometallurgical liquors arsenic was most likely to be in a trivalent form and thus the oxidation of arsenic should be examined and optimized with respect to temperature, pH, Fe(III)/As(V) ratio and arsenic concentration. Thus in their

13 work the Fe (III) to As(III, V) ratio was varied between 1:1 and 2:1 with a retention time of 5, 6, and 24 hours and a solution pH of 2.5 and 4 adjusted via KOH. Impurities in the form of Cr, Ni and Al were also studied to investigate their role on arsenic precipitation. Their results showed that to precipitate large amounts of arsenic (which is desired), the presence of ferric iron and As(V) is needed. Precipitates obtained at temperatures below 210 °C were amorphous(!), while crystalline compounds were obtained only at higher temperatures (>210 °C) when the initial pH was ≤ 2.5. When impurities were present (Cr, Ni and Al), the ferric arsenates obtained were contaminated with large quantities of those impurities. Unfortunately, only macroscopic characterization techniques were used (XRF, XRD and SEM) and thus no molecular formula or information was obtained for the crystalline phase(s) produced; moreover, very vague characterization/chemical descriptions of their phases were given. XRF analysis showed that their ferric arsenate solids had mainly Fe(III)/As(V) 1.5-2 molecular ratios in the solid and their XRD analysis indicated the presence of an “unknown ferric arsenate phase”, unlike that of scorodite or any other observed by Swash and Monhemius (1994) The most stable phase of interest (TCLP leachability of 3.3-4.8 mg/L As) was ironically the “unknown phase of ferric arsenate” which was produced from starting solutions having Fe(III)/As(III) ratio of two and pH of 2.5 at temperature greater than 210 °C. After the present work was initiated, Dutrizac and Jambor (2007) reported on an

extensive experimental program involving the precipitation of Fe (III)-AsO4-SO4 phases in the temperature range of 175-225 °C. Their study was undertaken in relation to refractory gold ore processing involving in situ precipitation of iron(III) arsenate phases in autoclaves operating at 190-250 °C. The initial part of their study involved the characterization of an autoclave arsenic containing residue from a Canadian gold mine (McCreadie et al., 2000), where the material gave an average chemical composition of

30% Fe, 34% AsO4 and 19 % SO4 and the XRD analysis indicated no match to any products obtained by Ugarte and Monhemius (1992) from pilot tests of arsenic bearing ores, nor from the studies conducted by Swash and Monhemius (1994). Hence they labeled their “new” ferric arsenate sulfate compound as “Phase 3”. In this program the effect of time (0-24 hours), initial acidity (0-0.71 M H2SO4), Fe(III) [0 - 0.7M] and As(V) [0-0.8M] concentrations were considered, corresponding to Fe(III)/As(V) ratios ranging

14 from 0.5 to 40. Characterization of the precipitated phases (which were all crystalline in

nature) was done mainly via powder XRD (λ Cu Kα = 1.5046 Å) and chemical analysis.

In their study pure scorodite (FeAsO4 · 2H2O) was found to form at 150-175 °C with an

Fe(III)/As(V) ratio of 3 in presence of 0.45 M H2SO4 (aq) and 5 hours. Therefore in their work it was decided to focus over the range of 180-250 °C to avoid the formation of scorodite. This was done so as the product found in the industrial arsenic-containing residue did not resemble scorodite chemically or structurally. Basic ferric sulfate

(FeOHSO4) was found to form over the temperature range 205-215 °C, in the presence of

0.41M H2SO4) at Fe(III)/As(V) ratios of 10, 20 and 40 and reaction times of 3-4 hrs. This

BFS (FeOHSO4) phase had the orthorhombic structure (Pnma) as reported by Johansson (1962).

Figure 7. Basic Ferric Sulfate MDO’s polytpes (left) orthombic and (right) monoclinic. The octahedra are the iron units and tetrahedra the sulfate. (Ventruti et al., 2005)

Recent studies by Ventruti et al. (2005) have reported, however, that in fact this phase is a superposition crystal structure that belongs to an Order-Disorder family of structures formed by equivalent layers of symmetry in which two Maximum Degree of Order polytypes are possible; one being an orthorhombic phase (Pnma) and the other a monoclinic (P21/C) polytype; the best fit (Rp = 0.009) to observe the powder pattern was obtained with a 69:31 ratio of monoclinic to orthorhombic polytypes. Dutrizac and Jambor in their study further reported on another “new” phase that they labelled “Phase

4”. The chemical formula of the latter phase was FeAsO4 · 3/4H2O- a phase originally

15 described by Jakeman et al. (1991). The formation of “Phase 4” was reported to occur at

reaction times of 3-4 hrs and 205-215 °C in the presence of 0.41 M H2SO4 and at Fe(III)/As(V) ratios of 0.5 to 1.33. Returning to “Phase 3” that they also mentioned in their work, it was reported to have a variable composition over the following range:

Fe(AsO4)0.25-0.54(SO4)0.46-0.75[(OH)0.36-0.75(H2O)0.25-0.64]. This phase was found to form as the sole ferric arsenate compound at 180-225 °C, from 0 to 6 hours reaction time in the presence of 0.1-0.70 M H2SO4 and at Fe(III)/As(V) ratios of 2 to `~4. They postulated their new Phase 3 to correspond to the monoclinic polytype of basic iron sulfate which is promoted by the solid-solution uptake of As (as arsenate) for S (as sulfate) and where the substitution of As results in a corresponding decrease in the OH content required to maintain the charge balance. The authors distinguished between “Phase 3” and basic iron sulfate (FeOHSO4) on the basis of a shoulder on the right of the most intense peak of the diffractogram (~ 2θ =28.4° or d =3.14Å); however no Rietveld refinement (via the GSAS program) nor one-dimensional disorder model was evaluated (via DIFFaX) was conducted to verify these findings. Furthermore, they stated that the basic iron sulfate

(FeOHSO4), which is predominantly an orthorhombic structure can accommodate up to 9

wt% of AsO4 in its orthorhombic structure. Beyond this composition limit they suggested the formation of Phase 3 to be favored. Finally, short term (40 hours) leachability tests in water (terminal pH in the range 3.5 to 4.5) yielded correspondingly < 0.1mg/L for “Phase 3” and 1-3 mg/L as for “Phase 4” that prompted the authors to suggest that “Phase 3” might be an acceptable carrier for the disposal of arsenic. From the above literature review it becomes evident that the hydrothermal Fe (III)-As

(V)-SO4 system is very complex resulting in different phases; the exact identity of which has not been well understood, nor the apparent discrepancies in the previous results by the different investigators have been addressed.

16 2.3 Previous research on the effect of co-ions on scorodite precipitation

Singhania et al. (2006) were the first to study the precipitation of scorodite (FeAsO4 · 2+ 2H2O) from mixed sulfate media at 95 °C and in the presence of third ion effects (Cu , Zn2+, Ni2+, Co2+, and Mn2+) as per the atmospheric scorodite process developed at McGill earlier method (Droppert et al, 1996; Filippou and Demopoulos, 1997). The focus of their study and of particular interest here, was to observe whether third ions-cations and anions (sulfates)-would incorporate in the crystal structure (resulting in losses of the valuable metals) and how they may affect their long term stability in terms of arsenic released. It was found that the scorodite precipitates obtained in the presence of Cu2+, Zn2+, and Ni2 had the visual characteristics (green color) of pure scorodite, while those in the presence of Co2+, and Mn2+ had a yellow-beige color. Chemical analysis of the solids determined

them to have the stoichiometric composition of pure scorodite (FeAsO4 · 2H2O) with ~ 1 wt. % of third cation impurity incorporation; this was further verified via powder XRD analysis which indicated only the presence of crystalline scorodite (FeAsO4 · 2H2O). Therefore in their studies it was concluded that the presence of foreign M2+ cations (Cu2+, Zn2+, Ni2+, Co2+, and Mn2+) had no apparent affect of the precipitation and formation of scorodite and no losses of these valuable metals was expected to occur under these conditions. Furthermore the stability (arsenic release) of the products made in the presence of the foreign M2+ cations did not have a measurable effect on their arsenic release. In terms of sulfate effect it was found only ~ 0.3 wt. % of SO4 incorporation. Furthermore no interference with the precipitation kinetics or the stability of the scorodite product obtained under various total sulfate concentrations was observed. However, typically the scorodite produced from sulfate media was found to have higher TCLP-type leachability than that produced from chloride (Demopoulos et al 1995) or nitrate media (Bluteau and Demopoulos, 2007), namely 1-2 mg/L vs. 0.1-0.3 mg/L. Fujita et al. (2008b) investigated the effect of Zn2+, Cu2+ and Na+ ions on the atmospheric precipitation of scorodite (95 °C) (Fujita et al., 2008a); in their process 2+ ferrous iron (Fe ) is present in the initial solution which is oxidized [via O2 sparging] to ferric state in the presence of arsenate leading to scorodite precipitation. This process scheme is a modification of the atmospheric scorodite process developed by Demopoulos

17 and colleagues at McGill University (Demopoulos et al., 1994; Demopoulos et al., 2003), which is based on supersaturation control via proper selection of the operating pH plus the use of scorodite seed. In the case of the process of Fujita et al. (2008a) supersaturation is controlled via oxidation as initially discussed conceptually in Singhania et al. (2006). From Fujita’s studies on the effect of Zn2+ and Cu2+ (Fujita et al., 2008b) it was concluded that the Zn2+ cation had little effect on scorodite production. On the basis of this observation they proposed the integration of their scorodite process (arsenic fixation) into zinc hydrometallurgical processes. Similarly, the addition of Cu2+ ion had no significant influence upon scorodite formation or its short term leachability (As release); however, they noticed (a) a decrease in pH, (b) increased ORP level and (c) elevated copper and sulfur (as sulfate) content in the scorodite product when copper was present in the starting reaction solutions. Again the authors offered the possibility of integrating their scorodite process to copper hydrometallurgical processes but only up to 70 g/L Cu by controlling the pH < 3, as above this pH copper would precipitate from solution in a

[Cu3(AsO4)3] form. Their latter statement of 70 g/L Cu, however, seems to be controversial as they also reported that when 40g/L Cu was present their scorodite product contained 1.9 - 2.35 % of Cu, which was deemed to represent a significant loss of valuable metal. In (1999) Monhemius and Swash reported on a hydrothermal (150 °C) processing route for removing arsenic from copper-refinery bleed streams. According to the first part of their study that involved test tubes copper (up to 20 g/L) had no effect on arsenic precipitation. In the second part of their study, however, that involved injection of iron and copper at temperature it was noticed copper to influence the rate and extent of arsenic precipitation like acid concentration did. For example at 190 °C, 20 g/L free acid and no copper, 80 % of the arsenic was precipitated as “scorodite” (No characterization data was presented to support this assertion). Under the same conditions but in the presence of 15 g/L Cu, with acid (and no acid), arsenic removal was found to be ~ 70 % (and ~ 80 %) and the product a mixture of scorodite and Type 1 phases (see section 2.2). Finally, Monhemius and Swash (1999) commented that under the conditions mentioned above (190 °C, 0 and 20 g/L free acid and 15 g/L Cu) copper affected the nucleation and growth of the iron arsenate phases (scorodite and Type 1 in this case) and that up to 3.4 % of

18 copper was present in the solids-a rather significant loss of valuable metal. By contrast the CESL copper leaching process, which operates at lower temperature (150 °C), is reported to be essentially free of similar type of copper losses.

2.4 Previous research conducted on the Ca (II) - Fe (III) - AsO4 system

The hydrothermal Ca-Fe-AsO4 system was investigated by P.M. Swash (1996) given its relevance to precipitation that occurs in industrial arsenical processing liquors upon neutralization with lime (CaO). To this end they synthesized a number of calcium iron arsenates and studied their stabilities.

In his study the Ca-AsO4 system was first explored.(Swash and Monhemius, 1995) In particular solids were precipitated at different pHs (1-12) and temperatures (25 °C-100

°C) and Ca:As molar ratio of one (added in the form of Ca(NO3)2·4H2O and As2O5). It

was found that at pH 7 and < 100 °C, crystalline type of CaHAsO4 compounds were precipitated and while the XRD analysis was inconclusive, IR comparison “matched’

those of weilite (CaHAsO4), haindingerite (CaHAsO4·H2O) and guerinite

(Ca5H2(AsO4)4·9H2O). Extra peaks observed for the haindingerite sample suggested impurities in the sample. At pH 8 and 20-75 °C the solid phase produced resembled that of guerinite; at temperatures > 100 °C the predominant phase appeared to be weilite, while at 225 °C a phase in the form of Ca3(AsO4)2·xH2O was inferred. Finally at pH 11,

and >100 °C, basic calcium arsenate (Ca5(AsO4)3(OH) was precipitated. The short term

TCLP leachability (arsenic release) of these Ca-AsO4 solids was extremely high (900- 4400 mg/L) and thus were deemed not acceptable for arsenic retention. Secondly and more relevant to our studies, the Ca-Fe-AsO4 system was explored via precipitation from three solution compositions with Ca:Fe(III)/As(V) ratios of 3:1:4, 1:1:2, and 1:3:4 (added

in the form of Fe(NO3)2·9H2O, Ca(NO3)2·4H2O, As2O5 and Li2SO4·9H2O). These solutions were treated at temperatures of 150-225 °C and pH of 1-7. It was found that at pH< 5, 150-225 °C and all the Ca:Fe(III)/As(V) ratios studied, scorodite and/or Type 1 was found to precipitate in preference to calcium arsenate phases. Some anomalies were found in the products’ XRD patterns, which the author (P.M. Swash) reported to be from calcium atoms incorporated into their crystal structures; however this statement is incorrect since the Ca2+ ion has a much greater ionic radius (100 pm) then that of Fe3+

19 which is 55 pm (low spin) or 64 pm (high spin-this is indeed the ferric spin states found in scorodite and Type 1/FAsH as discussed in later Chapter 3). At pH 5, Ca/Fe(III)/As(V) ratio of 1:1:2 and temperature of 150-225 °C, a compound designated as Type 4 was found and claimed to have distinct IR and XRD patterns. No significant hydroxyl vibrations were observed in the IR spectra thus it was given the following chemical

formula: CaFe2(AsO4)0.33(8-2x)(HAsO4)x, where x<1 from chemical analysis. In addition this so- called Type 4 was stated to be very different than the two naturally occurring Ca-

Fe-AsO4 minerals: arseniosiderite and yukonite [Ca3Fe4(OH)6(AsO4)4·3H2O and/or

Ca3Fe4(AsO4)2(OH)6·5H2O]. The solids synthesized at pH 7 were reported to be mixtures of calcium and iron arsenates but no further details were given. It was further found that the TCLP short term leachability of the phases precipitated at pH 3-9 and temperatures (25-200 °C) gave a range of arsenic concentration values from 7.4 mg/L to 3580 mg/L thus rendering these solids inadequate as stable arsenic carriers (EPA TCLP limit is now < 1mg/L and 5mg/L in 1994). In this study the only phases that had a low leachability (arsenic release) ≤ 7 mg/L were produced from solutions with Ca/Fe(III)/As(V) ratios of 1:1:2 and 1:3:4 and 225 °C. Apparently, those solids were found to be of low solubility since they were composed essentially of scorodite and/or Type 1.

Among the various calcium ferric arsenate minerals (give a general reference here) yukonite and arseniosiderite are the most common ones. Yukonite is a hydrated calcium ferric arsenate mineral relevant to this research, that was first found in Tagish Lake, Yukon, Canada by Tyrrell and Graham in 1913, who described its formula as . (Ca3Fe2)2(AsO4)2(OH)6 5H2O (Tyrrell and Graham, 1913). Jambor later (1966) re- examined the yukonite formula and proposed the following: . Ca6Fe16(AsO4)10(OH)30 23H2O. The occurrence of yukonite since then has been reported among other in the Sterling Hill mine, Ogdensburg, New Jersey, USA by Dunn (1982), in Saalfeld, Thuringen, Germany by Ross and Post (1997), Redziny, Sudetes, Poland by Pieczka et al. (1998) and very recently by Nishikawa et al. (2006) in Kamchatka, Russia and Garavelli et al. (2009) in Grotta della Monaca, Italy. Paktunc et al. (2003, 2004), Borba and Figueiredo (2004) and Garavelli et al. (2009) reported mineralogical evidence of the coexistence of scorodite and yukonite phases in oxidized arsenic bearing ores (with

20 arsenopyrite as the primary arsenide mineral). In several ore samples, a replacement of

scorodite by Ca-Fe(III)-AsO4 phases was observed (Paktunc et al., 2003, 2004) where

yukonite (Ca2Fe3(AsO4)4(OH)·12H2O) and arseniosiderite (Ca2Fe3(AsO4)3O2·3H2O)

were among the Ca-Fe(III)-AsO4 mineral phases detected. Finally, very recently, yukonite was also found in the tailings disposal of a gold mining operation in Nova Scotia, Canada by Walker et al. (2005 and 2009).

Figure 8.Yukonite mineral from Tagish Lake, Canada.

Yukonite has been reported to occur most often as intensively fractured, gel-like aggregates of dark brownish color in hand-specimens; in smaller fragments, it is purple reddish and slightly translucent. The chemical composition of this rare mineral has been

reported to vary substantially in terms of its Fe, Ca, As and H2O content, as can be observed from previous reports (Tyrrell and Graham 1913; Jambor 1966; Dunn 1982; Swash, 1996; Ross and Post 1997; Pieckza et al., 1998; Nishikawa et al., 2006; Paktunc et al., 2003, 2004; Becze and Demopoulos 2007; Walker et al., 2009; Garavelli et al., 2009). Its structural characterization via lab based X-ray diffraction gives broad reflections typical of a mineral with poorly ordered (semi-crystalline) structure (Ross and Post 1997; Garavelli et al., 2009). Electron diffraction measurements (Nishikawa et al., 2006) suggested it exhibits single crystal diffraction of an orthorhombic or hexagonal nature. EXAFS measurements conducted by Paktunc et al 2003, 2004 proposed a local molecular structure for yukonite with As-O, As-Fe and As-Ca coordination numbers of 4,

21 3.24 and 4.17 respectively; but the molecular and structural nature of this phase still remains largely unknown. In terms of arsenic stability, Krause and Ettel (1989) found after equilibrating natural yukonite samples for 197 days in water, ~ 6-7 mg/L As released at pH 6.15. Swash and Monhemius (1994), also tested natural yukonite by equilibrating it for 7 days only (i.e. no equilibrium reached) and found it to give approximately 2-4 mg/L As over the pH range 5-9. Becze and Demopoulos in (2007) conducted short term (24 hr) leachability testing was conducted on synthetic yukonite, which was found to give an arsenic release of 1.16- 5.11 mg/L over a pH range of 7.5-8.8. Extension of these tests to longer period of time (> 450 days) was carried out both in gypsum-free and gypsum-saturated waters at constant pH 7, 8 and 9.5. The arsenic release was found to be significantly higher in gypsum-free (8.96 mg/L at pH 7, 47.8 mg/L at pH 8, and 276.4 mg/L at pH 9.5) compared to gypsum-saturated (0.75 mg/L at pH 7, 2 mg/L at pH 8, and 6.3 mg/L at pH 9.5) waters (Becze et al., 2010). Arseniosiderite is another closely related hydrated calcium ferric arsenate mineral relevant to these studies which was first reported by Koenig (1889) as mazapilite from the Jesus Maria Mine in Zacatecas, Mexico but was later shown by Foshag (1937) to be arseniosiderite by comparison with a specimen from Mapimi in Durango, Mexico. This mineral is commonly found in other regions of the world, apart Mexico, such as Greece, Namibia, Spain, France, Austria, Germany, England, USA, Bolivia and Australia. Paktunc et al. (2003) reported the occurrence of

arseniosiderite (Ca2Fe3(AsO4)3O2·3H2O) and yukonite (Ca2Fe3(AsO4)4(OH)·12H2O) in the Ketza River mine tailings produced from the operation of a former gold mine/mill in south central Yukon (Canada); in 2004 the same authors (Paktunc et al., 2004) found the presence of arseniosiderite in the gold ore from the Ketza River mine location. In a separate study Fillipi et al. (2004) investigated the arsenic forms in contaminated soil, mine tailings and waste dump profiles from the Mokrsko, Roudny and Kasperke Hory gold deposits in the Bohemian Massif found in the Czech Republic. In this study, they reported via chemical composition (EDAX) and XRD analysis that arseniosiderite and pharmocosiderite [KFe4(AsO4)4(OH)4·6-7H2O] were only present at the Mokrsko location in soils above the granodiorite bedrock. Later on Filippi et al. (2007) published an extensive report on the mineralogical speciation (via SEM-EDS/WDS, Electron

22 microprobe, XRD and Raman spectroscopy) of the natural soils contaminated by arsenic found in the Mokrsko-west gold deposit (Czech Republic), where pharmacosiderite and arseniosiderite (in addition to scorodite and jarosite) were identified as products of

arsenopyrite (FeAsS) and/or pyrite(FeS2) oxidation-weathering. Drahota and Filippi (2009) published a review on secondary arsenic minerals in the

environment, where they recognized that both Ca-Fe(III)-AsO4 minerals associate intimately with pharmacosiderite and less so with other ferric arsenates,such as scorodite. Drahota et al. (2009) reported frequent arseniosiderite replacements for pharmacosiderite and scorodite as Paktunc et al. (2004) and Fillipi et al (2007) did previously. This

observation seems to indicate that these Ca-Fe(III)-AsO4 minerals form during maturing of the parental environment following sulfide mineral reaction and consumption and increase in pH that renders scorodite and pharmacosiderite unstable. Several authors have commented on the properties of arseniosiderite (Larsen and Berman 1934 ; Palache et al. 1934;Chukhrov et al. 1958). Palache et al. (1951) listed the crystal system of arseniosiderite as hexagonal or tetragonal, based on optical data, while later on Moore and Ito (1974) reported arseniosiderite to have the following chemical

formula [Ca3Fe4(OH)6(H2O)3(AsO4)4] and to belong to a monoclinic crystal system with space group [A2/a], eight molecular formula units per unit cell (Z = 8) and be

isostructural with robertsite [Ca3Mn4(OH)6(H2O)3(PO4)4] and mitridatite

[Ca3Fe4(OH)6(H2O)3(PO4)4]. However, the authors stated that the precise formula of the latter could not be unambiguously defined without detailed knowledge of the crystal structure.

Figure 9. Basic structural units of mitridatite: (left) sheets of III -12 [Fe 9O6(AsO4)9] and (right) Ca2O10(H2O)2 dimers. (Moore and Ito, 1977a)

Therefore in 1977, Moore and Araki published (Moore and Araki, 1977a) an extensive crystallographic study on mitridatite (and as an extension on arseniosiderite)

where they reported mitridatite (Ca6(H2O)6[Fe9O6(PO4)9]·3H2O) to posses a monoclinic structure but this time with a space group [Aa] and Z = 4. Briefly, mitridatite (Figure 9) is III -12 built up with compact sheets of [Fe 9O6(AsO4)9] oriented parallel to the {100} plane and situated at the x ~ 1/4 and 3/4 crystallographic positions. These sheets are built up from octahedral edge sharing nonamers each defining trigonal rings which fuse at their trigonal corners to the edge midpoints of symmetry equivalent nonamers and are

decorated above, below and in the plane by the PO4 tetratehedra. In addition to these

octahedral sheets, a thick assembly of CaO5(H2O)2 polyhedra and water molecules occur as open sheets parallel to the {100} plane and situated at the x ~ 0 and ½ positions. The

water molecules in these CaO5(H2O)2 polyhedra define an OW-OW’ edge, which is

shared by two calcium atoms resulting in the formation of Ca2O10(H2O)2 dimers that contribute to the destruction of trigonal symmetry when they are placed on the octahedral sheets. This arrangement leads to the monoclinic cell symmetry of mitradatite. Robertsite and arseniosiderite are isotypes of mitradatite with arseniosiderite obtained from

isomorphic replacement of PO4↔AsO4. This replacement results in dilation along the “a” crystallographic direction because of the contribution in this direction by the larger

24 arsenate tetrahedra is not constrained by the relatively rigid octahedra ∞[Fe9O33] sheets. It is also worth noting that same year Moore and Araki (1977b) published a short synopsis where they stated that mitradatite (and as an extension arseniosiderite) had the same structure as previously mentioned but reported the space group to be A2/a. Finally, on this topic (the structure of arseniosiderite), it is worth noting that no real crystallographic structure on it has ever been fully reported as in the case of mitradatite. The stability of arseniosiderite in terms of arsenic leachability was first investigated by Krause and Ettel (1989). These authors found after equilibrating natural arseniosiderite mineral samples for 197 days in water, ~ 6-7 mg/L As at pH 6.85. Later, Swash and Monhemius (1994) also tested natural arseniosiderite samples by equilibrating them, this time for 7 days only (i.e. no equilibrium reached) and found them it to give 0.5-0.7 mg/L As over the pH range 5-9. No further arsenic stability tests for arseniosiderite exist to date.

25 2.5 Interaction of Electromagnetic Radiation with Matter

Figure 10.Wavelength and Frequency representation of Electromagnetic Radiation.

Electromagnetic radiation (“light”) exhibits the wave-particle duality, that is to say it behaves as a particle and wave like manner as demonstrated by the experiments of Huyen, Young, Fresnel, Maxwell, Plank and Einstein. These pioneering studies resulted in the development of De Broglie’s postulation that matter itself (not just EMR) exhibits a wave-particle duality, which was later advanced by the quantum mechanical work of Heisenberg's uncertainty principle.(Heisenberg 1927) For our purposes, the interaction of EMR and matter (“molecules”) consist of three basic processes: Absorption (non- resonant), absorption followed by spontaneous emission and scattering (elastic and inelastic). In scattering “elastic scattering” (referred to as Raleigh) refers to a process when the energy input is equal to the energy output; while “inelastic scattering”, referred to as “Stokes” and “anti-Stokes” occurs when the energy input is less then the energy output (molecules absorb energy) or when the energy input is greater then the energy output (molecules lose energy). Detailed discussion on photabsorption and photon scattering phenomena in low energy and higher energy photon based techniques may be found by the reader elsewhere (Stohr 1996; Hollas, 2004; Nakamoto, 2009).

26 It is worth mentioning at this point (since it applies to both IR and NEXAFS measurements) that the actual absorption of radiation is governed by the fundamental molecular property of absorbance. Absorption can be related to linear absorption

coefficient (μl - in units of 1/cm), which is dependent upon the thickness of the sample and may be expressed via Beer’s law:

I(z) = I0 exp [-μl z] (eqn 1)

where z is the sample thickness, I0 is the incident radiation and μl = μ*p where μ = 2 (Na/atomic mass)σA where μ is the mass absorption coefficient (cm /g), p the density 3 (g/cm ), Na is Avogadro’s constant and σA is the atomic absorption cross section. (Stohr 1996) The transition energy of a molecule can be simply approximated as the sum of four energy terms, originating from translation (displacement), and changes in energy states resulting from the interaction with electromagnetic radiation: electronic, vibrational and rotational. Translational energies of molecules can be ignored for our purposes as these are not quantized and as a result are not measurable.

Etotal = Etrans + Eelec+Evibr+Erot (eqn 2)

The basis for this difference lies in the fact that electronic transitions occur at much shorter time scales on one hand and on the other because it is much greater (as also is the energy required to excite these transitions) than say the vibrational and rotational transitions (Figure 10).

27 2.6 Theory of Vibrational (Infrared and Raman) Spectroscopies

In this section only a general description of the theoretical aspects of vibrational spectroscopy is dealt with, while further details such as detection modes, data treatment, theory and other aspects may be found by the reader in the suggested references. (Harris and Bertolucci, 1989; Coates, 1998; Hollas, 2004; Nakamoto, 2009) When a molecule (or matter) is under EMR (“light’), a transfer of energy from the EMR to the molecule will take place. This in simple terms occurs when the Bohr frequency condition is satisfied:

ΔE = hν (eqn 3) where ΔE is the energy difference between two quantized states, h is Plank’s constant and ν is the frequency of the EMR.

It is customary to rewrite Bohr’s condition in terms of the wavenumber (ā = 1/λ = ν/c in units of cm-1), which is the unit of choice used in vibrational spectroscopy.

ΔE = hcā (eqn 4)

where h is Plank’s constant, ā is wavenumber, c is speed of light and ΔE = E2-E1 and E2,

E1 are the energies of the excited and ground states respectively. More specifically in vibrational spectroscopy the model of a harmonic oscillator of a simple diatomic molecule is used to simplify the discussion and therefore the vibrational frequency (in wavenumber) may be expressed as

ā = 1/2πc(k/μ)1/2 (eqn 5)

where c-speed of light, k-force constant and μ = m1*m2/(m1+m2); m1 and m2 are the masses of the atoms forming the molecular bond.

28 Therefore, from quantum mechanics, the energy of a vibrational level is expressed as:

Evib = (ν+1/2)ā (eqn 6)

where h-Plank’s constant, ā-vibrational frequency and ν-vibrational quantum number. In general, the quantized vibrational energy levels are not equally spaced because the molecule deviates from ideality and thus it behaves as a non-harmonic oscillator and as a result we may express the energy of the different vibrational levels in a series:

2 3 Evib = (ν+1/2)āeq-(ν+1/2) āeqχe+(ν+1/2) āeqχe+…… (eqn 7)

where āeq = equilibrium vibrational frequency and χe = non-harmonicity constant

The probability to observe a transition from an energy state Eo to an excited state Ei can be derived from time dependent perturbation theory in quantum mechanics, in which the perturbation arises from the interaction with the EMR field assuming a dipole approximation. In the dipole approximation, we assume the wavelength of the EMR is much greater then the size of the molecule and that the wave function (ψi) that describes the system is small outside the molecule. Mathematically this is expressed via the Transition Dipole Moment (Rν) which serves as the basis for our selection rules for absorption or emission processes within the dipole approximation.

A transition by absorption or emission of radiation may occur if the Rν is not equal to zero. 2 Rν = │< Ψexcited │μ │ Ψground>│ (eqn 8)

Where, Ψ-Wave function of the ground and exited states; μ-Dipole moment operator = <

ΣQiXi, Σ QiYi, Σ QiZi >; and Xi,Yi,Zi are atomic coordinates in space and Qi is the potential energy for the interaction of the EMR.

Naturally our selection rules for vibrational spectroscopy (IR and Raman) are derived from the transition dipole moment, where the only difference is in the expression of the Taylor series used for the dipole moment operator (IR) and polarizability tensor (Raman) shown below:

IR (1-D): μ = μeqm+ x(dμ/dx) + x2(dμ/dx)2 +………. (eqn 9)

The dipole moment is therefore rewritten as

Rν = d μ /dx < Ψexcited │X│ Ψground>+… (eqn 10) and is non-zero for Δν = ± 1.

Raman (1-D): α = α eqm+ x(dα /dx) + x2(dα /dx) 2 +…… (eqn 11)

The dipole moment is rewritten as

Rν = d α /dx < Ψexcited │X│ Ψground>+…. (eqn 12) and is non-zero for Δν = ± 1.

Therefore in simple terms, an IR absorption (band/peak) is observed only if the dipole moment of a molecule is non zero (i.e. it leads to a change in dipole moment). On the other hand in Raman scattering, a band is observed when the polarizability changes as the molecule vibrates. Therefore in IR spectroscopy we can observe only heteronuclear

diatomics (CO but not Br2) and molecules that are polar and have C1, Cn, Cnv and Cs symmetry. In Raman spectroscopy both heteronuclear (CO) and homonuclear (Br2) diatomics may be observed. This is so because as the molecule swells and contracts under the excitation of EMR, the control of the nuclei over the electrons varies which leads to a polarization change, even among heteronuclear diatomics. For polyatomic molecules an IR vibration is observed if the normal mode leads to a change in dipole moment; while in Raman if a molecule contains a center of inversion then it will only be Raman active and not IR active. A schematic of the vibrational process that occurs in IR and Raman spectroscopy can be seen in the following Figure 11. (Coates, 1998)

Figure 11.Energy level schematic of the vibrational transition processes that are observed in vibrational spectroscopy.(Coates, 1998)

More specifically the use of group theory, character tables, direct products of non degenerate and degenerate groups and factor group analysis may be used to determine whether or not a vibration will be active in the IR and Raman spectra. Moreover, their activity will depend upon many factors including the simple selection rules stated above, as well as the site and crystal symmetry it exhibits in the systems under investigation, not to mention matrix effects that may occur in multi-component samples. However, this will not be discussed in detail here; rather the reader is directed to the following reference for a detailed description of this material. (Harris and Bertolucci, 1989; Coates, 1998; Hollas, 2004; Nakamoto, 2009)

31 2.7 Theory of Near Edge X-ray Absorption Fine Structure Spectroscopy (NEXAFS)

In this section only a general description of the theoretical aspects are dealt with while further details such as data processing, detection methods and other details may be found by the reader elsewhere.(Stohr 1996; de Groot 2005)

Figure 12. Energy level schematic of the electronic transition processes that are observed in (a) X-ray photoelectron spectroscopy and (b) X-ray Absorption Spectroscopy. (Nilsson 2002)

The absorption of EMR by a substance is the result of energy and angular momentum transfer from the EMR to the atoms and/or molecules present in the material of interest. For X-rays, the energy carried by the EMR ranges from few electron-Volts (eV) to several tens of keV; such energies may promote absorption of an electron from a core orbital into an unoccupied valence orbital or into the continuum (Figure 12; Nilsson 2002). The latter is usually what people refer to when dealing with X-ray Photoelectron Spectroscopy (XPS; Fadley, 2010) and the former with Near Edge X-ray Absorption Spectroscopy (NEXAFS) also referred to as X-ray Absorption Near Edge Spectroscopy (XANES). Although XANES can also probe continuum states this is done in a distinct manner then XPS but for our purposes only those transitions to the discrete states are to our interest. Throughout this manuscript both XANES and NEXAFS will be used interchangeably as they both relate to the same technique. The energy required for the former route depends on the binding energy of the core electron and since the binding energy is directly related to atomic number (Z), the energy required to excite an electron

32 from a core orbital, or a core shell (K, L, M…) is characteristic of the elements at hand in the absorber.

Figure 13. Energy level schematic of the electronic transition and decay processes that are observed during a Near Edge X-ray Absorption Fine Structure experiment. (Watts et al., 2006)

When a core electron is excited, there are different decay processes that occur during the X ray photo-absorption of core electrons in a NEXAFS experiment as shown by the schematic in Figure 13 (Watts et al., 2006). Naturally, the first step in this process is the photo-absorption of the X-rays and excitation of a core electron to an unoccupied molecular orbital which generates a pre-edge peak in the absorption spectrum, while the complete ionization of core electron(s) forms the edge jump in the absorption spectra (“white line”) the position of which corresponds to the binding energy of the electron. Soon after, the core excited state relaxes via non radiative (ejection of electrons such as Auger electrons) or radiative (emission of fluorescence photons) pathways.

Figure 14. The various regions of an X-ray Absorption Spectrum. (George, 1998; George 2006)

From such an experiment one produces the so called X-Ray Absorption Spectrum (XAS) as shown in Figure 14 (George, 1998; George 2006) which may be divided into two regions; the NEXAFS region and the EXAFS region. The NEXAFS region ranges from slightly below the edge jump to approximately 50 eV above the edge, while the EXAFS region generally ranges from 50 eV above the edge to well past the ionization edge. For our purposes the EXAFS region is not discussed here and is beyond the scope of this work. NEXAFS features are commonly attributed as electronic transitions from the core orbitals to specific unoccupied molecular orbitals, for example Ti 1s  3d quadrupolar transition to the t2g and eg orbitals, characteristic of TiO2 polymorphs. This is in contrast to EXAFS features, which are due to photo-electron scattering from neighboring atoms close to the core excited atom (Aksenov et al., 2006) as shown in Figure 15 below.

Figure 15. Scheme of interaction of a photoelectron with the atoms of the nearest environment (A is the atom absorbing an X-ray photon and B is a neighboring atom): (a) the energy-level diagram of electrons in a crystal lattice at different excitation energies corresponding to single-scattering (EXAFS) and multiple scattering (XANES) processes. (b) The emerging wave corresponding to a free electron and the interference between the emerging and scattered waves as well as (c) the energy dependence of X-ray absorption in the absence of scattering from neighboring atoms and in the presence of scattering. (Aksenov et al., 2006)

X-ray absorption can be described by the X-ray absorption cross-section (σx) defined as the number of electrons excited per unit time divided by the number of incident photons per unit time per unit area. The probability of transition (that we will see a “peak”) from the initial |a to the final |a’ state can be described by Fermi’s golden rule in the one-electron approximation (Stohr, 1996):

2 σx  |  a|V(t)|a’  pf (E) (eqn 13)

where pf (E) is the energy density of the final states and V(t) is the time dependent perturbation potential relating the interaction between the electromagnetic wave with potential B and an electron of charge -e, mass m and a linear momentum operator p such that

35 V(t) = (e/mc) B · p where c is the speed of light (eqn 14)

The vector potential can be written for a photon wave vector k and polarization electric field vector δ interacting with an electron at position r and expanded in a truncated Taylor series as

B = δ (1+ik·r) if k·r << 1 (eqn 15)

Thus by integrating equations 15 and 14 into 13 the following expression can be derived:

2 σx  |  a| δ · p+i (δ · p) (k·r) |a’  pf (E) (eqn 16)

The first term in the expansion | a| δ · p |a’ 2 can be related to the interaction between the electric field vector (δ) of the EMR with the transition dipole moment. This electric dipole interaction is responsible for most of the photo-electronic effects and in the dipole approximation, which is usually considered to describe electronic transition in NEXAFS, only this term is considered. The second term in the expansion, which contains the imaginary part |  a| i (δ · p) (k·r) |a’ 2 is the origin of higher order terms, such as the electric quadrupole moment and the magnetic dipole moment but these are only considered if one uses X-ray dischroism, i.e. X-ray Magnetic Linear Dichroism (XMLD) and/or X-ray Magnetic Circular Dichroism (XMCD).

36 2.8 References

Aksenov V. L., Koval’chuck M. V., Kuz’min A. Y., Purans Y. and Tyutyunnikov S. I. (2006) Development of methods of EXAFS spectroscopy on synchrotron radiation beams: A review. Crystallography Reports 15, 908-935.

Berezowsky R., Xue T., Collins M., Makwana M., Barton-Jones I., Southgate M., and Maclean J. Pressure leaching las cruces copper ore (1999) JOM, 51, 36-40.

Becze L., Gomez M. A., Petkov V., Cutler J. N. and Demopoulos G. P. (2010) The Potential Arsenic Role of Ca-Fe(III)-AsO4 Compounds in Lime Neutralized Co- Precipitation Tailings, In proceedings of Uranium 2010 and 40th Annual Hydrometallurgy Meeting, Uranium Processing -Tailings Section, CIM-MET SOC, Saskatoon, SK, Canada,Vol II, 327-336.

Becze L. andDemopoulos G.P. (2007) Synthesis and Stability Evaluation of Ca-Fe(III)- AsO4 Phases, In Extraction and Processing Proceedings, B. Davis and M. Free, Eds., TMS, Warrendale, PA, pp. 11-17.

Bluteau M-C. and Demopoulos G. P. (2007) The incongruent dissolution of scorodite - Solubility, kinetics and mechanism. Hydrometallurgy 87, 163-177.

Borba R.P. and Figueiredo B.R. (2004) The influence of geochemical conditions on arsenopyrite oxidation and on the mobility of arsenic in tropical surface environments. Revista Brasileira de Geociencias 34, 489-500.

Churkov F. V., Moleva Y.A. and Ermilova L.P. (1958) New data concerning mitridatite. Akad. Nauk S.S.S.R. Ser. Geol. 8, 16-26

Coates J. (1998) Vibrational Spectroscopy: Instrumentation for Infrared and Raman Spectroscopy. Applied Spectroscopy Reviews, 33, 267-425.

De Klerk R. and Demopoulos G. P. (2008) Continuous Circuit Production and Accelerated Ageing of Iron(III)-Arsenic(V) Coprecipitates - Probing Process- Stability Relationships, In Extraction and Processing Proceedings, S.M. Howard, Ed., TMS, Warrendale, PA, 3-10.

Dymov I., Ferron C.J. and Phillips W. (2004). Pilot Plant Evaluation of a Hybrid Biological Leaching-Pressure Oxidation Process for Auriferous Arsenopyrite/Pyrite Feedstocks. In Pressure Hydrometallurgy; Collins, M.J.; Papangelakis V.G.; (Eds), CIM: Montreal, Canada, pp. 735-750.

de Groot F. M. F. (2005) Multiplet effects in X-ray spectroscopy. Coord. Chem. Rev. 249, 31–63.

