Investigation of sodium silicate derived gels as encapsulating materials for scorodite

Transcription

1 Investigation of sodium silicate derived gels as encapsulating materials for scorodite Jessica Adelman Department of Mining and Materials Engineering McGill University Montreal Canada June 2013 A thesis submitted to McGill University in partial fulfilment of the requirements of the degree of Master of Engineering Jessica Adelman, 2013

2 ABSTRACT High content arsenic waste generated in the metallurgical industry can be converted into a synthetic mineral, scorodite, FeAsO 4 2H 2 O, and deposited into a waste disposal site. Scorodite is most stable in a certain ph range under oxic conditions. A novel way to enhance the range of stability for scorodite is to encapsulate it with an inert material. In this work, silicate gel is developed and investigated as a possible encapsulating material for scorodite. The work in this thesis includes 1) Synthesis of atmospherically precipitated scorodite, 2) Production, optimization and characterization of silicate gels, 3) Study of different silicate gel ageing treatments (sealed and unsealed at room temperature, drying at 40 C and hydrothermal treatment at 110 and 160 C), 4) Stability testing of silicate gel-coated scorodite (Scorogel) at ph 9 under oxic (600 mv E h ) and anoxic conditions (0 mv E h ) and 5) Characterization of Scorogel particles. As the initial method of gel formation (titration of 1M sulfuric acid into 1M sodium silicate) produced a silicate gel with high alkalinity (ph 10) that was incompatible with scorodite, a reverse titration method was developed that produced a gel with optimum ph profile (5-6.5). The produced gel however proved to have only a marginal effect on scorodite stabilization. This prompted investigation of different ageing techniques as a means of producing silica-like coatings with better stabilization potential. The ageing techniques included drying (22, 40 C) and hydrothermal (HT) treatment (110, 160 C) of Scorogels. Surprisingly these measures proved counterproductive as aged Scorogels showed a higher release of As than scorodite alone. Through surface-sensitive (XPS, X-ray photoelectron spectroscopy) depth profile analysis, and molecular-sensitive analysis (Fourier Transform Infrared (FTIR) and Raman mapping), it was discovered that the silicate engaged into an ion-exchange type reaction on the surface of scorodite by bonding to iron, hence the observed release of arsenic. I

3 RÉSUMÉ Les déchets avec de haute teneur en arsenic qui sont produits par l'industrie métallurgique peuvent être convertis en un minéral synthétique : la scorodite FeAsO 4 2H 2 O. Les déchets convertis vont être ensuite disposés dans un site d'élimination des déchets. Dans des conditions oxiques, la scorodite est plus stable pour une certaine gamme de ph. Une nouvelle façon d'améliorer la gamme de stabilité de la scorodite consiste à l encapsuler avec un matériau inerte. Dans ce travail, un gel de silicate est développé et étudié comme matériau d'encapsulation de la scorodite. Cette thèse comprend: 1) La synthèse de la scorodite par précipitation à pression atmosphérique; 2) La production, l'optimisation et la caractérisation des gels de silicate; 3) Étude de différents traitements de vieillissement des gels de silicate (scellés et non scellés à la température ambiante de 22 C, séchage à 40 C et un traitement hydrothermique à 110 et 160 C); 4) Essais de stabilité des scorodites revêtus de gel de silicate (Scorogel) à un ph de 9 dans des conditions oxiques (600 mv E h ) et anoxiques (0 mv E h ); 5) La caractérisation des particules de Scorogel. Le procédé initial de formation du gel de silicate (titrage avec 1 M d'acide sulfurique dans 1 M de silicate de sodium) produit un gel de silicate à haute alcalinité (un ph de 10) qui est incompatible avec la scorodite, une méthode de titrage inverse a été développée pour produire un gel avec un ph optimal entre Le gel ainsi produit c est avéré avoir un effet marginal sur la stabilisation de la scorodite. Ce résultat a amené l étude de différentes techniques de vieillissement comme un moyen de produire des revêtements de silicate avec une meilleure stabilisation de la scorodite. Les techniques de vieillissement incluent le séchage (à 22 et 40 C) et le traitement hydrothermique (à 110 et 160 C) des Scorogels. Étonnamment ces vieillissements se sont révélés contre-productifs car les Scorogels âgées ont exhibé une plus grande libération d arsenic que la scorodite sans revêtement. L analyse chimique de surface profondeur (XPS, la spectrométrie photoélectronique X) par profile en et l'analyse moléculaire (IRTF, la spectroscopie infrarouge à transformée de Fourier et la cartographie Raman) des Scorogels ont démontré que le silicate est impliqué dans une réaction de type "échangeur d'ions" à la surface de la scorodite par liaison avec le fer, ce qui explique la libération observée d'arsenic. II

4 ACKNOWLEDGEMENTS I first and foremost want to thank my supervisor Professor George P. Demopoulos for his guidance and immense knowledge that he shared with me. I am really thankful to have him as a supervisor, this Masters challenged and taught me the most. I want to thank the McGill HydroMET group members for all of their support and camaraderie, especially Lydia, Levente, Sonia, Cesar, Nick, Renaud, Cecile, Mario, Thomas, Christoph, Matthew, Marie-Christine, Liying, Jay and Micah. I am thankful for all of the technical help and support from Monique Riendeau, Ranjan Roy, Andrew Golsztajn, Petr Fiurasek and Lihong Shang. Your assistance saved me many times, thank you for your patience. I especially want to thank Samir Elouatik from Université de Montreal, Department de Chemie. The characterization work you assisted me with was critical to my project s success. A big thank you! Thank you to NSERC and Teck for the funding support and insight during sponsor meetings. Finally, a gigantic thank you to my family, my mom and dad Reisa and Bernie, and my bro Matt, Alex, Aunty Judy, Uncle Ed, Uncle David. I missed you guys this whole time, but your support and humour kept me going. Especially Amy and Lucy (and Joey and Blue)! III