37 Demopoulos, G.P., Droppert, D.J.,VanWeert, G. (1994) Options for the immobilization of arsenic as crystalline scorodite. In: Harris, G.B., Krause, E. (Eds.), Impurity Control and Disposal in Hydrometallurgical Processes. Canadian Institute of Mining, Metallurgy and Petroleum, Montreal, pp. 57–69.

Demopoulos G.P. (2005) in Arsenic Metallurgy, R.G. Reddy, V.Ramachandran, eds., TMS, Warrendale, PA, pp. 25–50.

Demopoulos G. P., Lagno F., Wang Q., Singhania S. (2003) The atmospheric scorodite process, In Copper 2003-Cobre 2003, Hydrometallurgy of Copper (Book 2), Riveros, P. A.; Dixon, D.; Dresinger, D. B.; Menacho., J (Ed.), The Canadian Institute of Mining, Metallurgy and Petroleum (CIM): Montreal, volume VI, pp.597-616.

Demopoulos G. P., Droppert D. J., Van Weert G. (1995) Precipitation of crystalline scorodite(FeAsO4•2H2O) from chloride solutions. Hydrometallurgy. 38, 245-261.

Drahota, P., Rohovec J., Filippi M., Mihaljevic M., Rychlovsky P., Cerveny V. and Pertold Z. (2009) Mineralogical and geochemical controls of arsenic speciation and mobility under different redox conditions in soli, sediment and water at the Mokrsko-West gold deposit, Czech Republic. Sci. Total. Environ. 407, 3372- 3384.

Drahota, P. and Filippi M. (2009) Secondary arsenic minerals in the environment: A review. Environment International. 35, 1243-1255.

Droppert, D.J., Demopoulos, G.P., Harris, G.B., 1996. Ambient pressure production of crystalline scorodite from arsenic-rich metallurgical effluent solutions. In: Warren, G.W. (Ed.), EPD Congress 1996. TMS, Warrendale, pp. 227–239

Dunn P.J. (1982) New data for pitticite and a second occurrence of yukonite at Sterling Hill, New Jersey. Mineral. Mag. 46, 261-264.

Dutrizac J. E. and Jambor J. L. (2007) Characterization of the iron arsenate–sulfate compounds precipitated at elevated temperatures. Hydrometallurgy 86, 147-163.

Elements: Arsenic, Chemistry and Mineralogy. 2006, vol 2(2), pp.65-128.

Fadley C. C. (2010) X-ray photoelectron spectroscopy: Progress and perspectives.J. Electron. Spectros.. 178-179, 2-32.

Fillipou D. and Demopoulos G. P. (1997) Arsenic Immobilization by Controlled Scorodite Precipitation. JOM. 49, 52-55.

38 Filippi M., Golias V., and Pertold Z. (2004) Arsenic in contaminated soils and anthropogenic deposits at the Mokrsko, Roudny, and Kasperske Hory gold deposits. Environ. Geol. 45, 716-730.

Filippi M., Dousova B., and Machovic V. (2007) Mineralogical speciation of arsenic in soils above the Mokrsko-west gold deposits, Czech Republic. Geoderma. 139, 154-170.

Foshac, W. F. (1937) Carminite and associated minerals from Mapimi, Mexico. Am. Mineral. 22, 479-484.

Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2008a) Novel atmospheric scorodite synthesis by oxidation of ferrous sulfate solution. Part I Hydrometallurgy, 90, 92-102.

Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2008b) Effects of zinc, copper and sodium ions on ferric arsenate precipitation in a novel atmospheric scorodite process. Hydrometallurgy 93, 30-38.

Garavelli A., Pinto D., Vurro F., Mellini M., Viti C., Balic-Zuni T. and Della Ventura G. (2009) Yukonite from the Grotta della Monaca Cave, Sant’Agata Di Esaro, Italy: Characterization and comparison with Cotype material from the Daulton ine, Yukon, Canada. Can. Mineral. 47, 39-51.

George, G.N., Hedman, B., Hodgson, K.O. (1998). An edge with XAS. Nature Structural Biology 5, 645-647.

Graham G. N. (2006) Hard X-ray Absorption Spectroscopy and Imaging. Saskatoon Synchrotron Summer School I.

Harris D. C. and Bertolucci M. D. (1989) Symmetry and Spectroscopy: An Introduction to Vibrational and Electronic Spectroscopy. Dove Publications, Inc., New York.

Heisenberg, W. (1927), "Über den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik", Zeitschrift für Physik. 43, 172–198

Hollas J. M.. (2004) Modern Spectroscopy. (6th edn); Wiley: New York.

Jakeman R. J. B., Kwiecien M. J., Reiff W. M., Cheetham K. and Torardi C. C. (1991) A new ferric orthoarsenate hydrate: structure and magnetic ordering of FeAsO4.·3/4H2O. Inorg. Chem. 30, 2806-2811.

Jambor J.L. (1966) Re-examination of yukonite. Can. Mineral. 8, 667-668.

Jia Y. F., Demopoulos G.P., Chen N. and Cutler J. N. (2005) Co-precipitation of As (V) with Fe (III) in sulfate media: solubility and speciation of arsenic. In Arsenic

39 Metallurgy, R.G. Reddy and V. Ramachandran (Editors), TMS, Warrendale, PA, pp. 137-148.

Jia Y. F. and Demopoulos G.P. (2005b) Adsorption of arsenite onto ferrihydrite from aquoes Solutions: influence of media (sulfate vs nitrate), added gypsum, and pH alteration. Environ. Sci. Technol. 39, 661-668.

Johansson, G. (1962). On the crystal structure of FeOHSO4 and InOHSO4. Acta Chem. Scand. 16, 1234–1244.

Koenig, G. A. (1889) VI. Neue amerikanische Mineralvorkommen.1. Mazapilit. Z. Kristallogr. 17, 85-88.

Krause E. and Ettel V. A. (1989) Solubilities and stabilities of ferric arsenate compounds. Hydrometallurgy. 22, 311-337.

Larsen, E. S. III (1940) Overite and montgomeryite: two new minerals from Fairfield, Utah. Am. Mineral. 25, 315-326.

Langmuir D., Mahoney J., MacDonald A. and Rowson. J. (1999) Predicting arsenic concentrations in the pore waters of buried uranium mill tailings. Geochim. Cosmochim. Acta. 63, 3379-3394.

Langmuir D., Mahoney J. and Rowson. J. (2006) Solubility products of amorphous ferric arsenate and crystalline Scorodite (FeAsO4•2H2O) and their application to arsenic behavior in buried mine tailings. Geochim. Cosmochim. Acta. 70, 2942–2956.

Le Berre J.F., Gauvin. R and Demopoulos, G. P. (2007) Characterization of poorly- crystalline ferric arsenate precipitated from equimolar Fe(III)-As(V) solutions in the pH range 2 to 8 Metall. Mater. Trans. B, 38, 751–762.

Le Berre J. F., Gauvin R. and Demopoulos G. P. (2007) Synthesis, Structure, and Stability of Gallium Arsenate Dihydrate, Indium Arsenate Dihydrate, and Lanthanum Arsenate. Ind. Eng. Chem. Res. 46, 7875-7882.

Mambote R.C.M., Reuter M.A., van Sandwijk A. and Krijgsman, P. (2000) Immobilization of arsenic in crystalline form from aqueous solution by hydrothermal processing above 483.15K. Miner. Eng. 14, 391-403.

McCreadie, H., Blowes, D.W., Ptacek, C.J., Jambor, J.L. (2000). Influence of reduction reactions and solid-phase composition on porewater concentrations of arsenic. Environ. Sci. Technol. 34, 3159–3166.

40 Michel F. M., Ehm L., Antao S. M., Lee P. L., Chupas P. J., Liu G., Strongin D. R., Schoonen M. A. A., Phillips B. L. and Parise J. B. (2007) The structure of ferrihydrite, a nanocrystalline material. Science 316, 1726-1729.

Monhemius A. J. and Swash P. M. (1999) Removing and Stabilizing As from Copper Refining Circuits by Hydrothermal Processing. JOM. 51, 30-33.

Moore P. B. and Ito J. (1974) I. Jahnsite, Segelerite and Robertsite, Three new transition metal phosphate species II. Redefinition of Overite and isotype segelerite III. Isotopy of Robertsite, Mitridatite and Arseniosiderite. Am. Mineral. 59, 48-59.

III Moore P. B. and Ito J. (1977a) Mintridatite, Ca6(H2O)6[Fe 9O6(PO4)9]·3H2O. A noteworthy octahedral sheet structure. Inorg. Chem. 16, 1096-1106.

Moore P. B. and Ito J. (1977b) Mintridatite, a remarkable octahedral sheet structure. Mineralogical Magazine. 16, 1096-1106.

Myneni S. C. B., Traina S. J., Waychunas G.A. and Logan T.J. (1998) Experimental and theoretical vibrational spectroscopic evaluation of arsenate coordination in aqueous solutions, solids, and at mineral-water interfaces. Geochim. Cosmochim.Acta. 62, 3285–3661.

Nakamoto K. (2009) Infrared and Raman Spectra of Inorganic and Coordination Compounds. Part A: Theory and Applications in Inorganic Chemistry (6th edn); Wiley: New York.

Nickel E.H. and Nichols M.C. (1991) Mineral Reference Manual. Van Nostrand, New York, N.Y.

Nickel E.H. and Nichols M.C. (2007) IMA/CNMNC List of Mineral Names.

Nilsson A. (20002) Applications of core level spectroscopy to adsorbates. J. Electron. Spectros.. 1-3, 3-42

Nishikawa O., Okrugin V., Belkova N., Saji I., Shiraki K. and Tazaki K. (2006) Crystal symmetry and chemical composition of yukonite: TEM study of specimens collected from Nalychevskie hot springs, Kamchatka, Russia and from Venus mine, Yukon Territory, Canada. Mineral. Mag. 70, 73-81.

Paktunc D., Foster A. and Laflamme G. (2003) Speciation and characterization of arsenic in Ketza river mine tailings using X-ray absorption spectroscopy. Environ. Sci. Techn. 37, 2067-2074.

Paktunc D., Foster A., Heald S. and Laflamme G. (2004) Speciation and characterization of arsenic in gold ores and cyanidation tailings using X-ray absorption spectroscopy. Geochim. Cosmochim. Acta 68, 969-983.

Palache C., Berman H. and Frondel C. (1951) The System of Mineralo'gy of Dana, Yol. 2, Seventh ed., John Wiley and Sons, Inc., New York, pp. 955-956.

Pieczka A., Golobiowska B. and Franus W. (1998) Yukonite, a rare Ca-Fe arsenate, from Redziny (Sudetes, Poland). Eur. J. Mineral. 10, 1367-1370.

Ross D.R. and Post, J.E. New data on yukonite.(1997) Powder Diffr. 12, 113-116.

Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2006) Acidity, valency and third-ion effects on the precipitation of scorodite from mixed sulfate solutions under atmospheric-pressure conditions, Metall. Mater. Trans.B 37 , 189–197.

Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2005) Temperature and seeding effects on the precipitation of scorodite from sulfate solutions under atmospheric-pressure conditions. Metall. Mater. Trans. B 36, 327–333.

Smedley P.L. and Kinniburgh D.G. (2002) A review of the source, behavior and distribution of arsenic in natural waters. Appl. Geochem. 17, 517-568.

Stohr, J. (1996) NEXAFS Spectroscopy; Springer-Verlag: Berlin.

Swash, P. M. and Monhemius A. J. (1994) Hydrothermal precipitation from aqueous solutions containing iron (III), arsenate and sulfate, In Hydrometallurgy ’94, Chapman & Hall, New York, N.Y., pp. 177-190.

Swash, P. M. and Monhemius, A. J. (1995) In Sudbury ‘95, Mining and the Environment Proceedings; Hynes, T., Blanchette, M., Ed.; CANMET: Ottawa, pp 17-28

Swash P. M. and Monhemius A. J. (1998) The scorodite Process: a technology for the disposal of arsenic in the 21st century. In Effluent Treatment in the Mining Industry (eds. Castro S. H., Vergara F. and Sanchez M. A.) Andros Ltd.: Chile, pp. 119-161.

Swash, P. M. (1996) The hydrothermal precipitation of Arsenic solids in the Ca-Fe- AsO4-SO4 system at elevated temperature, Ph.D. Dissertation, Imperial College of Science, Technology and Medicine, University of London, London.

Tyrrell J. B. and Graham R. P. D. (1913) Yukonite, a new hydrous arsenate of iron and calcium, from Tagish Lake, Yukon Territory, Canada; with a note on the Associated Symplesite. In Transactions of the Royal Society of Canada, Third Series, vol. VII. pp. 13–18.

Ugarte, F.J.G., Monhemius, A.J., 1992. Characterization of high temperature arsenic containing residues from hydrometallurgical processes. Hydrometallurgy 30, 69– 86.

Ventruti, G., Scordari, F., Shiugaro, E., Gualtieri, A.F., Meneghini, C. (2005). The order– disorder character of FeOHSO4 obtained from the thermal decomposition of 3+ metahohmannite, Fe 2(H2O)4[O(SO4)2]. Am. Mineral. 90, 679–686.

Waychunas G. A., Davis J. A., and Fuller C. C. (1995) Geometry of sorbed arsenate on ferrihydrite and crystalline FeOOH: Re-evaluation of EXAFS results and topological factors in predicting sorbate geometry, and evidence for monodentate complexes Geochim. Cosmochim. Acta. 59, 3655–3661.

Walker S. R., Jamieson H. E., Lanzirotti A., Andrade C. F. and Hall G. E. M. (2005) The speciation of arsenic in iron oxides in mine wastes from the giant gold mine, N.W.T.: application of synchrotron micro-XRD and micro-XANES at the grain scale. Can. Mineral. 43, 1205-1224.

Walker S. R., Parsons M. B., Jamieson H. E. and Lanzirotti A. (2009) Arsenic mineralogy of near-surface tailings and soils: Influences on arsenic mobility and bioaccessibility in the Nova Scotia gold mining districts. Can. Mineral. 47, 533- 556.

Watts B., Thomsen L. and Dastoor P.C. (2006) Methods in carbon K-edge NEXAFS: Experiment and analysis.J. Electron. Spectros. 151, 105-120.

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.

43 3. The Hydrothermal Fe (III)-AsO4-SO4 system at 150- 225°C

3.1 Abstract

In this work, the hydrothermal reaction chemistry and characterization of high temperature (150-225°C) ferric arsenate phases produced from sulfate media was studied. The effect of Fe (III)/As (V) molar ratio, temperature, and time on the phases formed was examined. Three major arsenate-bearing phases were produced in our studies: (a) sulfate- containing scorodite (Fe(AsO4)1-0.67x(SO4)x · 2H2O where x≤0.20) at an Fe(III)/As(V) molar ratio of 0.7-1.87, 150-175°C and 2-24 hours reaction time; (b) ferric arsenate sub-

hydrate (FAsH; Fe(AsO4)0.998(SO4)0.01 · 0.72H2O) at Fe(III)/As(V) molar ratio of 0.69- 0.93, 200-225°C and 10-24 hours reaction time; (c) basic ferric arsenate sulfate (BFAS;

Fe(AsO4)1-x(SO4)x(OH)x · (1-x)H2O, where 0.38 months) stability testing of FAsH and BFAS found FAsH to yield somewhat higher arsenic release than BFAS. The latter’s arsenic release potential was evaluated to be equivalent or slightly better than that of scorodite.

44 3.2 Introduction

Autoclave processing of copper and/or gold-bearing mineral feedstocks is associated with the in-situ precipitation of iron (III) arsenates (Berezowsky et al., 1999; Dymov et al., 2004). These precipitates report with the leach residues into tailings ponds. Therefore, characterization and evaluation of the arsenic release (leachability) of these iron (III) arsenates is of great environmental interest.

In contrast to the poorly crystalline Fe (III)-AsO4 waste solids produced by co- precipitation during normal neutralization of hydrometallurgical process effluents (Langmuir et al., 1999; Moldovan et al., 2003; Jia et al., 2008, Chen et al., 2009), controlled precipitation (Filippou and Demopoulos, 1997; Singhania et al., 2006) or autoclave processing leads to the precipitation of crystalline phases. Swash and Monhemius (1994) were the first to report on the precipitation and characterization of Fe

(III)-AsO4 compounds from sulfate solutions-the type of solutions encountered in industrial processes-under autoclave processing conditions. According to their work four

distinct crystalline phases were found to form: scorodite, FeAsO4 · 2H2O; basic iron

sulfate, FeOHSO4 (BFS); “Type1”, Fe2(HAsO4)3 · zH2O with z<4; and Type 2,

Fe4(AsO4)3(OH)x(SO4)y with x+2y=3. The formation of these phases was correlated to temperature (150° to 225°C) and Fe (III) to As (V) molar ratio (1/1 – 9/1) and fixed retention time of 24 hours. Thus according to these authors, Type 1 formed in the whole temperature range 150°C to 225°C, when Fe (III)/As (V) molar ratio < 1.5. Scorodite was reported to form at 150° C and 175 °C when Fe (III)/As (V) ratio >1.5. Finally Type 2, in mixture with basic iron sulfate, was reported to form at 200°C ≤ T ≤ 225°C and Fe (III)/As (V) ratio >1.5. Of the two new phases, only Type 2 ( < 0.34 mg/L As) was found to meet the TCLP leachability criterion exhibiting similar behavior with scorodite (< 0.8 mg/L As).

More recently, Dutrizac and Jambor (2007) reported on an extensive experimental

program involving the precipitation of Fe (III)-AsO4-SO4 phases in the temperature range of 175-225°C. In their program, the effects of time (1-24 hours), initial acidity (0 to 0.71

M H2SO4) and variable Fe (III), As (V) concentrations were considered. Characterization

45 of the precipitated phases, which were all crystalline in nature, identified two new phases

in addition to scorodite (FeAsO4 · 2H2O) and basic ferric sulfate (BFS: FeOHSO4). The two new phases, labelled as “Phase 3” and “Phase 4”, were determined to have the following stoichiometries respectively: Fe(AsO4)x(SO4)y(OH)v(H2O)w where x+y = 1 and 3 v+w = 1; and “Phase 4”, FeAsO4 · /4H2O. Phase 3 was proposed to be a monoclinic

polytype of basic ferric sulfate produced via solid-solution uptake of AsO4. Phase 3 precipitated at 175-210C, but mixtures of Phase 3 and BFS were found to form at higher temperatures from solutions with Fe/As ~ 4. At Fe/As molar ratio ~1 and 205°C, Phase 4 was found to form instead. Finally in the same work, it was found that scorodite (containing a small amount of sulfate) formed in the 150-175°C range from solutions with initial Fe/As molar ratio ~3 (as calculated by the present authors). Short term (40 hours) leachability tests (terminal pH in the range 3.5 to 4.5) that were conducted on the two new phases yielded < 0.1mg/L As for Phase 3 and 1 - 3 mg/L As for Phase 4. This observation led the authors to suggest that Phase 3 might be an acceptable carrier for the disposal of arsenic.

From the above brief review, it becomes evident that the hydrothermal Fe (III) - As

(V) - SO4 system is very complex resulting in the formation of different phases, the true identity and environmental stability of which is a matter of industrial importance. In this study the hydrothermal precipitation of iron (III) arsenate-sulfate phases is revisited (a preliminary brief communication was made during the Hydrometallurgy 2008 Conference by Gomez et al., 2008) with the objectives of identifying the true nature of the precipitated iron (III) arsenate phases via comprehensive characterization and correlating their formation to prevailing solution chemistry in terms of reaction stoichiometries and temperature. Furthermore, the obtained results are compared to those of Swash and Monhemius (1994) as well as Dutrizac and Jambor (2007) with the view of clarifying the apparent differences between the two previous studies and contributing to the fuller understanding of the overall chemistry of this system.

46 3.3 Experimental Methods

3.3.1 Precipitation procedure

For the preparation of the starting solutions, analytical-reagent grade As2O5·xH2O and

Fe2(SO4)3·xH2O were dissolved in water in the desired molar proportions (CFe = 0.30 M and CAs = 0.075 - 0.40 M) to give different starting Fe(III) to As(V) molar ratios. The resulting solutions (with natural pH ~1) were placed in a two-liter Parr autoclave equipped with a glass liner. The solutions were then heated (typical heat-up period ~45min) to the desired temperature (150-225°C) and held there for different times (1, 4, 10 and 24 hours) while continuously stirred at 400 rpm. The resulting slurries were then filtered after cooling using a pressure filter and a 0.1μm filter paper. Following thorough washing, the solids were subjected to characterization and leachability testing; the filtrate solutions were analyzed for determination of their Fe, As, and S concentrations. All filtrates were in addition titrated to determine their free sulphuric acid concentration.

3.3.2 Characterization methods

X-Ray Diffraction (XRD), ATR-IR, Micro-Raman, Thermo Gravimetric Analysis (TGA), Field Emission Gun Scanning Electron Microscopy (FEG-SEM), X-ray absorption near-edge spectroscopy and Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) techniques were used to identify the nature of the materials.

The XRD analysis was performed with a Rigaku Rotaflex D-Max diffractometer equipped with a rotation anode, a copper target (λ Cu Kα = 1.5046 Å), a monochromator composed of a graphite crystal and a scintillator detector. The diffractometer used 40 kV and 150 mA. The scans were recorded between 5 and 100◦ 2θ with a 0.1◦ step size and an acquisition time of three seconds per step. XRD simulations were done with the crystallographic software CaRIne (version 3.1) using our experimental calculated lattice parameters and the published atomic positions from Hawthorne (1976) for Scorodite and Jakeman et al. (1991) for FAsH.

Infrared spectra were obtained using a Perkin Elmer FTIR (Spectrum BX model) spectrometer with a Miracle single bounce diamond ATR cell from PIKE Technologies.

47 Spectra over the 2000–550 cm-1 range were obtained by the co-addition of 200 scans with a resolution of 4 cm-1 at the FWHM of the internal Polystyrene strongest C-H vibration. Additional details on methods used may be found in Appendix 8.2.

Raman Microscopy was conducted on a Renishaw Invia microscope using the 50x short distance objective and a polarized argon laser operating at 514 nm operating at 10% of the laser power at the microscope exit. An average of 10 scans was obtained from 4000 to 150 cm-1 to improve the resolution and the statistics of the collection. Additional details on methods used may be found in Appendix 8.2.

The number of crystallization waters was determined with a TGA Q500 from Thermal Analysis Instruments. The acquisition was done between 20 and 900°C with a heating rate of 10°C per minute. The purge gas used was nitrogen.

The morphological characterization of the three phases was done on the Field Emission Gun Scanning Electron Microscope (FEG-SEM) Hitachi S-4700. Prior to the morphological analysis, the produced solids were deposited on carbon double sided tape and coated with a thin layer of AuPd.

The X-ray Absorption Near Edge Structure (XANES) spectra were recorded using the sample current in total electron yield mode using the spherical grating monochromator (SGM - 11ID-1) beamline at the Canadian Light Source (University of Saskatchewan, Canada). The beam line is equipped with a Dragon-type spherical grating monochromator and was designed for high spectral resolution studies (Regier et al., 2007). The spectroscopic resolving power (E/ΔE) is estimated to be greater then 3200. Reagent grade

hematite (α-Fe2O3) and melanterite (FeSO4*7H2O) from Fisher Scientific (98%+ purity) were used as standards for Fe3+ and Fe2+ oxidation states. The Fe L-edge spectra were normalized using a single normalization method and the energy scale of our compounds

was calibrated to the main L3 peak of hematite (α-Fe2O3) occurring at 709.5 eV (Garvie et al., 1994; van Aken et al., 1998; van Aken and Liebscher, 2002; van Aken and Lauterbach, 2003; Otero et al., 2008). The spectra were obtained from 696 to 736 eV using a coarse step size of 0.5 eV before the first edge and a smaller step of 0.04 eV at the

48 main absorption edges. Additional details on methods used may be found in Appendix 8.2.

Solutions and digested solid products were subjected ICP-AES analysis with a Thermo Jarrel Ash Trace Scan machine. Standards of 0.5, 5.0 and 50 mg/L of each

element (Fe, As and S) and a blank (4% vol. HNO3) were use to calibrate the instrument. The standard deviation in each case was found to be less then 5% by running the same sample at least three times. Prior to ICP-AES analysis the solids were digested in HCl (25% vol.) solution (500 mg solids in 100 mL solution) heated at 70 °C. The dissolution was complete after 30 minutes.

3.3.1 Short and long term arsenic release measurements

Both short-term and long-term leachability tests were performed. The short-term tests involved multiple (up to 7) TCLP contacts of 24 hours at pH~5 as described previously (Bluteau and Demopoulos, 2007). Such multiple-contact TCLP testing ensures collection of data that better reflect the stability of the crystalline phase(s) under investigation. In addition to the TCLP tests, long term leachability tests at target pH values of 3, 5 and 7.5 were carried out on FAsH and BFAS for the purpose of evaluating their long term arsenic release behavior. The pH of these longer term tests was adjusted periodically to their

target value via the use of NaHCO3 or HNO3, and samples were taken periodically and analyzed for arsenic via the use of ICP-AES.

3.4 Results and Discussions

The precipitation of iron (III) arsenate-sulfate phases was investigated over the

temperature range 150-225C. Typically 0.3 M Fe(SO4)1.5 solutions containing arsenic (V) at various molar ratios (0.7≤Fe(III)/As(V)≤4) were treated from 1 to 24 hours. There were three iron (III) arsenate phases found to form: (1) Sulfate-substituted scorodite

[Fe(AsO4)1-0.67x(SO4)x ·2H2O] (where 0.00≤x≤0.22) (2) ferric arsenate sub-hydrate

(FAsH) [FeAsO4 ·0.75H2O], and (3) basic ferric arsenate sulfate (BFAS) [Fe(AsO4)1- x(SO4)x(OH)x ·wH2O]. First their precipitation chemistry and domain of formation (i.e. precipitation conditions in terms of temperature, Fe(III)/As(V) molar ratio, time and acid

49 concentration range) for each of the phases is described, followed by their structural and molecular characterization as well as their short and long term leachability response.

3.4.1 Precipitation Chemistry

3.4.1.1 Precipitation of Scorodite

The precipitation conditions associated with the formation of scorodite are summarized in Table 1 below. On the same Table, the precipitation efficiency for each of the elements Fe and As, as well as the molar ratios of precipitated As/Fe ratio and acid generated/arsenic precipitated are also shown. Table 1. Scorodite experimental and precipitation data

Scorodite was found to form (all characterization-identification data is presented in the following section) in the temperature range 150-175°C with Fe(III)/As(V) ratios 0.70 to 1.8 and independent of retention time (1 to 24 hours). It is also worthy to remark here that scorodite was found to form at 200°C, when the Fe(III)/As(V) = 1, as an intermediate (metastable) phase. Thus hydrothermal precipitation after 1 hour yielded scorodite with presence (25%) of FAsH. Extension of hydrothermal processing to 10 hours (more on this later on) yielded a predominantly FAsH product with minor presence of scorodite,

50 while at 24 hours there was only FAsH present. This means that from a kinetics point of view scorodite forms quickly but because of lack of thermodynamic stability (at > 175°C) it converts with time to FAsH by subsequent dehydration:

FeAsO4 ·2H2O → FeAsO4 · 0.75H2O + 1.25H2O (1)

Interestingly enough according to Swash and Monhemius (1994) scorodite was supposed to be produced only at Fe (III)/As(V) > 1.5 for similar otherwise conditions ([Fe3+] =0.3 M, 150-175°C, no initial acid addition and 24 hours). Instead these authors reported the formation of the so called Type 1 phase (Fe2(HAsO4)3 · zH2O) under the same conditions we found scorodite.

Table 2. Chemical composition data for Scorodite.

Fe and As Precipitation: Analysis of the precipitation data (Table 1) revealed that at least 90% of the arsenic had precipitated under all conditions tested except for the case with the Fe/As molar ratio in the initial solution being less than 1. On the other hand the degree of percent iron precipitation was much lower than that of arsenic when the Fe/As ratio in solution was above 1. This behavior upon further analysis in terms of molar ratio of “precipitated As/ Fe ratio” was determined to indicate that iron precipitated in tandem

51 with arsenic (at a molar ratio Fe/As essentially one), i.e. no precipitation of other iron(III) phases took place; this is further confirmed with the detailed characterization data reported later. This is better demonstrated with the help of Figure 1a, where the initial molar ratio of Fe/As in solution is plotted against the molar ratio of As/Fe precipitated. As it can be seen the collected experimental data fall into line with slope 1.3. As discussed later (refer to composition data in Table 2), the produced scorodite was slightly deficient in arsenate content due to small incorporation of sulfate thus explaining in part the deviation of the experimental data from the theoretical slope of 1 in Figure 1a.

2.0

1.8

1.6

1.4

y = - 0.31 +1.38x 1.2 R=0.98

Initial Fe/As molar ratio 1.0

0.8

0.6 0.6 0.8 1.0 1.2 1.4 1.6 Precipitated As/Fe molar ratio

Figure 1. (a) Plot of Fe/As molar ratio in initial solution against molar ratio of precipitated As/Fe molar ration in the case of scorodite precipitation (data from Table 1).

Figure 1. (b) plot of Fe/As molar ratio in initial solution against molar ratio of precipitated As/Fe in the case of FAsH precipitation (data from Table 3).

53 3.5

3.0

2.5 solid

(Fe/As) 2.0

y = 0.10+0.98x 1.5 R= 0.97

1.0 1.0 1.5 2.0 2.5 3.0 3.5

(Fe/As) ppt

Figure 1. (c) plot of (Fe/As) solid against (Fe/As)ppt for BFAS precipitation (data from Table 5).

Acid Generation: The precipitation of scorodite is associated with acid generation. As per data in Table 1, the acid concentration in the final solution varied from 0.27 to 0.56 M in comparison to an initial pH of ~1. The final acid concentration was found to be in direct correspondence to the amount of precipitated arsenic. Thus upon calculation of the molar ratio “H2SO4 generated/As precipitated” (refer to Table 1) a mean value of 1.470.11 was obtained corresponding to the following reaction stoichiometry (Eq. 2), where w=2 for scorodite:

Fe(SO4)1.5 + H3AsO4 +wH2O → FeAsO4 ·wH2O + 1.5H2SO4 (2)

Chemical Composition: Table 2 summarizes the composition data for the various scorodite products. The mean chemical composition was: 24.440.88% Fe, 57.223.73%

54 AsO4, 3.442.60% SO4, and 15.331.22% H2O yielding the following empirical formula:

Fe(AsO4)0.94(SO4)0.08 · 1.95H2O. It is interesting to note the incorporation of a small fraction of sulfate within the crystal structure of scorodite; this has been confirmed via vibrational spectroscopy as reported elsewhere (Chapter 4 and 5, Gomez et al., 2010a; Gomez et al., 2010b). As a consequence of the small substitution of sulfate into scorodite, the Fe/As molar ratio was slightly higher than one (refer to data in Table 2). The corresponding ratio Fe/(As+S) that takes into account the sulfate substitution, on the other hand, was essentially one: 0.980.06 (Table 2). Hence the general molecular

formula of the sulfate-substituted scorodite becomes: [Fe(AsO4)1-0.67x(SO4)x ·2H2O] (where 0.00≤x≤0.20).

3.4.1.2 Precipitation of Ferric Arsenate sub-Hydrate (FAsH)

Precipitation: The precipitation conditions associated with the formation of this crystallization water-deficient ferric arsenate (more on its characterization later) are summarized in Table 3. In this series of tests no external acid was added. On the same Table, the precipitation efficiency for each of the elements Fe and As, as well as the molar ratios of “precipitated As/Fe ratio” and “acid generated/arsenic precipitated” are shown. FAsH was found to form at both 200 and 225°C with Fe(III)/As(V) ratios 0.70 to 1.1 and reaction time 10-24 hours. At Fe(III)/As(V)1.7 and the same otherwise temperature/time range and no acid addition, basic ferric arsenate sulfate (BFAS-more on this later, Table 5) was formed. (As explained in section 3.1.3, external acid addition or reaction time extension were found to favor the formation of FAsH at the expense of BFAS). As such it may be deduced that the transition between FAsH and BFAS formation over the temperature range 200-225°C lies at a ratio ~1.3-1.5. As previously mentioned, at 200°C and reaction time 1 hour, scorodite formed ahead of FAsH, which apparently converted to the latter upon extension of the reaction time as per equation (1). Under the same conditions (200-225°C and Fe(III)/As(V)~1) but with 24 hours reaction time, Swash and Monhemius (1994) found Type 1 to form. On the other hand Dutrizac and Jambor (2007) found “Phase 4” to form after 3 hours processing at Fe (III)/As (V) ~ 1 and 205 °C in the presence of externally added acid (0.4M).

55 Table 3. Ferric Arsenate sub-Hydrate experimental and precipitation data.

Similar to scorodite, iron and arsenic precipitated at a ratio essentially one (see Fig 1b) with arsenic(V) combining with iron(III). No evidence of additional co-precipitated arsenic-free ferric (hydroxy) oxide phase was apparent. The precipitation of FAsH was associated once more with acid generation-refer to Table 3. According to the stoichiometry of reaction (2), 1.5 moles of H2SO4 were expected for each mole of arsenic precipitated. The experimental values collected were somewhat lower than that expected (refer to Table 3) with a mean at 1.320.2.

Table 4. Ferric Arsenate sub-Hydrate chemical composition data.

56 Chemical Composition: Table 4 summarizes the composition data for the various ferric arsenate sub-hydrate (FAsH) products. The mean chemical composition was:

26.750.95% Fe, 67.250.50% AsO4, 0.450.17% SO4, and 6.00% H2O yielding the

following empirical formula: Fe(AsO4)0.998(SO4)0.01 · 0.72H2O. It is interesting to note the much lower crystallization water content hence the chosen name “sub-hydrate” adopted

in this work in analogy of other hydrated crystals like those of CaSO4 · xH2O (Ling and Demopoulos, 2005). Another apparent difference with scorodite is the much lower substitution (practically nil) of sulfate within the crystal structure of FAsH, pointing to the significant structural differences between the two phases as discussed further later.

3.4.1.3 Precipitation of Basic Ferric Arsenate Sulfate (BFAS)

At Fe(III)/As(V) molar ratio > 1.5 (and up to 4-the maximum ratio tested) basic iron

(III) arsenate sulfate (Fe(AsO4)x(SO4)y(OH)z · wH2O) was found to form at temperatures 175°C (and higher) when no acid was externally added after 4 to 24 hours reaction time. Under the same conditions of Fe(III)/As(V) molar ratio and T ≥ 200°C, Swash and

Monhemius (1994) reported the formation of Type 2 plus basic ferric sulfate (FeOHSO4) with the amount of the latter phase increasing with the increase of Fe(III)/As(V) molar ratio. In their study, no acid was externally added to the initial solution while the retention time was 24 hours. On the other hand Dutrizac and Jambor (2007) who carried out their experiments in the presence of externally added acid (typically ~0.4M H2SO4), found Phase 3 to form after 3 to 4 hours reaction time at 205°C. At higher temperatures (225°C) or longer time (>6 hours) and Fe(III)/As V) ratio ~ 4 (or higher) basic ferric sulfate in addition to Phase 3 were reported to form. In this work however, we did not detect the presence of basic ferric sulfate using the techniques employed herein.

57 Table 5. Basic Ferric Arsenate Sulfate experimental and precipitation data.

Precipitation of Fe and As: The amount of arsenic precipitated was in the order of 90 to 93% for all tests except those performed at 225°C. This is rather unexpected as higher temperatures are known to promote precipitation. Hence, the 225°C data should be considered as tentative. Nevertheless, if any error were involved in the analysis this applied equally to iron, so comparative evaluation of the iron and arsenic precipitation is still possible. The amount of iron precipitated in general was lower than that of arsenic especially at the higher Fe/As ratio of ~4. Upon calculating the number of moles of Fe and As precipitated, the ratio (Fe/As)ppt was determined as shown in Table 5. It can be seen this ratio to vary from 1.5 to 3 reflecting the different amount of arsenate contained in this phase-to be discussed in the next section.

Acid generation: The precipitation of BFAS was accompanied by acid (H2SO4) generation. The molar ratios of acid generated over arsenic and iron precipitated were determined and reported in Table 5. According to this data, the precipitation reaction

generates ~ 2 moles of H2SO4 per mole of As(V) precipitated or ~ 1 mole acid per mole of iron(III) precipitated. To account for these observations the following reaction (3) is proposed. The stoichiometry of BFAS is validated in the next section:

Fe(SO4)1.5 + (1-x)H3AsO4 + wH2O → Fe(AsO4)1-x(SO4)x(OH)x ·wH2O + (1.5-x)H2SO4 (3)

58 The x is determined in Table 6. If the above reaction stoichiometry were correct then the ratios (1.5-x)/(1-x) and (1.5-x)/1 should be close to the experimentally measured acid/As and acid/Fe molar ratios respectively, which indeed is the case as can be verified with the data of Table 5.

Table 6. Basic Ferric Arsenate Sulfate chemical composition data.

Chemical Composition: Table 6 summarizes the composition data for the various basic ferric arsenate sulfate (BFAS) products. The percent iron and water contents were essentially constant at ~30% and ~10% respectively, while the arsenate and sulfate contents varied in connection to each other. The interrelationship of arsenate and sulfate variation is better appreciated with the plot presented in Figure 1c. It can be seen that as the Fe/As ratio in solution increases from ~1.7-2 to 3.5-4, the amount of arsenate in the solids decreases while the amount of sulfate correspondingly increases. This behavior is indicative of solid solution composition as was initially proposed by Dutrizac and Jambor (2007) and confirmed via molecular characterization recently by the present authors (Chapter 4, Gomez et al., 2010a).

59 Table 7. Empirical and theoretical formulae of the various BFAS products.

By determining the molar ratios of As/Fe (or equivalently Fe/As), SO4/Fe, Fe/(As+S)

and H2O/Fe of the various BFAS solids (data in Table 6), the empirical formulae summarized in Table 7 were obtained. In these formulae a fraction of the water content is assigned as OH for charge balance purposes. It can be seen, within the inevitable

deviation due to experimental measurement, that [AsO4] + [SO4] is near one which implies that OH from a charge balance point of view should have the same stoichiometric

coefficient with SO4, what we have chosen to call “x”. This stoichiometric coefficient was found to vary (data in Table 6) from 0.3 to 0.6. This differs as much as 15% from the

experimentally determined SO4/Fe ratio (data in Table 6) hence a rather reasonable

match. Using x and applying the restrictions [AsO4] + [SO4] = 1 and [OH] + [H2O] = 1, the theoretical formula was determined for each of these BFAS products as shown in

Table 7. The generic solid solution formula, therefore for BFAS becomes: Fe(AsO4)1-

x(SO4)x(OH)x · (1-x)H2O, where 0.3

By comparison Dutrizac and Jambor’s (2007) formula of their Phase 3 varied from

Fe0.8-1.1(AsO4)0.25(SO4)0.75(OH)y to Fe0.8-1.1(AsO4)0.54(SO4)0.46(OH)y, while that of Swash

60 and Monhemius (1994) Type 2 phase was [Fe4(AsO4)3(OH)x(SO4)y with x+2y=3]. It is clear that the formula proposed by Swash and Monhemius (1994) did not recognize the existence of the solid composition behavior as they used a fixed Fe/As ratio of 1.33. On the other hand Dutrizac and Jambor’s (2007) formula pointed elementally towards the solid solution composition as confirmed in the present work and further proved spectroscopically elsewhere (Gomez et al., 2010a). In the case of Dutrizac and Jambor’s (2007) formula the Fe(III)/As(V) ratio varied from 1.5-4.4 while in the present work varied from 1.4 to 3.

BFAS conversion to FAsH: As mentioned earlier BFAS was found to form at Fe(III)/As(V) ≥2 and T ≥ 200°C while at Fe(III)/As(V) < 2 FAsH was found to precipitate instead. In all of these tests, no acid was externally added to the starting solutions. Thus when the test at Fe(III)/As(V) = 1.7-1.8 and 225°C was repeated in the

presence of 0.3 M externally added H2SO4 it was found that FAsH formed instead of the expected BFAS after 10 hours retention time. The respective precipitation efficiencies were 85.3% As and 58.5% Fe. By comparison in the absence of externally added acid (data in Table 5) the corresponding efficiencies were 100% As and 88.7% Fe. Calculation of the molar ratio of precipitated (Fe/As)ppt gave 1.20 in the presence of acid (formation of FAsH) versus 1.77 in the absence of acid (formation of BFAS). This may imply that increasing acid concentration promotes the formation of FAsH at the expense of BFAS. In other words, it is hypothesized that the formation of BFAS is kinetically favored and that given the right hydrothermal conditions in terms of temperature and time would transform to FAsH. This was verified by performing an additional test at 200°C and Fe(III)/As(V) molar ratio of 2 but this time employing an initial acidity of 0.3M and 72 hours retention time. Analysis of the final product showed to be FAsH with the following empirical formula: Fe(AsO4)0.99 · 0.73H2O. The final acidity in this test was 1.1M H2SO4. The following reaction is proposed to describe the transformation of BFAS to FAsH:

Fe(AsO4)1-x(SO4)x(OH)x·(1-x)H2O + 0.5xH2SO4  (1-x)FeAsO4·0.75H2O + xFe(SO4)1.5 +(0.25+0.75x) H2O (4)

61 The above metastability of BFAS and its conversion to FAsH was not reported previously. Dutrizac and Jambor (2007) who studied the effect of acid concentration did that at 205°C and 3h retention time. Apparently under those conditions given the slow transformation of BFAS to FAsH this went unnoticed; similarly it may be projected that BFAS would not transform in industrial autoclaves given the short retention time employed (~ 60 min). However, this may not be the case if the operating temperature were pushed well above the 225°C tested here as seems to be the recent industrial trend. Under such elevated temperature (>225°C) conditions some FAsH may form at the expense of BFAS, an event that can have consequences on the arsenic release (leachability) of the produced precipitates-this is discussed in a later section.

3.4.1.4 Precipitation of arsenate-bearing Basic Ferric Sulfate

Dutrizac and Jambor (2007) were the first, extrapolating from the ground breaking work of Ventruti et al. (2005), to suggest that BFAS (Phase 3 in their case) is a monoclinic polytype of BFS formed via arsenate uptake by BFS. The same authors reported that the orthorhombic BFS compound could accommodate up to 9.8 mass %

AsO4. With the view of studying the similarities and differences between BFAS and arsenate containing BFS a sample of the latter was prepared under the following

synthesis conditions: Fe/As=12 ([Fe(SO4)1.5]=0.96M, [H3AsO4]=0.08M), no free acid, 220°C and 5h. The produced arsenate bearing BFS had the following composition: 30%

Fe, 52% SO4, 2% AsO4, 10% OH and 5% H2O. By comparison the theoretical

composition of BFS is 33.1% Fe, 56.8% SO4 and 10% OH.

62 3.4.1.5 Precipitation Diagram

Inspired from the earlier phase diagram published by Swash and Monhemius (1994) we have constructed a new precipitation diagram that gives the domain of formation of the three ferric arsenate phases in terms of the Fe(III)/As(V) molar ratio in the solution and the temperature. This is shown in Figure 2.

Figure 2. The Gomez - Becze - Demopoulos (“GBD”) Precipitation Diagram of the

arsenate phases found in the Fe (III) - AsO4 - SO4 system.

There exist certain significant differences with the one published by Swash and Monhemius (1994). Thus according to Swash and Monhemius (1994), scorodite formation occurs only at Fe (III)/As(V) > 1.5 in the same temperature range (150-175°C) as in our diagram. At Fe(III)/As(V) ratios of 1-1.5 and temperature of 150-175°C, these

authors reported the formation of the so called Type 1 phase (Fe2(HAsO4)3 · zH2O), which is shown in the characterization section to be the same with our FAsH. However, in our studies under the same conditions (Fe(III)/As(V) ratios of 1-1.5 and 150-175°C) only scorodite was found to form. At higher temperatures (200-225°C and Fe(III)/As(V)~1) there was agreement (considering always Type 1 and FAsH equivalent) between the two works. In contrast Dutrizac and Jambor (2007), found “Phase 4” to form after 3 hours processing at Fe (III)/As (V) ~ 1 and 205°C in the presence of externally

63 added acid (0.4M). As is shown in the characterization section Phase 4 is indeed the same with Type 1 and FAsH. .

Another difference between Swash and Monhemius (1994) diagram and the one built

in this work is that basic ferric sulfate (FeOHSO4) is reported to co-precipitate along their Type 2 phase (same as our BFAS are shown later) even at Fe(III)/As(V) ratio of 2. On the other hand Dutrizac and Jambor (2007) found their Phase 3 (equivalent to BFAS as shown later) to form after 3 to 4 hours reaction time at 205°C. At higher temperatures (225°C) or longer time (>6 hours) and Fe(III)/As(V) ratio ~ 4 (or higher) basic ferric sulfate (BFS) in addition to Phase 3 was reported to form. No BFS was observed in this work up to the Fe(III)/As(V) ratio tested using the analysis techniques employed in this study.

Finally, there appears to be a discrepancy between the present work and those previously published (Swash and Monhemius, 1994; Dutrizac and Jambor, 2007) when it comes to the system of Fe (III)/As (V) ≥2 at 175°C. Thus in this work (see diagram in Figure 2) BFAS was found to form even at 175°C (10h retention time) when the Fe (III)/As (V) ≥2, while Swash and Monhemius (1994) as well as Dutrizac and Jambor (2007) reported instead the production of scorodite. In the case of the former study (Swash and Monhemius, 1994) 24 hours retention time and no external acid was added, while in the latter (Dutrizac and Jambor, 2007) a shorter retention time of 5 hours and external acid (0.4M H2SO4) was added to the starting solutions. It is possible that the longer retention time and the use of acid promoted the conversion of BFAS to scorodite in the case of the other works. In this regard it is worthy to mention that other work by the present authors reported elsewhere (chapter 5, Gomez et al., 2010b) found evidence of the formation of an intermediate phase having the BFAS molecular characteristics during precipitation of scorodite at 150°C. These observations serve to indicate that the diagram of Figure 2 should not be considered as absolute but rather as general guide recognizing that the prevailing precipitation kinetics especially as far it concerns supersaturation and short retention times can lead to deviations due to metastable phase formation (Desiraju, 2007; Demopoulos, 2009).

64 3.4.2 Characterization

In addition to chemical composition analysis (ICP-AES), the solid products were further analyzed by several analytical techniques (TGA, XRD, SEM, TEM, ATR-IR, Micro-Raman, and XANES). The detailed ATR-IR and Raman vibrational analysis has been already published is presented in chapter 4 (Gomez et al., 2010a) and the crystallographic electron and X-ray synchrotron based work is currently under way. Emphasis is given in the characterization of the least known phases FAsH and BFAS and their comparison to the phases labeled previously as Type 1, Type 2, Phase 3 and Phase 4.

3.4.2.1 TGA, XRD, ATR-IR, Raman

Figure 3 shows the Thermogravimetric and X-ray diffraction curves for scorodite, FAsH, and BFAS. For scorodite, FAsH, and BFAS the average percent of crystallization water was found to be 14.8 %, 6.19 %, 9.80 %, respectively. These numbers are in close agreement with the theoretical water contents of scorodite (15.6 %) and FAsH (6.50 %)

and the OH/H2O content of the BFAS formula (Tables 6 and 7) determined in the previous section.

Figure 3. TGA (left) and XRD (right) analysis of Scorodite, Ferric Arsenate sub- hydrates (FAsH), Basic Ferric Arsenate Sulfate (BFAS).

65 The three distinct phases, scorodite, FAsH and BFAS were identified by XRD analysis (Figure 3). Arsenate-containing basic ferric sulfate was specifically synthesized for the purposes of comparison to BFAS to see their structural similarities and differences.

Figure 4. XRD patterns obtained for experimental scorodite and reference JCPD file (005-0216).

Scorodite: The sulfate containing scorodite material exhibited good crystallinity and matched closely the JCPDS file for scorodite (N° 00-05-0216) (Figure 4). The d spacing was determined using Bragg’s law, the analysis was extended to the determination of the scorodite lattice parameters assuming the orthorhombic structure and simulations via the CaRIne software was done as in the case of scorodite produced by hydrothermal synthesis from nitrate solutions. (Le Berre et al., 2007) These will not be discussed in detail herein and are placed in the Appendix files (Table S1-S2 in the Appendix section) for the reader as it is a widely known phase and as such no needed information is obtained. In general, there was excellent agreement between the synthetic scorodite produced and that of the data base as expected.

66 Ferric Arsenate sub-Hydrate (Fs-3)

Phase 4 (As-184)

Type 1 (Swash 1994)

5 10152025303540 2

Figure 5. (a) XRD of FAsH, Phase 4 and reference Type 1.

FAsH (Fs-3)

FAsH (Fs-4)

Phase 4 (As-184)

Phase 4 (As-172)

200 400 600 800 1000 1200 1400 Wavenumber (cm-1)

Figure 5. (b) Raman of FAsH and Phase 4.

67 FAsH (Fs-3) FAsH (Fs-4) Phase 4 (As-184) Phase 4 (As-172)

600 800 1000 1200 1400 1600 1800 2000 Wavenumber (cm-1)

Figure 5. (c) ATR-IR of FAsH and Phase 4.

Ferric Arsenate sub-Hydrate (FAsH): A similar procedure was followed for XRD characterization of the produced FAsH and simulation of its structural model. The

synthetic material was in excellent agreement with the JCPDS file for FeAsO4·3/4H2O (N° 01-081-1923); but in addition its XRD pattern showed striking resemblance to the patterns of Phase 4 (sample kindly supplied by Dutrizac and Jambor, 2007) and Type 1 (Swash and Monhemius, 1994) (Figure 5a). Table 8 lists the experimental peak positions (d and 2θ values) and intensities in comparison to those of the reference material. As it can be observed the reference and experimental positions, both are in excellent agreement with one another.

68 Table 8. XRD of experimental FAsH compared to the XRD data of

FeAsO4·3/4H2O (Jakeman et al.,1991).

The XRD analysis was extended to the determination of the lattice parameters assuming the triclinic structure and using the d-spacing equation for triclinic systems.

Table 9. Experimental FAsH and reference FeAsO4·3/4H2O lattice parameter data (Jakeman et al.,1991)

69 The obtained values are given in Table 9. The structure of our FAsH phase was further probed by using the software Match! from Crystal Impact software (Brandenburg

and Putz, 2008); once again it was found that FAsH corresponded to the FeAsO4 · 3/4H2O compound first described by Jakeman et al. (1991).

Table 10. Comparison between XRD patterns obtained from the experimental FAsH product, the structure simulated with CaRIne and the reference

FeAsO4·3/4H2O material (Jakeman et al.,1991)

Simulations were done with the software CaRIne (version 3.1) using the experimental lattice parameters determined in this study (Table 10) and the atomic positions given by Jakeman et al. (1991) with the purpose to validate the structural model of FAsH. It is worth noting that because of the way the atomic coordinates were represented by Jakeman et al. (1991) (i.e. greater then one) translations of ½ were made for all the coordinates before running the simulation. Our ferric arsenate sub-hydrate like the

FeAsO4.3/4H2O compound possesses a triclinic structure in which the framework of the

crystal cell is composed of FeO6 octahedra and AsO4 tetrahedra. However, unlike

70 scorodite (Hawthorne, 1976; Le Berre et al., 2007), this hydrated ferric arsenate (FAsH) contains clusters of four FeO6 octahedra through edge sharing to form Fe4O16(H2O)2 moieties, which are interconnected by corner sharing of AsO4 tetrahedra. The simulation and experimental data (only intensity peaks > 5 considered) is presented in Table 10. As it can be observed the simulated XRD patterns are in excellent agreement with our experimental work and the JCPDS reference pattern (N° 01-081-1923). Moreover, the

XRD patterns shown in Figure 5a prove that the FAsH phase [FeAsO4(0.99) · 0.72 H2O]

produced in this work, Type 1 (Fe2(HAsO4)3·zH2O; reported by Swash and Monhemius (1994) and Phase 4 (described by Dutrizac and Jambor, 2007) are in fact the same phase

having the structural characteristics of FeAsO4·3/4H2O compound originally described by Jakeman et al. (1991).

Moreover, to ensure that our FAsH was the same as the Phase 4 (for Type 1 unfortunately samples could not be obtained) at the molecular level not just in terms of crystal structure, it was decided to investigate the arsenate bonding environment via the use of ATR-IR and Raman spectroscopies. As can be observed in Figure 5b-5c, it was found that indeed our produced FAsH and Phase 4 contain arsenate units in the same molecular environment and exhibit the same site and factor group symmetry as that described in detail elsewhere (Chapter 4, Gomez et al., 2010a).

Basic Ferric Arsenate Sulfate (BFAS): To date there exist only limited structural and crystallographic data for this phase, thus the XRD analysis could only be made empirically in this publication without resorting to theoretical structural calculations. The detailed molecular (IR and Raman) characterization of BFAS has been reported elsewhere (Chapter 4, Gomez et al., 2010a). The crystallographic (high resolution lab synchrotron based XRD) and atomistic (TEM-SAED) characterization of the Order- Disorder (OD) behavior of this complex phase and its relation to BFS is planned for a future publication and out of the space and time restriction of this thesis.

71 Table 11. XRD data of experimental BFAS, Phase 3 (Dutrizac and Jambor, 2007), Arsenate-containing BFS and reference Type 2 (Swash and Monhemius, 1994)

Table 11 gives a comparison of the peak positions (d -spacings) and intensity of BFAS, Type 2 (Swash and Monhemius, 1994), Phase 3 (Dutrizac and Jambor, 2007) and our synthetic arsenate containing FeOHSO4.

72 Basic Ferric Arsenate Sulphate (Ba-5) x x

Phase 3 (As-68) x x Type 2 (Swash 1994)

Arsenate-FeOHSO (McGill) 4

FeOHSO (Swash 1994) 4

5 101520253035404550556065 2

Figure 6. (a) XRD of BFAS, arsenate containing-BFAS, Phase 3, reference Type 2 and BFS.

Phase 3 (As-68)

Phase 3 (As-72)

BFAS (Ba-5) BFAS (Ba-8)

200 400 600 800 1000 1200 1400 Wavenumber (cm-1)

Figure 6. (b) Raman of BFAS and Phase 3.

73 BFAS (Ba-5) BFAS (Ba-8) Phase 3 (As-72) Phase 3 (As-68)

600 800 1000 1200 1400 1600 1800 2000 Wavenumber (cm-1)

Figure 6. (c) ATR-IR of BFAS and Phase 3.

In general, it can bee seen from Figure 6a and Table 11, that Phase 3 and Type 2 are in excellent agreement with our BFAS in terms of peak position and intensities; moreover it is worthy to note that BFAS/Type2/Phase 3 are distinct in comparison to produced arsenate-containing basic ferric sulfate in this study. It should be noted that Ventruti et al.

(2005) found an OD (Order-Disorder) relationship for Johansson’s (1962) FeOHSO4 structure. Only two MDO (Maximum Degree of Order) polytypes were determined to be most favorable, a monoclinic and an orthorhombic structure; these are geometrically unique and have nearly the same activation energy of formation with the monoclinic form being slightly more favorable. The structure of MDO1 results from a regular alternation of stacking operators (21/2 and 2-1/2) and yields an orthorhombic structure. MDO2 on the

other hand results from a sequence of symmetry operators (21/2|21/2|21/2…) and yields a monoclinic structure.

This relation was claimed to have been observed as well for the new Phase 3 by Dutrizac and Jambor (2007) via the tracing of a small shoulder in their XRD; however, no peak positions were reported in that work. In BFAS/Phase 3, the polytypic proportions

74 were claimed by Dutrizac and Jambor (2007) to be influenced by solid-solution substitution of arsenate for sulfate, which is greater in the monoclinic than the orthorhombic type. Yet, neither molecular nor structural spectroscopic data was given to support these claims or ideas. The operation of such substitution (solid solution behavior) between arsenate and sulfate in the case of BFAS was proved recently in another publication via the collection of vibrational spectra (Chapter 4, Gomez et al., 2010a). In terms of powder XRD analysis, however, we found the same diffraction patterns, i.e. the same crystal structures of BFAS at various arsenate and sulfate contents from solids produced at Fe (III) to As (V) ratios equal to two or four and 175 to 225 °C; where in our case we appear to be observing the other end of the solid-solution, namely sulfate for arsenate substitution in the structure of BFAS.

Again to ensure that our BFAS was the same as the Phase 3 at the molecular level not just in terms of crystal structure, we decided to further probe the arsenate bonding environment via the use of ATR-IR and Raman spectroscopies. As can be observed in Figure 6b-6c, it was found that indeed our produced BFAS and Phase 3 (Dutrizac and Jambor, 2007) contain arsenate and sulfate units in the same molecular environment as that described in detail elsewhere (Gomez et al., 2010a). Again for Type 2, unfortunately samples could not be obtained and thus only Phase 3 could be compared at the molecular level but due to the structural similarities observed via XRD, not much difference is expected.

Mixtures of Fe(III)-AsO4-SO4 phases: As it was noted in section 3.1, the effect of time on the phase formed is an important factor that previous studies (Swash and Monhemius, 1994; Dutrizac and Jambor, 2007) neglected to consider when investigating the same system. It is the purpose of this section to provide the structural and spectroscopic proof of the mixed phases found in this study.

75 Sc-4

Sc-2

Fs-3

10 20 30 40 50 60 2

Figure 7. (a) XRD of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3).

Fs-3

Sc-2

Sc-4

200 400 600 800 1000 1200 1400 Wavenumber (cm-1)

Figure 7. (b) Raman of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3).

76 Fs-3

Sc-2

Sc-4

600 800 1000 1200 1400 1600 1800 2000 Wavenumber (cm-1)

Figure 7. (c) ATR-IR of Scorodite (Sc-4), Scorodite+FAsH(Sc-2) and FAsH(Fs-3).

As mentioned in the scorodite part (Chapter 3), scorodite was found to form at 200°C, when the Fe(III)/As(V) = 1, as an intermediate (metastable) phase. Thus hydrothermal precipitation after 1 hour under these conditions yielded scorodite with presence (<25%) of FAsH. The quantification of the phase was determined via XRD (Figure 7a) using the integrated intensity relationship following a similar procedure as that used in other published works (Kontoyannis and Vagenas, 1999; Dickinson and McGrath, 2001) while the use of ATR-IR and Raman spectroscopies (Figure 7b-7c) was only used to qualitatively observe the mixtures of the two phases that occurred in the samples. The reason for not using these vibrational techniques to quantify the amount of each phase in the mixtures is because it is well known for the ATR-IR technique, the intensity is not a fixed variable that may be controlled confidently (as concentrations vary) since the intensity of the ATR-IR spectra can vary with the amount of pressure applied to the sample, as well as the index of refraction of the material; in the case of Raman spectroscopy, polarization effects (orientations), the Raman cross sections and other factors may have an influence upon the intensity of the peaks observed and thus may not be such a reliable way to quantify binary or tertiary mixtures, as shown in the work of

77 Dickinson and McGrath (2001). However, vibrational spectroscopies (ATR-IR and Raman) in this case did indeed show that our mixture was composed of scorodite and FAsH but moreover that the “dominant” phase was scorodite with some FAsH present as can be observed from Figure 7b-7c.

Sc-4

Fs-5

Fs-3

10 20 30 40 50 60 2

Figure 8. (a) XRD of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3).

78 Fs-3

Fs-5

Sc-4

200 400 600 800 1000 1200 1400 Wavenumber (cm-1)

Figure 8. (b) Raman of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3).

Fs-3

Fs-5

Sc-4

600 800 1000 1200 1400 1600 1800 2000 Wavenumber (cm-1)

Figure 8. (c) ATR-IR of Scorodite (Sc-4), FAsH+Scorodite(Fs-5) and FAsH(Fs-3).

79 Upon extension of hydrothermal processing to 10 hours, at 200°C and Fe(III)/As(V) = 1, a predominantly FAsH product was obtained with some minor scorodite (~ 20%). Again the amount of scorodite in this mixture was determined via XRD (Figure 8a) while the vibrational work (Figure 8b-8c) was used to qualitatively observe the two phase components of the mixture. It is interesting to note here however, that in this case, the Raman spectra (Figure 8b) showed no real scorodite component while the XRD (Figure 8a) and ATR-IR (Figure 8c) analysis showed a contribution from both components.

Fs-3

mixture

Ba-3

10 20 30 40 50 60

2

Figure 9. (a) XRD of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3).

80 Fs-3

mixture

Ba-5

200 400 600 800 1000 1200 1400 Wavenumber (cm-1)

Figure 9. (b) Raman of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3).

Fs-3

mixture

Ba-5

600 800 1000 1200 1400 1600 1800 2000

Wavenumber (cm-1)

Figure 9. (c) ATR-IR of FAsH (Fs-3), FAsH+BFAS(mixture) and BFAS(Ba-3).

81 As we reported in section 3.1.3 BFAS was found to convert to FAsH upon an increase in acid concentration or reaction time. However, the formation and occurrence of these two phases together has never been reported by any previous work on this system (Swash and Monhemius, 1994; Dutrizac and Jambor, 2007). The co-existence of these two high temperature phases is substantiated with the aid of the XRD and spectroscopic data presented in Figure 9a-9c. In this case owing to the few diffraction lines BFAS exhibits, the XRD (Figure 9a) of the mixture shows only small peaks for BFAS since the FAsH phase has so many lines in similar regions, making it difficult to determine the amount of each phase confidently. Nevertheless, the ATR-IR and Raman analysis clearly elucidated the presence of both phases as shown in Figure 9b-9c.

3.4.2.2 Crystalline Particle Morphology

The different synthetic phases were characterized further by FEG-SEM microscopy to investigate their particle size and morphology.

Figure 10. FEG-SEM images of Scorodite (Sc-4) at (a) 5x and (b) 20x magnification.

For Scorodite, the FEG-SEM examination (Figure 10a-10b) showed at low and high magnifications the particles to consist of aggregates of individual sub-micron orthorhombic crystallites.

Figure 11. FEG-SEM images of FAsH (Fs-3) at (a) 5x and (b) 20x magnification.

The Ferric Arsenate sub-Hydrate (FAsH), the FEG-SEM study (Figure 11a-11b) showed again the particles to consist of aggregates but this time the individual crystallites were generally less than 1 μm having the triclinic system features. It should be noted that in spite of the structural and magnetic information reported by Jakeman et al. (1991) on

FeAsO4·3/4H2O, no SEM work was ever published on their particle morphology until now.

For Basic Ferric Arsenate Sulfate (BFAS) (Figure 12a-12c), the morphology and size of the particles was distinct in terms of overall look in comparison to scorodite and FAsH. Here again we observe aggregates of particles forming; however, now the individual crystallites were larger in size (~ 2 μm). Furthermore, the physical appearance of the crystals, that is the outward form of the internal structure of this phase, appears in

83 two shapes: a rounded type particle and a distinct monoclinic-orthorhombic like crystal. From Figure 12b, it can be deduced that indeed they are the same phase (Figure 12c), with one grown to its full monoclinic-orthorhombic like form and the others re- crystallizing to form its well defined crystal shape as to our delight captured via our FEG- SEM studies (Figure 12b).

3.4.1.3 Fe L2,3 NEXAFS

Finally the three iron (III) arsenate phases were characterized using synchrotron- based techniques; in particular Fe L-edge X-ray Absorption Near Edge Structure (XANES) was employed. XANES spectroscopy is an element-specific finger printing technique, which probes the empty density of states in the system and sensitive to the particular chemical and electronic environment of the compound of interest (Koningsberger and Prins 1988; Colliex et al., 1991; de Groot, 1998; Otero et al., 2008; Huse et al., 2010) In general, the Fe L-edge edge is influenced by crystal field effects, which in turn are affected by the local symmetry around the absorbing iron atom in addition to atomic multiplet and charge transfer effects (de Groot et al., 1990; de Groot, 2005). Thus, the Fe L-edge XANES may be used for qualitative information of the local molecular environment around the absorbing atom, in addition to oxidation state and electronic or magnetic information (dichroism) (Otero et al., 2008; van Aken and Lauterbach, 2003; Huse et al., 2010)

Figure 13. Fe L-edge XANES of Scorodite, FAsH, BFAS, α-Fe2O3 and

FeSO4·7H2O

3+ Figure 13 presents the Fe L-edge XANES spectra of reagent grade Fe (α-Fe2O3) and 2+ Fe (FeSO4·7H2O) in addition to those of the high temperature synthetic Fe (III) - AsO4 -

SO4 phases (scorodite, FAsH and BFAS), the former two were used to serve as references.

Scorodite: The iron exists in a trivalent state and although much research has been conducted on the As and Fe K-edges of scorodite (Waychunas et al., 1993; Sherman and Randall, 2003; Moldovan et al., 2003; Chen et al., 2009) there is little or no information on the Fe L-edge of this common phase. However, as it has been shown for hematite (van Aken et al., 1998; van Aken and Liebscher, 2002; van Aken and Lauterbach, 2003), the Fe L-edge XANES may give new insights or confirm other observations on the electronic/magnetic nature, ligand environment and coordination geometry of the Fe3+ in this ferric arsenate phase. The L2,3 maxima of the produced synthetic scorodite (features

D, B in Figure 13) were observed at 723.2 eV(L2) and 709.5 eV(L3) respectively; their

corresponding pre-peaks (features C, A) occurred at 721.3eV(L2) and 708.2eV (L3)

85 respectively (Figure 13). The transitions of the L3 edges have been assigned as Fe: 2p3/2

→ Fe: 3d (t2g; eg) transitions; those of the L2 have been suggested as transitions from Fe:

2p1/2 → Fe: 3d. The valency of the Fe in scorodite as expected was trivalent. The approximate separation between the L2 and L3 edges due to spin-orbit coupling (see Appendix 8.2) was found to be 13.7 ± 0.1eV, in agreement with values reported for Fe +3 phases (Garvie et al., 1994; van Aken et al., 1998; van Aken and Liebscher, 2002; van Aken and Lauterbach, 2003). For more precise spin-orbit values, Fe 2p XPS analysis is recommended but for our purposes this was not necessary to obtain (see Appendix 8.2).

From the Fe L-edge XANES spectra of scorodite it is clear that the sharp pre-peak at low energy ~ 705eV, which is indicative of octahedral low spin Fe +3 compounds (Otero et al., 2008; Huse et al, 2010), is not present in scorodite. Thus we can infer that the Fe +3 6 2 in scorodite in a paramagnetic high spin state (ground: A1g+ T2g) in an octahedral type of 3 2 crystal field and an electronic configuration of (t2g) (eg) with a corresponding to an 6 approximate S5/2 ground state configuration (de Groot et al., 1990; de Groot, 2005). Moreover, since we have a high spin state we can further deduce that in this case the electron repulsions should dominate the ligand field effects and as such we should have a

weak field and small Δo. In addition, we can infer that as in the case of hematite (van Aken and Lauterbach, 2003), the ferric state in scorodite should be of a paramagnetic high spin anti-ferromagnetic character. The metal-ligand coordination based on the above findings should therefore be in a six fold environment in agreement with previous reported crystallographic and EXAFS data (Hawthorne, 1976; Chen et al., 2009).

Ferric Arsenate sub-Hydrate (FAsH): The L2,3 maxima of the produced FAsH

(features D, B) were observed at 722.8 eV(L2) and 709.5 eV(L3); their corresponding pre-

peaks (features C, A) occurred at 721.1eV(L2) and 708.0eV (L3) respectively (Figure 13).

The transitions of the L3 edges in FAsH have been assigned as Fe: 2p3/2 → Fe: 3d (t2g; eg)

transitions while those of the L2 have been suggested as transitions from Fe: 2p1/2 → Fe: 3d. The valency of iron in our produced synthetic FAsH is trivalent in agreement with independent Mossbauer studies (Jakeman et al., 1991); furthermore the approximate

separation between the L2 and L3 edges was found to be 13.3 ± 0.1eV in agreement with

86 other Fe +3 phases (Garvie et al., 1994; van Aken et al., 1998; an Aken and Liebscher, 2002; van Aken and Lauterbach, 2003).

The sharp pre-peak diagnostic of octahedral low spin Fe +3 compounds (Otero et al., 2008; Huse et al, 2010) is not present in the Fe L-edge XANES spectra of FAsH (Figure 13); once again suggesting that the FAsH phase should exhibit a paramagnetic high spin anti-ferromagnetic character as in scorodite and hematite (van Aken and Lauterbach, 2003) which is in agreement with Mössbauer studies conducted by Jakeman et al. (1991). Again the ferric iron state in FAsH is expected to be in an octahedral type of crystal field 3 2 6 and an electronic configuration of (t2g) (eg) with a corresponding to an approximate S5/2 ground state configuration (de Groot et al., 1990; de Groot, 2005) and since a high spin character is observed, we expect that the electron repulsions should dominate the ligand field effects and as such we should have a weak field and small Δo.

Basic Ferric Arsenate Sulfate (BFAS) is still a relatively unknown phase structurally, but as mentioned recent studies have suggested that the crystal structure could be monoclinic (Dutrizac and Jambor, 2007). However, no crystallographic data has ever been published to confirm these results; in our case, the synthetic products are fine powders and as a result obtaining single crystal diffraction was not possible. Therefore, by looking at the Fe L-edge of this phase we may obtain new insights into the local chemical environment around the iron (coordination geometry, high spin or low spin, ligand environment) of this largely unknown phase.

In the case of the synthetic BFAS, the L2,3 maxima (features D, B in Figure 13) were observed at 722.8 eV(L2) and 709.5 eV(L3) while their corresponding pre-peaks (features

C, A) occurred at 721.1eV(L2) and 708.0eV (L3) respectively. The L3 edges as before have been proposed as Fe: 2p3/2 → Fe: 3d (t2g; eg); those of the L2 were assigned to a Fe:

2p1/2 → Fe: 3d transition. The valency of the Fe in BFAS was trivalent. The approximate separation between the L2 and L3 edges was found to be 13.3 ± 0.1eV in agreement with that of Fe +3 phases (Garvie et al., 1994; van Aken et al., 1998; an Aken and Liebscher, 2002; van Aken and Lauterbach, 2003).

87 As in the case of the other high temperature Fe (III)-AsO4-SO4 phases, the iron state in BFAS was found to be in a high spin state (weak field, small Δo) and an in octahedral 3 2 crystal type of field with a possible (t2g) (eg) electronic configuration and a metal (iron)- ligand (oxygen) coordination that should be around six. Again the predicted ground state +3 6 configuration of the Fe in BFAS should be ~ S5/2 and we expect a paramagnetic high spin anti-ferromagnetic character as in hematite, scorodite and FAsH (van Aken and Lauterbach, 2003).

3.4.3 Short and Long Term Leachability Response

3.4.3.1 Short term arsenic release

Figure 14 below shows the sequential TCLP-like leachability response of scorodite, FAsH and BFAS. As it can be seen after the first 24 hour contact, scorodite and BFAS were well below 1 mg/L As but FAsH yielded about one order higher i.e. ~5 mg/L. However, with subsequent contacting all samples yielded values < 0.1 mg/L.

Figure 14. TCLP-like sequential test on Scorodite, FAsH and BFAS.

The initial higher values measured for FAsH are attributed to possible trace contamination (via co-precipitation of a poorly crystalline phase or adsorption) which upon multiple contacting was removed and representative data of the bulk phase was obtained. Similar behavior was observed earlier by Bluteau and Demopoulos (2007) for

88 scorodite and Le Berre et al. (2007) for mansfieldite. Swash and Monhemius (1994) showed similar results in terms of short term arsenic release for their Type 2 and scorodite phases produced, namely 0.34 mg/L and 0.8 mg/L of arsenic released respectively, but much higher leachability (11.9 mg/L As) values were obtained in the case of the Type 1 phase. On the other hand Dutrizac and Jambor (2007) reported 0.1 mg/L of arsenic release for Phase 3 and 1-3 mg/L for Phase 4 after 40 hours leachability testing with a terminal pH of 3.2-3.8 and 4.2-4.5 respectively. The higher leachability levels reported by Swash and Monhemius (1994) and Dutrizac and Jambor (2007) for their phases corresponding to FAsH (Type 1 and Phase 4) may reflect the different inherent stability of those phases or simply was the outcome of the applied testing procedure. In their studies no pretreatment of the solids prior to leachability testing was done as it was in this work (Figure 14). However, it is understood that the TCLP- type of testing provides only a comparative measure but does not tell us much about the long- term arsenic release of a particular phase and as such longer arsenic release measurements were conducted, something not undertaken previously.

3.4.3.2 Long term arsenic release

Figure 15a-15c shows the long term arsenic release measurements at nominal pH 3, 5 and 7.5 for the FAsH and BFAS phases along with their corresponding measured pH values. In the case of the BFAS two samples were tested, one with low sulfate content (Ba-4) and one with high sulfate content (Ba-7). The purpose of this was to observe whether the arsenic release of the BFAS phase was influenced by the solid-solution behavior (composition) of the starting material. In all cases essentially near equilibrium values were obtained as evident from the plateau of the various curves, with the exception perhaps of the test run at pH 8 (sample Ba-8, in Fig. 15c).

89 pH 3.0 pH 3.0 0.7 pH 3.0 pH 3.0 pH 3.0

0.6 FAsH (Fs-3)

0.5 pH 3.0 pH 3.3 BFAS (Ba-5) BFAS (Ba-8) 0.4

As (mg/L) pH 3.1 pH 3.1 pH 3.1 0.3 pH 3.2 pH 3.2 pH 2.8 0.2

0.1 pH 3.3 pH 2.7 pH 2.5 pH 2.6 pH 2.6 0.0 pH 2.9 pH 2.6 0 20 40 60 80 100 120 140 160 180 200 220 240 260 Time (days)

Figure 15. (a) Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 3.

At pH 3 (Figure 15a) a similar behavior to the TCLP test was observed, namely the FAsH phase was found to yield slightly higher arsenic release (0.67 mg/L) than that of the low and high sulfate content BFAS. In terms of the BFAS, the values ranged from 0.27 mg/L (low sulfate phase, Ba-4) to 0.1 mg/L (high sulfate phase, Ba-7).

pH 5.1 pH 5.1 pH 5.0 pH 5.1 2.5 pH 5.3

2.0 FAsH (Fs-3) pH 5.3 BFAS (Ba-5) pH 4.9 BFAS (Ba-8) 1.5

1.0 As (mg/L)

0.5

pH 5.3 pH 5.3 pH 5.2 pH 5.2 pH 5.2 pH 5.0 0.0 pH 2.8 pH 3.2 pH 2.8 pH 2.9 pH 3.3 pH 3.3 0 20 40 60 80 100 120 140 160 180 200 220 240 260 Time (days)

Figure 15. (b) Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 5.

90 At pH 5 (Figure 15b), the FAsH again gave the highest arsenic release (2.57 mg/L) with the two BFAS samples giving identical values of only 0.1 mg/L (it should be noted though that the pH of the high sulfate BFAS system drifted and stabilized at 3.3 instead the target value of 5, in spite of the fact it was always adjusted to the target pH of 5). By comparison the equilibrium solubility of the sulfate-substituted scorodite (Sc-4) was determined to be 0.6 mg/L As at pH 4.8 (equilibrium reached after 140 days; Gomez et al., 2008). This is slightly higher than the value determined for the scorodite produced by hydrothermal precipitation from nitrate media, namely 0.35 mg/L As at pH 5 (Bluteau and Demopoulos, 2007).

pH 8.1 pH 8.0 pH 8.0 60 pH 8.0 FAsH (Fs-3) BFAS (Ba-5) 40 BFAS (Ba-8)

As (mg/L) 20 pH 7.4 pH 7.3 pH 7.3 pH 6.9 pH 7.2 pH 7.3 pH 7.6 pH 7.3 pH 7.3 pH 7.4 pH 7.4 0 pH 3.5 pH 6.9

0 20 40 60 80 100 120 140 160 180 200 220 240 260 Time (days)

Figure 15. (c) Long term arsenic release measurements for FAsH and BFAS with low (Ba-5) and high (Ba-8) sulfate content at pH 7.5.

Finally, at nominal pH 7.5 (Figure 15c), FAsH and the low sulfate content-BFAS (Ba-4) drifted to slightly lower equilibrium pH, namely 7.3 and 7.4 respectively at which the corresponding stable final arsenic concentrations were 16.2 and 13.1 mg/L. On the other hand the high sulfate content BFAS (Ba-7) system drifted initially down to pH 3.5, which upon re-adjustment stabilized at pH 6.9 much lower than the target pH 7.5. This prompted another pH adjustment that led to overshooting and stabilization at pH 8. At this pH the arsenic concentration was 66.2 mg/L. By comparison the corresponding long- term leachability values of scorodite as obtained via interpolation from the work of

91 Bluteau and Demopoulos (2007) are 13 and >100 mg/L at pH 7.3 and 8 respectively. This implies that BFAS independent of its sulfate content is equally or slightly more stable than scorodite, while FAsH is less so. Nevertheless, in all cases it may be deduced from the presented data that the disposal of these phases should be done at pH ~ 7 or lower to avoid significant arsenic release. It should be noted that the characterization of the stability products via XRD, Raman and ATR-IR analysis did not show any evidence of change of phase transformation in any case. Moreover, the presence of ferrihydrite as predicted in the literature for ferric arsenates (Bluteau and Demopoulos, 2007) was not observed in any of the phases here.

3.5 Summary and Conclusions In this work, the hydrothermal reaction chemistry, detailed characterization and experimental arsenic release behavior of ferric arsenate phases produced in the temperature range 150-225°C from sulfate media was reported. Here are the major conclusions from this research: (1) The domain of formation of these phases depends upon four main experimental variables: temperature, Fe(III)/As(V) molar ratio, acidity and time. (2) There were four arsenate-bearing phases produced according to the following conditions: (a) sulfate-containing scorodite (Fe(AsO4)1-0.67x(SO4)x ·2H2O where x≤0.20) at an Fe(III)/As(V) molar ratio of 0.7-1.87, 150-175°C and 2-24 hours reaction time; (b)

ferric arsenate sub-hydrate (FAsH; Fe(AsO4)0.998(SO4)0.01 · 0.72H2O) at Fe(III)/As(V) molar ratio of 0.69-0.93, 200-225°C and 10-24 hours reaction time; (c) basic ferric arsenate sulfate (BFAS; Fe(AsO4)1-x(SO4)x(OH)x · (1-x)H2O, where 0.3

92 (5) Our FAsH and BFAS were found to be identical to the Type 1 and Type 2 phases of Swash and Monhemius as well as the “new” Phase 4 and Phase 3 compounds produced by Dutrizac and Jambor. (6) Fe L-edge XANES of the produced ferric arsenate phases (scorodite, FAsH and BFAS) indicated that they all exhibited a paramagnetic high spin anti-ferromagnetic character similar to hematite. (7) Short term TCLP-like (pH 5 and 24 hours) arsenic release measurements showed FAsH to be more soluble than scorodite after 24 hours, while after subsequent TCLP contacts all phases were found to yield negligible amounts of arsenic (< 1mg/L). (8) Long term arsenic release testing (~8 months duration) determined FAsH to yield higher arsenic release than BFAS, independent of the sulfate content of the latter. Thus at pH 5 the arsenic release level by FAsH was 2.5 mg/L but only 0.1 mg/L in the case of BFAS. The corresponding equilibrium value for sulfate-substituted scorodite was 0.6 mg/L. At pH 7.3-7.4 the corresponding values for FAsH and BFAS were 16 and 13 mg/L respectively while that of scorodite is 13 mg/L. Hence for safe disposal of these phases a pH of ~7 or lower are recommended.

3.6 References Berezowsky R., Xue T., Collins M., Makwana M., Barton-Jones I., Southgate M. and Maclean J. (1999) Pressure Leaching of Las Cruces copper ore. JOM 51, 36-40. Bluteau M-C. and Demopoulos G. P. (2007) The incongruent dissolution of scorodite - Solubility, kinetics and mechanism. Hydrometallurgy 87, 163-177. Boudias, C. and Monceau D. (1998) Software CaRIne Crystallography, version 3.1. CaRine Crystallography, Senlis, France. Brandenburg K. and Putz Holger (2007) Crystal Impact-Software for Chemist and MaterialScientist.Match http://www.crystalimpact.com/match/Default.htm Chen N., Jiang D.T., Cutler J., Kotzer T., Jia Y.F., Demopoulos G.P. and Rowson J. W. (2009) Structural characterization of poorly-crystalline scorodite, iron(III)―arsenate co-precipitates and uranium mill neutralized raffinate solids using X-ray absorption fine structure spectroscopy. Geochim. Cosmochim. Acta 73, 3260-3276.