5 TABLE OF CONTENTS ABSTRACT... I RÉSUMÉ... II ACKNOWLEDGEMENTS... III LIST OF FIGURES... VI LIST OF TABLES... IX CHAPTER 1 INTRODUCTION... 1 CHAPTER 2 LITERATURE REVIEW The Arsenic Problem Arsenic Fixation Methods Scorodite Production of Scorodite Atmospheric Production of Scorodite Other Approaches to Producing Scorodite Stability of Scorodite Industrial Application of Scorodite Encapsulation of Scorodite The Concept and Previous Studies Silica-based Encapsulation Methods Sodium Silicate-derived Gels Characterization of Sodium Silicate-derived Gels Summary CHAPTER 3 EXPERIMENTAL METHODS Chemicals Procedures Production of Scorodite Production of Silicate Gel Washing and Blending Ageing IV

6 3.2.5 Stability Testing Analysis and Characterization Elemental Analysis FTIR and Raman Spectroscopy SEM, TGA and XRD Analysis X-Ray Photoelectron Spectroscopy (XPS) Process Flow Chart CHAPTER 4 RESULTS AND DISCUSSION Stability of Atmospheric Scorodite Silicate Gel Formation Reaction of Sodium Silicate with Sulfuric Acid Silicate Gel Washing - Removal of Sodium and Sulfate Silicate Gel Ageing Silicate Gel as Encapsulant for Scorodite General Initial Testing- High ph Silicate Gel Reverse Titration- Low ph Silicate Gel Oxic and Anoxic Stability Protocols Oxic and Anoxic Stability Response Effect of Modified Ageing on Silicate Gel Transformation and Scorogel Stability Ageing Scorogel by Drying at 22 C and 40 C Ageing by Hydrothermal Treatment of Silicate Gel and Scorogel (110 C and 160 C) How Does Silicate Affect the Release of Arsenic from Scorodite? Molecular and Surface Characterization of Scorogel What Can be Learned from this Study and the Way Forward CHAPTER 5 CONCLUSION LITERATURE CITED APPENDIX A V

7 LIST OF FIGURES Figure 2.1- Graphical representation of the McGill supersaturation-controlled based scorodite precipitation process: Scorodite is produced within the metastable zone, where heterogeneous nucleation occurs on seed particles (Demopoulos et al., 2003)... 9 Figure 2.2- Solubility data of scorodite under different ph conditions at 22 C. A comparison of previous studies with the results from M.C. Bluteau s research (Bluteau and Demopoulos, 2007) Figure 2.3- The release of arsenic from atmospheric scorodite as a function of E h (ph 7, gypsum-saturated water; NaOH base, 22 C) (Verdugo et al., 2012) Figure Aluminum phosphate coating around a single scorodite particle (Lagno et al. 2010) Figure 2.5- E h -ph diagram of Si-H 2 O at 25 C, Si=10-6 mol/kg, calculated with OLI 3.2 (MSE model) Figure 3.1- Applikon reactor used to produce atmospheric scorodite Figure 3.2- Atmospheric scorodite synthesis results with A) XRD results and reference spectrum displaying a crystalline scorodite spectrum and B) Total dissolved arsenic released (mg/l) over 5 24 hour washing cycles Figure 3.3- Silicate gel (from 1M sodium silicate solution) Figure 3.4- Description of autoclave (acid digestion vessel- Image by Parr Instrument Company) Figure 3.5- Flow chart, summarizing the various procedural steps followed from gel formation, washing, blending and ageing to stability testing of Scorogel material. Note: Different concentrations of sodium silicate solutions were prepared (0.25, 1 and 5M) however 1M is used in the flow chart as this was the concentration used for most series of work Figure 4.1- Hydrothermally produced scorodite that was used as seed in atmospheric scorodite production (Leetmaa, 2008) Figure 4.2- Images of scorodite produced by atmospheric precipitation at 85 C- (from left to right): Scorodite after filtration (image by C.Verdugo, HydroMET Laboratory) and SEM images of scorodite particles in silicate gel matrix (0.5:1 Si/As molar ratio) at different magnification. 37 Figure 4.3- Comparison of scorodite stability for two differently produced crops: Atmospheric vs. hydrothermal. Total dissolved arsenic (mg/l) released over 4 weeks of stability testing at ph 9 under oxic conditions from atmospheric scorodite (this study) and hydrothermal scorodite (Bluteau, 2004); ph adjustment done with Ca(OH) 2 in this study and NaOH in Bluteau study.. 38 Figure 4.4- Silicate tetrahedron (left) and molecular structure of sodium silicate (right). Grey spheres= silicon atoms, red spheres= oxygen atoms, purple spheres= sodium atoms. H 2 O groups are not shown. (McDonald and Cruickshank, 1967; Mills, 2008) Figure 4.5- Image of silicate gel prepared by titration of 1M sulfuric acid (30 ml) into 1M sodium silicate solution (60 ml) VI