93 Colliex C., Manoubi T., and Ortiz C. (1991) Electron-energy-loss-spectroscopy near-edge fine structures in the iron-oxygen system. Phys.Rev. B 44, 11402-11 420. de Groot F. M. F., Fuggle J. C. , Thole B. T. and Sawatzky G. A. (1990) 2p x-ray absorption of 2d transition-metal compounds: An atomic multiplet description including the crystal field. Phys. Rev. B: Condens. Matter, 42, 5459-5468. de Groot F. M. F. (2005) Multiplet effects in X-ray spectroscopy. Coord. Chem. Rev. 249, 31-63. de Groot F. M. F. (2008) Ligand and metal X-ray absorption in transition metal complexes. Inorg. Chim. Acta. 361, 850-856. Demopoulos G. P. (2009) aqueous precipitation and crystallization for the production of particulate solids with desired properties. Hydrometallurgy 96, 199-214. Desiraju, G. R. (2007) Crystal Engineering: A Holistic View. Angew. Chem. Int. Ed. 46, 8342 – 8356. Dickinson S. R. and McGrath K. M. (2001) Quantitative determination of binary and tertiary calcium carbonate mixtures using powder X-ray diffraction. Analyst 126, 118- 1121. Dutrizac J. E. and Jambor J. L. (2007) Characterization of the iron arsenate-sulfate compounds precipitated at elevated temperatures. Hydrometallurgy 86, 147-63. Dymov I., Ferron C. J. and Phillips W. (2004) Pilot Plant Evaluation of a Hybrid Biological Leaching-Pressure Oxidation Process for Auriferous Arsenopyrite/Pyrite Feedstocks. In Pressure Hydrometallurgy (eds. Collins M. J. and Papangelakis V.G.). CIM, Montreal, Canada, pp. 735-750. Filippou D. and Demopoulos G. P. (1997) Arsenic immobilization by controlled scorodite precipitation. JOM 49, 52-55. Garvie L.A.J., Craven A. J. and Brydson R. (1994) Use of electron-energy loss near-edge fine structure in the study of minerals. Am. Mineral. 79, 411-425. Gomez M. A., Becze L., Bluteau M. C., Le Berre J. F., Cutler J. N. and Demopoulos G.

94 Gomez M. A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G. P. (2010a) Vibrational spectroscopy study of hydrothermally produced scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) and basic

ferric arsenate sulfate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O ). J. Raman Spectros. 41, 212-221. Gomez M. A., Becze L., Celikin M. and Demopoulos G. P. (2010b) The effect of copper

on the formation of scorodite (FeAsO4·2H2O) from aqueous hydrothermal conditions: Evidence of a hydrated ferric cupric arsenate-sulfate short lived intermediate. Inorg. Chem. (In review) Hawthorne F. C. (1976) The Hydrogen Positions in Scorodite. Acta Crystallogr. Sect. B: Struct. Sci. B32, 2891-92. Huse N., Kim T. K., Jamula L., McCusker J. K., de Groot F. M. F., and Schoenlein R. W. (2010) Photo-induced spin-state conversion in solvated transition metal complexes probed via time-resolved soft X-ray spectroscopy. JACS 132, 6809-6816. Jakeman R. J. B., Kwiecien M. J., Reiff W . M., Cheetham K. and Torardi C. C. (1991) A new ferromagnetic orthoarsenate hydrate: structure and magnetic ordering of

FeAsO4·3/4H2O Inorg. Chem. 30, 2806-2811. Jia Y. F. and Demopoulos G. P. (2008) Coprecipiation of arsenate with iron (III) in aqueous sulfate media: Effect of time, lime as base and co-ions on arsenic retention. Water Res. 42, 661-668.

Johansson, G. (1962) On the crystal structure of FeOHSO4 and InOHSO4. Acta Chemica Scandinavica. 16, 1234–1244. Koningsberger D. C. and Prins R. (1988). X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES. New York: John Wiley & Sons. Kontoyannnis C. G. and Vagenas N. V. (2000) Calcium carbonate phase analysis using XRD and FT-Raman spectroscopy. Analyst 125, 251-255. Langmuir D., Mahoney J., MacDonald A. and Rowson. J. (1999) Predicting arsenic concentrations in the pore waters of buried uranium mill tailings. Geochim. Cosmochim. Acta 63, 3379-3394.

95 Le Berre, J. F., T. C. Cheng, R. Gauvin and G.P. Demopoulos. 2007. Hydrothermal synthesis and stability evaluation of mansfieldite in comparison to scorodite. Can. Metall. Q. 46, 1-10. Ling Y. and Demopoulos G. P. (2005) Preparation of alpha-Calcium Sulfate Hemihydrate by Reaction of Sulfuric Acid with Lime, Ind. Eng. Chem. Res., 44, 715-724. Moldovan B. J., Jiang D. T. and Hendry M. J. (2003) Mineralogical characterization of arsenic in uranium mine tailings precipitated from iron rich hydrometallurgical solutions. Environ. Sci. Technol. 37, 873-879. Otero E., Wilks R. G., Regier T., Bylth R.I.R, Moewes A. and Urquhart S. G. (2008) Sustituent effects in the iron 2p and carbon 1s edge Near-Edge X-ray Absorption Fine Structure (NEXAFS) spectroscopy of ferrocene compounds. J. Phys. Chem. A. 112, 624-634. Regier T., Krochak J., Sham T.K., Hu Y.F., Thompson, J. and Blyth R.I.R. (2007) Performance and Capabilities of the Canadian Dragon: The SGM Beamline at the Canadian Light Source. Nucl. Instr. Meth. A. 582, 93-95. Sherman D. M., and Randal S. R. (2003) Surface complexation of arsenic (V) to iron (III) (hydr)oxides: Structural mechanism from ab initio molecular geometries and EXAFS spectroscopy. Geochim. Cosmochim. Acta 67, 4223-4230. Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2006) Acidity, valency and third-ion effects on the precipitation of scorodite from mixed sulfate solutions under atmospheric-pressure conditions, Metall. Mater. Trans.B 37 B, 189–197. Swash P. M. and Monhemius A. J. (1994) Hydrothermal precipitation from aqueous solutions containing iron (III), arsenate and sulfate, In Hydrometallurgy ’94 (eds. Chapman & Hall). New York, N.Y., pp. 177-190. van Aken P. A., Liebscher B. and Styrsa V.J. (1998) Quantitative determination of iron

oxidation states in minerals using Fe L2,3-edge electron energy-loss near-edge structure spectroscopy. Phys. Chem. Miner. 25, 323-327. van Aken P. A. and Liebscher B. (2002) Quantification of ferrous/ferric ratios in

minerals: new evaluation schemes of Fe L2,3 electron energy-loss near-edge spectra. Phys. Chem. Miner. 29, 188-200.

96 van Aken P. A. and Lauterbach S. (2003) Strong mangnetic linear dichroism in Fe L2,3 and O K electron energy-loss near-edge spectra of antiferromagnetic hematite α-

Fe2O3. Phys. Chem. Miner. 30, 469-477. Ventruti G., Scordari F., Shiugaro E., Gualtieri A. F. and Meneghini C. (2005) The

order–disorder character of FeOHSO4 obtained from the thermal decomposition of +2 metahohmannite, Fe3 (H2O)4[O(SO4)2]. Am. Mineral. 90, 679–686. 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.

97 4. Vibrational Spectroscopic study of hydrothermally produced Fe (III)-AsO4-SO4 phases

4.1 Abstract

Three crystalline ferric arsenate phases: (1) scorodite; FeAsO4·2H2O, (2) ferric

arsenate sub-hydrate (FAsH; FeAsO4·0.75H2O) and (3) basic ferric arsenate sulfate

(BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O) synthesized by hydrothermal precipitation 3- 2- (175-225 °C) from Fe(III)-AsO4 -SO4 solutions have been investigated via Raman and 3- Infrared spectroscopy. The spectroscopic nature of these high temperature Fe(III)-AsO4 - 2- SO4 phases has not been extensively studied despite their importance to the hydrometallurgical industrial processing of precious metal (Au and Cu) arsenical sulphidic ores. It was found that scorodite, FAsH and BFAS all gave rise to very distinct arsenate, sulfate and hydroxyl vibrations. In scorodite and FAsH, the distribution of the internal arsenate modes was found to be distinct with the factor effect being more predominant in the crystal system. For the crystallographically unknown BFAS phase, vibrational spectroscopy was used to monitor the arsenate ↔ sulfate solid solution behaviour that occurs in this phase where the molecular symmetry of arsenate and sulfate

in the crystal structure is reduced from an ideal Td to a distorted Td or C2/C2v symmetry. With the new collected vibrational data of the pure phases, the use of ATR-IR spectroscopy was finally extended to investigate the nature of the arsenate in an industrial residue generated by pressure oxidation of a gold ore, where it was found that the arsenate was present in the form of BFAS.

98 4.2 Introduction

Arsenic is a common toxic element in mineral feedstocks which may mobilize during metallurgical recovery operations, and cause environmental problems.(Juillot et al., 1999) The stability-solubility of arsenic-containing waste solids depends on the crystallinity and the type of its arsenic bearing phases.(Riveros et al., 2001) Hence employment of advanced spectroscopic techniques for the characterization of industrially relevant synthetic ferric arsenate phases as demonstrated here is of great scientific and practical importance. In particular, the vibrational spectroscopy (Raman and IR) of the three ferric arsenate phases that have been observed to form when arsenical mineral feedstocks are treated in autoclaves under hydrothermal conditions (150 to 230 °C) for the recovery of gold or copper is studied. The three ferric arsenate phases in question (all crystalline) are:

(a) scorodite; FeAsO4·2H2O, (b) ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75H2O) and (c) basic ferric arsenate sulfate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)]x·wH2O). While scorodite is a natural mineral which has been the subject of several mineralogical (Kitahama et al., 1975; Hawthorne, 1976; Krause and Ettel, 1988) and processing (Demopoulos, 2005; Singhania et al., 2005; Singhania et al., 2006; Fillipou and Demopoulos, 1997; Demopoulos et al., 1995; Bluteau and Demopoulos, 2007; Fujita et al., 2008) studies, little is known about the other two phases that form in autoclaves by 3- 2- hydrothermal treatment of Fe(III)-AsO4 -SO4 solutions. Swash and Monhemius (1994) 3- were the first investigators to study the high temperature (150 to 225 °C) Fe(III)-AsO4 - 2- SO4 system. According to these investigators two new ferric arsenate phases labelled

Type 1 and Type 2 forms in addition to scorodite (FeAsO4·2H2O) and basic ferric sulfate

(FeOHSO4). Based on XRD and chemical analysis these two new phases were identified as: Fe2(HAsO4)3·zH2O with z < 4 (Type 1), and Fe4(AsO4)3(OH)x(SO4)y with x + 2y = 3 (Type 2). In 2007, Dutrizac and Jambor (2007) revisited this system and reported two

new distinct phases that they labelled “Phase 3” (Fe(AsO4)0.6 (SO4)0.4(OH)0.6(H2O)0.4) and

“Phase 4” (Fe(AsO4) · 3/4H2O).

It was lately realized via a comprehensive characterization program that Phase 3 and Phase 4 reported by Dutrizac and Jambor are the same with the Type 1 and Type 2 compounds identified by Swash and Monhemius.(Gomez et al., 2008) These phases

99 identified now by their generic chemical names: ferric arsenate sub-hydrate (FAsH;

FeAsO4·0.75H2O) and basic ferric arsenate sulfate (BFAS; Fe[(AsO4)1- x(SO4)x(OH)x]·wH2O, where 0.3 < x < 0.7 and w = 0.2 - 0.5) (Gomez et al., 2008) have been characterized mainly on the basis of XRD analysis. The characterization of BFAS becomes even more challenging because of its apparent solid solution composition. (Dutrizac and Jambor, 2007; Gomez et al., 2008) However, as we showed elsewhere (Gomez et al., 2008; Gomez et al., 2010) X-ray diffraction is not a molecular sensitive probe, thus vibrational and absorption-based spectroscopies are better suited to monitor solid solutions in these phases and for characterizing complex industrial mineral processing residues. It is the purpose of this paper to report the detailed vibrational 3- 2- spectroscopic information of the high temperature Fe(III)-AsO4 -SO4 phases, namely scorodite, ferric arsenate sub-hydrate and basic ferric arsenate sulfate to further advance the understanding of the molecular characteristics of the ferric arsenate phases and demonstrate the value of the obtained spectroscopic data in characterizing the mineral form(s) of arsenate in industrial processing residues. (Riveros et al., 2001)

4.3 Experimental Methods

4.3.1 Precipitation procedure

3- 2- For the synthesis of the Fe(III)-AsO4 -SO4 phases, analytical-reagent grade

As2O5·xH2O and Fe2(SO4)3·xH2O were dissolved in water in the desired molar proportions (CFe = 0.30 - 0.40 M and CAs = 0.09 - 0.40 M) to give different starting Fe(III) to As(V) molar ratios. The resulting solutions were left at their natural pH (≤ 1) and were placed in a two liter Parr titanium autoclave and heated to the desired temperature (175 - 225 °C) and held there for different times (1 - 24 hours). The resulting slurries were then filtered after cooling using a pressure filter with a 0.1μm filter paper. The solids were then washed with deionized water and dried at 40 °C until a constant weight was achieved. (Gomez et al., 2008; Gomez et al., 2010)

The experimental synthesis conditions for the various phases characterized in this study were:

100 Syn-scorodite: 175 °C, 10 hours, 0.3M As(V) and 0.28 M Fe(III).

Syn-ferric arsenate sub-hydrate (FAsH): 225 °C, 10 hours, 0.3M As(V) and 0.27 M Fe(III).

Syn-basic ferric arsenate sulfate (BFAS): 200 °C, 10 hours, 0.3M As(V) and 0.15M Fe(III).

Arsenate-containing industrial residue: 230 °C, 1 hour, this industrial residue supplied by Barrick Gold Corporation was generated in a pilot plant-scale autoclave by pressure oxidation of the Dolin Creek gold ore.

The formation of these iron (III) arsenate-sulfate phases can be described by the following general reaction (Gomez et al., 2008; Gomez et al., 2010):

Fe(SO4)1.5 + H3AsO4 +(w+z/2)H2O → Fe(AsO4)x(SO4)y(OH)z ·wH2O + (1.5-y)H2SO4

Solid composition analyses for the various phases characterized in this study are provided in the supplementary material (Table S1).

4.3.2 Characterization methods

The Raman laser excitation was provided by a polarized argon laser operating at 514 nm. The laser delivered 25 mW at the laser exit and 8mW at the sample using the 50x short distance objective. An average of 10 scans was collected from 150 to 4000 cm-1 to improve the resolution and the statistics of the collected data. The energy resolution was 4 cm-1 at the full width half max of the internal Si reference peak. The scans generally were collected at 10 % of the laser output at the microscope exit, to avoid radiation damage. The sample was in addition inspected with the microscope objective after each scan. The system was calibrated to the 520 cm-1 Silicon peak (for position and intensity) before the collection of any data. Data collection and spectral treatment were performed with the WiRE 2.0 software from Renishaw. Additional details on methods used may be found in Appendix 8.2.

101 Infrared spectra were obtained using a Perkin Elmer FTIR (Spectrum BX model) spectrometer with a Miracle single bounce diamond ATR cell from PIKE Technologies. Spectra over the mid-IR (4000–550 cm-1) range were obtained by the co-addition of 200 scans with a resolution of 4 cm-1 and a mirror velocity of 0.6329cm/s. Spectral treatment such as baseline adjustment, ATR correction and data collection were performed using the Spectrum software (version 5.02) from Perkin Elmer. Additional details on methods used may be found in Appendix 8.2

The XRD analysis was performed on a Rigaku Rotaflex D-Max diffractometer equipped with a rotation anode, a copper target (λ Cu Kα = 1.5406 Å), a monochromator composed of a graphite crystal and a scintillator detector. The diffractometer used 40 kV and 150 mA. Scanning took place between 10 and 100 deg 2θ with a 0.1 deg step with an acquisition time of 3 seconds by step.

4.4 Results and Discussions The hydrothermal conditions associated with the precipitation-production of the three phases characterized have been the subject of previous publications.(Gomez et al., 2008; Gomez et al., 2010) and subject of Chapter 1. Briefly the three phases, scorodite;

FeAsO4·2H2O, ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) and basic ferric

arsenate sulfate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O) were found to form under the following conditions. The scorodite precipitation is favored in the temperature range 150 - 175 °C and Fe(III)/As(V) molar ratio in the staring solution of 0.75 to 1.5. At 200 °C, scorodite was found to form only as an intermediate phase (1 hr reaction time) converting with time to ferric arsenate sub-hydrate. The basic ferric arsenate sulfate tends to form when the Fe(III)/As(V) molar ratio > 1.5 and temperature 175 - 225 °C. Finally, ferric arsenate sub-hydrate is the stable phase at temperatures ≥ 200 °C and Fe(III)/As(V) molar ratio of 0.75 to 1.5.

102 4.4.1 Scorodite (FeAsO4·2H2O)

Scorodite (FeAsO4·2H2O) exists in the orthorhombic crystal system with a space

group D2h and Z = 8. (Kitahama et al., 1975; Hawthorne, 1976) Fragmented vibrational

spectra analysis of this hydrated ferric arsenate (FeAsO4·2H2O) has been reported by many previous authors over the last few decades. (Griffith, 1970; Kitahama et al., 1975; Hawthorne, 1976; Coleyshaw and Griffith, 1994; Baghurst et al., 1996; Myneni et al., 1998; Ondrus et al., 1999; Filippi et al., 2007) In all these works, suggested assignments were made based on the usual stretching and bending vibrations expected for the ideal tetrahedral arsenate molecule.(Vansant et al., 1973) Vibrational analysis has been routinely used to predict the distribution of the internal modes of the arsenate molecules in the crystal system (i.e. site-group or factor-group analysis) of mineral arsenate forms(Martens et al., 2003; Frost et al., 2004; Frost et al., 2007) due to the intermolecular interactions and symmetries of molecules that occur in the lower symmetry states of mineral crystal systems, unlike that of ideal states (gaseous). This change in symmetry may cause the removal of degeneracy of vibrational modes, activate (IR and/or Raman) and/or render inactive vibrational modes. Moreover, interionic (intermolecular) exchanges in the crystal lattice may cause shifts in the energy positions of vibrations relative to those reported in the ideal gas state or for natural mineral samples. Furthermore the vibrational spectrum obtained is also composed of lattice modes (vibrations due to rotation and translations of the molecule in the crystal lattice) at ≤ 300cm-1. Our analysis focuses on the distribution of the internal modes of arsenate and

water molecules. It should be noted that all the arsenate and water molecules occupy a C1 3- site symmetry. The internal modes of the H2O and the AsO4 molecules of scorodite

(FeAsO4·2H2O) are presented in correlation schemes in the supplementary material (Table S2 and S3). Factor group analyses used to predict the distribution of the internal 3- modes of the AsO4 and the H2O molecules were found to be as follows:

3- Γ(AsO4 ) = 9(Ag (Ra)+ B1g (Ra)+ B2g (Ra)+ B3g (Ra)+ Au (In.)+ B1u (IR)+B2u (IR)+

B3u(IR)).

Γ(H2O) = 3(Ag (Ra)+ B1g (Ra)+ B2g (Ra)+ B3g (Ra)+ Au (In.)+ B1u (IR)+B2u (IR)+

B3u(IR)).

103

The IR and Raman spectra of synthetic scorodite (FeAsO4.2H2O) are shown in Fig. 1 and the proposed band assignments are given in Table 1 shown below.

Table 1. Raman and IR Spectral assignment for Scorodite, FAsH and BFAS.

104 4.4.1.1 Vibrations of OH units and H2O molecules

Wavenumber(/cm-1) Wavenumber(/cm-1)

1000 900 800 700 600 3600 3200 2800

580 900

720

Trasmission Intensity Trasmission Intensity Intensity Trasmission 3516 795 2960

800 3516

830 ) ) 3080

870 315 181 450 420 292 880 340 280 483 Raman Intensity Raman

380 Raman Intensity 250 235 215 1000 800 600 400 200 3600 3200 2800 -1 Wavenumber(/cm ) Wavenumber(/cm-1)

Figure 1. ATR-IR (top) and Raman (bottom) spectra of synthetic Scorodite.

Figure 1 displays two IR and Raman active bands at around 3506 (sharp) and 2949 cm-1 (broad). These two vibrations correspond to hydroxyl stretching vibrations from water molecules in scorodite. In scorodite both of these water molecules are cis bonded to iron octahedra and are placed in the tunnel structure along the c - axis. (Hawthorne, 1975; Xu et al., 2007) Moreover, these two vibrations in the hydroxyl stretching region can be correlated to two distinct types of hydrogen bonding that occurs in the crystal system between the water molecules (H-bond donors; Lewis acid) and the arsenate group (H- acceptor; Lewis base) in the crystal structure of scorodite and other similar isostructural III M AsO4·2H2O derivatives (Fig. 2). (Le Berre et al., 2007)

105

These electrostatic interactions, which redistribute electronic charge, are important in molecular crystal systems as they give rise to more order in the structure(Xu et al., 2007) and thus tend to stabilize the system giving rise to tunnel-like structures along the c - axis in scorodite and other related hydrated metal arsenates.(Summi de Portilla, 1974; Jakeman et al., 1991) The OH stretch at lower wavelengths is produced from the first type of H-bonding in the scorodite system which is a strong type of H-bonding (shorter O-H lengths) that occurs between the crystalline water groups and the oxygen of the arsenate molecules. The higher wavelength sharp OH stretch can be correlated to the weaker H- bonding (longer O-H bond lengths) between the oxygen atoms. All these vibrational correlations in terms of bond lengths can be verified with existing crystallographic information for scorodite.(Hawthorne, 1976; Krause and Ettel, 1988) In agreement with the present analysis, Baghurst et al. (1996) reported two similar bands at 3511 (sharp) and 2927 cm-1 (broad) corresponding to hydroxyl stretching vibrations from water molecule in their synthetic scorodite. Previous works(Swash, 1996; Ondrus et al.,

1999; ,31) have reported the OH stretching and H2O stretching frequencies at 3516 and

106 3300 cm-1 for natural and synthetic scorodite using transmission mode IR. However, in the case of the natural mineral scorodite, impurities such as phosphates were reported but no link on how the energy positions and over all appearance of the spectra can be affected by impurities was mentioned. Similarly, synthetic scorodite and other iron arsenate phases characterized using transmission IR mode(Swash, 1996) relevant to this study lacked the fine structure and resolution to extract any vibrational information for the various iron arsenate phases produced. It is noted here that the mode of collection is very important to consider as different samples give rise to better signals for detection in different IR collection modes. The ATR mode is especially suited for opaque powders, thin films and solutions while transmission mode often tends to “smear” vibrations due to the less light able to transmit or be detected in opaque solids, films and/or dilute solutes (i.e. poorer signal is collected by the detector). The shift to lower wavenumbers’ in our

case for the hydroxyl stretches may be due to the strong hydrogen bonds in FeAsO4.2H2O and/or more likely small variations in wavenumbers’ encountered from sample to sample due to small variations in the elemental compositions, not to residual organic materials as claimed by Ondrus et al. (1999) since these bands have been observed in both natural and synthetic products of previous works.(Hawthorne, 1976; Baghurst et al., 1996) The bending vibration of the water molecule in our case occurs at 1620 cm-1 in reasonable agreement with values reported by Ondrus et al. (1999)(1619 cm-1) and Baghurst et al. (1996)(1587 cm-1) for the natural and synthetic scorodite, respectively.

107 3- 4.4.1.2 Arsenate (AsO4 ) stretching and bending modes

3- The Analysis of the vibrational spectra gives the following results. The free AsO4 ion has the four normal modes of vibration under ideal tetrahedral symmetry, these are: -1 -1 υ1(A1) symmetric stretching (818 cm ), υ3(F2) antisymmetric stretching (786 cm ), υ2(E) -1 - symmetric bend (350 cm ) and the υ4(F2) antisymmetric bending mode (405 cm 1 3- ).(Vansant et al., 1973) However, in the case of scorodite the ideal AsO4 tetrahedral symmetry is not kept in the site symmetry nor under the crystal field and as a result the 3- distribution of the internal modes of the AsO4 observed belong to the representations

6(Ag + B1g + B2g + B3g) and 6(B1u + B2u + B3u) which are Raman and infrared active respectively (Table S3). These modes appear in the infrared spectrum as shoulder band at 900 cm-1 and in Raman spectrum as a medium intensity band at 880, 870 and 830 cm-1.

The most intense bands observed in the Raman spectrum at 800 cm-1 and in the -1 3- infrared spectrum at 795 cm are assigned to the symmetric (νsAsO4 ) stretching vibration, while the shoulder which appears at 720 cm-1 is attributed to As-O-Fe -1 stretching vibration. The band that appears at 575 cm is attributed to Fe-OH2 stretching vibration.(Ondrus et al., 1999) The bands observed in the Raman spectrum at 483, 450 -1 3- and 420 cm correspond to υ4(AsO4 ) antisymmetric bending, while those observed at -1 3- 380 and 340 cm correspond to υ 2(AsO4 ) symmetric bending. All bands observed below 340 cm-1 are attributed to external modes.

The scorodite orthoarsenate group would be expected in crystal site symmetry to have six frequencies, both IR and Raman active in the As-O bond stretching and bending 3- 3- region (the υ3 and υ4 region of AsO4 ion). The crystal site symmetry of the AsO4 ion indicates that all of the frequencies should be IR and Raman active. A comparison of the IR and Raman data shows indeed that there is much agreement. This effect can be attributed to site group splitting, which would not give rise to mutual exclusion between the IR and Raman frequencies. This point may be illustrated by the wavenumbers’ 3- derived from the υ1 symmetric band of the AsO4 ion where the Raman assignments are consistently 5 cm-1 higher than the IR assignments. For scorodite crystals, nine 3- frequencies were observed in the Raman spectrum. Since the AsO4 groups are located at

108 sites of C1 symmetry, site group analysis would yield 9A mode both IR and Raman active, while factor group analysis yields 36 Raman and 27 IR active frequencies, as stated above. Clearly in scorodite there is considerable site group splitting since a comparison of our vibrational data (Figure 1 and Table 1) shows a concurrence between the IR and Raman bands.

4.4.2 Ferric Arsenate sub-Hydrate (FAsH; FeAsO4·0.75H2O)

The ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) crystallizes in the triclinic system(Jakeman et al., 1991), this was confirmed in this study (Chapter 3) via

XRD(Gomez et al., 2010) with a space group Ci and Z = 4. The arsenate atoms lie on two different C1 crystallographic symmetries and all the H2O molecules occupy a C1 site. The 3- internal modes of H2O and AsO4 of FAsH are presented in the correlation scheme given in Tables S4 and S5 of the supplementary material information. Factor group analyses 3- predict the distribution of the internal modes of the AsO4 and H2O of FAsH to be as follows:

3- Γ(AsO4 ) = 18 Ag (Ra)+ 18Au (IR)

Γ (H2O) = 6Ag (Ra)+ 6Au (IR)

109

Wavenumber(/cm-1) Wavenumber(/cm-1)

1050 950 850 750 650 550 3700 3300 2900 2500

2882 898

860 963 790 3557

841 Intensity Trasmission Trasmission Intensity 3498 3325 740

850 3510 3341 291 810 265 943 495 3568 246

927 ) ) 406 215 3080

519 158 785 454 378 734 875 Raman Intensity Raman Raman Intensity Raman 345 327 181 1000 800 600 400 200 3700 3300 2900 2500 -1 Wavenumber(/cm ) Wavenumber (/cm-1)

Figure 3. ATR-IR (top) and Raman (bottom) spectra of FAsH.

The Raman and IR spectra of FAsH are shown in Fig. 3 and the proposed band assignments are given in Table 1. The interpretation of the vibrational spectra of FAsH 3- will be made on the basis of characteristic vibrations of H2O and AsO4 molecular

groups.

110 4.4.2.1 Vibrations of OH units and H2O molecules

The stretching vibration bands of hydroxyl units in water molecules of FAsH appear in the infrared spectrum as two sharp bands observed at 3557 and 3498 cm-1 and a broad one at 3325 cm-1 and in the Raman spectrum as a doublet in 3510 and 3560 cm-1 (Fig. 3). These two sets of bands correspond to hydroxyl stretching vibrations from the water molecules in FAsH. One water molecule is bonded to an iron octahedron while the other water is related to hydration (placed at a half occupied crystallographic site); both of these water molecules are found in the tunnel structure along c-axis of FAsH.(Jakeman et al., 1991) In addition, these two sets of hydroxyl stretches from water can be correlated to the distinct types of hydrogen bonding that occur in the crystal system between the water molecules (H-bond donors; Lewis acid) and the arsenate groups (H-acceptor; Lewis base) found in the tunnel structure along the c -axis of FAsH (Fig. 4).

111

In FAsH (FeAsO4·0.75H2O), the degree of hydration and therefore the crystal structure produced is totally distinct in spite of the significant chemical similarities to

scorodite (FeAsO4·2H2O) and the anhydrite analogue (FeAsO4).(Xu et al., 2007; Reiff et al., 1993) These structural characteristics reflect the importance that the degree of hydration (and H-bonding) has on the type of crystal structure and its chemical properties. This is especially true with closely related phases such as our ferric arsenate hydrates (FAsH and scorodite) but also observed in other closely related systems such as the gypsum derivatives. Thus, the use of vibrational spectroscopy to analyze the various hydrogen bonding interactions through the hydroxyl stretches in FAsH and scorodite can be used for further fingerprinting purposes but more importantly to see the difference of H-bonding environments between the water and arsenate molecular groups in the crystal system.

As mentioned above, these electrostatic interactions (H-bonding), which redistribute electronic charge in the structure, are important for molecular crystal systems (such as FAsH or scorodite) as they give rise to more order (i.e. stabilize) in the crystal structure.(Xu et al., 2007) In these hydrated ferric arsenates some of the order in the structure is expressed through the formation of tunnel structures along a crystallographic c - axis.(Jakeman et al., 1991) The OH stretches at lower wavelengths are thus produced from the first type of H-bonding in the system which is a strong type of H-bonding (shorter O-H bond length), likely from the crystalline water OH units (Lewis acid) and the oxygen arsenate groups (Lewis base). The higher wavelength stretches can be correlated to weak H-bonding (longer O-H bond length) that likely occurs between the crystalline and hydration water (OH units) and the arsenate groups along the tunnel structure parallel to the c - axis. All these vibrational correlations can be verified with the existing crystallographic information.(Jakeman et al., 1991) The bending vibration of the water molecule in FAsH appears to be active in both the infrared (1620 cm-1) and Raman (1623 cm-1) spectra but is not very distinct in structure with respect to scorodite or any other phase analyzed.

112 3- 4.4.2.2 Arsenate (AsO4 ) stretching and bending modes

The bands observed in the infrared spectrum at 963, 898, 860 and 841 cm-1 3- correspond to υ3 antisymmetric stretching vibrations of both AsO4 groups, while in the

Raman spectrum the antisymmetric stretching vibration υ3 is observed at 943, 927, 875 and 850 cm-1. The doublet observed at 810 and 785 cm-1 in the Raman spectrum is due to 3- υ 1 symmetric stretching vibrations of two different AsO4 groups, while in infrared, it appear as weak band at 790 cm-1. The strong and broad band observed in the infrared spectrum at 740 cm-1 is due to As-O-Fe symmetric stretching, while the equivalent -1 3- vibration in the Raman spectrum appears at 734 cm . The AsO4 bending modes are observed only in Raman spectra because their region of wavenumbers’ is not covered by -1 mid-infrared spectroscopy (i. e. below 550 cm ). The antisymmetric bending mode υ 4 -1 3- vibrations appear at frequencies 519, 495, 454 and 406 cm for both types of AsO4 -1 groups, while the symmetric bending mode υ2 appear at frequencies 378 and 345 cm . All the bands observed below 345 cm-1 can be attributed to external modes. All the

observed bands have Au symmetry species in the infrared spectrum and Ag in the Raman spectrum as expected for a centrosymmetric group.

The full influence of factor group and site effects on the internal vibrations of the arsenate anions is almost observed for all modes in the IR and Raman spectra. For 3- instance, the antisymmetric stretching vibration (νasAsO4 ) should be split into three IR-

active (3A) in full site group effect and three IR-active (3Au) and three Raman (3Ag) in full factor group effect. However, given that we have two types of arsenate groups, this means one can expect to observe the double of these bands. Only four bands in both infrared and Raman spectra (Fig. 3) are observed. Moreover, only four bands are 3- observed for the antisymmetric bending υ4(AsO4 ). This can be explained by the fact that two of the bands are either overlapped with others or too weak to be observed.

Both factor group and site group effects of the internal vibration modes of the arsenate anions have the same wavenumber in both infrared and Raman spectra. That means we cannot determine if the effect site or factor is predominant, on the basis of the observed bands in infrared and Raman spectra. However, the space group for FAsH is

113 centrosymmetric which means no coincidences are expected to occur between the IR and Raman band positions. Most of the observed IR bands are not coincident in position with those observed in the Raman spectrum (Table 1) as expected for a centrosymmetric crystal, which also means the factor effect is predominant in the crystal system. Similar behaviour has been observed for the vibrational spectra of hydrated rare earth orthophosphates.(Assaaoudi et al., 2001)

4.4.3 Basic Ferric Arsenate Sulfate (BFAS; Fe[AsO4]1x(SO4)x(OH)x·wH2O])

The crystal structure of BFAS (Fe(AsO4)1-x(SO4)x(OH)x·wH2O) has not been determined. Previous investigations have alluded to having a possible orthorhombic- monoclinic type of crystal system.(Swash and Monhemius, 1994; Dutrizac and Jambor 2007; Gomez et al., 2008; Gomez et al., 2010) The spectroscopic and crystallographic information of this phase is still very limited and unknown. As a result of this lack of structural information, a full vibrational treatment of the spectral modes could not be fully accomplished like in scorodite and FAsH. However, the vibrational modes and their correlation to the molecular symmetries (bonding) in the crystal system (specifically site substitution of similar molecular groups) are discussed in this report and will be used for the crystallographic determination of BFAS using (high resolution) powder diffraction methods. Infrared and Raman spectra of BFAS are shown in Figure 5.

114 Wavenumber(/cm-1) Wavenumber(/cm-1) 1400 1200 1000 800 600 3700 3300 2900 2500

639 1040 586 1112

Trasmission Intensity Intensity Trasmission Trasmission Intensity Intensity Trasmission 772 914 828 3405

1094 485 3416 1060 225 859 428 180 ) ) 1167 790

928 572

641

378 Raman Intensity Raman Raman Intensity Raman

1400 1200 1000 800 600 400 200 3700 3300 2900 2500 -1 Wavenumber(/cm ) Wavenumber(/cm-1)

Figure 5. ATR-IR (top) and Raman (bottom) spectra of BFAS.

The band appearing at 570 cm-1 in the Raman spectrum and 580 cm-1 in the infrared spectrum is probably due to Fe-OH2 stretching vibration.(Ondrus et al., 1999; Summin de Portilla, 1974) The hydroxyl vibrations of BFAS were found to consist only of a broad band around 3400 cm-1 but to be distinctive with respect to those of FAsH and scorodite. The higher wavenumber vibrations are often attributed to weak hydrogen bonding and the broadness of the vibrations are often reflective of a diffuse network in correlation to weak/diffuse type of H-bonding. It should be noted that the degree of hydration is slight in BFAS (w ~ 0.2 to 0.5) and therefore the water bending (~ 1600 cm-1) vibrations were only observed in some samples but were indistinguishable in vibrational structure/form with respect to those of scorodite and FAsH.

2- 3- A free SO4 ion is the same as AsO4 ion under tetrahedral symmetry and both exhibit four fundamental vibrations; the non-degenerate symmetric stretching mode υ

115 1(A1), the doubly degenerate bending mode υ2(E), the triply degenerate antisymmetric

stretching mode υ3(F2), and the triply degenerate antisymmetric bending mode υ 4(F2). All the modes are Raman active, whereas only υ3 and υ4 are active in the IR under ideal 2- tetrahedral symmetry. The average ideal frequencies for SO4 under ideal symmetry for these modes occur at 983, 450, 1105 and 611 cm-1 respectively (Rasmussen, et al., 2004), 3- -1 and those for AsO4 at 818, 350, 786, 405 cm .(Vansant et al., 1973)

2- 3- In crystal systems, both SO4 and AsO4 ions may occupy lower site symmetries

resulting in the IR inactive υ1 and υ2 modes to become active and the degeneracies of υ 2, 2- 3- υ3 and υ4 modes to be removed. The degenerate υ3 mode of both SO4 and AsO4 is found to be split into three and two components respectively at 1060, 1094 and 1167 cm-1 2- -1 3- for SO4 and 928 and 859 cm for AsO4 (see Raman spectrum of BFAS in Fig. 5).

Moreover, the appearance of the υ1 band which is IR inactive, has been observed at 1060 -1 2- -1 3- cm for SO4 , and 772 cm for AsO4 in the infrared spectrum due to the symmetry

lowering of sulfate and arsenate ions from Td to C2, 2v in the crystal system. The υ4 mode appears as a weak band in both the Raman and infrared spectra at 641 cm-1 for the sulfate -1 3- group and at 453 cm in the Raman spectrum for AsO4 group.

116 4.4.3.1 Molecular solid solution behavior

Figure 6. Raman spectra of three BFAS samples with various arsenate and sulfate contents. All spectra have been normalized to one for comparison.

The Raman and IR of various BFAS samples synthesized will be shown including the BFAS sample presented in Table S1 and corresponding Figure 5. The Raman spectra as we have previously shown appear to be less sensitive to the symmetry/degeneracy and

activity of the υ3 and υ1 arsenate and sulfate vibrations (peak shape and splitting) and more sensitive to the various concentrations in the sample as demonstrated by the change of intensities of the vibrational bands and the supporting elemental analysis via ICP- AES.(Gomez et al., 2008) The respective Raman spectra are shown in Figure 6. It should be noted that the Raman collection of the BFAS samples with various sulfate and arsenate concentrations, were taken under the exact same conditions (laser power, number of scans and accumulation time), after which the data was normalized (to one) to compare the relative intensities. According to the data of Figure 6 there is a striking 3- 2- correspondence between the intensities of the AsO4 and SO4 bands and their wt. %

117 content. Now new data, this time ATR-IR spectra (Fig. 6) is presented further to explore the correspondence between band symmetry, intensity and composition.

Figure 7. ATR-IR spectra of BFAS phases with various arsenate and sulfate contents. The spectra above (samples 1-3) correspond to BFAS phases with higher arsenate content, and the dashed spectra below (samples 4-6) for solids with higher sulfate content.

Figure 7 shows the IR spectra in solid lines for the BFAS phases of higher arsenate content (wt. %) relative to the sulfate and the dotted black lines are for BFAS phases with slightly higher sulfate (lower arsenate) content (wt. %). The observed vibrations (cm-1) and suggested assignments can be found in Table 2.

118 Table 2. IR spectral assignments for BFAS samples with various sulfate and arsenate content.

It should be noted that quantitative analysis using the relative ATR mode intensities of the bands is less precise as there are several variables (such as amount of pressure applied on sample) which may affect the intensities of the IR bands. IR analysis was instead employed qualitatively, as it offers rich symmetry information (related to bonding symmetries) obtained by monitoring the degeneracy and activity of the υ3 and υ1 arsenate and sulfate vibrations in the mid-IR range. The described behaviour of reduction of 3- 2- symmetry from Td to C2 or C2v symmetry for the AsO4 and SO4 group still occurs as the concentration of both groups vary in the BFAS structure as observed in Figure 7.

Here we can see that the degeneracy and activity of the IR υ3 and υ1 arsenate (sulfate) modes is more prominent as the concentration of that molecular group is greater. Thus,

the arsenate (sulfate) group symmetry is reduced from a Td → C2/C2v (bidentate type of

bonding) while that of the sulfate (arsenate) group remains ~ distorted Td or C2/C2v and vice versa. This molecular trend as observed via IR spectroscopy further supports and 3- 2- proves spectroscopically the isomorphic AsO4 ↔ SO4 solid solution behavior of BFAS in which each molecular group can play each other’s role while keeping the crystal system the same (aside from slight lattice expansion) as observed via XRD.(Gomez et al., 2008; Gomez et al., 2010) Similar bonding behavior and symmetry for sulfates has been monitored via ATR-IR spectroscopy in various systems by previous works.(Peak et al., 1999; Zhang and Peak, 2007) This new collected molecular information of arsenate and sulfate molecular group’s behavior in the crystal structure of BFAS, will aid in the future determination of the crystal structure currently under progress.

119 4.4.4 Characterization of an Industrial Arsenate-Containing Residue

The gold mining industry employs autoclaves operating at 190 to 240 °C for the oxidation of arsenical ores. During hydrothermal processing, arsenic (as arsenate) and iron (as ferric) report in a residue which after gold recovery is disposed of in waste (tailing) management sites. Since the form by which ferric arsenate precipitates may impact on the release arsenic from the waste material, it is imperative to develop molecular sensitive methods for its characterization. Till recently such characterization have been biased to XRD analysis.(Swash and Monhemius, 1998; Monhemius and Swash, 1999; McCreadie et al., 2000; Harris, 2000; Dutrizac and Jambor, 2007) The vibrational spectroscopic analysis reported here is shown to be superior versus XRD when it comes to multi-component industrial residue characterizing. To this end, a sample of an industrial arsenate-containing residue given in Table S1 with the analysis 3- was examined. As it can be seen its % As and AsO4 content was only 0.6 and 1.1 % respectively. XRD analysis of the residue and comparison to the three ferric arsenate phases is presented in Figure S1 of the supplementary material. The major peak in the XRD pattern of the residue (at ~ 27° 2θ degrees) was due to the presence of silica. The 3- low % AsO4 content of the unknown ferric arsenate phase was below the detection limit of the XRD and thus was not observed. In contrast ATR-IR proved to be a powerful enough to detect the arsenate form as it can be deduced from the data of Figure 8.

120

Figure 8. The ATR-IR spectra (arsenate and sulfate υ3 antisymmetric stretching region) of an arsenate containing residue and of the different synthetic high 3- 2- temperature Fe(III) - AsO4 -SO4 phases. The spectra have been set to a relative scale for easier comparison.

The Raman spectrum was not possible to collect from the industrial residue in spite of the use of 4 laser wavelengths (488, 514, 632 and 785 nm) and two different spectrometers (bulk and micro). This lack of Raman signal is attributed to fluorescence problems encountered and the inability to employ a wider range of wavelengths to analyze the industrial sample (such as the 232 nm or 1000 nm). Hydroxyl, arsenate and sulfate bands were found to be active and show structure in the mid-IR range. The arsenate and sulfate concentrations (1.1 wt % and 14 wt % respectively-refer to Table S1) were well within the detection limits of the ATR technique. Thus, it is reasonable to assign the bands at ~ 800 cm-1 in Figure 8 as coming from the arsenate moiety in the residue sample. The ATR-IR spectrum of the industrial residue shown in Figure 8 appears to have the same vibrational symmetry with the arsenate and sulfate bands as

121 found in our BFAS phase. Neither scorodite nor FAsH, appear plausible as observed from the ATR-IR (Figure 8) and XRD (Figure 9) spectra.

Figure 9. XRD spectra of the scorodite, FAsH, BFAS and of an arsenate containing industrial residue The shifted position for the arsenate band in the industrial arsenate containing residue vis-a-vis that of the synthetic BFAS phase is likely attributed to impurities in the industrial system. Similar effect is observed if one compares the IR spectra of natural and synthetic scorodite.(Griffith, 1970; Baghurst et al., 1996) It should be noted that the formation of BFAS as explained earlier is favored at high Fe(III)/As(V) molar ratios, which is the case of the examined industrial residue (Table S1).

122 4.5 Summary and Conclusions

Vibrational (IR and Raman) spectroscopy has been used to fingerprint the ferric

arsenate phases, scorodite (FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH;

FeAsO4·0.75 H2O) and basic ferric arsenate sulfate (BFAS; Fe[(AsO4)1- x(SO4)x(OH)x]·wH2O) that may be encountered during the autoclave processing of 3- 2- Fe(III)-AsO4 -SO4 solutions by monitoring the characteristic arsenate, sulfate and hydroxyl vibrational modes. The identification of these phases is an important topic due to the lack of existing spectroscopic data and because these phases are relevant to the stabilization of arsenic (as arsenate) in waste solids generated in hydrometallurgical recovery operations. Vibrational spectroscopy has been shown to be useful in identifying hydrated ferric arsenate phases very similar in chemical nature (composition) such as scorodite and FAsH due to the different molecular environments and symmetries that each molecular group exhibits under the crystal field. In the case of the BFAS phase, we observe distinct arsenate and hydroxyl vibrations but also the occurrence of sulfate vibrational modes, indicative of a basic ferric arsenate sulfate compound unlike that of scorodite and FAsH. Spectroscopic molecular evidence (similar reduction of symmetry) of the isomorphic solid solution behaviour that occurs between the arsenate and sulfate groups in BFAS was observed via Raman and ATR-IR analysis. With the vibrational information for the pure phases gathered herein, the use of vibrational spectroscopy was further extended to the ATR-IR analysis of an industrial pressure oxidation arsenate 3- containing residue (1.1% AsO4 ). Despite the low arsenate content via ATR-IR we were able to determine its arsenate nature, namely the basic ferric arsenate sulfate phase was found, something that was not possible by conventional XRD.

123

4.6 References

Assaaoudi H., Ennaciri A. and Rulmont A. (2001) Vibrational spectra of hydrated rare earth orthophosphates. Vib. Spectrosc. 25, 81-90. Baghurst D. R., Barrett J., Coleyshaw E. E., Griffith W. P. and Mingos M. P. (1996) Microwave techniques for the synthesis and deuteration of minerals, with

particular reference to scorodite, FeAsO 4·2H2O. Mineral. Mag. 60, 821-828. Bluteau M. C. and Demopoulos G. P. (2007) The incongruent dissolution of scorodite — Solubility, kinetics and mechanism. Hydrometallurgy 87, 163-177. Coleyshaw E. E., Griffith W. P. and R. J. Bowell (1994) Fourier-transform Raman spectroscopy of minerals. Spectrochim. Acta. 50A, 1909. Dutrizac J. E. and Jambor J. L. (2007) Characterization of the iron arsenate–sulfate compounds precipitated at elevated temperatures. Hydrometallurgy 86, 147-163. Demopoulos G. P. (2005) On the preparation and stability of scorodite, In Arsenic

Metallurgy, (eds. Reddy R. G. and Ramachandran V.) TMS, Warrendale,

PA, 25-50.

Demopoulos G. P., Droppert D.J. and Van Weert G. (1995) Precipitation of crystalline

scorodite (FeAsO4 · 2H2O) from chloride solutions. Hydrometallurgy 38, 245- 261. Fillipou D. and Demopoulos G. P. (1997) Arsenic Immobilization by Controlled Scorodite Precipitation. JOM. 49, 52-55. Filippi M., Dousova B. and Machovic V. (2007) Mineralogical speciation of arsenic in soils above the Mokrsko-west gold deposit, Czech Republic. Geoderma. 139, 154-170. Frost R. L., Weier M. L., Williams P. A., Leverett P. and Kloprogge J. T. (2007) Raman spectroscopy of the sampleite group of minerals, J. Raman Spectrosc. 38, 574- 583.

124 Frost R. L., Kloprogge J. T. and Martens W. N. (2004) Raman spectroscopy of the arsenates and sulfates of the tsumcorite mineral group. J. Raman Spectrosc. 35, 28-35. Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2008) Novel atmospheric scorodite synthesis by oxidation of ferrous sulfate solution. Part I. Hydrometallurgy 90, 92-102. Gomez M. A., Becze L., Bluteau M. C., Le Berre J. F., Cutler J. N. and Demopoulos G.

P. (2008) Autoclave Precipitation and Characterization of Fe (III) - AsO4 -SO4 phases, In Hydrometallurgy’08 (eds. Young C. A., Taylor P. R., Anderson C. G. and Choi Y.), SME, Phoenix, Az, pp. 1078-1085. Gomez M. A., Becze L., Cutler J. N. and Demopoulos G. P. (2010) On the hydrothermal reaction chemistry and characterization of ferric arsenate phases precipitated from

Fe2(SO4)-As2O5-H2SO4 solutions. Hydrometallurgy. (Submitted) Griffith W. P. (1970) Raman studies on rock-forming minerals. Part II. Minerals

containing MO3, MO4, and MO6 groups. J. Chem. Soc. A, 286-291. Harris B. (2000) The removal and stabilization of arsenic from aqueous process solutions: past, present and future. In Minor Elements. (eds. Young C.A.) SME, Littleton, CO, pp. 30-20. Harris B. (2003) The removal of arsenic from process solutions: theory and practice. In Minor elements (eds. Young C.) TMS, Vancouver, Canada, pp. 1889-1902. Hawthorne F.C. (1976) The hydrogen positions in scorodite. Acta Cryst. B32, 2891- 2892. Jakeman R. J. B., Kwiecien M. J., Reiff W. M., Cheetham K. and Torardi C. C. (1991) A new ferric orthoarsenate hydrate: structure and magnetic ordering of

FeAsO4·3/4H2O. Inorg. Chem. 30, 2806-2811. Juillot F., Ildefonse P., Morin G., Calas G., de Kersabiec A. M. and Benedetti M. (1999) Remobilization of arsenic from buried wastes at an industrial site: mineralogical and geochemical control. Appl. Geochem. 14, 1031-1048. Kitahama K., Kiriyama R. and Bala Y. (1975) Refinement of the crystal structure of scorodite. Acta Cryst. B31, 322-324.

125 Krause E. and Ettel V.A. (1988) Solubility and stability of scorodite, FeAsO4·2H2O: New data and further discussion. Am. Mineral. 73, 850-854. Le Berre J. F., Gauvin R. and Demopoulos G. P. (2007) Synthesis, Structure, and Stability of Gallium Arsenate Dihydrate, Indium Arsenate Dihydrate, and Lanthanum Arsenate. Ind. Eng. Chem. Res. 46, 7875-7882. Martens W., Frost R. L. and Williams P. A. (2003) Molecular structure of the adelite group of minerals—a Raman spectroscopic study. J. Raman Spectrosc. 34, 104- 111. McCreadie H., Blowes D. W., Ptacek C. J. and Jambor J. L. (2000) Influence of Reduction Reactions and Solid-Phase Composition on Porewater Concentrations of Arsenic. Environ. Sci. Technol. 34, 3159-3166. Monhemius A. J. and Swash P. M. (1999) Removing and Stabilizing As from Copper Refining Circuits by Hydrothermal Processing. JOM. 51, 30-33. Myneni S. C. B., Traina S. J., Waychunas G. A. and Logan T. J. (1998) Experimental and theoretical vibrational spectroscopic evaluation of arsenate coordination in aqueous solutions, solids, and at mineral-water interfaces. Geochim. Cosmochim.Acta. 62, 3285-3300. Ondrus P., Skala R., Viti C., Veselovsky F., Novak F. and Jansa J. (1999) Parascorodite,

FeAsO4·2H2O—a new mineral from Kanˇ k near Kutná Hora, Czech Republic. Am. Peak D., Ford R. G. and Sparks D. L. (1999) An in Situ ATR-FTIR Investigation of Sulfate Bonding Mechanisms on Goethite. J. Colloid Interface Sci. 218, 289-299. Rasmussen S. B., Boghosian S., Nielsen K., Eriksen K. M. and Fehrmann R. (2004)

Crystal Structure and Spectroscopic Properties of CsVO2SO4. Inorg. Chem. 43, 3697-3701. Riveros P. A., Dutrizac J. E. and Spencer P. (2001) Arsenic Disposal Practices in the Metallurgical Industry. Can. Metall. Q. 40, 395-420. Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2005) Temperature and seeding effects on the precipitation of scorodite from sulfate solutions under atmospheric-pressure conditions. Metall. Mater. Trans. B. 36, 327-333.

126 Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2006) Acidity, valency and third-ion effects on the precipitation of scorodite from mixed sulfate solutions under atmospheric-pressure conditions. Metall. Mater. Trans. B. 37, 189-197. Sumin de Portilla V. I. (1974) Infrared spectroscopic investigation of the structure of some natural arsenates and the nature of H-bonds in their structures. Can. Mineral. 12, 262-268. Swash, P. M. and Monhemius A. J. (1994) Hydrothermal precipitation from aqueous solutions containing iron (III), arsenate and sulfate, In Hydrometallurgy ’94, Chapman & Hall, New York, N.Y., pp. 177-190. Swash, P. M. (1996) The hydrothermal precipitation of Arsenic solids in the Ca-Fe-

AsO4-SO4 system at elelvated temperature, Ph.D. Dissertation, Imperial College of Science, Technology and Medicine, University of London, London. Swash P. M. and Monhemius A. J. (1998) The scorodite Process: a technology for the disposal of arsenic in the 21st century. In Effluent Treatment in the Mining Industry (eds. Castro S. H., Vergara F. and Sanchez M. A.) Andros Ltd.: Chile, pp. 119-161. Vansant F. K., Van Der Veken B. J. and Desseyn H. O. (1973) Vibrational analysis of arsenic acid and its anions: I. Description of the Raman spectra. J. Molec. Struc. 15, 425-437. Xu Y., Zhou G. P. and Zheng X. F. (2007) Re-determination of iron (III) arsenate dihydrate. Acta Crystallogr, Sect.E. 63, i67-i69.

Zhang G. Y. and Peak D. (2007) Studies of Cd(II)–sulfate interactions at the goethite– water interface by ATR-FTIR spectroscopy. Geochim. Cosmochim. Acta. 71, 2158-2169.

127 5. The Hydrothermal Cu (II)-Fe(III)-AsO4-SO4 System at 150°C

5.1 Abstract

The effect of copper sulfate on scorodite precipitation and its mechanism of formation at 150 °C were investigated. It was determined that scorodite was the dominant phase

formed in all conditions explored (0.7

CuSO4, 0-0.3 M MgSO4 and >90 min reaction time). The produced scorodite was found

to incorporate up to 5% SO4 in its structure: Fe(AsO4)1-0.67X(SO4)X·2H2O, where 01) was found to slow down the precipitation kinetics of scorodite. The presence of divalent cations (Cu or Mg) did not interfere in terms of kinetics or yield the precipitation of scorodite. There was less than 5% of copper precipitating along scorodite with the latter containing < 1% Cu or Mg in the solids. Precipitation under short times and lower temperatures (30 min - 60 min and

90 - 135 °C) revealed for the first time the formation of a Cu-Fe-AsO4-SO4-H2O short lived intermediate (“liquid crystal”) that closely resembled a basic ferric arsenate sulfate (BFAS) type of phase before ultimately converting to the most stable scorodite phase. Finally investigation of an industrially produced via copper concentrate pressure leaching (1 hr and 150 °C) arsenic (0.5%)-containing residue revealed the arsenic form (as arsenate) to resemble that of BFAS.

128 5.2 Introduction Arsenic is a common toxic element encountered in the non-ferrous metallurgical industry and its removal-fixation remains an important topic (Riveros et al., 2001; Miretzky and Cirelli-Fernandez, 2010) as many valuable metals (Cu,Ni,Zn,Co,U, and Au) are extracted from arsenic-containing mineral feedstocks. Of interest to the present work is the precipitation behaviour of arsenic (V) during processing of ores or concentrates in autoclaves under hydrothermal conditions (>100 °C). In particular of relevance is the pressure leaching (PL) of copper sulphide concentrates as exemplified by the CESL proces.(Defreyne et al., 2006; Mayhew et al., 2010) The CESL process typically employs relatively mild pressure oxidation conditions (150 °C, operating pH 1-

3, 1400 kPa oxygen pressure and 60 min retention time) in relatively low acid (H2SO4) environment. During autoclave processing of sulphide feedstocks, arsenic (as arsenate) combines with ferric iron to form crystalline ferric arsenate phases such as scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FeAsO4·0.75H2O, FAsH; previously labeled

as “Type1” or “Phase4”) or basic ferric arsenate sulfate (Fe[(AsO4)1- x(SO4)x(OH)x]·wH2O), BFAS; previously labeled as “Type2” or “Phase3”).(Swash and Monhemius, 1994; Dutrizac and Jambor, 2007; Gomez et al., 2008; Gomez et al., 2010a) By far, the ferric arsenate phase that has been most widely considered and advocated for the immobilization of arsenic (V), especially from arsenic-rich industrial feedstocks is

scorodite (FeAsO4·2H2O).(Filippou and Demopoulos, 1997; Riveros et al., 2001)Detailed information on scorodite’s (natural and synthetic) crystal structure(Hawthorne, 1976; Kitahama et al., 1975; Dutrizac and Jambor, 1988; Xu et al, 2007; Gomez et al., 2010a) and stability-solubility evaluation.(Krause and Ettel, 1988; Langmuir et al., 2006; Bluteau and Demopoulos, 2007) In a previous paper7 we have reported scorodite to be the dominant form in the temperature range 150-175 °C - the temperature of relevance to Cu processing (via pressure leaching) - with Fe (III)/As (V) ratios 0.75 to 1.5 and retention time of 2 to 24 hours. In addition to the type of ferric arsenate phase forming, of particular interest is the possible loss of valuable metals (in this case Cu) via co- precipitation, as well as the effect copper can have on arsenic precipitation and phase formed. The effect of cations (such as Cu, Zn, Ni, Co, Mn) and excess sulfate anion on

129 the precipitation-production of scorodite has been examined mainly at atmospheric conditions (~95 °C).(Singhania et al., 2006; Fujita et al., 2008) Under the latter

conditions scorodite was found to contain up to 1% Cu and 4% SO4 with no measurable effect on its precipitation kinetics/yield. Some limited data on the interaction of copper with autoclave precipitation of scorodite has been reported by Monhemius and Swash (1999). Their tests were done though at 190 °C (and not 150 °C-the temperature at which the CESL process operates) using Fe/As =1, 7 g/L As(V) solutions, without and with the presence of 15 g/L Cu and

20 g/L H2SO4. Under those conditions the authors reported copper to have a negative effect on the kinetics and yield of scorodite precipitation. For example arsenic precipitation efficiency decreased from ~95 to ~80% in the presence of copper, this being further decreased when acid was added. Moreover, they reported the scorodite to have been contaminated with the “Type1” (FAsH) compound and to contain up to 3.4% copper.

It is the aim of this study to: (a) investigate if scorodite (FeAsO4·2H2O) formation is influenced by the presence of foreign cations such as Cu2+ or Mg2+ at 150 °C; (b) verify if copper co-precipitates with scorodite (either as a substituent element or as a separate copper arsenate-sulfate phase); (c) investigate the mechanism of scorodite

(FeAsO4·2H2O) formation and in particular report on an intermediate precursor phase; and (d) relate the findings to industrial practice via the analysis of a bench scale residue generated in a CESL copper pressure leaching test work program.

130 5.3 Experimental Methods

5.3.1 Materials and procedure

The precipitation tests were conducted using 1L aqueous solutions made of reagent

grade chemicals (As2O5·xH2O, Fe2(SO4)3·xH2O, CuSO4·5H2O, MgSO4·7H2O) from Sigma-Aldrich and a 1.6L mechanically agitated (300 rpm) Buchi glass autoclave. All solutions contained 0.3 M Fe(III) and variable amounts of As(V) and Cu(II). The Fe(III)/As(V) molar ratio was varied from 0.7 to 2. The heat up time from room temperature to the target temperature of 150°C took ~ 2.5 hours. The progress of the precipitation reactions was monitored via collection and analysis of samples at various times: 2.5 (t=0 at 150°C), 3.5, 4.5, 7.5 and 12.5 hours.

Table 1. Initial solution compositions and precipitation efficiencies at the end of the reaction period (12.5 hrs).

The initial pH of all solutions was ≤ 1 (free acidity 0.16 M(15.2-17.2 g/L) H2SO4). The experimental conditions for the various precipitation tests-cases investigated are summarized in Table 1. The industrial residue investigated in this study was provided by Teck Metals and was produced in a bench scale pressure oxidation autoclave under the

131 CESL conditions (200 g/L solids, 15g/L Cu, 15g/L free acid, 60 min, 3,000 kPa total pressure and 150 ºC) using a chalcopyrite-bornite concentrate from their Antamina mine in Peru. The major elemental components of the residue were: 0.54% As, 29.3% Fe, 1.92% Cu, 36.3 % S(tot), 32.1% S0.

5.3.2 Characterization methods

All solutions were analyzed for As, Fe, Cu, and Mg concentrations by ICP-AES using a Thermo Jarrel Ash Trace Scan machine. The chemical composition of the final solids was determined by HCl digestion followed by ICP-AES analysis. The powder and selected area X-ray diffractograms (XRD) were recorded with a

Bruker D8-DISCOVER diffractometer equipped with a cobalt target (Co Kα1 radiation, λ = 1.78897 Å), a crystal graphite monochromator, and a CCD detector. The diffractometer used 40 kV and 35 mA. The scans were measured from 5º to 113º 2θ with 0.02o step and acquisition time of 70 seconds per frame. To collect over the whole two theta range (5º to 113º) three frames were taken. The morphological characterization of the produced solids was done with a Variable Pressure Scanning Electron Microscope (VP-SEM) Hitachi S-3000N. Transmission electron microscopy (TEM) images were obtained using a Philips CM-200 microscope operating at 200 kV (λ ~ 0.025 Ǻ). The samples were prepared by dropping dilute solutions of the particles in ethanol onto 400-mesh carbon-coated copper grids and letting the solvent evaporate at room temperature. Selected area electron diffraction (SAED) was also conducted on selected particles. The gels produced were analyzed using polarized light microscopy and were prepared on standard microscopy slides for observation. Pictures were taken with a Nikon DS Camera control unit DS-U2 on a Nikon Polarizing Microscope Eclipse LV100POL. A full-wave retardation plate into the optical path of the microscope was used to allow the observation of different orientations of nanocrystals (texture) and the optical properties of the material (isotropic versus anisotropic materials), further information of the use and application of this technique and microscope refer to Nikon website (see references).

132 Infrared spectra were obtained using a Perkin Elmer FTIR (Spectrum BX model) spectrometer with a Miracle single bounce diamond ATR cell from PIKE Technologies. Spectra over the 4000–550 cm-1 range were obtained by the co-addition of 100 scans with a resolution of 4 cm-1 at the FWHM of the internal Polystyrene strongest C-H vibration. Additional details on methods used may be found in Appendix 8.2. Raman spectra were collected by an InVia Raman microscope from Renishaw in normal and Confocal mode. Laser excitation was provided by a polarized operating at 514 nm. The laser beam produced a spot size of approximately ≤ 5 μm in diameter using the 50x short distance objective. Averages of 10 scans were obtained from 1700 to 150 cm-1. The energy resolution was 4 cm-1 at the full width half max (FWHM) of the internal Si reference peak. The scans were collected at 10 % of the laser output at the microscope exit to avoid radiation damage. Additional details on methods used may be found in Appendix 8.2.

5.4 Results and Discussions

5.4.1 Precipitation of Scorodite

5.4.1.1 Equilibrium results

A summary of the precipitation results based on the analysis of the final solutions and solids can be found in Tables 1 and 2. It should be noted that these results correspond to the solutions and solids after 10 hours of reaction at 150 C (or 12.5 hrs including the heat up period). Given the elevated temperature at which the tests were run, these results for all practical purposes are considered to reflect near equilibrium conditions. Essentially nearly complete arsenic and iron precipitation (> 90%) was observed in all tests (Case 2, 4 and 5) with Fe/As molar ratio equal to one (corresponding to the stoichiometry of scorodite). In contrast to Monhemius and Swash (1999), the presence of copper (or Mg; compare Case 2 to Case 4 or 5 in Table 1) did not appear to influence the arsenic precipitation yield.

133 Table 2. Composition of precipitated solids at the end of the reaction period (12.5 hrs).

With reference to iron and arsenic precipitation as a function of the Fe/As molar ratio we observe the following. In Case 1, corresponding to nominal Fe/As ratio of near 2, the highest % As precipitation yield was observed (97.6%) with the equivalent %Fe yield being only ~60%. Exactly the opposite happened in Case 3, where a Fe/As ratio < 1 (~0.7) was used. This suggests that iron “follows” arsenic, i.e. the two elements precipitate as a single compound, scorodite (more on its characterization later) as evidenced by a number of factors. First the molar ratio of the amounts of iron and arsenic precipitated (with some deviation from Case 1 & 3) was found to be around 1 (see Table 1). Similarly, the unity molar Fe/As ratio was obtained from the analysis of the solids (Table 2) which corresponds to the theoretical scorodite Fe/As molar ratio.

134 5.4.1.2 Kinetic results

According to the data plotted in Figure 1 (below) the precipitation of arsenic and iron was essentially complete once the target temperature of 150 °C was reached (2.5 hrs), independent of the absence (Figure 1a and b) and presence of copper (Figure 1c and d). There may be seen only in Case 1 in which the Fe/As ratio was above 1 (1.87), that precipitation was not yet complete (90.2% after 2.5 hrs, vs. 95.1% after 3.5 hours and 97.6% after 12.5 hours). Such retardation effect on the precipitation of scorodite as a result of the presence of elevated (>1) Fe/As molar ratio was also observed previously by Monhemius and Swash (1999). At this point it is worthy to point out that Monhemius and Swash (1999) have wrongly labeled their tests as iron injection experiments at temperature as stated in their manuscript: “experiments were carried out by heating the arsenic bearing solutions (without iron) in the autoclave until the desired temperature was reached, then a solution containing the required amount of iron was injected into the autoclave in order to start the precipitation reactions”. However, upon consulting the dissertation thesis of P.M. Swash (1996) it was found that the figures published in their kinetic studies (Monhemius and Swash 1999) were indeed the exact same ones with those that involved heating-up of the Fe-As solution as done in this work. The experiments involving iron injection at temperature (Swash, 1996) exhibited much faster kinetics than the corresponding ones involving heat-up of the solution. This explains the agreement between the results of the present work (that involved the heat-up period) and those reported in Monhemius and Swash (1999).

135

18 (a) Fe 16 As 14 12

Concentration (g/L) 4

0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Reaction time (hrs)

Figure 1. (a)Concentration profiles as a function of reaction time for the Case 1. The reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached.

24 22 (b) Fe 20 As 18 16 14 12

10 8

Concentration (g/L) 6 4 2 0

024681012

Reaction time (hrs)

Figure 1. (b) Concentration profiles as a function of reaction time for the Case 2. The reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached.

136

35 Fe 30 As

25 Cu

Concentration (g/L) 10

0 024681012 Reaction time (hrs)

Figure 1. (c) Concentration profiles as a function of reaction time for the Case 3. The reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached.

30 Fe 25 As Cu 20 Mg

10 Concentration (g/L)

024681012

Reaction time (hrs)

Figure 1. (d) Concentration profiles as a function of reaction time for the Case 5. The reaction time of 0 hrs corresponds to the beginning of the experiment at 25 °C and 2.5 hrs corresponds to the time the target temperature of 150 °C was reached.

137 In terms of copper and magnesium there was no significant variation of their solution concentration with time (Figure 1c and d) suggesting little (5.8% Cu and 3% Mg according to Table 1) or no precipitation involvement for them. Thus in the Case 3 experiment involving low Cu concentration (4.3 g/L), there was about 9% of Cu coprecipitated when temperature had reached 150 °C (2.5 hrs) but this was decreased to 4.6% at the end of the reaction (12.5 hrs) with only 0.4 % detected in the solid. The final %Cu co-precipitation numbers for the higher copper concentration tests (Case 4 and 5) were 3.2, 6.8% and 0.3, 0.8% found in the final solid respectively-refer to Table 1 and Table 2). However, more careful analysis of samples collected during the heat-up period from the test of Case 4 revealed the temporary co-precipitation of copper along iron and arsenic to be more significant. This is investigated in section 3.2.

5.4.1.3 Characterization of Scorodite

The chemical composition of all precipitated scorodite products is shown in Table 2. Their ferric and arsenic content fell between 21.1 % to 24.1 % and 28.8 % to 35.1 %

respectively, which on average are close to the theoretical values (Fe 24 % and As 32.3 %). All scorodite products were found to contain certain amount of sulfate (from 0.3 to

4.8%-Table 2) due to the sulfate ↔ arsenate substitution: Fe(AsO4)1-0.67x(SO4)x ·2H2O. The presence of sulfate in the scorodite structure was further verified via infrared spectroscopy, as discussed in a recent paper by Gomez et al. (2010a) and shown below. There was good correspondence between the above stoichiometric formula in terms of x

(which varied from 0 to 0.16) and the experimentally determined SO4 composition (refer to data in Table 2). Similarly, the degree of sulfate incorporation was observed previously for the case of atmospheric production of scorodite from sulfate media.(Singhania et al., 2005; Singhania et al., 2006) Concerning Cu and Mg there were only minor amounts that found to report as impurities in the scorodite solids, namely less than 0.6%-in contrast to the high copper content (3.4%) reported by Monhemius and Swash (1999). The corresponding % Cu content of atmospherically produced scorodite has been reported to be ~1% and 2%.(Singhania et al., 2006; Fujita et al., 2008) The low (0.6%) Cu content in the present work most likely reflects structural incorporation within scorodite while the

138 higher amounts reported by Monhemius and Swash (1999) and Fujita et al. (2008) may reflect separate copper arsenate co-precipitated phase(s) as suggested in their works. No such copper arsenate or sulfate phases were detected in the present work via either XRD or vibrational spectroscopy. The incorporation of the divalent cations into the scorodite(Kitahama et al., 1975; Hawthorne, 1976; Xu et al., 2007) crystal structure (Cu2+ or Mg2+ for Fe3+, Table 2) apparently is not favorable in contrast to trivalent cation incorporation discussed elsewhere.(Le Berre et al., 2007; Gomez et al., 2010b) Thus one could rationalize the extra affinity of M3+ ions such as Fe3+, In3+, Ga3+, and Al3+ towards arsenic (in solution or in the final solid covalent crystal asymmetric repeating units) versus that of M2+ ions (such as Cu2+, Mg2+ in this study or others encountered in the CESL processes Ni2+,Co2+, Zn2+ ) may arise from the difference in the charge of the metal cation (+2 vs. +3) not being sufficient to neutralize the arsenate oxygen charges once the metal (M2+) coordination to arsenate is formed through stable molecular complexes in solution, and until the until the end of crystallization process; where as in scorodite (or other M3+ isostructural phases (Le Berre et al., 2007; Gomez et al., 2010b), the metal (M3+) arsenate complexes are stable in solution and until the end of crystallization reaction via the asymmetric repeating units. (Kitahama et al., 1975; Hawthorne, 1976; Xu et al., 2007; Gomez et al., 2010b) Alternatively, the lower of affinity of the M2+ cation (Mg2+ and Cu2+ vs. Fe3+) towards the arsenate ions may also arise from the electronegativity difference of the metal cations which can coordinate to 3+ 2+ +2 2+ 2+ the arsenate ligands where the electronegativities (χ) of Fe , Mg , and Cu , Co , Zn , Ni2+ are reported to be 1.68, 1.20, 1.51, 1.46, 1.50, and 1.42 respectively.(Zhang, 1982) Therefore it can be seen that ferric iron exhibits the highest electronegativity relative to the divalent cations and should be then the most stable complex in solution and then onto the asymmetric repeating unit during crystallization.

139

Case 1 Case1 Case 2 Case 2 Case 3 Case 3 Case 4 Case 4 Case 5 Case 5 Standard Standard

Arb. Intensity Arb. Intensity

10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 two theta two theta

Figure 2. XRD of the solids produced at 150 °C once the target temperature was reached at (left) 2.5 hrs and (right) at the end of the reaction period (12.5 hrs). All were found to be scorodite

Positive identification of the formation of crystalline scorodite was made by XRD (Figure 2) and molecular analysis [ATR-IR (Figure 3) and Micro-Raman (Fig S1) shown in the Appendix]. As it can be deduced from the XRD patterns, all precipitates consisted of crystalline scorodite independent of the time of production: 2.5 hrs (end of heat-up period-Fig 2(left)) or 12.5 hrs (Fig 2(right)).

140 However, the molecularly-sensitive ATR-IR analysis (Fig 3(left)) showed that in Case 1 (Fe/As=1.9) at 2.5 hrs (once the target temperature was reached), not all the H- bonding in the system had been fully developed as evidenced by the broad band in the high wavenumber region (3000-4000 cm-1). The importance of H-bonding in the crystal structure of scorodite is discussed elsewhere.(Gomez et al., 2010a; Gomez et al., 2010b) This implies that the crystallization kinetics of scorodite were slower in this case (high Fe/As ratio~1.9) as opposed to the other cases with Fe/As  1). This observation is in agreement with the slower precipitation kinetics measured (refer to section 5.1.2) and also reported by Monhemius and Swash (1999). It may be tentatively postulated that the presence of excess ferric iron alters the nature of the precursor Fe(III) - sulfato (and arsenate) complexes hence the crystallization kinetics of scorodite are affected.( Disarajoue, 2007; Demopoulos, 2009)

The particle morphology of the scorodite precipitates was evaluated by SEM (Figure 4 and S2 of Appendix). In Case 1 (Fe/As =1.9, no Cu) an outward crystal habit

141 (orthorhombic like crystal) was exhibited by the product once the target temperature was reached (Fig 4a, Fig S2a; 150 °C and 2.5 hrs). But in addition other particles are seen just starting to form their observed outward geometrical crystal shapes. At the end of the reaction (Fig. 4f, Fig. S2f ; 150 °C and 12.5 hrs) most of the particle are observed to have taken their full external orthorhombic - dipyramidal crystal form but with not so well defined edges. This particle morphology evolution is consistent with the observation of slower precipitation (Figure 1a) and crystallization (Figure 3(left)) kinetics noted earlier. In contrast, when the Fe/As ratio was ~1 (as in Case 2), the particle morphology appeared to have fully developed upon reaching the target temperature of 150 °C after 2.5 hrs (Fig. 4b, Fig. S2b) with most of the particles exhibiting good external orthorhombic - dipyramidal crystal shape analogous to that of particles produced at higher temperature and longer times (Fig S2c and Fig S2i; 160 °C and 24 hrs). The particle morphology of the scorodite produced in the presence of copper (Case 3 and 5 - see Figures 4c,h, and 4e,j; Fig S2d,S2S and Fig S2f,S2l) was found to consist of rounded particles made up of small crystallites, which were distorted with reference to the orthorhombic - dipyramidal outward crystal form observed in Cases 1 and 2 (Figure 4a,f and 4b,g; Fig S2a,S2f and S2b,S2h). This may reflect alteration of the surface properties of the growing scorodite crystals by the presence of the copper (and magnesium) cations. However, the scorodite of Case 4 produced again in the presence of copper did not manifest the same trend. Thus the scorodite produced after 2.5 hours in Case 4 (Figure 4d, 4i and Figure S2e, S2k) consisted of rounded type particles showing the beginning of their outward crystal form (Fig 4d, S2e), while at the end of the reaction (12.5 hrs-Fig 4i, S2k) the particles were characterized by rounded-cauliflower morphology made up of small well defined orthorhombic crystallites, similar to those found by Fujita et al. (2009) (where As =20g/L, initial pH 0.5 ) as well as in Case 2 of the present work (Fig 4g and Fig S2’c, S2’i). In conclusion, copper seems to exercise a rather complex influence on the crystallization of scorodite as far particle /crystal habit is concerned without however, in any way reducing the crystallinity of the final product (refer to Figures 2 and 3).

142 5.4.2 Formation of a Cu(II)-Fe(III)-AsO4-SO4 intermediate 5.4.2.1 Solution Changes A benefit of using a glass autoclave as done in this work is that it makes possible to monitor visually the solution changes as the reaction goes through the various precipitation (nucleation, growth, transformation etc) stages. Thus during the heat up period of Case 4 test (Fe/As = 0.90, Cu= 0.29 M (18.6 g/L)) the presence of a gelatinous type of precipitate was briefly observed, which subsequently disappeared once the target temperature (150 °C) was reached, at which point well formed scorodite particles appeared (Figures 2 left, 3 left and 4d). This prompted the closer look of this phenomenon via the collection of solution and gel samples during the heat up period. Collection was achieved by taking a sample from the autoclave at different temperatures and times and immediately centrifuging for 3 minutes at 1000 rpm to separate the gel phase from solution. It should be noted that all experiments conducted in this chapter and

40 Fe As S 30 Cu Acid

Concentration (g/L) Concentration 10 Case 4

02468101214

Reaction time (hrs)

143 thesis were conducted at least two times to ensure reproducibility; however, the observation of such a strange phase made the current investigator repeat this a 3rd time to ensure that indeed this phase formed. The collected data is summarized in Table 3 while the obtained solution component concentration profiles are shown in Figure 5 (pg 143).

Table 3. Analysis of solid/gel and solution during the heat up period and beyond of Case 4 experiment.

In general the arsenic and iron concentration kinetic profiles (Figure 5 in pg 143) during the heat up period and beyond followed the trend of precipitation seen previously with the other tests performed (Figure 1). Thus it can be seen that iron and arsenic (99 % Fe and 90 % As) had precipitated entirely once the target temperature of 150 °C was reached (2.5 hrs) and remained approximately the same until the end of the reaction (Figure 5), a behavior we have observed in the previous cases.

144

The copper, sulfate (S), and acid profiles however, were peculiar and unlike any of the other cases (Figure 1). Thus once the target temperature of 150 °C was reached (after the 2.5 hr heat up), copper and sulfur (as sulfate) had been removed from solution up to 71.5 % (5.3 g/L) and 63.2 % (9.7 g/L) respectively soon after returning to solution again (Figure 5). As already mentioned (section 5.4.1.1) the solids obtained at 2.5 and 12.5 hours found to consist only of scorodite. This implies that any copper (and sulfate) containing precipitate that had formed (and responsible for the temporary drop in Cu and S (as sulfate) concentrations) re-dissolved upon solid collection and not appearing in the final scorodite product at 2.5 or 12.5 hrs. Of interest is also the acid concentration profile. The precipitation of scorodite leads to acid generation.(Gomez et al., 2008; Gomez et al. 2010c) In Case 4, though, the acidity decreased from 15.2 g/L at 25 °C to 10.3 g/L at 150°C (after the 2.5 hr heat up), then increased to 35.3 g/L (150°C) at 3.5 hrs and finally to 39.7 g/L at 4.5 hrs and remained constant thereafter signaling the end of the reaction process.

5.4.2.2 Characterization of Cu(II)-Fe(III)-AsO4-SO4 intermediate

Gels separated from the collected samples were analyzed and the results are reported in Table 3. The provided analysis should be considered as semi-quantitative because of the inherent complexity involved in sampling and analyzing such highly hydrated gelatinous substance. The amount of hydration according to TGA analysis was at least 30-40 % hydration. This degree of hydration is also evident from the IR analysis (shown below) which show clear strong bands at 1633 cm-1 and 3000-3500 cm-1 indicative of

H2O and OH/H2O groups. The results nevertheless provided evidence that at least the initial gel that formed (90 C and 30 min) contained significant amounts of sulfate and copper in addition to iron and arsenic as expected.

145

Figure 6. Photographs of products formed at (a) 30 min, 90 °C; (b) 40 min, 101 °C ; (c) 60 min, 125 °C and (d) 96 min, 135 °C during the heat up period of Case 4 test.

A visual examination of the formed gel and its evolution with time can be made with the images in Figure 6. It can be seen that the initially formed precursor gelatinous dark green material (30 min, 90 °C) evolved to a mixture of gel and nuclei of a solid light yellow powder precipitate (40 min, 101 °C); the latter developing to a more solid light yellow powder with a small surrounding gelatinous layer (60 min, 127 °C); and finally assuming the usual pale green color of scorodite (96 min, 135 °C) with all the copper at this point being rejected back into the solution as evident from the blue color of solution. It should be noted that during the heat up period, the color of the solution changed from dark to light blue-green color and then to the final pure blue color (observed after the 96 min period) that remained the same until the end of the test (12.5 hrs). As most of the samples collected were observed to be of a gelatinous or amorphous nature (Figure 6), it was decided to first investigate their molecular structure using techniques (ATR-IR and Raman) which are independent of whether the material is amorphous, semi-crystalline or crystalline.(Horn and Sully, 1999; Deb et al., 2001; Tang et al., 2002; Kohl et al., 2009) Figure 7 and 8 display the ATR-IR and Raman spectra of the gelatinous intermediate products until the formation of scorodite (FeAsO4·2H2O).

146 BFAS

30 min, 90 C

40 min, 101 C

60 min, 125 C MOH SO4 2 OH/H2O H2O AsO4 96 min, 135 C

Scorodite

3550 3050 2550 2050 1550 1050 550

Wavenumbers/cm-1

From the ATR-IR spectra (Figure 7) we can observe that the products obtained from 30 – 60 minutes (90-127 °C) are essentially all the same at the molecular level and exhibit distinct arsenate (771-801 cm-1), sulfate(114-1064 cm-1), water (1625cm-1) and wide diffuse hydroxyl (3052-3350 cm-1) bonding environments which are not the same as in scorodite.(Hawthorne, 1976; Xu et al., 2007; Gomez et al., 2010a; Gomez et al., 2010b ) Scorodite was observed for the first time to appear at 96 min and 135 °C (Fig. 7-8). From the above observations, we can infer that the product formed during the heat up

147 period up to 60 minutes and 127 °C is not scorodite but rather a hydrated arsenate sulfate intermediate which contains various amounts of copper (Table 3).