8 Figure 4.6- TGA analysis of 1M sodium silicate gel, A) Not dried B) Air dried for 7 days Figure 4.7- Percent total sulfate (left) and sodium (right) removed after 14 hours of washing (water-to-gel w/w ratio of 10:1) Figure 4.8- Appearance of silicate gel before and after washing Figure 4.9- Scorodite manually blended with silicate gel at 0.5:1 Si/As molar ratio (Scorogel) 42 Figure SEM images and EDS mapping of silicate gel coated scorodite particles (Scorogel) (1:0.1 Si/As molar ratio) A) Thin section image of scorodite coated particles B) Silicate gel coated particle C) Hole observed in silicate gel coating, D) EDS mapping of Scorogel particle 43 Figure Reverse titration curves (sodium silicate sulfuric acid solution) showing gelation ph point for A) 0.25M B) 1M and C) 5M sodium silicate solutions Figure Stability testing of Scorogel in an Erlenmeyer flask filled with DI water, 40:1 w/w ratio water to scorodite Figure Short term stability testing of Scorogel compared to scorodite alone- Total As (mg/l) released over 8 days (ph 9, oxic conditions- 600 mv E h ; 1:1 Si/As molar ratio, 1M silicate gel prepared using reverse titration method) Figure Medium term stability testing of Scorogel compared to scorodite alone for the case of low concentration silicate gel- Total As (mg/l) released over 30 days: A) Oxic conditions (600 mv E h, ph 9) and B) Anoxic conditions (0 mv E h, ph 9); Scorogel produced from 0.25 M sodium silicate solution by reverse titration (0.1:1 Si/As molar ratio) Figure Silicate gel coated scorodite particles (Scorogel) at 1:1 molar ratio. Washing and ageing occurred under standard conditions (3 cycles of 2 hours, 2:1 water: solid w/w ratio, closed sealed 22 C ageing) Figure Effect of drying on silicate gel ageing and Scorogel stability- Total As (mg/l) released over 15 days: Scorogel air dried at room temperature (22 C) for 2 weeks or for 24 hours at 40 C (using washed 1M silicate gel, 0.1:1 Si/As molar ratio) compared to scorodite alone under A) Oxic conditions and B) Anoxic conditions. ph and E h levels are displayed for Day 15 (before adjustment to ph 9 and 0 mv E h (anoxic samples only)) Figure ph (A) and E h (anoxic samples) (B) profiles during stability testing for scorodite and Scorogel samples aged at 40 C for 24 hours and at RT (22 C) for 2 weeks (measured and adjusted values shown) Figure FTIR spectra of washed 1M silicate gel before and after hydrothermal treatment in an autoclave at 110 C for 8 hours. A) Si-OH group; B) H 2 O group; C) and D) Possible Si-based groups Figure Effect of hydrothermal treatment on ageing and Scorogel stability- Total As (mg/l) released over 15 days: Scorogel (using washed 1M silicate gel, 0.1:1 Si/As molar ratio) compared to scorodite alone under A) Oxic conditions and B) Anoxic conditions. Silicate gel and Scorogel were hydrothermally (HT) treated at 110 C and 160 C for 8 hours. ph and E h levels are listed for Day 15 (before adjustment to ph 9 and 0 mv E h ) VII

9 Figure The polymerization of silicic acid (Si(OH) 4 ) adsorbed onto Fe(III) surface bonding sites of ferrihydrite, A) Adsorption of Si(OH) 4 monomer, B) Polymerization, C) Siloxane links (Si-O-Si) between Si(OH) 4 monomers (Swedlund and Webster, 1999) Figure SEM images of HT treated (8 hours at 110 C) Scorogel particles (5M silicate gel, 0.5:1 Si/As molar ratio) and after 30 days of stability testing at ph 9 under oxic conditions Figure Si and As concentration in solution during stability testing (at ph 9) of hydrothermally (HT) treated Scorogel at 110 and 160 C (1M sodium silicate solution and 1M sulfuric acid, 0.1:1 Si/As molar ratio) Figure XPS depth profile analysis (As and Si peaks) of Scorogel particle before and after sixth etching cycles (Scorogel aged at 40 C for 24 hours) Figure FTIR mapping displaying a cross section of a Scorogel particle (aged at 40 C for 24 hours, and subjected to 15 days stability testing) and the associated intensity of arsenic depicted in color (red represents high As presence and blue no presence). Peak identified at cm -1 ) Figure Raman mapping displaying the presence of Si along a cross sectioned Scorogel particle (aged at 40 C for 24 hours, and underwent 15 days stability testing) and the associated intensity of silicon depicted in color (white represents high Si presence). Si peak identified at 1000 cm -1 ) Figure Depiction of a Scorogel particle (scorodite coated with silicate gel), with the possible formation of an iron-silicate phase VIII

10 LIST OF TABLES Table 2.1- Summary of arsenic release data from previous encapsulation studies under oxic conditions Table 2.2- Summary of arsenic release data from previous encapsulation studies under anoxic conditions Table 2.3- Summary of spectroscopic characterization data relevant to the analysis of sodium silicate and amorphous silica Table 3.1- Chemicals used during this study Table 3.2- List of equipment in reactor setup used to produce atmospheric scorodite (Figure 3.1) Table 4.1- Stability constants of Fe 3+ silicate and Fe 3+ arsenate complexes Table Series A.1- As (mg/l), ph and E h results (for samples under anoxic conditions only (0 mv) for stability test of washed gels versus unwashed gels (0.5:1 Si/As molar ratio, 1M sodium silicate) Table Series A.2- As (mg/l), ph and E h results for initial reverse titration short term stability test (1:1 Si/As molar ratio, 1M sodium silicate) Table Series A.3- As (mg/l), ph and E h results for the ageing as drying series- RT 2 weeks and 40 C 24 hours (0.1:1 Si/As molar ratio, 1M sodium silicate) Table Series A.4- As (mg/l), ph and E h (mv) data for ageing hydrothermally treated Scorogels and gels at 110 and 160 C, 8 hours (0.1:1 Si/As molar ratio, 1M sodium silicate) IX