With the view of identifying this intermediate complex we decided to examine if it resembles one of the other ferric arsenate phases (FAsH and/or BFAS) that are known to form besides scorodite under hydrothermal conditions.(Swash and Monhemius, 1994; Dutrizac et al., 2007; Gomez et al., 2008; Gomez et al., 2010a-2010c) This examination revealed that the arsenate and sulfate bonding environment as expressed in the IR spectra of these gelatinous products (Figure 7) exhibits a similar symmetry and profile as the basic ferric arsenate sulfate (BFAS) described elsewhere.(Gomez et al., 2010a) In particular it is seen the samples of interest (30 – 60 minutes and 90-127 °C) to exhibit the characteristic BFAS vibrational features where some degree of degeneracy of the -1 -1 triply degenerate ν3 (F2) - arsenate (730-806 cm ) and sulfate (1051-1108 cm ) modes are removed in the infrared spectrum. This results from a decrease in the symmetry of the arsenate and sulfate molecules which go from their “ideal” tetrahedral symmetry (Td) in solution to a lower symmetry (C2,2v) in the intermediate gel product to a final C1 site symmetry (96-760 min, 135-150 °C) as that found in scorodite.(Hawthorne, 1976; Xu et al., 2007; Gomez et al., 2010a) Furthermore, we can also infer from the vibrational symmetry of the arsenate and sulfate molecules that a C2,2v - bidentate binuclear or bidentate mononuclear metal bonding coordination(Peak et al., 1999; Zhang and Peak, 2007) exists in the phase formed at 30 – 60 min (90-135 °C) similar to that found in the BFAS phase.

148 AsO4

BFAS SO4

30 min, 90 C

40 min, 101 C

60 min, 125 C

96 min, 135 C

Scorodite

200 400 600 800 1000 1200 1400 1600

Wavenumbers/cm-1

The BFAS features (similar sulfate and arsenate molecular environment) of the gels (samples of 30 – 60 min from 90-127 °C) were also evident in the collected Raman spectra (Fig 8). At 96 min (135 °C), only arsenate vibrations from the scorodite product were observed and most of the sulfate has been expelled from the solid (Table 3). Similar Raman spectroscopic monitoring has been recently reported for the kinetic mechanism of aluminum phosphate synthesis(Fan et al., 2009) but never for the scorodite formation. In particular, the arsenate and sulfate profile of the bands at 700-900 cm-1 and 1047 -1202 cm-1 are quite broad in the Raman spectra, indicative of a diffuse/disordered arsenate and

sulfate type of environment. The degenerate ν3(XO4) mode of both arsenate and sulfate

149 groups was found to split (removed due to lowering of symmetry from ideal Td) into two −1 2− −1 3− components respectively at 986 - 1252 cm for SO4 and 780 - 962 cm for AsO4 suggestive of a C2,2v type of symmetry for arsenate and sulfate groups similar to that found in BFAS and in agreement with our IR analysis.(Gomez et al., 2010a) Furthermore, it can be observed from Figure 8, that the lower wavenumber Raman active -1 3+ 2+ -1 arsenate and sulfate modes (~450 cm ), M-OH2 where M = Fe , Cu (~450 cm ) and lattice type of modes (<300 cm-1) of the gels collected at 30 – 60 min and 90-127 °C are the same as in BFAS; while the corresponding modes of the product obtained at 96 min and 135 C are the same found in scorodite. (Gomez et al., 2010a)

Figure 9. Polarized microscopy images of the intermediate gelatinous product obtained at 40 min and 101 °C. The scale on the images is 100 μm.

The intermediate green gel product (40 min, 101 °C) was further analyzed with

polarized microscopy, selected area XRD (with Co Kα1, λ = 1.78897 Å), and TEM- SAED. Polarized microscopy (with a full wave retardation plate) images of the gel product are shown in Figure 9. This technique was chosen as it is useful for determining whether materials are isotropic or anisotropic; as well it is useful when investigating the structure of anisotropic solids and liquid crystals on a mm scale.(Bellare et al., 1990; Gray, 2008; Hanley, 2010) The alignment of crystals and the types of crystals under the polarized microscope can be observed based on the difference of refraction index each plane or components it may contain. For the gel material in question (Figure 9), it is observed that there are regions where texture (expressed as different colors, blue and yellow in the material) and orientation of ordered domains (likely as nano-crystals) are

150 similar to those encountered in birefringent anisotropic liquid crystals.(Bellare et al., 1990; Gray, 2008; Hanley, 2010) But in addition there exists a gelatinous isotropic component (observed as transparent and of the same color as the full wave retardation plate-“pink”) with the same optical properties in one direction, i.e. with only one index of refraction, indicating that no texture or nor crystal orientation are present in these regions.

Selected area XRD analysis (Figure 10) on two different spots of the gel material showed an amorphous signal typical of a material with poor long range order (as expected) in one region but a diffraction pattern indicative of a semi-crystalline phase on the other region (in agreement with the polarized microscopy results). Calculated d spacings of the XRD diffraction peaks (using λ=1.78897Ǻ) for the amorphous and semi- crystalline components of the gel material are reported in Table S1 of Appendix. More

151 importantly, the XRD pattern of the semi-crystalline component of the gel (Figure 10) was not as of scorodite nor any other phase found in the JCPS database.

TEM and SAED images collected from the gel product are shown in Figures 11 and Figure S3. It can be seen that the particles appear to be quite dense in some regions but transparent in others. The SAED of the particles showed both amorphous and polycrystalline features in different regions suggesting that some atomic planes and order must exist in this gelatinous material (Figure 11). EDS analysis (Figure 11) indicated that the gel contained Fe, As, S, O and Cu (not from grid) atomic units in agreement with the ICP-AES elemental (Table 3) and IR/Raman molecular analysis (Fig. 7-8).

152

Figure 12. (a) TEM of the semi-crystalline component of the gel product (40 min and 101 °C) showing the formation of ordered atomic planes from atoms in the unordered phase to the lattice fringes for one particle. All lattice fringes that were observed were measured to be 3.27 Å, which are close to the lattice spacing observed in BFAS/Type2/Phase 3 (3.25Å).( Swash, 1996; Dutrizac and Jambor, 2007; Gomez et al., 2010c) The corresponding lattice spacings in scorodite are usually 3.17 and 4.45 Å. (Swash, 1996; Dutrizac and Jambor, 2007 Gomez et al., 2010c)

153

Figure 12. (b) TEM of the semi-crystalline component of the gel product (40 min and 101 °C) showing the formation of ordered atomic planes from atoms in the unordered phase to the lattice fringes for a distinct particle. All lattice fringes that were observed were measured to be 3.27 Å, which are close to the lattice spacing observed in BFAS/Type2/Phase 3 (3.25Å).( Swash, 1996; Dutrizac and Jambor, 2007; Gomez et al., 2010c) The corresponding lattice spacings in scorodite are usually 3.17 and 4.45 Å. (Swash, 1996; Dutrizac and Jambor, 2007 Gomez et al., 2010c)

154 To verify if these atomic planes and order were indeed present in the gelatinous intermediate phase, higher magnification TEM images were taken (Figures 12 and Figure S3-S4 in Appendix). From these images lattice fringes of atomic planes were indeed revealed. These were found to have spacings of 3.27 Å close to the 3.25 Å spacing of BFAS. In an astonishing way, a closer look of the TEM images (Figure 12 and Figure S3- S4 in Appendix) revealed for the first time, the nanoscale organization of disordered atoms to atomic planes and the creation of lattice fringes out of the amorphous gel for an inorganic crystal system. It is further noted that according to TEM images (Fig S5a-S5b) this gel exhibits different lattice fringe directions, which may explain the texture observed in the polarized microscopy images (Fig 9). In addition the gel appeared to be highly porous (Fig, S5b-S5d). At this point is worth noting that the gelatinous products (Fig. 6) were stable at ambient conditions after 1 week of producing them (Figure S6). The yellow type of solid powder eventually decomposed but interestingly enough the dark green gel product seemed to remain stable even after 7 days (Figure S6). From the above described analysis it is apparent that this gelatinous dark green phase formed in Case 4 (30 – 60 min, 90-127 °C) (shown in Figure 6) is a precursor/intermediate copper-carrying ferric arsenate sulfate hydrate species, which converts (via a dissolution-recrystallization mechanism) at longer times (96 min-12.5 hrs) and higher temperatures (135-150 °C) to scorodite via the exclusion of copper and sulfate.(Desiraju et al., 2007; Demopoulos, 2009) The formation of this type of short- lived intermediate complex has not been observed/reported in previous scorodite precipitation systems either under hydrothermal (Monhemius and Swash, 1999) or atmospheric pressure conditions (Singhania et al., 2006; Fujita et al., 2008).

155 5.4.3 Characterization of a copper pressure leach residue

Figure 13. Picture of the as-received CESL pressure leaching arsenic-containing residue (left) and the same residue after the extraction of elemental sulfur with toluene (right).

In order to correlate the findings of our laboratory study that involved scorodite precipitation from synthetic solutions to actual copper concentrate leaching conditions as done in Teck ’s CESL process (Defreyne et al., 2006; Defreyne and Cabra, 2009) an industrial pressure leaching arsenic- containing residue (Figure 13, kindly provided by Teck Metals) was characterized. During pressure leaching of typical copper concentrates (as is the Antamina concentrate treated to produce the residue analyzed here) the majority of ferric iron precipitates as various iron oxides-hydroxide phases (mainly as hematite). (Defreyne et al., 2006; Defreyne and Cabra, 2009) Similarly, any arsenic present in the concentrate upon oxidation reports to the residue but its exact form is not known hence the present investigation. The residue studied here, principally composed of iron (29.3%) and sulfur (36.3%) with the latter the bulk being in elemental form (32.1%) and the remaining 4 % as sulfate.

The residue contained only 0.5% As (1 % AsO4), which explains the difficulty associated with its mineralogical identification for which no information was available prior to this study. Essentially all (96.5%) arsenic contained in the concentrate reported to the residue with only 3.5% remaining in the final leach solution.

156 According to previous hydrothermal studies scorodite is expected to form at 150 C-the temperature at which the CESL process operates.(Swash, 1996; Dutrizac and Jambor, 2007; Gomez et al., 2008; Gomez et al., 2010c) However, in those studies (that involved direct precipitation from solution) the Fe/As molar ratio was much lower than the nominal Fe/As ratio (~ 80) encountered in the Antamina concentrate (24%Fe, 0.4% As) investigated here and the reaction time in general longer than one hour. Under the conditions the residue was produced in this study (150 C, Fe/As=80 and 1 hr) arsenic (as arsenate) may also precipitate in addition to scorodite as BFAS as observed for high Fe/As solutions at temperatures above 175 C or alternatively simply co-precipitate via adsorption on the main iron oxide precipitate.

Figure 14. (a) XRD of the CESL industrial pressure leaching arsenic-containing residue before and after the extraction of elemental sulfur with toluene. Reference XRD of hematite is also shown for comparison.

157

Figure 14. (b) Raman of the CESL industrial pressure leaching arsenic-containing residue before and after the extraction of elemental sulfur with toluene. Reference Raman hematite spectra is also shown for comparison.

Figure 14. (c) ATR-IR of the CESL industrial pressure leaching arsenic-containing residue before and after the extraction of elemental sulfur with toluene. Reference hematite is also shown for comparison in the inset.

158 Thus, to determine the arsenic (arsenate) form in the residue, PXRD, Micro-Raman and ATR-IR analysis was conducted. From the XRD analysis (Figure 14a and Figure S7 in Appendix) it was observed that although the sample was of a crystalline form, it was not completely crystallized as evident from the high type of amorphous background. Running this PXRD through the JCPDS database (Fig S7) gave us matches for hematite and elemental sulfur as expected but the confidence in the matches was very low for each case (≤ 45 %) and this was only if there were a restriction of elements of interest in the database search (for examples by only choosing, Fe, S, O, H and As). Moreover, the match to any type of arsenic or arsenate bearing phase (such as scorodite) was not successful even when the appropriate restrictions in the search were imposed (Fe, As, S, O and H). This apparently reflects the complexity of the sample and in particular the minor occurrence of arsenic (0.5%). Thus, after this PXRD analysis was conducted, the molecularly-sensitive Raman Microscopy technique was employed. The Raman spectra of the arsenic-containing residue can be found in Figure 14b. At first glance, no arsenate vibrational modes were apparent at low or high wavenumbers nor the presence of elemental sulfur was detected.(Eckert et al., 1996; Becze et al., 2009) The only phase that was observed via Raman spectroscopy was hematite as observed from comparison to reagent grade hematite from Sigma-Aldrich and literature.(de Faria et al., 1997) To confirm that elemental sulfur (S0) was present, the residue was subjected to elemental sulfur extraction in toluene as described elsewhere.(Becze et al., 2009)

159

The extracted elemental S0 was analyzed via Raman spectroscopy (Fig. 15a). The amount of elemental sulfur present was determined to be ~ 30 %, in agreement with the analysis provided by Teck Metals. Interestingly enough the visual (color) appearance of the residue and its PXRD pattern did not change much upon sulfur extraction suggesting that the latter was amorphous (Figure 13, 14a, S7). But more importantly there was no arsenate phase detected by Raman even after the elemental sulfur was removed from the residue. Therefore ATR-IR spectroscopy was employed. At this point it is worth noting that because the arsenate molecules are not observed/active in one method (Raman in this case), it does not mean that they will not be observed in the complementary IR method.(Nakamoto, 2009) This is so because the “peaks” which are observed in the spectra (transition between vibrational energy states) are governed by selection rules that determine their activity (i.e. whether a ``peak`` is visible). In crystal systems (such as 0 FeAsO4·2H2O, α-Fe2O3 or S ), this is determined by their ion symmetry, their site symmetry and factor group symmetry (Nakamoto, 2009; Gomez et al., 2010a) not to mention other effects that may occur when more than one phase are present in the same sample.(Becze at al., 2009; Filippi et al, 2009) Figures 14c and 15b show the ATR-IR spectra of the residue before and after sulfur removal. As it can be seen, the ATR-IR spectra do not change much after sulfur removal

160 (in agreement with PXRD analysis) and all the molecular groups of interest remain. In general, the whole mid-IR (500-4000cm-1) range (Figure 14c) shows strong sulfate vibrations (1000-1200 cm-1), the water bend mode (~1630 cm-1) and a strong broad absorption due to hydroxyl modes (~3378 cm-1). Small vibrations in the IR region of the -1 ν3 (F2) antisymetric arsenate stretch (786 cm ) were also observed but due to the large signals of the other molecular groups, they could not be clearly deciphered. Thus we decided to focus (“tune”) only on the arsenate vibrations by running multiple scans over -1 the 700-900 cm region. From Figure 15b, it can be observed that tuning to the ν3 (F2) arsenate mode gives a clear description of the arsenate bonding environment. In addition it shows that the removal of S0 does not affect the arsenate or any other molecular species of interest (Figure 14c).

Figure 16. ATR-IR spectra of the ν3 (AsO4) mode, for the CESL industrial pressure leaching arsenic-containing residue compared to other relevant high temperature ferric arsenate phases.(Gomez et al., 2010a)

More importantly, it can be observed that upon comparison to our previous vibrational work (Gomez et al., 2010a) on high temperature phases [such as scorodite, ferric arsenate sub-hydrate (FAsH) and basic ferric arsenate sulfate (BFAS)] that the

161 arsenate form in the Teck residue is not that of scorodite but rather resembles that of BFAS phase (Fig 16) and the short lived intermediate phase described earlier (Fig 7). Another possibility that should also be considered is one where the arsenate (and

sulfate) molecules have a lower symmetry (C2,2v) and thus give rise to the exact same type of IR bands as in the case of covalently adsorbed onto the hematite via bidentate bridging surface complexes.(Lefevre and Fedoroff, 2006; Catalano et al., 2007) However, the striking resemblance of the arsenate IR footprint with that of BFAS on one hand and the short time (60 min) employed in the CESL process make, in light of the findings of this work, quite probable the occurrence of arsenic in the particular industrial residue as BFAS-like phase. In the case of higher arsenic content concentrates processed in continuous autoclaves (where no heat-up period exists) the formation of scorodite is expected to dominate as reported recently.(Mayhew et al., 2010)

5.5 Summary and Conclusions From this study the following conclusion can be made: (1) Scorodite was found to be the equilibrium (most stable) phase in all solutions and

conditions tested (0.7

MgSO4 and >90 min reaction time). The produced scorodite [Fe(AsO4)1-2/3X(SO4)X·2H2O] was found to incorporate up to 5% SO4 in its structure as determined via elemental and vibrational spectroscopy. (2). The presence of excess ferric iron in the initial solution (Fe/As = 2) was found to slow down the kinetics of scorodite precipitation in comparison to the equimolar concentrations (Fe/As = 1) tests. (3) The presence of Cu or Mg was not found to interfere in terms of kinetics or yield with scorodite precipitation. The precipitated scorodite contained < 1% Cu or Mg in the final product and the formation of separate copper phases was not observed. (4) In the presence of equimolar (0.3 M each) iron, arsenic and copper concentrations and under short times (30 min - 60 min) and lower temperatures (90 - 135 °C) it was

discovered that a short lived intermediate Cu-Fe-AsO4-SO4-H2O gel resembling a liquid crystal type of form. This intermediate was characterized to have BFAS-like phase

162 features that kinetically transformed after 60 min and 127 °C to the more stable scorodite phase. (5) Characterization of an arsenic -containing industrial residue produced via copper concentrate pressure leaching at 150 °C and 1 hr (CESL Process) revealed the presence 0 of hematite (Fe2O3), elemental sulfur (S ) and an arsenate (SO4 and H2O/OH) form which closely resembled that the BFAS or gel type of intermediate phase. The low arsenic content of the residue (0.5 % As) in this case cannot preclude the possibility of arsenate adsorbed onto hematite. Solutions containing higher concentrations of arsenic are expected to produce scorodite.

5.6 References Bellare J. R., Davis H. T., Miller W. G. and Scriven L. E. (1990) Polarized optical microscopy of anisotropic media: Imaging theory and simulation. J. Colloid Interface Sci. 136, 305-326.

Bezce L., Gomez M. A., Le Berre J. F., Pierre B. and Demopoulos G. P. (2009) Formation of Massive Gunningite-Jarosite Scale in an Industrial Zinc Pressure Leach Autoclave: A Characterization Study. Can. Metall. Q. 48, 99-108.

Bluteau, M.-C.; Demopoulos, G. P. (2007) The incongruent dissolution of scorodite — Solubility, kinetics and mechanism. Hydrometallurgy. 87, 163-177.

Catalano J. G., Zhang Z., Park C., Fenter P. and Bedzyk M. J. (2007) Bridging arsenate surface complexes on the hematite (0 1 2) surface. Geochim. Cosmochim. Acta. 71, 1883-1897.

Deb S. K., Wilding M. C., Somayazulu M. and Mcmillan, P. F. (2001). Pressure-induced amorphisation and an amorphous-amorphous transition in densified porous silicon. Nature 414, 528-530 de Faria D. L. A., Silva S. V. and de Oliveira M. T. (1997) Raman microspectroscopy of some iron oxides and oxyhydroxides. J. Raman Spectros. 28, 873-878.

163 Defreyne J., Grieve W., Jones D. L. and Mayhew K. (2006) The Role of Iron in the CESL Process, In Iron Control Technologies (eds. Dutrizac J. E. and Riveros P. A.), CIM, Montreal, Canada, pp. 205-221. Defreyne J. and Cabra T. (2009) EARLY COPPER PRODUCTION RESULTS FROM VALE’S HYDROMETALLURGICAL CESL REFINERY, ALTA conference, Utah

Demopoulos G. P. (2009) Aqueous precipitation and crystallization for the production of

particulate solids with desired properties. Hydrometallurgy 96, 199-214.

Desiraju, G. R. (2007) Crystal Engineering: A Holistic View. Angew. Chem. Int. Ed. 46,

8342 – 8356.

Dutrizac J. E. and Jambor J. L. (2007) Characterization of the iron arsenate–sulfate compounds precipitated at elevated temperatures. Hydromtallurgy 86, 147-163.

Dutrizac, J. E.; Jambor, J. L. (1988) The synthesis of crystalline scorodite, FeAsO4 ·

2H2O. Hydrometallurgy 19, 377-384.

Eckert B., Albert H. O., Jodl H. J. and Foggi P. (1996) Raman Studies of Sulfur Crystal

(α-S8) at High Pressures and Low Temperatures. J. Phys. Chem. 100, 8212-8219.

Fan F., Feng Z., Sun K., Guo M., Guo Q., Song Y., Li W. and Li C. (2009) In Situ UV Raman Spectroscopic Study on the Synthesis Mechanism of AlPO-5. Angew. Chem. Int. Ed. 48, 8743 –8747

Filippou D. and Demopoulos G. P. (1997) Arsenic Immobilization by Controlled Scorodite Precipitation. JOM, 49, 52-55.

Filippi M., Machovič V., Drahota P. and Böhmová V. (2009) Raman Microspectroscopy as a Valuable Additional Method to X-ray Diffraction and Electron Microscope/Microprobe Analysis in the Study of Iron Arsenates in Environmental Samples. Appl. Spectrosc. 63, 621-626.

164 Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2008) Effects of zinc, copper and sodium ions on ferric arsenate precipitation in a novel atmospheric scorodite process. Hydrometallurgy 93, 30-38.

Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2009) Effect of pH on atmospheric scorodite synthesis by oxidation of ferrous ions: Physical properties and stability of the scorodite. Hydrometallurgy 96, 189-198.

Gomez M. A., Becze L., Bluteau M. C., Le Berre J. F., Cutler J. N. and Demopoulos G.

Gomez M. A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G. P. (2010) Vibrational spectroscopy study of hydrothermally produced scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75H2O) and basic

ferric arsenate sulfate (BFAS; Fe[(AsO4)1−x(SO4)x(OH)x]·wH2O) J. Raman Spectros. 41, 212-221.

Gomez M. A., Le Berre J-F., Assaaoudi H. and Demopoulos G. P. (2010) J. Raman Spectros. in print. DOI 10.1002/jrs.2639.

Gray D. G. (2008) Transcrystallization of polypropylene at cellulose nanocrystal surfaces. Cellulose 15, 297-301.

Hanley I. W. (2010) Liquid crystal phase formation by biopolymers. Soft matter 6, 1863- 1871.

Hawthorne F. C. (1976) The Hydrogen Positions in Scorodite. Acta Crystallogr., Sect. B32, 2891-2892.

Horn A. B. and Sully K. J. (1999) ATR-IR spectroscopic studies of the formation of sulfuric acid and sulfuric acid monohydrate films. Phys. Chem. Chem. Phys. 1, 3801-3806.

165 Kitahama K., Kiriyama R. and Bala Y. (1975) Refinement of the Crystal Structure of Scorodite. Acta Crystallogr. B31, 322-324.

Krause E. and Ettel V.A. (1988) Solubility and stability of scorodite, FeAsO4·2H2O: New data and further discussion. Am. Mineral. 73, 850-854. Kohl I., Winkel K., Bauer M., Liedl K. R., Loerting T. and Mayer E. (2009) Raman Spectroscopic Study of the Phase Transition of Amorphous to Crystalline β- Carbonic Acid. Angew. Chem. Int. Ed. 48, 2690 –2694

Langmuir D., Mahoney J. and Rowson, J. (2006) Solubility products of amorphous ferric

arsenate and crystalline scorodite (FeAsO4 · 2H2O) and their application to arsenic behavior in buried mine tailings. Geochim. Cosmochim. Acta. 70, 2942-2956.

Lefevre G. and Fedoroff M. (2006) Sorption of sulfate ions onto hematite studied by attenuated total reflection-infrared spectroscopy: Kinetics and competition with other ions. Phys Chem Earth 31, 499-504.

Mayhew K., Parhar P. and Salomon-de-Friedberg H. (2010) CESL Process as Applied to Enargite-Rich Copper Concentrates. In Copper2010: Hamburg

Miretzky P. and Cirelli-Fernandez A. (2010) Remediation of Arsenic-Contaminated Soils by Iron Amendments: A Review. Crit. Rev. Env. Sci. Tec. 40, 93-115.

Monhemius A. J. and Swash P. M. (1999) Removing and Stabilizing As from Copper Refining Circuits by Hydrothermal Processing. JOM. 51, 30-33.

Nakamoto K. (2009) Infrared and Raman Spectra of Inorganic and Coordination Compounds. Part A: Theory and Applications in Inorganic Chemistry (6th edn); Wiley: New York.

Peak D., Ford R. G. and Sparks D. L. (1999) An in Situ ATR-FTIR Investigation of Sulfate Bonding Mechanisms on Goethite. J. Colloid Interface Sci. 218, 289-299.

166 Riveros P. A., Dutrizac J. E. and Spencer P. (2001) Arsenic Disposal Practices in the Metallurgical Industry. Can. Metall. Q. 40, 395-420.

Swash P. M. and Monhemius A. J. (1994) Hydrothermal precipitation from aqueous solutions containing iron (III), arsenate and sulfate. In Hydrometallurgy ’94; Chapman & Hall, Eds.: New York, p. 177-190.

Swash P. M. (1996) The hydrothermal precipitation of Arsenic solids in the Ca-Fe-AsO4-

SO4 system at elevated temperature, Ph.D. Dissertation, Imperial College of Science, Technology and Medicine, University of London.

Tang X. C., Pikal M. J. and Taylor L. S. (2002) A Spectroscopic Investigation of Hydrogen Bond Patterns in Crystalline and Amorphous Phases in Dihydropyridine Calcium Channel Blockers. Pharm. Res. 19, 477-483.

Xu Y., Zhou G. P. and Zheng X. F. (2007) Re-determination of iron (III) arsenate dihydrate. Acta Crystallogr, Sect.E. 63, i67-i69.

Zhang Y. (1982) Electronegativities of elements in valence states and their applications. 1. Electronegativities of elements in valence states. Inorg. Chem. 21, 3886-3889.

Zhang G. Y. and Peak D. (2007) Studies of Cd(II)–sulfate interactions at the goethite– water interface by ATR-FTIR spectroscopy. Geochim. Cosmochim. Acta. 71, 2158-2169.

167 6. The Hydrothermal Ca (II)-Fe (III)-AsO4 System at 95°C

6.1 Abstract

This chapter describes the first synthesis method of yukonite, its thorough molecular and structural analysis along with natural specimens originating from Tagish Lake (Canada) and Grotta della Monaca (Italy) for comparison, and its structural relation to arseniosiderite. The synthetic and natural yukonites were found to have a range of composition according to the general formula Ca2Fe3-5(AsO4)3(OH)4-10·xH2O where x = 2-11. The synthetic yukonite was found to be equivalent at the atomic, molecular and structural level to the Tagish Lake yukonite. At the molecular level, arseniosiderite, via vibrational spectroscopy, was found to have an H-bonding system as in scorodite and 2- exhibit an extra arsenate mode indicative of HAsO4 groups. Heating experiments along with ATR-IR analysis indicated the presence of structural water and hydroxyl units in arseniosiderite. In yukonite in contrast, a wide diffuse H-bonding environment was observed with only arsenate groups. The presence of both structural water and hydroxyl groups was further verified via ATR-IR spectroscopy. The As K, Fe 2p and Ca 2p XANES spectra of yukonite and arseniosiderite are identical confirming that the local nature of the As, Fe and Ca atoms in these structures is the same. TEM and diffraction (X-ray and electron) showed that yukonite consists of nano-crystalline domains while in the case of arseniosiderite micro-size single crystal domains exist.

168 6.2 Introduction Gold, base metal, and uranium ore deposits are commonly associated with arsenic in

the form of (sulfo) arsenide minerals like arsenopyrite (FeAsS), enargite (Cu3AsS4), gersdorffite (NiAsS) ot nicolite (NiAs) (Riveros et al., 2001). Upon processing of these ores by leaching to extract their valuable metal content, arsenic reports to aqueous process-effluent streams as arsenate species. The removal and immobilization of arsenic(V) from such solutions typically involves co-precipitation with iron(III) (Fe(III)/As(V) molar ratio ≥ 3) combined with lime neutralization (Mahoney et al., 2007; Jia and Demopoulos, 2008). This practice results in the disposal of a mixture of poorly crystalline ferric arsenate, arsenic-bearing ferrihydrite and gypsum in a tailings management facility (Langmuir et al., 1999; Jia et al., 2005; Moldovan and Hendry, 2005; Chen et al., 2009). Alternatively, in the case of arsenic-rich and iron-deficient solutions, fixation of arsenic in the form of scorodite by controlled precipitation may be adopted (Filippou and Demopoulos, 1997; Singhania et al., 2005, 2006; Fujita et al., 2008).

The disposal of these iron(III) arsenate precipitates along with gypsum in the lime neutralized tailings raises the question of possible formation of Ca-Fe-AsO4 compounds such as yukonite and the implications this may have on arsenic retention. Laboratory research recently reported by our group did indeed find evidence of phase transformation of both Fe(III)-As(V) co-precipitate and crystalline scorodite to yukonite

(Ca2Fe3(AsO4)4(OH)·12H2O). For example, a poorly crystalline ferric arsenate co- precipitate (Fe(III)/As(V) molar ratio = 2), synthesized by lime neutralization at 22oC and pH 8, was found by Jia and Demopoulos (2008) to transform to yukonite after 7 weeks accelerated aging at 75oC. In a parallel study, Bluteau et al. (2009) found crystalline

scorodite (FeAsO4·2H2O) to partially transform to yukonite (Ca2Fe3(AsO4)4(OH).12H2O) upon equilibration in gypsum saturated solution at 75oC and pH 7 and 9.

Yukonite is a hydrated calcium ferric arsenate mineral, first found in Tagish Lake, Yukon, Canada by Tyrrell and Graham in 1913, who described its formula as . (Ca3Fe2)2(AsO4)2(OH)6 5H2O (Tyrrell and Graham, 1913). Jambor later (1966) re-

169 examined the yukonite formula and proposed the following: . Ca6Fe16(AsO4)10(OH)30 23H2O. The occurrence of yukonite since then has been reported among other in the Sterling Hill mine, Ogdensburg, New Jersey, USA by Dunn (1982), in Saalfeld, Thuringen, Germany by Ross and Post (1997), Redziny, Sudetes, Poland by Pieczka et al. (1998) and very recently by Nishikawa et al. (2006) in Kamchatka, Russia and Garavelli et al. (2009) in Grotta della Monaca, Italy. Yukonite has been reported to occur most often as intensively fractured, gel-like aggregates or dark brownish in hand- specimens; in smaller fragments it is purple reddish and slightly translucent. Its characterization via lab based X-ray diffraction method gives broad reflections typical of a mineral with poorly ordered structure (Garavelli et al., 2009). It is difficult to separate the pure yukonite phase from its subtle intergrowths with other minerals (arseniosiderite, scorodite, goethite, quartz) via XRD. Paktunc et al. (2003, 2004), Borba and Figueiredo (2004) and Garavelli et al. (2009) reported mineralogical evidence of the coexistence of scorodite and yukonite-like phases in oxidized arsenic bearing ores (with arsenopyrite as the primary arsenide mineral). Interestingly enough, in several ore samples,

a replacement of scorodite by Ca-Fe(III)-AsO4 phases was observed (Paktunc et al.,

2003, 2004). Arseniosiderite (Ca2Fe3(AsO4)3O2·3H2O) and yukonite

(Ca2Fe3(AsO4)4(OH)·12H2O) were among the mineral phases detected. Finally, very recently yukonite was also found in the tailings of a gold mining operation in Nova Scotia, Canada by Walker et al. (2009).

The aim of this chapter is to report on a detailed molecular-sensitive and structural investigation of both synthetic and natural cotype specimens (Tagish Lake, Canada and Grotta della Monaca, Italy) of yukonite and compare its molecular and crystal structure to arseniosiderite. This study constitutes the first report on the laboratory synthesis of yukonite, its detailed spectroscopic (via ATR-FTIR and Micro-Raman, and XANES), and structural (X-ray, TEM and SAED) analysis in comparison to natural yukonite and arseniosiderite.

170 6.3 Experimental Methods

6.3.1 Synthesis of Yukonite

The synthesis of yukonite was accomplished via a procedure developed in our laboratory (Becze and Demopoulos, 2007). Disodium hydrogen arsenate

(Na2HAsO4.7H2O) was dissolved in deionized water followed by pH adjustment to

approximately 0.5 with nitric acid (HNO3). Subsequently, appropriate amount of iron . . (III) sulfate (Fe2(SO4)3 5H2O) and gypsum (CaSO4 2H2O) were added to the arsenic containing solution and agitated for 24 hours at room temperature. The initial Ca/Fe/As

molar ratio was 0.5/0.75/1 (CAs=1.4 or 12.8 g/L). After that the solution was heated to 95ºC under agitation (300 rpm) and its pH value was brought to 8 in about 15 minutes using sodium hydroxide (NaOH). The pH was kept constant at 8 by addition of NaOH or

HNO3 as required for variable synthesis times (t = 24, 72, and 120 h). After that the precipitate was filtered using 0.1 µm pore size membrane in a pressure filter at 50 psi and washed 3 times with deionized water before drying (T = 50oC) and thorough characterization. Further details of the synthesis conditions for each synthetic phase made in this study can be found in Table 1.

6.3.2 Origin of Natural yukonite and Arseniosiderite Samples

Two natural yukonite samples were examined. One specimen originated from the prehistorically exploited copper-iron mine at Grotta della Monaca cave, Sant’Agata di Esaro, Cosenza, Italy (Garavelli et al., 2009). The other sample, which was in the form of single crystals (National Mineral Collection of Canada, Ref. # 064815), was collected in 1984 from Venus Mine, Windy Arm, Tagish Lake, Yukon Territory, Canada. For further comparative characterization study a natural arseniosiderite sample was used, which originated from Romanêche, Near Maĉon, Saone-et-Loirse, France and donated to us by the Redpath Museum of McGill University.

171 6.3.3 Chemical Analysis

The chemical analysis of the synthetic and the as-received yukonite samples (dried at 50oC for 24 hours) was made by electron-microprobe technique using a JEOL JXA- 8900L instrument. The operating conditions were: voltage 15 kV, beam current 20 nA, beam size 2μm, counting time 20 s. Ten spots were analysed from various regions of each sample in order to achieve statistical accuracy and to ensure homogeneity; standards used for EMP analysis were: CoNiAs for As, Fe2O3 for Fe, Ca3Fe2Si3O12 for Ca, BaSO4 for S

and NaAlSi3O8 for Na.

6.3.4 Characterization (mineralogical analysis) methods

The powder X-ray diffraction were recorded with a Philips PW1710 diffractometer equipped with a copper target (Cu Kα1 radiation, λ = 1.54060 Å), a crystal graphite monochromator and a scintillation detector. The diffractometer used 40 kV and 20 mA. The scans were measured from 2° to 100° 2θ with 0.1° step and acquisition time of 3 seconds per step. To ensure no impurities occurred in the Tagish Lake sample as in the Grotta della Monaca sample, additional analysis was conducted at the APS diffraction 11BM beamline (Wang et al., 2008) , a high resolution synchrotron powder diffraction data were using an average wavelength of 0.458(2) Ǻ . Discrete detectors covering an angular range from -6 to16° 2θ are scanned over a 34° 2θ range, with data points collected every 0.001° 2θ and scan speed of 0.01° s-1. Data are collected while continually scanning the diffractometer 2θ arm. A mixture of NIST standard reference

materials, Si (SRM 640c) and Al2O3 (SRM 676) are used to calibrate the instrument, where the Si lattice constant determines the wavelength for each detector.

For the morphological characterization of the yukonite samples, a Variable Pressure Scanning Electron Microscope (VP-SEM) Hitachi S-3000N equipped with an energy dispersive spectrometer for X-ray analysis was used.

Transmission electron microscopy (TEM) images were obtained using a Philips CM- 200 microscope operating at 200 kV. The samples were prepared by dropping dilute solutions of the particles in ethanol onto 400-mesh carbon-coated copper grids and

172 evaporating the solvent. Selected area electron diffraction (SAED) was also conducted on selected particles. Raman spectra were collected by an In Via Raman microscope from Renishaw in normal and Confocal mode. Laser excitation was provided by a polarized He-Ne laser operating at 632 nm. The laser beam produced a spot size of approximately ≤ 5 μm in diameter using the 50x short distance objective. Averages of 10 scans were obtained from 1400 to 150 cm-1. The energy resolution was 4 cm-1 at the full width half max (FWHM) of the internal Si reference peak. The scans were collected at 10 % of the laser output at the microscope exit to avoid radiation damage. Additional details on methods used may be found in Appendix 8.2.

Infrared spectra were obtained using a Perkin Elmer FTIR (Spectrum BX model) spectrometer with a Miracle single bounce diamond ATR cell from PIKE Technologies. Spectra over the 4000–550 cm-1 range were obtained by the co-addition of 200 scans with a resolution of 4 cm-1 at the FWHM of the internal Polystyrene strongest C-H vibration. Additional details on methods used may be found in Appendix 8.2.

X-ray Absorption Near Edge Structure (XANES) spectra were collected at the Canadian Light Source facility located in the University of Saskatchewan campus. The As K-edge (11867eV) was collected at the 06ID-1 (HXMA) beam line using a fixed exit double crystal monochromator with Si (111) crystals. This beamline configuration provides an energy range of 3-27 keV with a resolving power of ~7000. The energy scale

for the monochromator was referenced to the inflection point of the Au LIII-edge (11919.7 eV) of a thin gold foil. The relative energy scale was reproducible within ±0.1eV and all spectra were averages of at least two reproducible scans. Calcium and iron 2p-edge spectra were collected on the high resolution SGM - 11ID-1 beam line (Regier et al., 2007) which is equipped with a 1200 lines/mm grating that provides high- resolution (>5000) light for the photon energy ranges 270 to 700 eV. Spectra were collected using the 50 μm exit slits and the total electron yield (TEY). All samples were crushed and placed on carbon conductive tape before placing them in the absorption chamber. The Ca L-edge spectra were collected from 344 to 360 eV, while for the Fe L-

173 edge the spectra were obtained from 696 to 736 eV. In both cases a coarse step size of 0.5 eV was taken before the first edge and a fine step of 0.04 eV at the main absorption edges. The energy scale for the Fe L-edge XANES measurements was calibrated to the

highest intensity L3 peak of hematite (α-Fe2O3) occurring at 709.50 ± 0.01 eV (Garvie et al., 1994; van Aken et al., 1998, 2002, 2003; Otero et al., 2008). The energy scale of the Ca L-edge XANES spectra was not calibrated against an absolute standard as in other works (Naftel et al., 2001; Flee and Liu, 2009). However, it should be noted that the

positions of the L3 and L2-edges observed (349.86 ± 0.01 eV and 353.16 ± 0.01 eV respectively) for Ca(OH)2 correspond to those reported by Roooos et al. (1982) by adding 0.96 eV. The relative energy scale was reproducible within ±0.1eV or better and all spectra were normalized using a single normalization with averages of two reproducible 5+ 3+ scans. Reagent grade chemicals from Sigma Aldrich As2O5(As ), As2O3(As ), 3+ 2+ 2+ Fe2O3(Fe ), FeSO4·7H2O(Fe ) and Ca(OH)2(Ca ) were used as standards for oxidation states. Additional details on methods used may be found in Appendix 8.2.

6.4 Results and Discussions

6.4.1 Chemical Composition

The chemical composition obtained by microprobe analysis of the various synthetic and natural yukonite mineral samples (plus an arseniosiderite sample) is summarized in Table 1. Our synthetic yukonite samples had an arsenate content varying from 40.0 to 43.1% close to that obtained from the Tagish Lake yukonite sample (~42.7%). In contrast 3- the Grotta della Monaca yukonite specimen was richer in arsenic (~48.9% AsO4 ). Essentially all samples were free of any sulfate content but our synthetic yukonite contained a minor amount of sodium (<1%) apparently due to interstitial contamination arising from the use of NaOH in synthesis. By comparison the arseniosiderite sample 3- (Romanech) had an intermediate composition, namely 44.2% AsO4 (close to that of Tagish Lake yukonite), 21.7% Fe (close to that of Grotta della Monaca yukonite) and 10.3% Ca (higher than all yukonites). According to the determined molar ratios Ca/Fe/As the synthetic and natural yukonites appear to converge to the following stoichiometries

174 (numbers rounded to nearest integer and charge balanced by changing the OH/H2O ratio-

the OH/H2O content was determined by difference):

Formula I-Ca2Fe5(AsO4)3(OH)10·2-5H2O in the case of Synthetic 1, 3 and Tagish Lake

yukonites; and Formula II-Ca2Fe3(AsO4)3(OH)4·5-11H2O in the case of Synthetic 2 and Grotta della Monaca yukonites. Formula I and II were obtained after multiplying x3 the experimentally determined Ca/Fe/As molar ratios (refer to Table 1) and rounding them to nearest integer. This worked very well for Formula I. For Formula II, however, the stoichiometric number for Fe fell between 3 and 4 (3.51); so Formula II is considered

tentative. Alternatively Formula III-Ca2Fe5(AsO4)4(OH)7·xH2O (obtained by multiplying x4 the Ca/Fe/As molar ratios-Table 1), which has also been reported by Nickel and Nichols (1991) may replace Formula II.

Table 1. Solid composition of systhetic and natural yukonites, in addition to the natural arseniosiderite and chemical formulas.

It is interesting to note that among the various stoichiometries reported in literature for yukonite (Garavelli et al., 2009) Formula I is essentially the same with the ones reported for Tagish Lake yukonite by Jambor (1966) and Nickel and Nichols (2007) (Ca6-

7Fe15-16(AsO4)9-10(OH)30-32·23-25H2O if divided by 3). On the other hand Formula II agrees with the stoichiometry reported for yukonite by Nishikawa et al. (2006) and

175 Garavelli et al. (2009). Hence the determined composition for our synthetic yukonite falls within the range found in natural mineral specimens. At this point is worthy to comment further on the wide range of composition (in terms of Fe, As, Ca and H2O/OH content) and the many different formulae that have been proposed by previous authors (Tyrrell and Graham 1913; Jambor 1966; Dunn 1982; Ross and Post 1997; Pieckza et al., 1998; Nishikawa et al., 2006; Paktunc et al., 2003, 2004; Walker et al., 2009; Garavelli et al., 2009) including the present study. Thus, one may postulate that this difference in composition may arise from the type of analytical technique used (Electron Microprobe, XRF, TEM-EDS and XPS) or as a result of non- homogeneous samples. In our studies for each sample (natural and synthetic), a 10 spot analysis was undertaken in distinct random locations to ensure homogeneity and statistical accuracy. The difference in chemical composition was found to be in the range of 0.01-2.68 % and the variation of chemical composition for each sample along with their standard deviation are shown in the Table 1. Therefore, these small variations indicate that the samples (natural and synthetic) analyzed were homogeneous in composition. Furthermore, although granular texture was observed in the Grotta della Monaca sample studied by Garavelli et al., (2009) no such granular texture was observed in our Grotta della Monaca sample. In addition as it can be deduced from Table 1, the agreement between the synthetic samples and the natural samples is very reasonable where these granular texture effects in the synthetic samples are less of an issue. Moreover, it can be stressed that our EMP analysis of the Grotta della Monaca sample gave the same chemical formula as that determined via TEM-EDS and XPS by Garavelli et al. (2009). Thus any differences in composition of observed do not derive from the analytical techniques used but rather the chemical nature of yukonite itself, which apparently may accommodate various amounts of Fe, Ca, As and H2O/OH in its structure.

176

6.4.2 X-ray Diffraction and SEM analysis

After-24 h(Syn-1) After-72 h (Syn-2) After-120 h (Syn-3)

Arb. Intensity

10 20 30 40 50 60 70 80 90 100 Two Theta (degree)

Figure 1. (a) X-ray diffraction patterns of the synthetic yukonite formation at different times

    Grotta della Monaca  

Tagish Lake

Arb. Intensity

Synthetic - 1

Romanech Arseniosiderite 10 20 30 40 50 60 70 80 90 100 Two Theta (degree)

177 Figure 1. (b) X-ray diffraction patterns of the synthetic and natural (Grottal della Monaca and Tagish Lake) yukonites and arseniosiderite (Romanech). The symbol Δ in the Grotta della Monaca sample (left) indicates the presence of scorodite impurity. Powder X-ray diffraction (PXRD) analysis (Fig. 1a and 1b) revealed the synthetic materials to be similar to their natural yukonite counterparts and distinctly different than arseniosiderite (Ca2Fe3(AsO4)3O2·3H2O) - a mineral with similar composition to that of yukonite (Table 1). In the case of our arseniosiderite sample, the following chemical

formula Ca3Fe5(AsO4)4(OH)9·8H2O was found to best match the composition data found - -2 2- in Table 1. Here we used OH rather than O or [(O2) ] groups for charge balance purposes because as per our spectroscopic analysis presented in section 6.3 and 6.5, it - 2- contains OH units in its structure and no peroxo [(O2) ] groups.

Figure 2. SEM images of synthetic and natural yukonites [(a) Synthetic (Syn-1); (b) Tagish Lake; (c) Grotta della Monaca] and arseniosiderite[(d) Romanech]. The scale on the images is 5 μm.

178

According to the PXRD patterns of the three newly synthesized yukonite products their order of crystallinity increased with reaction synthesis time-all synthesized at 95C for 24, 72 and 120 hours respectively becoming equivalent to that of the natural mineral. In comparison to arseniosiderite both the synthetic and natural samples of yukonite were found not to have fully expressed their crystal structure and morphology at the macroscopic level as observed via SEM images shown in Fig. 2. By far the synthetic yukonite’s PXRD pattern, purity and morphology resemble closer that of the Tagish Lake yukonite (Fig. 1 and S1a). To ensure that no impurities were present in the Tagish Lake sample, synchrotron based PXRD was conducted at λ =0.458Ǻ (Fig. S1a). Additional lab based single crystal diffraction (λ =0.710Ǻ) was conducted in an attempt to determine its lattice parameters but only a diffuse signal was observed (Fig. S1b). In the case of the Grotta della Monaca yukonite it was detected via XRD, SEM-EDS and ATR-IR analysis to be contaminated with K2SO4 and FeAsO4·2H2O (Fig. 1, 2 and 3). The coexistence of scorodite with yukonite was also reported by Garavelli et al. (2009) and earlier by Pactunk et al. (2003, 2004) but with some difference. As reported in previous papers by Bluteau and Demopoulos (2007) and Bluteau et al. (2009) scorodite is metastable transforming to yukonite in calcium-rich environments. This is consistent with Paktunc et al. (2003, 2004) who observed a replacement of scorodite by Ca–Fe Fe arsenate phases. This was also observed during the course of our yukonite synthesis work as we have reported elsewhere (Becze and Demopoulos, 2007). However, Garavelli et al. (2009) have wrongly interpreted the presence of scorodite in their Grotta della Monaca yukonite sample as the result of yukonite transformation rather than the other way around.

6.4.3 ATR-IR Spectroscopy

The molecular structure of yukonite has not been thoroughly investigated aside from a recent FTIR report by Garavelli et al. (2009). The vibrational spectra (ATR-IR and Raman) of yukonite (natural and synthetic) and natural arseniosiderite are compared in Fig. 3a and 3b. It can be observed that the natural (Tagish Lake and Grotta della Monaca) yukonite and the laboratory synthesized product are in excellent agreement in terms of

179 vibrational structure as it was the case with the XRD patterns (Fig. 1 below). The ATR- IR spectra of yukonite (natural or synthetic) and natural arseniosiderite exhibit the 3- common molecular signals (Fe-O and/or Fe-OH2, H2O, OH and AsO4 vibrations; Table 2 below) expected for hydrated ferric arsenates.

Figure 3. (a) ATR-FTIR spectra of synthetic (Syn-1) and natural (Grotta della Monaca and Tagish Lake) yukonite and arseniosiderite (Romanech)

180 Figure 4. (b) Micro-Raman spectra of synthetic (Syn-1) and natural (Grotta della Monaca and Tagish Lake) yukonite and arseniosiderite (Romanech)

Table 2. ATR-IR and Micro-Raman band assignment for yukonite and arseniosiderite.

3- The infrared active ν3(F2) AsO4 mode in particular was observed in all yukonite and 3- arseniosiderite samples; the absence of the ν1(A1) mode indicates an AsO4 environment

with Td symmetry (Myneni et al., 1998). The IR spectrum of arseniosiderite is similar to that of yukonite but displays an additional strong vibration at higher wavenumbers (919 -1 2- cm ) typical of HAsO4 units (C3v ideal symmetry) (Myneni et al., 1998; Frost et al.,

2006; Sejkra et al., 2009). The much lower frequencies reported for the ν3(F2) mode in the case of yukonite (815 cm-1) may indicate “slightly longer” As-O bonding that occurs

181 -1 in yukonite versus those higher frequencies ν3(F2) vibrations (825-840 cm ) reported in other ferric arsenate minerals, such as scorodite and parascorodite (Griffith, 1970; Hawthorne, 1976; Coleshaw and Griffith, 1994; Ondrus et al., 1999).

Table 3. Hydrogen bond lengths calculated using the regression functions from Libowitzky (1999) and the ATR-IR hydroxyl stretching frequencies observed in this study.

The infrared hydroxyl region (3000-4000 cm-1) in the cases of hydrated ferric arsenates may be richer in bonding information than the arsenate (700-850 cm-1) region as shown for scorodite and parascorodite (Ondrus et al., 1999). The infrared hydroxyl region has the advantage of giving us insights into the intermolecular hydrogen bond nature of minerals (Summin de Portilla, 1974; Hawthorne, 1976). For example in scorodite, where two distinct types of intra-molecular hydrogen bonding occur between the crystal water molecules of metal octahedra and the arsenate groups of the form MO- H•••OAs, it is possible to relate each specific water (OH) stretching to a particular vibrational band of the two distinct crystallographic water sites (Hawthorne, 1976; Gomez et al., 2010a). A good estimate of the relative strength of H-bonding interactions

182 between the two distinct water molecules of scorodite can be inferred by observing the IR hydroxyl bands and their relative positions. In addition H-bond distances (O•••O and H•••O) can be calculated from vibrational positions using Libowitzky’s correlation functions (Libowitzky, 1999) (Table 3). This type of analysis has been found to be a good estimate of bond lengths as shown in recent reports of other arsenate minerals (Cejka et al., 2009, Frost et al., 2009). “Strong” (or Type I) hydrogen bonding in the crystal structure of hydrated ferric arsenates is expressed in the IR wavenumber region from 2900 to 3600 cm-1 while “weak” (Type II) hydrogen bonding tends to absorb at higher wavenumbers (~3650 cm-1) (Gomez et al., 2010a). In the case of arseniosiderite (Fig. 3a) two hydroxyl stretches (3100, 3576 cm-1) are observed, almost identical to that found in scorodite, indicating that two distinct types of H-bonding environments exist (one strong- Type I and one weak-Type II (Gomez et al., 2010b); Table 3). Yukonite on the other hand exhibits only a diffuse band (3111-3215cm-1) indicative of disordered type of H- bonding, typical of its glassy nature.

Figure 4. ATR-IR of arseniosiderite (Romanech) (left) in the arsenate and water stretch region and (right) the hydroxyl and/or water region at 25, 450 and 650 °C.

To further investigate if yukonite (Tagish Lake) and arseniosiderite (Romanech) contained molecular water groups and hydroxyl groups (not from water) in their structure, samples were annealed at 450°C and 650°C following a similar procedure as that conducted by Garavelli et al. (2009). For arseniosiderite, it was observed (Figure 4) -1 that at 450°C the amount of water (as evidenced via the δH2O mode at ~1624cm ) had

183 3- -1 decreased; in addition the ν3(AsO4 ) at 778 cm had increased in band width (due to 2- -1 disorder from heating the sample) and energy, while the ν1(HAsO4 ) mode at 919 cm had decreased and shifted to higher energy. More importantly, the hydroxyl environment in arseniosiderite upon heating had completely changed (and shifted to higher energies, indicative of weaker H-bonding) suggesting that the weakly bonded water (observed at 3576cm-1) has been completely removed at 450°C. At higher temperatures (650°C) 3- 2- similar effects on ν3(AsO4 ) and ν1(HAsO4 ) modes were evident. Moreover, the amount -1 of water (as evidenced via the δH2O mode at ~1617cm ) was completely removed, while the hydroxyl environment (as represented by the broad band at 3000-3500cm-1) was decreased but not eliminated. This proves that in addition to molecular water units, hydroxyl units also exist in the arseniosiderite structure as originally thought by Moore and Ito (1974) but subsequently revised opting instead for O2- groups (Moore and Araki 1977). We note, however, that they arrived at this option by investigating bond length- bond strength variations via XRD. However, X-ray diffraction (used by Moore and Akari 1977) is not particularly sensitive to H-atoms (as say neutron diffraction) thus the

differentiation between H2O and OH is rather difficult, especially for such a complicated structure as arseniosiderite. On the other hand the combined heating and molecular- sensitive spectroscopic methodology applied in the present work clearly is a much more dependable technique for distinguishing between these two molecular groups of interest.

Figure 5. ATR-IR of yukonite (Tagish Lake), (left) in the arsenate and water stretch region and (right) the hydroxyl and/or water region at 25, 450 and 650 °C.

184 In the case of yukonite (Tagish Lake) similar features with those found in arseniosiderite were determined. Thus as it can be observed in Figure 5, the amount of -1 molecular water (as evidenced via the δH2O mode at ~1634cm ) had decreased at 450°C as well as the hydroxyl mode. Upon heating to 650°C the amount of water (evidenced via -1 the δH2O mode at ~1617cm ) was completely removed while the hydroxyl environment had significantly decreased and shifted to higher energies, indicating once again (as observed for arseniosiderite) that in addition to molecular water units there exist hydroxyl groups in yukonite’s structure in agreement with the previous molecular formulae proposed and the heating studies conducted by Garavelli et al. (2009). These spectral changes with temperature were also observed for the Grotta della Monaca yukonite sample (Garavelli et al., 2009) but were perhaps not as clear as in our Tagish Lake sample, likely due to the impurities present in the former sample (scorodite, K2SO4) or as a result of the fact that yukonite only exhibits a broad hydroxyl band unlike that of arseniosiderite, which exhibits more distinct hydroxyl features.

185 6.4.4 TEM and Electron Diffraction

Figure 6. TEM and SAED images of natural yukonite (Tagish Lake) (a, b, e, f), and arseniosiderite (Romanech) (c, d).

The degree of long range order in the crystal structures of three closely related Ca-Fe- arsenates (yukonite and arseniosiderite) greatly differs as it can be deduced from the XRD (Fig. 1) and TEM/SAED data (Fig. 6). Our TEM analysis (Fig. 6a, 6b, 6e, and 6f) suggests that yukonite is a nano-domained ordered (hypocrystalline) crystalline material,

186 whose nanocrystallites (with physical size ≥ 100nm) show internal order (lattice distances) only over short nanometer scale (1-15nm). The limited growth of yukonite domains appears to occur via preferential elongation of its (300) atomic planes (PDF35- 0553) corresponding to a d-spacing of 0.32 nm, consistently observed in all the nano- domained crystals examined (Fig. 6e). Selected area electron diffraction (SAED) analysis of yukonite at the 100 nm scale showed very weak and diffuse rings (Fig. 6a) but after zooming at 5nm level stronger diffraction rings and more atomic planes were visible (Fig. 6b). The latter observation is in agreement with Garavelli et al. (2009) but in contrast to the findings by Nishikawa et al. (2006), who observed only single crystal diffraction.

Arseniosiderite on the other hand is a fully-crystalline compound of which all particles exhibit a very thin plate morphology (Fig. 2d) with long range periodic atomic order (micron size domains) (refer to TEM work data in Fig. 4c and 4d). The size of the arseniosiderite flat plate particles is ≥ 1μm with their growth occurring via preferential elongation of its (002) atomic planes using the PDF026-1002, corresponding to our measured d-spacing of 0.54 nm as consistently observed for the lattice fringes in all different crystals examined. SAED showed a single crystal diffraction pattern at both 100 nm or 5 nm scale with an hexagonal type of lattice symmetry never before observed in arseniosiderite; further tilt orientations (up to 30°) were also collected to observe the Kikuchi lines and possibly different lattice orientations but none other then the hexagonal symmetry was observed. Studies by Nishikawa et al. (2006) stated that their SAED of yukonite exhibited orthorhombic symmetry for some crystals and hexagonal for others with the unit meshes having the same dimensions; thus we conclude that arseniosiderite impurities were likely present in their sample that went undetected. At this point it should be reiterated that the arseniosiderite sample analyzed in this study was ensured to be free of impurities as it could be verified via XRD, ATR-IR and Micro-Raman analysis) and with comparison of the JCPD database file 0261002 and X-ray data from Moore and Ito (1974) for arseniosiderite.

187 For the crystallographically unknown yukonite, only short range order analysis has been reported through As K-edge EXAFS (Paktunc et al., 2003, 2004) but no consideration has been given on how the expressed long range order and H-bonding affect the nature of these phases, namely yukonite and arseniosiderite. Yet, the degree of hydration and therefore the hydrogen bonding of crystalline water/OH molecules greatly

impacts the crystal structure that is observed [FeAsO4 (anhydrous) (Reiff et al., 1993), monoclinic; FeAsO4•0.75H2O (FAsH) (Jakeman et al., 1991; Gomez et al., 2010a), triclinic; and FeAsO4•2H2O (parascorodite and scorodite) (Hawthorne, 1976;Ondrus et al., 1999; Gomez et al., 2010b), hexagonal and orthorhombic respectively], and therefore the order (both short and long range) that follows. Thus, our TEM, X-ray and electron diffraction and ATR-IR results show that the degree of long range order in arseniosiderite versus yukonite is different and may arise from the type and the degree of H-bonding network that occurs throughout the system. According to the IR spectra yukonite only exhibits a wide hydroxyl stretch while arseniosiderite has two distinct hydroxyl stretching vibrations similar to those observed in scorodite (Hawthorne, 1976; Gomez et al., 2010a,b). These types of vibrations are indicative of two distinct crystallographic water sites, pointing to two distinct types of hydrogen bonding with different H-bond strengths and lengths (Fig. 3 and Table 3). Therefore, from the TEM/SAED and IR results, it may inferred that the higher degree of long range order (micron-domains) in arseniosiderite in comparison to yukonite (nano-domains) is due to the different degree of hydrogen network that occurs in their crystal structures.

188 6.4.5 Raman Spectroscopy

The Raman spectrum of yukonite has not been reported previously and thus we show for the first time the Raman spectra of natural (Tagish Lake and Grotta della Monaca) and synthetic yukonite in comparison to that of natural arseniosiderite (Fig. 3b). All the 3- expected Raman symmetric and asymmetric bend and stretching vibrations for the AsO4 group are observed in the Raman spectrum of yukonite, indicating that the Td symmetry is preserved and no hydrogen arsenate units exist, in agreement with IR results (Table 2). Arseniosiderite in contrast, exhibits an additional vibration at higher wavelength (929 cm- 1 2- ) indicative of HAsO4 units (Myneni et al., 1998; Frost et al., 2006; Sejkora et al., 2009) as observed in the IR spectra. The Raman spectrum of yukonite (Fig. 3) has the advantage of being simple and characteristic enough to be used as a quick qualitative and complementary finger printing tool when investigating the relevant phases encountered in mines and mill tailings (scorodite, ferrihydrite, arsenical ferrihydrite, yukonite, and arseniosiderite) on site or in the laboratory (Fig. S2).

The use of the Confocal Raman Microscope offers the additional ability to get micron (1μm or less depending on laser source and microscope objective) spatial resolutions (Allakhverdiev et al., 2009) for better analysis and chemical speciation of smaller grains with relatively low detection limits (~1 wt%).

189

Figure 7. Micro-Raman spectra of natural yukonite (Tagish Lake) and arseniosiderite (Romanech) showing the arsenate modes. (On the bottom spectrum,

the FWHM is also shown for the ν2 arsenate band).

3- In this study, the most intense AsO4 (ν1 and ν2) stretching vibrations were used to 3- examine the relative crystallinity (AsO4 order) of yukonite and arseniosiderite (Fig. 7). The Raman vibrational bands (Fig. 3 and 7) were found to be broader in the case of yukonite (synthetic and natural) in comparison to the sharper more intense vibrational bands observed for arseniosiderite. For example, the FWHM for the ν2 arsenate band was found to be 34 cm-1 for yukonite vs. 32 cm-1 for arseniosiderite (Fig. 7) confirming that 3- the AsO4 molecules in arseniosiderite are in higher long range order than in yukonite; similar conclusions were observed from the XRD and TEM data (Fig. 1, 6 and S1). 3- It should be noted that the AsO4 groups in arseniosiderite (no full crystallographic structure exists for arseniosiderite), as in its suspected isostructure Mitridatite, form

intermolecular H-bonds between the calcium structural water (as open sheet CaO5(H2O)2 polyhedra) and the arsenate-iron octahedral sheet layers which gives rise to perfect basal

190 cleavage, parallel to {100} direction. In contrast, yukonite contains less calcium content and more weakly bound water (TGA indicated 11 - 14 wt% loss below 300°C indicating some of the water units occurring as structural and other as weakly bound interlayer groups between the sheet structures) than arseniosiderite. Thus we may infer yukonite exhibits less ordered H-bonding due to interlayer water groups (observed at lower temperatures via TGA) interfering with the intermolecular H-bonding needed to order the overall crystal lattice, and hence we observed lower long range order (Fig. 2, 6) expressed. In other words more intermolecular H-bond networks enhances the long range order expressed by arseniosiderite (Fig. 2, 6) while in yukonite excess water molecules disturbs the needed intermolecular H-bonding to express long range order.

Finally, it is worthy to mention that from the Raman spectra (Fig. 3b, 7) of natural and synthetic yukonite we observe no peroxo nor any superoxo vibrational bands -1 -1 occurring at 825-870 cm or 1131-1580 cm as in studtite (UO2(O2)·4H2O) (Burns and Hughes, 2003; Bastians et al., 2004), suggesting that the appropriate molecular formula for yukonite should not include peroxo units in agreement with previous works (Tyrrell and Graham, 1913; Jambor 1966; Ross and Post 1997; Pieczka et al., 1998; Paktunc et al., 2003; Nishikawa et al., 2006; Becze and Demopoulos 2007; Walker et al., 2009; Garavelli et al., 2009). Similarly according to our Raman spectra of arseniosiderite (Fig. 3 and 7) and previous work (Filippi et al., 2007) there does not appear to contain peroxo groups either but OH groups as reported in section 6.3, hence our preference for the new

formula Ca3Fe5(AsO4)4(OH)9·8H2O rather then Ca2Fe3(AsO4)3O2•3H2O as the latter one gives the wrong impression that peroxo units exist in arseniosiderite structure as in that of studtite.

191 6.4.6 X-ray Absorption Analysis

The electronic and molecular structure of yukonite has been mainly investigated through As K-edge EXAFS analysis (Paktunc et al., 2003, 2004) and in one brief report via XPS analysis (Garavelli et al., 2009). Paktunc et al. (2003) demonstrated that gold ores and exposed mine tailings from the Ketza River Mine were rich in arsenate - minerals such as arsenical-ferrihydrite, scorodite, arseniosiderite and/or yukonite. It was reported that the coordination number and bond lengths of the As-O and As-Fe units were essentially identical in yukonite and arseniosiderite, and as mentioned above, the only difference was the As-Ca coordination number (4.17 and 2.44 respectively). Under suspicion here is the As-Ca coordination number 2.44 listed by Paktunc et al. (2003). It is noted that Paktunc et al. (2003) failed to reference any of the extensive work done by Moore and Ito (1974), and later Moore and Araki (1977a,b) on the structure of mitridatite (and as extension arseniosiderite) but still somehow managed to fit the EXAFS structure of arseniosiderite and came up with a As-Ca coordination number of 2.44. But Moore and Araki (1977a) had clearly stated that “each Ca unit is coordinated to four phosphate/arsenate oxygens”. In a subsequent publication the same researchers (Paktunc et al., 2004) reported different As-Ca coordination numbers for arseniosiderite, namely 2.44, 3.60 and 5.5, thus creating an uncertainty as to the true molecular identification of these relevant arsenic minerals. In this work after a brief presentation of the As-K and Fe- 2p X-ray Absorption Near Edge Structure (XANES) we focus on the Ca-2p XANES of yukonite and arseniosiderite that has not been studied before aiming to resolve the current contradicting information (Paktunc et al., 2003, 2004) concerning the local calcium electronic and coordination environment.

192

Figure 8. (a) As K-edge XANES (left) spectra of synthetic (Syn-1) and natural

(Tagish Lake) yukonite, arseniosiderite (Romanech), reagent grade As2O3 and

As2O5.

Figure 8. (b) EXAFS (right) spectra of synthetic (Syn-1) and natural (Tagish Lake) yukonite, arseniosiderite (Romanech).

193 The As-K XANES was used to monitor the arsenic oxidation state in synthetic and Tagish Lake yukonite and arseniosiderite minerals (Fig. 8a). In all minerals, arsenic was found to exist at the As5+ oxidation state in agreement with previous work (Paktunc et al., 2003, 2004; Cances et al., 2008; Walker et al., 2005, 2009). No significant difference in As5+ position (±0.8 eV) was observed between the natural and synthetic yukonite, which was prepared from a pure As5+ source. Furthermore, in agreement with previous reports where the As-K edge EXAFS spectra of the “pure” phases were fitted (Paktunc et al., 2003), we observed that the As K-edge XANES and non-fitted EXAFS spectra (Fig. 8b) are qualitatively identical in terms of the local arsenic state in both yukonite and arseniosiderite.

Figure 9. (a) Fe L-edge XANES spectra of synthetic (Syn-1) and natural (Tagish Lake) yukonite, arseniosiderite (Romanech), reagent grade hematite.

194

Figure 9. (b) Ca L-edge XANES spectra of synthetic (Syn-1) and natural (Tagish Lake) yukonite, arseniosiderite (Romanech), reagent grade calcium hydroxide.

The Fe L-edge XANES spectra of yukonite and arseniosiderite (Fig. 9a) shows the 3- valency of the iron state of these Ca-Fe-AsO4 phases to be trivalent as expected. The

approximate separation between the L2 and L3 edges for synthetic, Tagish Lake yukonite and natural arseniosiderite was found to be 13.3 ± 0.1eV in agreement with that of Fe3+ phases such as hematite (Garvie et al., 1994; van Aken et al., 1998, 2002, 2003) and other hydrated ferric arsenate phases (scorodite, FAsH-ferric arsenare sub-hydrate) (Gomez et al., 2009). Again if more accurate spin-orbit splitting values are needed the reader is refered to conduct XPS measurements (see Appendix 8.2). The Fe3+ state in yukonite (natural and synthetic) was deduced to be in a paramagnetic high spin state (ground: ~ 6 2 3+ A1g+ T2g) since no feature (at ~ 705eV) characteristic of octahedral low spin Fe III compounds such as K3[Fe (CN)6] was observed (Otero et al., 2008). Therefore, the ferric iron in yukonite (and arseniosiderite) is expected to have an octahedral-type of

195 3 2 6 environment with a possible (t2g) (eg) electronic configuration and an ~ S5/2 electronic ground state configuration (deGroot et al., 1990; deGroot, 2005); and to exhibit a paramagnetic high spin anti-ferromagnetic character as observed in the case of hematite(Fe2O3) (van Aken et al., 2003), scorodite (FeAsO4·2H2O) (Hawthorne, 1976;

Jakeman et al., 1991) and FAsH (FeAsO4·0.75H2O) (Jakeman et al., 1991).

The Ca L-edge XANES has not been as extensively investigated on mineral species as their usual Fe L-edge XANES or 43Ca MAS-NMR (Lin et al., 2004) counter parts. With complex “crystal” systems such as arseniosiderite and yukonite, the Ca-2p XANES provides a sensitive probe to determine the local coordination structure at the selected atomic site within 4-5 Å (Borg et al., 1992) from the core atom. Hence it can provide specific information on the local character of the 3d0 unoccupied states via the 2p53dn excited electronic state (deGroot et al., 1990; deGroot, 2005). The Ca L2,3 XANES spectra of octahedral coordination compounds with a positive crystal field have been described to consist of spin-orbit splitting which gives rise to two main peaks from the

(a2) 2p3/2-eg, (b2) 2p1/2-eg final states along with a number of smaller peaks features

(a1)2p3/2-t2g and (b1)2p3/2-t2g from transitions into the other empty states (deGroot et al., 1990;Himpsel et al., 1991; Borg et al., 1992; deGroot, 2005). The multi-peaks observed are due to the crystal field arising from the symmetry of the atoms surrounding the Ca2+ ions leading to atomic and charge transfer multiplet effects. These effects have been shown to add extra features and redistribute the intensity over all the lines in the XANES spectrum (Naftel et al., 2001; deGroot, 2005). The splitting of these features can be related nonlinearly to the value of the crystal field splitting parameter. Therefore, the measured splitting (a2-a1) and (b2-b1) has been shown to be related to the coordination and site symmetry around the calcium atom in various calcium organic and inorganic phases; while the (b1-a1) and (b2-a2) splittings are related to the spin-orbit splitting at the

2p level. Furthermore, the relative intensities of a1 to a2 and b1 to b2 have been shown to give a qualitative indication of the magnitude of the crystal field. For example the more

intense the a1 and b1 features are (relative to the a2 and b2), the larger the magnitude of the crystal field (10Dq) is expected; that is for 10Dq = 0 the a1 and b1 features are not observed (deGroot et al., 1990; Naftel et al., 2001; deGroot, 20005; Flee and Liu, 2009).

196 Table 4. Peak positions for the Ca-Ledge XANES analysis of the

Ca-Fe-AsO4 phases (eV)*

In phases such as portlandite [Ca(OH)2], the coordination around the calcium atom has been demonstrated (in agreement with results by Roooose et al. (1982)) to be of an octahedral nature via 43Ca MAS-NMR (Lin et al., 2004) and neutron diffraction (Pavese et al., 1997) measurements - a fact confirmed here via the Ca 2p XANES (Figure 9b and Table 4). However, determination of coordination information via 43Ca MAS-NMR spectroscopy for calcium ferric arsenate compounds such as yukonite and arseniosiderite is not practical due to the high spin paramagnetic character of the ferric state in these phases (as indicated in the Fe L-edge section) and as a result other techniques such as XAS must be employed. The Ca L-edge XANES spectra of yukonite, arseniosiderite and chemical reagent grade Ca(OH)2 are shown in Fig. 9. The Ca(OH)2 was used as a oxidation reference source and quality check. The Ca L-edge XANES spectra of yukonite

and arseniosiderite confirm that the calcium is in divalent oxidation state as in Ca(OH)2.

In this study, the (a2-a1) and (b2-b1) were found to be 1.0 eV for both yukonite and arseniosiderite, suggesting that the coordination of the calcium in yukonite should be as

in arseniosiderite. Similarly, the spin-orbit splitting at the 2p level (b1-a1) and (b2-a2) was found to be 3.32 in both yukonite and arseniosiderite (Table 4). According to the Ca 2p

197 XANES spectra of yukonite and arseniosiderite, the intensity of the a1 and b1 features is quite weak in comparison to that of the a2 and b2 features suggesting that the magnitude of the 10Dq crystal field is positive and not very large around the calcium site as in

phases such as aragonite (CaCO3) (Flee and Liu, 2009). This is in contrast to cases with a

strong crystal field such as CaF2 (Himpsel et al., 1991; Naftel., 2001; deGroot, 2005; Flee and Liu, 2009). Thus, from the Ca L-edge XANES spectra (Fig. 9b) it may be inferred that the electronic and local calcium coordination state (Ca-As) and thus coordination number in yukonite is very similar or exactly the same as in arseniosiderite, in contrast to previously reported EXAFS data (Paktunc et al., 2003, 2004), and that the only difference at the molecular level is observed in the vibrational spectra. Therefore yukonite has similar local structural environment with arseniosiderite as evident from the Fe L-edge, Ca L-edge and As K edge XANES data presented in Fig. 8 -9 in this study.

6.5 Summary and Conclusions

In this report we presented the first synthesis method for yukonite; we compared the synthetic yukonite to different natural yukonites from Canada and Italy and we studied the structural relation of yukonite to arseniosiderite. The major findings of this work are: (1) The synthetic yukonite-produced at 95C from gypsum-saturated aqueous ferric sulfate-arsenate suspension at pH 8 over a period of 24 hr- was found to be in excellent agreement in terms of molecular and crystal structure with the natural Tagish Lake yukonite specimen as determined by ATR-IR, Micro-Raman, XRD, TEM, and XANES. The Grotta della Monaca specimen was found to be contaminated with scorodite which has been shown to be metastable vis-a-vis yukonite at pH> 7 (Bluteau et al., 2009). The synthetic and natural yukonite samples were found to have a range of composition

according to the general formula Ca2Fe3-5(AsO4)3(OH)4-10·xH2O where x = 2-11. The synthetic sample used in the extensive characterization study had a very similar

stoichiometric formula, namely Ca2Fe5(AsO4)3(OH)10·2-5H2O, with the natural yukonite sample originating from the Tagish Lake area in Yukon, Canada, where yukonite was first discovered.

198 (2) Based on the Ca L-edge, Fe L-edge and As K edge XANES, it is apparent that yukonite exhibits the same local calcium, iron and arsenic structural environment and units encountered in arseniosiderite but with less long range order in the crystal lattice domains (as observed via X-ray and electron diffraction). The lower order in yukonite is proposed to arise from lack of developed hydrogen bonding environment (observed via ATR-IR) in contrast to arseniosiderite’s extended well ordered H-bonding structure. The lower long range order in yukonite is expressed physically as nano-domain randomly oriented lattices, as opposed to arseniosiderite which forms micro-domain single crystal lattices as revealed via SEM, TEM and SAED images. The molecular structure of 2- arseniosiderite was in addition found to contain HAsO4 units (via IR and Raman) and a hexagonal lattice never before observed. The arseniosiderite sample (Romanech) analyzed in this work was found to have the following formula:

Ca3Fe5(AsO4)4(OH)9·8H2O. The presence and use of OH was verified via combined heating and ATR-IR tests. In contrast Raman analysis did not detect the presence of

peroxo groups, hence the published formula of arseniosiderite (Ca2Fe3(AsO4)3(O)2·3H2O) is not considered to be accurate from a molecular point of view as shown in our spectroscopic work.

6.6 References Allakhverdiev K. R., Lovera D., Altstadt V., Schreier P. and Kador L. (2009) Confocal Raman Microscopy: Non-destructive materials analysis with micrometer resolution. Rev.Adv.Mater,Sci. 20, 77-84. Baghurst D. R., Barrett J., Coleyshaw E. E., Griffith W. P. and Mingos M. P. (1996) Microwave techniques for the synthesis and deuteration of minerals, with particular

reference to scorodite, FeAsO4·2H2O. Mineral. Mag. 60, 821-828. Bastians S., Crump G., Griffith W. P. and Withnall R. (2004) Raspite and studtite: Raman spectra of two unique minerals. J. Raman Spectros. 35, 726-731. Becze L. and Demopoulos G.P. (2007) Hydrometallurgical synthesis, characterization

and stability of Ca-Fe-AsO4 Compounds. In Extraction and Processing Proceedings, B. Davis and M. Free, Eds., TMS, Warrendale, PA, pp. 11-17.

199 Bluteau M-C. and Demopoulos G. P. (2007) The incongruent dissolution of scorodite - Solubility, kinetics and mechanism. Hydrometallurgy 87, 163-177. Bluteau M-C., Becze L. and Demopoulos G.P. (2009) The dissolution of scorodite in

gypsum-saturated waters: Evidence of Ca-Fe-AsO4 mineral formation and its impact on arsenic retention. Hydrometallurgy 97, 221-227. Borba R.P. and Figueiredo B.R. (2004) The influence of geochemical conditions on arsenopyrite oxidation and on the mobility of arsenic in tropical surface environments. Revista Brasileira de Geociencias 34, 489-500. Borg A., King P. L., Pianetta P., Lindau I., Mitzi D.B., Kapitulnik A., Soldatov A. V.,

Della Longa S. and Bianconi A. (1992). Ca 3d unoccupied states in Bi2Sr2CaCu2O8

investigated by Ca L2,3 x-ray-absorption near-edge structure. Phys. Rev. B: Condens. Matter 46, 8487-8495.

Burns P.C., and Hughes K. A. (2003) Studtite, [(UO2)(O2)(H2O)2](H2O)2: The first structure of a peroxide mineral. Am. Mineral. 88, 1165-1168. Cances B., Juillot F., Morin G., Laperche V., Poly D., Vaughan D. J., Hazemann J.-L., Proux O., Brown Jr G. E., and Calas G. (2008) Changes in arsenic speciation through a contaminated soil profile: A XAS based study. Sci. Total Environ. 397, 178-189. Cejka J., Sejkora J., Frost R. L. and Keeffe E. C. (2009) Raman spectroscopic study of

the uranyl mineral metauranospinite Ca[(UO2)(AsO4)2]·8H2O. J. Raman Spectros. 40, 1786546-1790. Chen N., Jiang D.T., Cutler J., Kotzer T., Jia Y.F., Demopoulos G.P. and Rowson J. W.

(2009) Structural characterization of poorly-crystalline scorodite, iron(III)―arsenate co-precipitates and uranium mill neutralized raffinate solids using X-ray absorption fine structure spectroscopy. Geochim. Cosmochim. Acta 73, 3260-3276.

Coleyshaw E.E., Griffith W.P. and Bowell R.J. (1994) Fourier Transform Raman Spectroscopy of Minerals. Spectrochim. Acta, Part A 50, 1909-1918.

de Groot F. M. F., Fuggle J. C. , Thole B. T. and Sawatzky G. A. (1990) 2p x-ray absorption of 2d transition-metal compounds: An atomic multiplet description including the crystal field. Phys. Rev. B: Condens. Matter, 42, 5459-5468. de Groot F. M. F. (2005) Multiplet effects in X-ray spectroscopy. Coord. Chem. Rev. 249, 31-63.

200 Dunn P.J. (1982) New data for pitticite and a second occurrence of yukonite at Sterling Hill, New Jersey. Mineral. Mag. 46, 261-264. Filippi M., Dousova B. and Machovic V. (2007) Minerological speciation of arsenic in soils above the Mokrsko-west gold deposit, Czech Republic. Geoderma, 139, 154- 170. Filippou D. and Demopoulos G. P. (1997) Arsenic immobilization by controlled scorodite precipitation. JOM 49, 52-55.

Fleet M.E. and Liu X. (2009) Calcium L2,3-edge XANES of carbonates, carbonate apatite and oldhamite (CaS). Am. Mineral. 94, 1235-1241. Frost R. and Weier M. and Martens W. (2006) Raman microscopy of

synthetic goudeyite (YCu6(AsO4)2(OH)6·3H2O). Spectrochim. Acta, Part A. 63, 685-689. Frost R. L. and Sejkora J. and Čejka J. and Keeffe E. C. (2009) Raman spectroscopic study of the mixed anion sulfate-arsenate mineral parnauite

Cu9[(OH)10|SO4|(AsO4)2]·7H2O. J. Raman Spectros. 40, 1546-1550. Fujita T., Taguchi R., Abumiya M., Matsumoto M., Shibata E. and Nakamura T. (2008) Novel atmospheric scorodite synthesis by oxidation of ferrous sulfate solution. Part I. Hydrometallurgy 90, 92-102. Garavelli A., Pinto D., Vurro F., Mellini M., Viti C., Balic-Zuni T. and Della Ventura G. (2009) Yukonite from the Grotta della Monaca Cave, Sant’Agata Di Esaro, Italy: Characterization and comparison with Cotype material from the Daulton ine,

Yukon, Canada. Can. Mineral. 47, 39-51. Garvie L.A.J., Craven A. J. and Brydson R. (1994) Use of electron-energy loss near-edge fine structure in the study of minerals. Am. Mineral. 79, 411-425. Golebiowska B., Pieczka A. and Zabinski W. (2001) Spectroscopy study of yukonite (Ca- Fe arsenate mineral). In Mineralogy and Spectroscopy, Proceedings of 4th European Conference, 13, 21-22. Gomez M. A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G. P. (2010a) Vibrational spectroscopy study of hydrothermally produced scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) and basic

201 ferric arsenate sulfate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O ), J. Raman Spectros. 41, 212-221. Gomez M. A., Le Berre J-F., Assaaoudi H. and Demopoulos G. P. (2010b) Raman spectroscopic study of the hydrogen and arsenate bonding environment in 3+ 3+ isostructural synthetic arsenates of the variscite group - M AsO4·2H2O (M = Fe, Al, In and Ga)- implications for arsenic release in water, J. Raman Spectros. in

print. DOI 10.1002/jrs.2639 Griffith W. P. (1970) Raman studies on rock-forming minerals. Part II. Minerals

containing MO3, MO4 and MO6 groups. J. Chem. Soc. (A), 286-291. Hawthorne F. C. (1976) The Hydrogen Positions in Scorodite. Acta Crystallogr. Sect. B: Struct. Sci. B32, 2891-92. Himpsel F. J., Karlsson U. O., Mclean A. B., Terminello L. J., de Groot F. M. F., Abbate M., Fuggle J. C., Yarmoff J. A, Thole B. T. and Sawatzky G. A. (1991) Fine structure of the Ca 2p x-ray-absorption edge for bulk compounds, surfaces, and interfaces. Phys. Rev. B: Condens. Matter 43, 6899-6907. Jakeman R. J. B., Kwiecien M. J., Reiff W . M., Cheetham K. and Torardi C. C. (1991) A new ferromagnetic orthoarsenate hydrate: structure and magnetic ordering of

FeAsO4·3/4H2O Inorg. Chem. 30, 2806-2811. Jambor J.L. (1966) Re-examination of yukonite. Can. Mineral. 8, 667-668. Jia Y., Demopoulos G.P., Chen N. and Cutler J. (2005) Co-precipitation of As(V) with Fe(III) in Sulfate Media: Solubility and Speciation of Arsenic. Arsenic Metallurgy, Reddy R.G., and Ramachandran V., Eds., TMS, Warrendale, PA, pp. 137-148. Jia Y. and Demopoulos G.P. (2008) Coprecipitation of arsenic with iron(III) in aqueous sulfate media: Effect of time, lime as base and co-ions on arsenic retention. Water Res. 42,661-668. Langmuir D., Mahoney J., MacDonald A. and Rowson J. (1999) Predicting arsenic concentrations in the porewaters of buried uranium mill tailings. Geochim. Cosmochim. Acta 63, 3379-3394. Libowitsky E. (1999) Correlation of O-H stretching frequencies and OH•••O hydrogen bond lengths in minerals. Monatsh. Chem. 130, 1047-1059.