11 1. CHAPTER 1 INTRODUCTION Arsenic is commonly found in ores processed in gold, silver, copper and uranium milling and metallurgical operations that require proper immobilization and control to avoid water contamination problems. Federal regulations (MMERs) stipulate <1 mg/l As in mineral processing effluents (Anonymous, 2012) or in leachability tests of hazardous wastes as is the US Environmental Protection Agency s (EPA) TCLP (Toxicity Characteristic Leaching Procedure) protocol (Anonymous, 1992). Therefore it is critical to produce an arsenic-bearing material that will remain stable when deposited into a waste disposal site. The common approach to arsenic removal from effluent solutions is through the combination of co-precipitation with ferric ions and lime neutralization. With this technique there is a high iron requirement (minimum Fe/As molar ratio>3) (De Klerk et al., 2012) and therefore it is not appropriate for high arsenic content waste. For high arsenic content waste like roaster flue dust, acid plant effluents or arsenical residues, crystalline scorodite, FeAsO 4 2H 2 O, produced under atmospheric pressure is desirable due to the 1:1 As/Fe molar ratio and good filtration properties (Filippou and Demopoulos, 1997). Scorodite, when subjected to TCLP testing (at ph 5 for 24 hours), releases <1 mg/l As hence is acceptable for disposal of high arsenic content waste material. In fact, the production of atmospheric scorodite is underway at the industrial level in Chile (Ecometales Ltd.) and Japan (Dowa Metals and Mining Co. Ltd.) (Kubo et al., 2010; Lagno et al., 2009). However, the stability of scorodite is limited to the ph range between ph 3 and 7 and oxic conditions (Bluteau and Demopoulos, 2007). Scorodite shows decreased stability under anoxic conditions (Verdugo et al., 2012). Therefore it is of interest to seek means for enhancing the stability of scorodite by encapsulation with an inert material. Previous McGill HydroMET work has included scorodite encapsulation research into aluminum phosphate (Lagno et al., 2010), hydroxyapatite and fluoroapatite (Katsarou, 2011) and aluminum hydroxyl gels (Leetmaa, 2008). The present work looks into an alternative scorodite encapsulation system featuring silicate gels produced by the reaction of sodium silicate with sulfuric acid as possible coating/encapsulant materials. 1

12 The main objectives of the work described in this thesis are to: 1. Study the formation of silicate gels by different reaction sequences. 2. Study different gel ageing methods in an effort to promote condensation and transformation into a silica-like matrix. 3. Evaluate silicate gel s effectiveness as protective scorodite coating via stability testing at ph 9, under oxic (600 mv E h ) and anoxic (0 mv E h ) conditions. 4. Characterize silicate gel-coated scorodite (Scorogel) particles by surface and molecular sensitive techniques in order to elucidate the stabilizing or absence thereof role of silicate gels. This thesis consists of the following chapters: 1. Chapter 2- Literature Review- Which includes discussion of arsenic fixation techniques, past research on scorodite production and stability, previous scorodite encapsulation materials and methods, the chemistry of sodium silicate, previous sodium silicate encapsulation studies, silicate gel production and characterization methods. 2. Chapter 3- Experimental Methods- Chemicals and reagents used, experimental procedures for the production of scorodite, silicate gel, washing, blending, ageing and stability testing of Scorogel are given. The different characterization techniques applied are also described. 3. Chapter 4- Results and Discussion- Comprising atmospheric scorodite stability, characterization of silicate gel, results from washing and ageing silicate gels and stability testing and characterization of Scorogel materials. 4. Chapter 5- Conclusion- Final concluding remarks made. 2

13 2. CHAPTER 2 LITERATURE REVIEW This thesis deals with scorodite production by atmospheric precipitation and its encapsulation using sodium silicate as a precursor material. This encapsulation method is an attempt at enhancing scorodite stability in terms of resistance to dissolution at ph >7 (oxic conditions) or under reducing potential (anoxic conditions) hence providing a robust arsenic fixation process for safe long-term disposal. As a result the present literature review, following a brief introduction to the arsenic problem in the mining industry, examines why scorodite production is a suitable solution to arsenic fixation of high arsenic content waste by giving an overview of previous work done in particular in relation to atmospheric precipitation and stability. Further, the review is extended into the background chemical and environmental factors and previous relevant work that potentially make sodium silicate a suitable material for encapsulating scorodite. 2.1 The Arsenic Problem It is well known that arsenic is toxic and poisonous and long term exposure to it at very low water concentrations (50 µg/l) can lead to liver, lung, kidney and bladder cancer (Smith et al., 1992). The 2006 Health Canada drinking water guidelines state the allowed maximum concentration of arsenic to be 0.01 mg/l or 10 ppb (Anonymous, 2006). The Mining Metal Effluent Regulations state that the maximum authorized arsenic concentration in a grab effluent sample must be 1 mg/l, or 1 ppm (Anonymous, 2012). Arsenic is found in naturally occurring minerals with an average distribution of 2 mg per kg within the earth s crust. Arsenic is found in sulfide minerals such as arsenopyrite, FeAsS, enargite, Cu 3 AsS 4, orpiment, As 2 S 3, and realgar, As 4 S 4. These sulfides are commonly found in ores containing gold, silver, copper and uranium. Arsenic is mobilized due to the natural weathering of these arsenic-bearing minerals (Tanaka, 1988). In Bangladesh thousands of people are suffering from arsenic poisoning-related health problems including cancer and an additional 29 million people are at risk (Nordstrom, 2000). The source of the arsenic is natural 3

14 stemming from the weathering of arsenopyrite and its mobilization resulting in groundwater contamination (Nordstrom, 2000). Within the mining and metallurgical industry arsenic waste is generated from processes such as roasting, smelting and leaching of ores. Arsenic trioxide dust is the main form of arsenic waste of roasting and/or smelting processes. Arsenic-rich wastes also come in the form of acid plant effluents, copper-electro-refining bleed streams and even arsenic contaminated soils (Demopoulos, 2005). Storing these arsenic-rich wastes is dangerous because of their high solubility. This was observed in the case of Giant Mine in Yellowknife, Northwest Territories (Riveros et al., 2001). Giant Mine was a gold mine operating from 1948 to When the mine closed down, it went into receivership, and mine ownership was transferred to the federal Department of Indian and Northern Affairs. Over the years the roasting operation of the mine had generated 237,000 tonnes of arsenic trioxide dust, As 2 O 3 that had been stored without proper containment in underground chambers and mined out stopes. Currently permafrost is surrounding these chambers meaning there is no water entering the chambers and dissolving the As 2 O 3. However the permafrost is at risk of melting due to increasing annual temperatures. A remediation plan was developed in order to maintain the frozen conditions underground, the method is known as the Frozen Block method. This method consists of pipes being installed around each chamber and stope. The pipes are connected to an above ground plant that pumps super cooled carbon dioxide liquid that will freeze the areas in and around the chambers and stopes. The cost of setup for this remediation plan was estimated at $200 million dollars and 10 years to implement (Anonymous, 2007). However recent reports have raised the costs to almost $1 billion that prompted the authorities to re-evaluate their options (Anonymous, 2013). This example provides clear evidence of the significance of developing cost-effective and safe arsenic fixation processes to deal with this type of arsenical waste. One such arsenic fixation process that has been proposed/demonstrated on laboratory scale by Demopoulos and co-workers (Debekaussen et al., 2001) involves conversion of arsenic trioxide to crystalline scorodite. 4

15 2.2 Arsenic Fixation Methods As earlier mentioned, many metallurgical processes result in large amounts of arsenical waste. In managing this waste, the main goal in industry is to form a stable solid residue that is safe for long-term disposal. The disposal environment itself needs high consideration, as this environment can exist under different ph and redox conditions (Harris, 2000). Within this disposal environment, acid mine drainage, biological activity, and surrounding mineralogy (eg. limestone) can all have an effect (Riveros et al., 2001). Commonly cement based stabilization/solidification technologies have been used to treat arsenic wastes. Cement was used to treat iron hydroxide, high content arsenic sludge. The cement consisted of 1 part Portland cement and 3 parts concrete sand (Jing et al., 2003). Portland cement is a mixture of gypsum, calcium silicate, calcium aluminate, and calcium aluminoferrite (Hydrometallurgy Group, 2008). After 18 hours of TCLP (Toxicity Characteristic Leaching Procedure) testing, at ph 6.8, the arsenic released into water was reduced from 7050 µg/l at ph 3.6 to 16 µg/l. At this ph, the As(V) was converted into a calcium arsenate precipitate. The issue with the use of the cement, was when the material was exposed to air, carbonation of the cement occurred and calcium was released thereby reducing the effectiveness of the material (Jing et al., 2003). Also, the high ph of the Portland cement (ph 12.5) may not be suitable for all arsenic waste (Hydrometallurgy Group, 2008). Another common method for arsenic fixation is co-precipitation with ferric ions upon neutralization in the presence of excess iron (III), typically [Fe(III)]/[As(V)] molar ratio > 3. Co-precipitation in this case, refers to the process where the neutralization of an acidic Fe(III)- As(V) solution results in the precipitation of arsenate along an iron(iii) compound, such as ferrihydrite. Ferrihydrite, FeOOH 0.5H 2 O, has absorption properties for cations and anions which include AsO 3-3 and AsO 3-4 (Demopoulos, 2009). Jia and Demopoulos determined that co-precipitation (Jia and Demopoulos, 2008) is a better method of arsenic fixation compared to simple adsorption (Jia and Demopoulos, 2005), a process that involves the adsorption of arsenate onto pre-synthesized ferrihydrite. Both systems had an Fe(III)/As(V) molar ratio= 4. During solubility studies conducted at ph 8, adsorption onto ferrihydrite released 1 order of magnitude greater arsenic concentration into solution compared to the co-precipitation method. Detailed characterization studies involving initially 5

16 XRD (Jia and Demopoulos, 2005; Jia et al., 2003) and later synchrotron methods (Chen et al., 2009) revealed that in addition to ferrihydrite the majority of arsenate reported in the solids as a poorly crystalline ferric arsenate especially at ph<8. The effectiveness of the co-precipitation process was further determined to improve in sulfate (typical industrial case) vs. nitrate (typical laboratory case) solutions (Jia and Demopoulos, 2008; Twidwell et al., 2005) and by using lime instead of NaOH as base (Jia and Demopoulos, 2008). Another important finding was the realization that continuous co-precipitation and in particular staged co-precipitation is more effective at arsenate removal from solution than batch co-precipitation (De Klerk et al., 2012). According to the latter study continuous co-precipitation in two stages (first stage at ph ~4 and second stage at ph ~8) allows for the bulk of arsenic to precipitate as poorly crystalline ferric arsenate with the remaining soluble fraction fixed via surface adsorption on ferrihydrite. The disadvantages of the co-precipitation process include the formation of a sludge-like precipitate that is difficult to filter (Twidwell et al., 2005). Also, the high iron content (because of the Fe/As ratio>3) and neutralizing agent requirement makes this co-precipitation method unsuitable for high arsenic content waste such as the arsenic trioxide dust found at Giant Mine, which consists of 60% arsenic (Anonymous, 2007; Riveros et al., 2001). Another method used in the past to remove arsenic from solution is lime precipitation. Neutralization of As(V) with lime leads to the formation of different calcium arsenate compounds (Bothe and Brown, 1999) such as Ca(OH) 2 (AsO 4 ) 2 which is comparable to the naturally occurring mineral guerinite, Ca 5 H 5 (AsO 4 ) 4 9H 2 O. However, these compounds are susceptible over the long run to CO 2 attack and conversion to calcium carbonate with concomitant release of arsenic to the environment hence this method is not acceptable for arsenic fixation (Harris, 2003; Riveros et al., 2001). For arsenic waste with a high arsenic content, the production of scorodite, a naturally occurring mineral, FeAsO 4 2H 2 O, is a suitable solution because of its lower iron requirement, high arsenic content (25-30%) and low solubility (<1mg/L) at ph 5 (Bluteau and Demopoulos, 2007; Demopoulos et al., 2003; Riveros et al., 2001). 6