202 Lin Z., Smith M. E., Sowrey F. E. and Newport R. J. (2004) Probing the local structural environment of calcium by natural-abundances solid-state 43Ca NMR. Phys. Rev. B: Condens. Matter, 69, 224107-1 -7. Mahoney J., Slaughter M., Langmuir D., and Rowson J. (2007) Control of As and Ni releases from a uranium mill tailings neutralization circuit: Solution chemistry, mineralogy and geochemical modeling of laboratory study results. Appl. Geochem. 22, 2758-2776. Moldovan B. J. and Hendry M. J. (2005) Characterizing and quantifying controls on arsenic solubility over a pH range of 1–11 in a uranium mill-scale experiment, Environ. Sci. Technol. 39, 4913–4920. Moore P. A. and Ito J. (1974) Isotypy of Robertsite, Mitridatite, and Arseniosiderite. Am. Mineral. 59, 48-59. III Moore P. A . and Araki T. (1977a) Mitridatite, Ca6(H2O)6[Fe 9O6(PO4)9]·3H2O. A noteworthy octahedral sheet structure. Inorg. Chem. 16, 1096-1106. Moore P. A . and Araki T. (1977b) Mitridatite: a remarkable octahedral sheet structure. Mineral. Mag. 41, 527-528. Myneni S. C. B.,Traina S. J., Waychunas G. A. and Logan T. J. (1998) Experimental and theoretical vibrational spectroscopic evaluation of arsenate coordination in aqueous solutions, solids and at mineral-water interfaces. Geochim. Cosmochim. Acta 62, 3285-3300. Naftel S. J., Sham T. K., Yiu Y.M. and Yates B. W. (2001) Calcium L-edge XANES study of some calcium compounds. J. Synch. Rad. 8, 255-257. Nickel E.H. and Nichols M.C. (1991) Mineral Reference Manual. Van Nostrand, New York, N.Y. Nickel E.H. and Nichols M.C. (2007) IMA/CNMNC List of Mineral Names. Nishikawa O., Okrugin V., Belkova N., Saji I., Shiraki K. and Tazaki K. (2006) Crystal symmetry and chemical composition of yukonite: TEM study of specimens collected from Nalychevskie hot springs, Kamchatka, Russia and from Venus mine, Yukon Territory, Canada. Mineral. Mag. 70, 73-81.

203 Ondrus P., Skala R., Viti C., Veselovsky F., Novak F. and Jansa J. (1999) Parascorodite,

FeAsO4·2H2O- a new mineral from Kank near Kutna Hora, Czech Republic. Am. Mineral. 84, 1439-1444. Otero E., Wilks R. G., Regier T., Bylth R.I.R, Moewes A. and Urquhart S. G. (2008) Sustituent effects in the iron 2p and carbon 1s edge Near-Edge X-ray Absorption Fine Structure (NEXAFS) spectroscopy of ferrocene compounds. J. Phys. Chem. A. 112, 624-634. Paktunc D., Foster A. and Laflamme G. (2003) Speciation and characterization of arsenic in Ketza river mine tailings using X-ray absorption spectroscopy. Environ. Sci. Techn. 37, 2067-2074. Paktunc D., Foster A., Heald S. and Laflamme G. (2004) Speciation and characterization of arsenic in gold ores and cyanidation tailings using X-ray absorption spectroscopy. Geochim. Cosmochim. Acta 68, 969-983. Palmer, S. E., Frost, R. L., Ayoko, G. and Nguyin, T. (2008) Synthesis and Raman 2- 2- spectroscopic characterisation of hydrotalcites with CO3 and (MO4) anions in the interlayer. J. Raman Spectros. 39, 395-401. Pavese A., Catti M., Ferraris G. and Hull S. (1997) P-V equation of state of Portlandite,

Ca(OH)2, from powder neutron diffraction data. Phys. Chem. Miner. 24, 85-89. Pieczka A., Golobiowska B. and Franus W. (1998) Yukonite, a rare Ca-Fe arsenate, from Redziny (Sudetes, Poland). Eur. J. Mineral. 10, 1367-1370. Roooos R. E., Madcte, A. A. and Makcemob, U. A. (1982) Atomic and solid state effects

in L2,3 absorption spectra of calcium in some compounds. Izvestiya Akademii Nauk SSSR, Seriya Fizicheskaya. 6, 789-792. Regier T., Krochak J., Sham T.K., Hu Y.F., Thompson, J. and Blyth R.I.R. (2007) Performance and Capabilities of the Canadian Dragon: The SGM Beamline at the Canadian Light Source. Nucl. Instr. Meth. A. 582, 93-95. Reiff W. M, Kwiecien M.J., Jakeman R. J. B., Cheetham A. K. and Torardi C. C. (1993)

Structure and Magnetism of anhydrous FeAsO4: Inter- vs Intradimer magnetic exchange interactions. J. Solid State Chem. 107, 401-412. Riveros P.A., Dutrizac J.E. and Spencer P. (2001) Arsenic disposal practice in the metallurgical industry. Can. Metall. Q. 40(4), 395-420.

204 Ross D.R. and Post, J.E. New data on yukonite.(1997) Powder Diffr. 12, 113-116. Sejkora J., Hawthorne F. C., Cooper M. A., Grice J. D., Vajdak J. and Jambor J. L.

(2009) Burgessite, Co2(H2O)4[AsO3(OH)]2(H2O), a new arsenate mineral from the Keely mine, south Lorrain township, Ontario, Canada .Can. Mineral. 47, 159-164. Singhania S., Wang Q., Filippou D. and Demopoulos G. P. (2005) Temperature and seeding effects on the precipitation of scorodite from sulfate solutions under atmospheric-pressure conditions. Metall. Mater. Trans. B. 36B,327-333. SinghaniaS., Wang Q., Filippou D. and Demopoulos G. P. (2006) Acidity, valency and third-ion effects on the precipitation of scorodite from mixed sulfate solutions under atmospheric-pressure conditions, Metall. Mater. Trans.B 37 B, 189–197. Sumin de Portilla, V. I. (1974) Infrared spectroscopic investigation of the structure of some natural arsenates and the nature of H-bonds in their structures. Can. Mineral. 12, 262-268. Tyrrell J.B. and Graham R.P.D. Yukonite, a new hydrous arsenate of iron and calcium, from Tagish Lake, Yukon Territory, Canada; with a note on the Associated Symplesite. In Transactions of the Royal Society of Canada, Third Series, Vol. VII, 1913, pp. 13-18. van Aken P. A., Liebscher B. and Styrsa V.J. (1998) Quantitative determination of iron

minerals: new evaluation schemes of Fe L2,3 electron energy-loss near-edge spectra. Phys. Chem. Miner. 29, 188-200. van Aken P. A. and Lauterbach S. (2003) Strong magnetic linear dichroism in Fe L23 and

O K electron energy-loss near-edge spectra of antiferromagnetic hematite α-Fe2O3. Phys. Chem. Miner. 30, 469-477. Walker S. R., Jamieson H. E., Lanzirotti A., Andrade C. F. and Hall G. E. M. (2005) The speciation of arsenic in iron oxides in mine wastes from the giant gold mine, N.W.T.: application of synchrotron micro-XRD and micro-XANES at the grain scale. Can. Mineral. 43, 1205-1224.

205 Walker S. R., Parsons M. B., Jamieson H. E. and Lanzirotti A. (2009) Arsenic mineralogy of near-surface tailings and soils: Influences on arsenic mobility and bioaccessibility in the Nova Scotia gold mining districts. Can. Mineral. 47, 533-556. Wang J., Toby B. H., Lee P.L., Ribaud L., Anyao S. M., Kurtz C., Ramanathan M., Von Dreele R. B. and Beno M. A. (2008) A dedicated powder diffraction beamline at the Advanced Photon Source: Commissioning and early operational results Rev. Sci. Instrum. 79, 085105 1-7.

206 7. Conclusions and Perspectives

7.1 Conclusions and Contributions to Original Knowledge

The first part of this research and the core topic of this PhD thesis involved the

investigation of the hydrothermal high temperature (150-225 °C) Fe(III)-AsO4-SO4 system, the phases that are produced, and their arsenic stability into the aqueous environment. This research was conducted, due to its direct impact and industrial relevance to the copper and gold industry.

In the first part of the research it was found that by investigating the effect of precipitation conditions [Fe (III)/As (V) molar ratio, temperature, time and acidity] on the type of phases formed, that three main arsenate phases were formed. New phases (Phase 3 and 4) reported to be produced in this system by Dutrizac and Jambor were found to be the same ones (Type 2 and Type 1) earlier reported by Swash and Monhemius and also our work, at the elemental, molecular and structural level. Thus for the first time resolution to previous contradictions on the matter and unifying theories on the phases that are formed in this system was provided. Given this, more chemically descriptive names [Ferric Arsenate sub-Hydrate (FAsH) = Type 1 = Phase 4 and Basic Ferric Arsenate Sulphate (BFAS) = Type 2 = BFAS] were given to the phases in question and a new revised phase diagram for this system was created to replace the previous one created by Swash and Monhemius. For the first time it was also found in this study that the general reaction which leads to the formation of these phases, lead to acid generation, and that the reaction time (kinetics) plays a key role in the formation of these phases. For example, scorodite (FeAsO4 · 2H2O) was found to be metastable at shorter reaction times

with FAsH (FeAsO4 · 0.75H2O). The kinetic aspect of these reactions naturally was found to be an important property of the reaction never before considered. Of the phases

produced in this study, arsenic stability tests showed that scorodite (FeAsO4 · 2H2O) and BFAS were found to yield low arsenic release while FAsH yielded much higher arsenic release and thus unacceptable as an arsenic carrier.

207 As part of this research, in addition to identifying the phases formed via numerous techniques (ICP-AES, XRD, SEM) a more detailed vibrational (IR and Raman) spectroscopic investigation of these phases was conducted. This was done to further investigate the molecular groups of interest in these phases such as arsenate, water and sulfates. In this part of the research, the key vibrational signal of each phase, and factor group analyses of modes was for the first time given for the phases found in this system. This type of molecular analysis then that lead to the realization that the chemical formula

given by Swash and Monhemius for “Type1: Fe2(HAsO4)3 · zH2O” could not be possible

and as such the better suited molecular formula for this phase was FeAsO4 · 3/4H2O given which supported the crystallographic information but also the vibrational assignment of the observed spectra. For the crystallographically unknown BFAS phase, vibrational spectroscopy was employed to prove for the first time that a molecular solid solution occurs between AsO4↔SO4 in the structure of this phase, information to be used later when the crystal structure of this phase (BFAS) was solved but is unfortunately out of the scope of this thesis due to time and space restriction. The last part of this study was used the molecular information gathered of the pure system, to investigate for the first time using this technique (vibrational spectroscopy), the nature of the arsenic form of an industrial residue (donated by the Barrick Gold Corporation) which could not be identified otherwise due to the nature of the sample. The arsenic (as arsenate) was found to present in the form of the BFAS form thus indicating that this phase may be indeed produced as an arsenic byproduct during the processing of gold.

In the second part of the research, the Cu(II)-Mg(II)-Fe(III)-AsO4-SO4 system was investigated due its relevance in the copper industry and the CESL process employed by Teck Metals. In this part of the research it was observed that in all cases scorodite

(FeAsO4·2H2O) was the principal thermodynamically and most stable arsenate phase that was formed at all Fe/As ratios studied, and in the presence of foreign divalent cations (Cu2+ or Mg2+) after 60 minutes (and up to 10 hr) reaction times. The formation of an intermediate kinetically favored gel type (“liquid crystal”) of basic cupric ferric arsenate sulfate phase was for the first time observed during the heat up period (below 60 minutes) going from 25°-150 °C when an equimolar concentration of arsenic, copper and iron was

208 present. No divalent (Cu2+ or Mg2+) co-precipitates in an arsenate or sulfate form were detected and the amount of the divalent cation found in the solids was < 1 %, which meant no loss of precious metals occurred, this was in agreement with the industrially produced results of the CESL process. It was proposed here for the first time that the chemical reasoning for this lack of metal substitution in scorodite and therefore arsenic affinity may arise from the divalent nature of the cations charge not being sufficient enough to neutralize the oxygen charges or as a result of the higher affinity (electronegativity) of the Fe3+ cations has towards arsenate versus that of the divalent cations of interest in this study (Cu2+ or Mg2+) and in the industry (Co2+, Zn2+, Ni2+). For the first time investigation of an arsenic containing residue (donated by the Teck Metals Corporation) produced under industrial CESL conditions was undertaken, this revealed that their produced industrial residues had the presence of hematite (α-Fe2O3), elemental sulfur (S0) and an arsenate form which closely resembled that of our BFAS or gel type of intermediate phase found in this study as a result of short reaction times (~60 minutes) employed.

The last part of my research for my PhD thesis involved the investigation of the

Ca(II)-Fe(III)-AsO4-SO4 system (synthetic and natural). The motivation for this came from the parallel work conducted by our research group and evidence of these types of phases found in the disposal tailing sites of gold mines. It was found from these studies that the synthetic yukonite was found to be equivalent at the atomic, molecular and structural level to the natural yukonite. At the molecular level for the first time, arseniosiderite, via vibrational spectroscopy (IR and Raman) was found to have an H- 2- bonding system as in scorodite and exhibit an extra arsenate mode indicative of HAsO4 groups. In yukonite in contrast, a wide diffuse H-bonding environment was observed with only arsenate groups present. The As K, Fe 2p and Ca 2p XANES spectra of yukonite and arseniosiderite were found to be identical, confirming that the local nature of the As, Fe and Ca atoms in these closely related but distinct phases is the same correcting previous erroneous literature information. TEM and diffraction (X-ray and electron) showed for the first time that yukonite consists of nano poorly crystalline domains while in the case of arseniosiderite micro-size single crystal domains exist. Thus a chemical and

209 structural relation between these mysterious yet widely occurring industrial arsenate compounds was established for the first time, something that has been debated for some time since the known existence of these two minerals.

7.2 Future Directions

For the Fe(III)-AsO4-SO4 pure system some issues still remain to be addressed that could not be all included in the scope of this work. The first is to determine the crystal structure of the BFAS (Type 2 = Phase 3) phase, which has shown to have satisfactory arsenic short and long term arsenic retention properties and correlate its structure with its functionality (i.e. arsenic release). In this aspect, the structural relation between BFAS

and BFS (FeOHSO4) is something that is still unknown but an important piece of information to gather in further establishing the domain of formation of these high temperature ferric arsenate phases, especially at higher Fe(III)/As(V) molar ratios than that investigated here ( > 4). In our case the Fe(III)/As(V) molar ratio of the starting solutions was adjusted by usually varying the As(V) concentration and fixing the Fe(III); however the effect of fixing the As(V) concentration and varying the Fe(III) was not investigated here and thus it may be worth to investigate. Precipitation by in-situ 2+ 3+ oxidation [O2(g)] of Fe and As 2O3 systems may be explored at various process conditions to observe its effect on the mode of arsenic precipitation and phases formed. The phases formed from the in-situ oxidation should be compared (using several of elemental, electronic, molecular and structural techniques) to the phases generated in the present study. The in-situ oxidation may be looked at to get a more realistic insight into the change in solution and solid chemistry that occurs during the industrial process before ending up with the final residues (which were analyzed here). Lastly, as mentioned in these studies after the long term stability measurements were finished (at 25°C) no evidence of phase transformation occurred at the low or high pH conditions. This may be due to the small amount of material that reacted (only ppm) or due to other factors that needs to be further investigated.

210 In-situ precipitation during leaching of arsenide mineral mixtures such as

enargite(Cu3AsS4), arsenopyrite(FeAsS), chalcopyrite(CuFeS2), pyrrhotite(FeS) and

pyrite(FeS2) may be also investigated to study if leaching affects the formation of the

Fe(III)-As(V)-SO4 precipitates.

Finally for the Ca(II)-Fe(III)-AsO4-SO4 system, again the crystal structures of the two main and relevant phases (yukonite and arseniosiderite) are still unknown, this is largely due to their structural complexity. In the case of yukonite due to its nano-poorly crystalline short range order, the use of specialized techniques such as high energy XRD and Pair Distribution Function data analysis should be employed to solve its 3D structure so we may get insight into its structure and functionality (i.e. arsenic release). This is recommended as it is still not well understood what controls its arsenic retention properties, is it its structure or a solution co-ion effect? Moreover, the use of this specialized technique can be employed to analyze the products obtained after the long- term arsenic stability measurements to observe if any phase transformation has occurred. Lastly, something that should be done is to investigate the nature and form of the arsenic containing residue (s) from the uranium industry in a similar manner as that done in this thesis for the gold and copper industrial residues.

211 8. Appendix

8.1 Additional material from chapters

8.1.1 (Chapter 3) The Hydrothermal Fe (III)-AsO4-SO4 system at 150-225°C

Table S1. Experimental (this work) and reference Scorodite lattice parameter data.

◦ ◦ ◦ Orthorhombic a (A ) b (A ) c (A ) Le Berre [23] 10.36 10.04 9.00 JCPDS 10.32 10.03 8.94 (005-0216) This work 10.30 10.00 8.93

212 Table S2. Comparison between XRD patterns obtained from the experimental Scorodite product, the structure simulated with CaRIne and the reference data.

Scorodite Simulation Scorodite Plane Experimental With Scorodite JCPDS Index Pattern Atom Positions (00-005-0216) d(A°) d(A°) d(A°) (hkl) 5.58 5.59 5.56 111 4.99 5.00 4.95 020 4.45 4.46 4.44 201 4.07 4.07 4.06 211 3.78 3.79 3.78 112 3.33 3.37 3.36 202 3.17 3.17 3.16 122 3.05 3.05 3.05 311 2.99 2.99 2.98 131 2.75 2.75 2.75 113 2.67 2.67 2.67 231 2.58 2.57 2.58 400 2.50 2.50 2.50 213 2.31 2.31 2.31 331 2.18 2.18 2.18 412 2.14 2.13 2.13 114 2.11 2.11 2.11 332 2.04 2.04 2.04 422 1.94 2.00 2.00 124 1.83 1.83 1.84 052 1.80 1.79 1.80 440 1.76 1.75 1.75 522 1.66 1.66 1.66 611 1.65 1.64 1.65 144 1.58 1.58 1.58 253 1.51 1.51 1.51 533 1.47 1.47 1.47 613 1.40 1.40 1.40 444 1.28 1.28 1.28 416 1.23 1.23 1.23 812 1.21 1.20 1.21 281 1.19 1.19 1.19 644 1.11 1.11 1.11 814 1.05 1.05 1.05 932 1.02 1.02 1.02 933 1.01 1.01 1.01 1002

213 8.1.2 (Chapter 4) Vibrational Spectroscopic study of hydrothermally produced Fe (III)-AsO4-SO4 phases

Table S1. Solid composition analyses for the various phases characterized.

Table S2. Spectral predictions and correlation schemes for the internal modes of

H2O in Scorodite.

214 Table S3. Spectral predictions and correlation schemes for the internal modes of 3- AsO4 in Scorodite.

Table S4. Spectral predictions and correlation schemes for the internal modes of

H2O in Ferric Arsenate sub-Hydrate.

215 Table S5. Spectral predictions and correlation schemes for the internal modes of 3- AsO4 in Ferric Arsenate sub-Hydrate.

216 8.1.3 (Chapter 5) The Hydrothermal Cu (II)-Fe(III)-AsO4-SO4 System at 150°C

Table S1. XRD (λ = 1.78897 Å) of semi-crystalline component of the green gel formed in Case 4 (40 minutes and 101 °C) shown in Figure 10.

217

218

Figure S3. TEM and EDX of the semi-crystalline component of the gel product (this is a distinct particle then that shown in Fig 12). Lattice spacing was found to be 3.27 Å.

219

Figure S6. Photograph of intermediate gel products after 7 days exposed to air.

220

221 8.1.4 (Chapter 6) The Hydrothermal Ca (II)-Fe (III)-AsO4 System at 95°C

Figure S1. X-ray analysis of natural and synthetic (Syn-1) yukonites at different X- ray wavelengths: (a) Powder x-ray diffraction using lab based (λ=1.54Ǻ) and synchrotron based radiation (λ=0.458Ǻ) and (b) lab based (λ=0.71Ǻ) single crystal diffraction on the Tagish Lake sample.

222

223 8.2 Additional details on methods used

8.1.1 (Chapter 3) The Hydrothermal Fe (III)-AsO4-SO4 system at 150-225°C

Attenuated Total Reflectance - Fourier Transform Infrared (ATR-IR): Through out this thesis, the method of choice we used in terms of IR spectroscopy was the ATR mode. This was done in comparison to other methods such as the transmission mode often found in IR measurements (Coates, 1998). ATR was employed as a result of the fact that in all cases, the synthetic (or natural) products analyzed were opaque in nature and thus doing the IR measurement in the Transmission mode would result in a ”smearing” of vibrational structure. In contrast ATR is well suited for opaque materials and as a result we observe far more vibrational structure which is important in assigning the modes and group analysis. The ATR mode also offers the advantage of conducting this analysis on solids (amorphous or crystalline) and liquids. In our case the solid sample powders could be measured as they were. Similar pressures were applied to all samples to ensure we had good intimate contact with the surface. Good contact is important so that the critical angle criterion is maintained and therefore interaction will occur between the internally reflected radiation and the sample occurs, giving a good signal. In the ATR-IR mode the intensity may vary as a result of the materials index of refraction or the amount of pressure exerted, as such we never used the ATR-IR intensities to make any correlations and only used the spectral features to interpret the data. The accumulation and data processing of the spectra were conducted using the Spectrum software (version 5.3.1) in a standard fashion. In terms of data processing, the spectra once collected were treated wit the software to remove the background while the normalization (background to the data) is done automatically by the software. In terms of artifacts that may not be due to the vibrational structure of the samples when doing these measurements (ATR -1 mode), there is always a small amount of CO2 observed at 2300 cm which comes from the air and sometimes the ATR crystal (in our case diamond observe at 2000 cm-1). In our case, the choice of using the diamond ATR crystal was appropriate since our vibrational bands of interest (arsenate, hydroxyl, and sulfate) did not occur anywhere these artifacts occured. The reproducibility of the measurements was ensured by taking 200 individual reproducible scans and then adding them together; the sums of all the reproducible scans

224 are the IR spectra shown throughout this thesis. Moreover to ensure reproducibility of the spectra, some samples were checked and measured in two distinct ATR-IR machines; one in the chemistry department (CSACS) and the other at the materials engineering department of McGill University (Professor Nazhat).

Raman Microscopy: Through out this thesis the method of choice to use in terms of Raman spectroscopy was in the Microscope with the use of the 514 nm wavelength for all products (chapters 3 - 5). The choice of the 514 nm laser was used for all these products as a result of the fact it gave the best signal (counts) and no fluorescence background. In the beginning of the test trials to get the best spectra (high counts and little fluorescence background), the use of the 632 nm and 785 nm lasers were used. However, the higher wavelengths gave rise to higher background signals and as a result were not chosen to be suitable for the analysis of these materials. In case of arseniosiderite and yukonite, the use of the 514 nnm laser gave no signal or structure and only a broad fluorescence band was observed. As a result of the latter, higher wavelengths were employed (632 nm laser) to get a good spectra. To ensure no radiation damage occurred in our inorganic metal oxide solids, all measurements were conducted at 10 % of the laser output (~ 2mW) at the microscope exit. The power at the laser and microscope exit is always measured in the instrument as part of the Centers’ (at the Chemistry Department of the University of Montreal) way to check their instrument. Moreover in the initial stages of analysis, all the solid samples were tested at various laser power outputs (10%, 50 % and 100 %) and their spectra observed and over plotted to ensure that no spectral distortions or shifts in positions had occurred. The energy calibration and counts were ensured to be satisfactory using the 520 cm-1 silicon peak using both internal and external silicon crystal standards. Data processing and collection was done with Wire (version 3.2) software. No spectral artifacts are observed in these measurements from the instrument or components. Samples were mounted on microscope glass slides and made as flat as possible so that it was easier to focus the microscope objective. In the beginning of the measurements, a bulk mode Raman instrument was used to observe the Raman structure of these phases. However due to the low counts we obtained alternative Raman detection modes such as when the microscope

225 was used. The choice of the microscope was done as opposed to a bulk mode as a result of the fact that we obtained much higher signal counts and better spectral resolution but also as our synthetic samples were always homogeneous in nature, the use of the microscope was reasonable. In addition in the initial test trials to obtain the best spectra of these phases, the measurements were done using the 5x (spot size 10’s of microns), 10x, 50x and 100x (~1 micron with the 514 nm laser) objectives to ensure the exact same spectra was obtained regardless of the objective, which in all cases it was. As a result we decided to conduct our measurements using the 50x objective as it gave the best results (counts and large enough area). Additional spots on random locations of the sample were also done to ensure tha the same spectra was obtained, which in all cases it was.

X-ray Absorption Near Edge Structure (XANES): In these studies measurements in both in the hard X-ray (As K-edge) and soft X-ray (Ca and Fe L-edge) regime were conducted. For the As K-edge, all samples were well grounded and spread evenly in the sample holders (CLS customized for HXMA) to avoid any pin hole effects, kapton tape was used to seal the powder. The amount of mass mixed mixed with an inert powder (BN) were such that we calculated to have 2 absorption lengths for each of the samples, so we wold have as uniform samples as possible and avoid any thickness effect on the collected spectra. The densities used were from literature values. Transmission yield was the choice of detection method for our As K-edge measurements as a result that we had sufficient amount of concentration of arsenic in all the phases (> 25 wt. %) and as such we need not to resort to other detection methods such as the Fluorescent yield (used for lower concentrations). In terms of beam damage (reduction or decomposition) to the samples, we did not expect (similar studies have been conducted in the literature for phases such as those studied here) or observe any such effects to occur in the collected spectra (at least two reproducible scans were taken and then compared). All data processing (calibration, sum of spectra, normalization, spline addition and Fourier transform) were conducted with the software Athena in the IFFEFIT package. For the Fe and Ca L-edge measurement, the samples were pressed into conducting carbon tape then placed in the vacuum chamber of the SGM line and always using 50 µm at the exit slits. This size of the exit slits was found to give best result in terms of signal

226 and resolution. The spectroscopic resolving power (E/ΔE) for measurements at the SGM is estimated to be better than 3200 and as such is ideal for monitoring oxidation states. It is noted at this point that all of our compounds are inorganic transition metal oxides which are very stable at room conditions or under vacuum. The TEY yield was chosen to be the detection mode of choice as the electron yield for that element is higher then its fluorescence yield and as such is better suited to get a better signal. Again in our compounds there was no radiation damage to be expected as numerous researches has been previously conducted in the literature similar experiments on similar systems as in this study. All samples were also visually inspected after each collection to ensure no burning occurred and each scan lasted maximum 5- 10 minutes. Standard reagent grade (Fischer Scientific) hematite (Fe3+) and melanterite (Fe2+) were used to check that our

energy calibration and oxidation states using as a reference the main L3 peak of hematite as reported throughout the literature (see references in experimental section of chapters 3 and 6). Charging of the samples were also not observed as no spikes or differential structure was observed Furthermore no reduction of our ferric states were observe in any of our samples, as their observed positions were exactly identical with that of hematite in all scans measured before and after energy calibration indicating that no reduction effects had taken place in our samples. XAS energies are typically found to not be accurate better then 0.1 eV and as such we can assume that the accuracy in the measurement should be less then or equal to 0.1eV. All data processing (energy calibration and normalization) were conducted with the Athena software from IFFEFIT package. The spin orbit splitting reported here were estimated from the simple difference

between the main L2 and L3 edges (spin-orbit splitting). In our case the spin-orbit splitting was reported as additional information to further verify the oxidation state but the reader is noted that these numbers are only approximate. To verify that this method was reasonable to apply to our compounds, the spin-orbit splitting of reagent grade

hematite and melanterite were checked against the literature values. The L2,3 maximums

of hematite in this study were observed at 722.6 eV(L2) and 709.5 eV(L3) and their

corresponding prepeaks at 721.1eV(L2) and 708.1eV (L3) respectively. The separation

between the L2 and L3 edges due to spin-orbit coupling was found to be 13.1 ± 0.1eV in agreement with the Fe L-edge XANES and ELNES values reported by previous authors

227 for Fe +3 compounds (see references in experimental section of chapters 3 and 6). As noted in the text for our purposes this approximation of the spin-orbit splitting was sufficient enough for use to have as this was mainly used to verify the oxidation state of the phases along with the position. However, if more exact spin-orbit splitting values are needed, the reader is referred to conduct XPS analysis where more accurate numbers may be obtained as a result of the fact that in XPS we observe more atomic like states then in XAS; in the latter the features are dependent upon oxidation state, coordination state, and the ligands present to name a few.

8.1.2 (Chapter 5) The Hydrothermal Cu (II)- Fe (III)-AsO4-SO4 system at 150°C For ATR-IR and Raman measurements see section 8.1.1 as the same methods and analysis were undertaken upon collection of the measurements for these solids.

8.1.3 (Chapter 6) The Hydrothermal Cu (II-) Fe (III)-AsO4 system at 150°C For ATR-IR, Raman and XAS measurements see section 8.1.1 as the same methods and analysis were undertaken upon collection of the measurements for these solids.

228 8.3 Complete list of publications, conference proceedings and projects that have been produced during the coarse of PhD studies

A. Articles published in refereed journals 1. Becze L., Gomez M. A. and Demopoulos G. P. (2009) Characterization and Dissolution Behaviour of a Chloride-containing Silver Material in Nitric Acid Solution. Erzmetall 62, 255-262 2. Becze L., Gomez M. A., Le Berre J. F., Pierre B. and Demopoulos G.P. (2009) Formation of Massive Gunningite-Jarosite Scale in an Industrial Zinc Pressure Leach Autoclave – a Characterization Study. Canadian Metall. Quarterly. 48, 99-108. 3. Gomez M.A., Assaaoudi H., Becze L., Cutler J. N. and Demopoulos G.P. (2010) Vibrational spectroscopy study of hydrothermally produced scorodite

(FeAsO4·2H2O), ferric arsenate sub-hydrate (FAsH; FeAsO4·0.75 H2O) and basic

ferric arsenate sulphate (BFAS; Fe[(AsO4)1-x(SO4)x(OH)x]·wH2O ) Journal of Raman spectroscopy. 41, 212-221. (Impact Factor 3.5) 4. Gomez M. A., Le Berre J. F., Assaaoudi H. and Demopoulos G. P. (2010) Raman spectroscopic study of the hydrogen and arsenate bonding environment in 3+ 3+ isostructural synthetic arsenates of the variscite group-M AsO4•2H2O (M = Fe, Al. In and Ga)-implications for arsenic release in water. Journal of Raman Spectroscopy. DOI: 10.1002/jrs.2639. (Impact Factor 3.5) 5. Gomez M. A., Becze L., Blyth R.I.R., Cutler J. N. and Demopoulos G.P. (2010) Molecular and structural investigation of yukonite (synthetic & natural) and its relation to arseniosiderite. Geochimica et Cosmochimica Acta. 74, 5835-5851. (Impact Factor 4.4) 6. Lee K. E., Gomez M. A., Elouatik S. and Demopoulos G. P. (2010) Further understanding of the adsorption of N719 complex on Anatase TiO2 films for DSSC applications using Vibrational spectroscopy and Confocal Raman imaging. Langmuir, 26, 9575-9583. (Impact Factor 4.0)

229 7. Charbonneau C., Lee K. E., Shan G. B., Gomez M. A., Gauvin R. and Demopoulos G. P. (2010) Preparation and DSSC performance of mesoporous film photoanodes based on hydroxyl-rich anatase nanocrystallites. Electrochemical and Solid-State Letters, 13, H257-H260. (Impact Factor 2.1)

B. Articles submitted or to be submitted in refereed journals before end of PhD period

8. Gomez M. A., Becze L., Ceklin M. and Demopoulos G. P. (2010) The effect of copper

on the formation of Scorodite (FeAsO4·2H2O) from aquoes hydrothermal conditions: Evidence of a hydrated cupric arsenate sulfate lived intermidiate. Inorganic Chemistry, (Submitted and in review) 9. Gomez M.A., Becze L., Bluteau M.C., Le Berre J.F., Cutler J. N. and Demopoulos G.P. (2010) On the hydrothermal reaction chemistry and characterization of ferric

arsenate phases precipitated from Fe2(SO4)3-As2O5-H2SO4 solutions” Hydrometallurgy (Accepted 2011) 10. Lee K. E., Gomez M. A., Regier T., Hu Y. and Demopoulos G. P. Further Understanding of the Electronic Interactions between N719 Sensitizer on Anatase

TiO2 Films: A Combined XAS and XPS study. Journal of Physical Chemistry C, (Submitted and in review) 11. Gomez M. A., Ventruti G., Assaaoudi H., Ceklin M., Putz H. and Demopoulos G. P. (2010) The nature of synthetic Basic Ferric Sulfate and Basic Ferric Arsenate Sulfate: A crystal and molecular structure determination and applications of their materials properties. Inorganic Chemistry (In progress).

230 C. Other refereed contributions Conference Papers 12. Gomez M. A., Becze L., Bluteau M. C., Le Berre J. F., Cutler J. N. and Demopoulos

G. P. (2008) Autoclave Precipitation and Characterization of Fe (III) - AsO4 -SO4 phases. In Hydrometallurgy’08 (eds. Young C. A., Taylor P. R., Anderson C. G. and Choi Y.), SME, Phoenix, Az, pp. 1078-1085. 13. Becze L., Gomez M.A., and Demopoulos G.P. (2009) Characterization and Dissolution Behaviour of a Chloride-Containing Silver Material in Nitric Acid Solution. In Proceedings of European Metallurgical Conference EMC 2009, Innsbruck, Austria. 14. Lagno F., Garcia I., Ledesma C., Demopoulos G.P., Becze L., Gomez M.A., Katsarou L. (2009) Fixation of Arsenic and Antimony from Copper Smelter Flue Dust and Electrorefinery Bleed-Off as Crystalline Scorodite,. In Proceedings of HydroCopper 2009, Antofagasta, Chile. 15. Gomez M.A., Becze L., Cutler J. and Demopoulo G.P. (2009) Hydrothermally

Produced Synthetic and Industrial Fe(III)-AsO4-SO4 Phases: Molecular Characterization and Environmental Stability Evaluation. In Proceedings of 218th National ACS Fall Meeting, Geochemistry Section,Washington, D. C., USA. 16. Demopoulos G. P., Charbonneau C., Lee K. E., Shan G. B., Gomez M. A. and Gauvin R. (2009) Aqueous Synthesis of Anatase Nanocrystallites and Their Application to Dye-Sensitized Solar Cells. presented at IC4N 2009, Int’l Conference from Nanoparticles and Nanomaterials to Nanodevices and Nanosystems, Rhodes, Greece. 17. Demopoulos G. P., Charbonneau C., Lee K. E., Shan G. B., Gomez M. A. and Gauvin R. (2009) Synthesis of Hydroxyl-Rich Anatase Nanocrystallites, their Characterization and Performance as Photoanode in Dye-Sensitized Solar Cells. ECS Transactions, 21, pp. 23-34. 18. Gomez M. A., Becze L., Lee K.E., Charbonneau C., Demopoulos G. P., Cutler J. N., Assaaoudi H. and Elouatik S. (2010) Raman Spectroscopy in the Hydrometallurgical and Materials Engineering World. In Proceedings of the American Institute of Physics 1267, pp. 546-547.

231 19. Becze L., Gomez M. A., Petkov V., Cutler J. N. and Demopoulos G. P. (2010) The

Potential Arsenic Role of Ca-Fe(III)-AsO4 Compounds in Lime Neutralized Co- Precipitation Tailings. In proceedings of Uranium 2010 and 40th Annual Hydrometallurgy Meeting, Uranium Processing -Tailings Section, CIM-MET SOC, Saskatoon, SK, Canada,Vol II, pp. 327-336. 20. Lee K. E., Gomez M. A., Elouatik S., and Demopoulos G. P. (2010) Confocal Raman

Imaging of N719 Complex Adsorbed onto Anatase TiO2 Films in High Wavenumber Region: Examining the involvement of Ti-OH Surface Groups. ECS transactions, 28, pp. 1-7. 21. Demopoulos G. P., Charbonneau C., Lee K. E., Shan G. B., Gomez M. A., and Gauvin R. (2010) New Cost-effective Manufacture of DSSC TiO2 Paste with Enhanced Sensitizer Anchoring Properties, Materials for Energy, DECHEMA, Karlshrue, Germany pp. 372-374.

Conference Posters 22. Gomez M. A., Becze L., Demopoulos G. P. and Cutler J. N. (2007) Autoclave

Precipitation of Fe(III)-AsO4-SO4 Phases-Characterization and Stability Evaluation. 10th Annual Users Meeting, Canadian LightSource, Inc. Saskatoon, Canada. 23. Becze L., Le Berre J. F., Gomez M. A. and Demopoulos G. P. (2007) Arsenic control

in hydrometallurgy via the formation of Ca-Fe-AsO4 compounds, Conference of Metallurgist (COM 2007), Toronto, Canada. First Prize (best paper award). 24. Gomez, M. A. Becze L., Leberre J. F., Cutler J. N. and Demopoulos G. P. (2008) Spectroscopy in Hydrometallurgical Research. CLS Earth and Environmental Science Theme Workshop, Canadian LightSource, Inc. Saskatoon, Canada. 25. Gomez M.A., Becze L., Demopoulos G.P., Assaaoudi H., Cutler J. N. and Blyth R. I.

R. (2009) Characterization of Ca-Fe(III)-AsO4-SO4 Phases relating to Arsenic Disposal in Hydrometallurgical Operations. 12th Annual Users Meeting, Canadian LightSource, Inc. Saskatoon, Canada.

232 26. Charbonneau C., Lee K. E., Shan G. B., Gomez M. A., Gauvin R., and Demopoulos G. P. (2009) High-performance Dye-sensitized Solar Cell based on Large Surface Area Aqueous Synthesized Titania Nanoparticles. Materials Research Society Fall Meeting, Boston, USA.

Technical Company Reports 27. Demopoulos G.P., Becze L., Li Z., Le Berre J-F., Gella V. and Gomez M.A. Investigation of the Crystallization Behaviour of Zinc and Iron(II) Sulphates in Xstrata Copper’s Kidd Creek Zinc Pressure Leach Autoclave. Submitted to Xstrata Copper, February 2008. 28. Gomez M.A., Becze L. and Demopoulos G.P. Characterization of arsenic-containing precipitates originated from Teck Cominco’s Process Streams. Submitted to Teck Cominco, September 2008. 29. Gomez M.A., Becze L. and Demopoulos G.P. Characterization of arsenic containing discharge slurry originated from Barrick’s Donlin Creek continuous pressure oxidation pilot plant. Submitted to Barrick Gold Corporation, October 2008. 30. Becze L., Gomez M.A., Katsarou L. and Demopoulos G.P. Characterization and medium term stability of scorodite produced by atmospheric step-wise precipitation in Ecometales’ Ltd’s pilot plant. Submitted to Ecometales Ltd., December 22, 2008. 31. Gomez M.A. Raman spectroscopy of Ilmenite Concentrates. Submitted to Rio Tinto Iron and Titanium, May 2008. 32. Gomez M.A., Becze L., Demopoulos G.P., Assaaoudi H., Cutler J. N. and Blyth, R.

I. R. (2008) Characterization of Ca-Fe(III)-AsO4-SO4 Phases relating to Arsenic Disposal in Hydrometallurgical Operations. Canadian Light Source, Inc. Activity Report. 33. Becze L., Gome M.A., and Demopoulos G.P. Project AAA-Abatement of arsenic and antimony: Characterization of scorodite produced by atmospheric step-wise precipitation in Ecometales’ Ltd’s demonstrative plant. Submitted to Ecometales Ltd., February 22, 2010.

233 34. Becze L., Gome M.A., and Demopoulos G.P. Characterization of copper smelter flue dust and a leaching residue formed in Ecometales’ Ltd’s pilot plant, Submitted to Ecometales Ltd., August 27, 2010.

234