17 2.3 Scorodite Production of Scorodite Crystalline scorodite may be synthesized either in an autoclave at temperatures above 100 C (~160 C) Hydrothermal scorodite - Or under atmospheric conditions (below 100 C) - Atmospheric scorodite. The first synthesis of scorodite was reported in 1988 by Dutrizac and Jambor, who employed nitrate media at 160 C. Later Swash and Monhemius studied extensively the hydrothermal precipitation chemistry of iron (III) arsenates in sulfate media over a wider range of conditions, such as different Fe/As molar ratios and temperature ( C). These authors determined that scorodite formed at T<200 C and Fe/As< 2. In addition to scorodite they found two unknown phases that formed that they labeled Type 1 and Type 2. Subsequent work by Dutrizac and Jambor (2007) and Gomez et al. (2011) further clarified the identity of these phases that were properly named ferric arsenate sub-hydrate (FAsH- FeAsO 4 3/4H 2 O) for Type 1 and basic ferric arsenate sulfate (BFAS- Fe(AsO 4 ) 1- x(so 4 ) x (OH) x (1-x)H 2 O) for Type 2. Gomez et al. (2011) determined that hydrothermal scorodite could be produced at 150 and 175 C after 2 hours, although at shorter times or lower temperatures an intermediate BFAS-like phase was found to form prior to scorodite (Gomez et al., 2011a). While crystalline scorodite can be produced in an autoclave, the method would be difficult to establish in large scale industrial production due to the high capital costs associated with the use of an autoclave (Dutrizac and Jambor, 1988; Filippou and Demopoulos, 1997). 7

18 2.3.2 Atmospheric Production of Scorodite Pioneering work by Demopoulos and co-workers demonstrated that it is possible to produce crystalline scorodite without the use of an autoclave under atmospheric pressure if the supersaturation during scorodite precipitation is properly controlled (Demopoulos et al., 1995; Droppert et al., 1996). Since that breakthrough several studies have been reported on the subject of atmospheric precipitation of scorodite that recently culminated to the industrial implementation of this process and the development of other similar processes. The supersaturation-controlled process developed at McGill (Demopoulos, 2005) can be described with the aid of Figure 2.1. By operating within the metastable zone heterogeneous nucleation on the surface of scorodite seed is promoted producing well grown crystalline scorodite particles that do not adhere to the reactor walls and have excellent filterability. This scorodite production process is performed at C, the higher the temperature the faster the kinetics. The metastable zone is defined by the lower C eq line representing the solubility of scorodite as a function of ph and the upper C cr line representing the critical supersaturation line. The latter line demarks the two nucleation regimes: homogeneous nucleation (above C cr ) associated with production of amorphous material and heterogeneous nucleation (below C cr ) characterized by surface deposition on seed (Demopoulos, 2009). The C cr and C eq values depend on the solution media, the concentration of arsenic, temperature and ph. According to Figure 2.1 scorodite precipitation is accomplished via step-wise ph adjustment. For the example of Figure 2.1, the diagram predicts arsenic concentration being reduced from 10 g/l down to 0.02 g/l after two steps (of ph adjustment). Industrial implementation of the step-wise procedure involves a series of continuous stir tank reactors (CSTR) corresponding to the respective steps as described elsewhere (Demopoulos et al., 2003). 8

19 Figure 2.1- Graphical representation of the McGill supersaturation-controlled based scorodite precipitation process: Scorodite is produced within the metastable zone, where heterogeneous nucleation occurs on seed particles (Demopoulos et al., 2003) Many factors can affect the quality of atmospheric scorodite. Singhania et al studied the effect of temperature on scorodite precipitation. When the temperature under which scorodite precipitated was increased from 85 C to 100 C, not only was there a 3 to 4 times increase in rate of precipitation, but also the solubility of the scorodite product decreased. The same authors studied different kinds of seeding material. The seeds tested included hydrothermally (HT) produced scorodite (following the method of Dutrizac and Jambor 1988), atmospherically produced scorodite, hematite and gypsum. The scorodite produced on hydrothermal (HT) seed was more effective than the scorodite produced with atmospheric seed when the same mass of each was used because of the larger specific surface area of the former. Foreign types of seed like hematite and gypsum also proved effective. This time, however, a slow initial nucleation period was observed (induction period) that resulted in S-shape precipitation curves exhibited by autocatalytic reactions. In other words after the initial slow formation of freshly nucleated scorodite on the foreign seed s surface the scorodite nuclei acted as a catalyst causing acceleration of precipitation rate. Among the two foreign seed materials, hematite was more effective because of its finer particle size (larger specific surface area). It was 9

20 further noticed that scorodite could nucleate not only on externally added gypsum crystals but also on gypsum crystals that formed in-situ. Another interesting finding was that the leachability of scorodite was slightly better (1 to 2 mg/l As) in the presence of gypsum seed compared to scorodite alone (3 mg/l As) (Singhania et al., 2005), an observation that was further verified and studied in subsequent stability studies (Bluteau et al., 2009). In another study, Singhania et al. (2006) reported on the effect of ph control. These authors found that once the proper initial ph of the solution is chosen to be within the metastable zone (refer to Figure 2.1) there is no need to control ph. For example in an experiment where scorodite was produced without ph control it was revealed that scorodite could still be produced efficiently, with 95% As precipitating in the first 2 hours and only 0.2 g/l remaining in solution after 4 hours. In other words, it can be seen that this system has a tolerance for ph changes (Singhania et al., 2006). Fugita et al found that a ph range of 0.3 to 1.0 is favourable for synthesizing scorodite in agreement with the metastable zone determined at McGill (Figure 2.1). The Fe/As molar ratio also has an effect on scorodite precipitation and solubility. The yield of scorodite produced decreased as the Fe/As ratio increased: Fe/As=1.0, 85% yield, Fe/As=1.25, 83% yield, Fe/As=1.5, 80% yield, Fe/As=2.0, 50% yield, Fe/As=3.0, 20% yield. However, with higher Fe/As molar ratios, more stable (in terms of TCLP leachability) precipitates were reported to form (Fe/As=3, 0.5 mg/l solubility, Fe/As= mg/l) (Singhania et al., 2006). Interestingly, the presence of Fe(II) in initial iron concentration decreases the solubility of scorodite (0.12 mg/l with Fe(II), 2.4 mg/l without Fe(II)). This is because particle growth is favoured by low supersaturation that is regulated by the oxidation rate of Fe(II) to Fe(III) (Fujita et al., 2008a; Singhania et al., 2006). As can be seen, many factors affect the quality of the final scorodite material produced, and there are always new techniques focussed on improving the solubility of scorodite. 10

21 2.3.3 Other Approaches to Producing Scorodite A variation of the atmospheric precipitation process is the method developed by Fujita and co-workers that involves oxidation of Fe(II) by oxygen sparging at 95 C and ph<1 (Fujita et al., 2008a; Fujita et al., 2008b; Fujita et al., 2009). The concept behind this method was first investigated by Singhania et al. (2006) where H 2 O 2 was used to oxidize ferrous in the presence of As(V) at ph<1 and 95 C. Recently the oxidation concept of regulating supersaturation in the acidic region (corresponding to the metastable zone (see Figure 2.1) was also successfully demonstrated this time relying on micro-organisms to oxidize Fe(II) to Fe(III) (Gonzalez- Contreras et al., 2009). The method however in contrast to the use of oxidants like oxygen or hydrogen peroxide that can treat high As(V) concentration solutions is slow hence applied to arsenic solutions at ~1 g/l. Nanocrystalline scorodite has also been produced under ambient conditions (70 C, Fe/As= 1, ph<2, 3 days equilibration). The final product was 45 nm in length in the form of stubby prismatic crystals. At ph 2 the solubility of the product was low, 0.23 mg/l. However as ph increased so did the release of arsenic (ph 7, 5.4 mg/l). The solubilities of nanocrystalline scorodite were similar to that of hydrothermally produced scorodite (Paktunc and Bruggeman, 2010). Nanocrystalline scorodite may not be a viable solution to scorodite production in industry due to anticipated filtration issues due to extremely small particle size and very long retention times. Employed as a seed product it has potential, with its very high surface area, it could be an option. Le Berre et al. (2008) demonstrated that crystalline scorodite can also be produced through the transformation of poorly crystalline ferric arsenate. Thus starting with an Fe/As=1 molar ratio slurry neutralized to ph 2 at room temperature poorly crystalline ferric arsenate was produced More on the characterization of this type of ferric arsenate is given in Le Berre et al. (2007) And heating the resultant slurry to 80 C transformation to scorodite was observed to take place. The higher the temperature and lower the ph the faster the transformation kinetics were. It was determined that the crystallization of scorodite occurred within the amorphous 11

22 precursor phase with localized dissolution allowing for mass transport within the aggregate (in agreement with the Avrami-Erofeev nucleation-growth model) Stability of Scorodite In earlier studies (e.g. Singhania et al. (2005) the stability of synthetic scorodite was evaluated on the basis of a TCLP type test involving ~ 24 hour contact of scorodite particles with water (L/S=20) of ph 5 adjusted with common acids (HNO 3 or H 2 SO 4 ) or bases (NaOH) instead of the EPA prescribed acetic acid buffer solution (Anonymous, 1992). According to the TCLP protocol an arsenical material is stable if its leachability is <5 mg/l As. Bluteau and Demopoulos 2007 undertook to study the leachability of scorodite over wider ph range and longer equilibration times for better evaluation of its true intrinsic stability. In addition, elevated temperatures (40 and 70 C) were tested. In this study the authors used hydrothermally produced scorodite (as per Dutrizac and Jambor 1988 method) and not atmospheric scorodite. Thus, as the ph and temperature increased so did the release of arsenic from scorodite. The release of arsenic via an incongruent dissolution mechanism was found to reach equilibrium (at room temperature) after 24 weeks at ph 7 or lower but not ph 8 and 9. The equilibrium As concentration at ph 7 was 5.8 mg/l, 1 mg/l at ph 6 and 0.3 mg/l at ph 5. In contrast at ph 8 > 100 mg/l As was released (no equilibrium reached) after 64 weeks at 22 C. From these experiments, as well as previous studies, scorodite can be seen to be stable (< 1 mg/l solubility at 22 C) only over the ph range of approximately 3 to 7 (Figure 2.2) hence the importance of exploring additional measures such as encapsulation to expand its stability region. 12

23 Bluteau and Demopoulos 2007 Figure 2.2- Solubility data of scorodite under different ph conditions at 22 C. A comparison of previous studies with the results from M.C. Bluteau s research (Bluteau and Demopoulos, 2007) In hydrometallurgical processes, lime is used as a neutralization agent hence gypsum would be produced along with scorodite. Bluteau et al. (2009) studied the stability of scorodite in gypsum-saturated waters and found calcium ions to lower the release of arsenic. For example at ph 7 and 22 C the equilibrium As concentration was 3.6 mg/l instead of 5.9 mg/l in the absence of gypsum. At ph 9 and 70 C the authors observed partial transformation of scorodite to a Ca-Fe(III) arsenate phase resembling the mineral yukonite (Ca 2 Fe 5 (AsO 4 ) 3 (OH) H 2 O) (Gomez et al., 2010b). If such transformation also happens at 22 C and ph 7 to explain the lower observed solubility is not known and further work is required to resolve this issue. In addition to scorodite stability decreasing with ph increasing above the neutral value, scorodite stability also decreases when exposed to reducing conditions. This is of concern as anaerobic bacteria in tailings impoundments, or the presence of unreacted sulfide minerals can reduce Fe(III) to Fe(II) and in the process decompose scorodite as was observed by McCreadie et al. (1998) in the case of autoclave generated ferric arsenate solids placed in a tailings impoundment area. Scorodite undergoes reductive decomposition to Fe(II) and As(III) when the solution E h decreases below 100 mv and this E h level is typical in water covered areas with greater than 2 meters depth (Rochette et al., 1998; Welham et al., 2000). The reduction rate of 13

24 arsenate is ph dependant. The rate of decomposition is more rapid under acidic conditions. Hydrogen sulfide (H 2 S) can reduce arsenate approximately 300 times faster at ph 4 than at ph 7 (Rochette et al., 2000). In laboratory studies abiotic chemical reducing conditions have been generated by using sodium sulfite (Na 2 SO 3 ) and sodium sulfide (Na 2 S) but only the latter was capable of lowering the E h level to 100 mv (Katsarou, 2011; Lagno, 2005; McCreadie et al., 2000). Some typical reductive dissolution data reproduced from Verdugo et al. (2012) are shown in Figure 2.3. As it can be seen the release of arsenic becomes dramatic when E h is < 200 mv (Verdugo et al., 2012). Hence once more additional protection of scorodite via encapsulation explored in the present work is required. Figure 2.3- The release of arsenic from atmospheric scorodite as a function of E h (ph 7, gypsum-saturated water; NaOH base, 22 C) (Verdugo et al., 2012). 14

25 2.3.5 Industrial Application of Scorodite Different arsenical waste materials or solutions may be processed to fix arsenic in the form of scorodite, such as acid plant effluents, residues, sludges, flue dusts, or other by-products (Wang et al., 2000a). This can be done in an autoclave (Swash and Monhemius, 1998) as was commercially demonstrated in the early 90s by Geldart et al. (1992). In this case arsenic trioxide sludge blended with roaster calcine waste and arsenical gold flotation concentrate was processed at 210 C producing iron arsenate precipitates that were claimed to be scorodite. The recovery of gold compensated for the cost of the process. The precipitates were shown to be stable after 13 weeks equilibration at ph 7.5 (~0.13 mg/l As). At McGill atmospheric scorodite has been produced in laboratory studies (Demopoulos et al. 2003) from arsenic trioxide flue dust (Debekaussen et al., 2001; Wang et al., 2000b) and acid plant effluent solution (Filippou and Demopoulos, 1997). In both cases the arsenic was in the form of As(III) and therefore had to be oxidized prior to conversion to scorodite. This was done initially using hydrogen peroxide (H 2 O 2 ). Hydrogen peroxide was shown to have high oxidation efficiency through slow addition rate and temperature elevation (95 C). This method worked, however H 2 O 2 is expensive and prompted the development of an alternative oxidation process making use of an SO 2 /O 2 gas mixture (Debekaussen et al., 2001; Wang et al., 2000a). For oxidation of As(III) to be successful using SO 2 /O 2, elevated temperature (90 C), high agitation rate (1000 rpm), and presence of Fe(III) that acts as a catalyst needs to be applied (Demopoulos, 2005). Recently the McGill-developed atmospheric scorodite process was implemented commercially in a mine in Chile following its successful piloting (Lagno et al., 2009). The Ecometales scorodite plant treats 5,000 tonnes of arsenic per year in the form of flue dust. The process involves dissolution in acid, oxidation by H 2 O 2, addition of iron in the form of magnetite and precipitation of scorodite in two steps as per Figure 2.1. Another industrial development is the Dowa demonstration plant in Japan where scorodite is produced at 95 C from Fe(II)-As(V) acidic solution (<ph 1) by oxygen sparging (Kubo et al., 2010). In view of the increased commercial interest in scorodite developing an encapsulation process that will strengthen its stability becomes an important issue. 15

26 2.4 Encapsulation of Scorodite The Concept and Previous Studies Encapsulation as a concept entails the formation of a sort of coating or matrix made of an inert material to protect scorodite. By inert material it is implied resistance to dissolution in terms of ph variation and reducing E h conditions Refer to scorodite stability section. This is a new concept of arsenic stabilization research that was initiated some 10 years ago at McGill s HydroMET laboratory (Lagno, 2005). Stability results from previous encapsulating projects of the HydroMET group are summarized in Tables 2.1 and 2.2. Initial methods for scorodite encapsulation involved microencapsulation whereby aluminum phosphate and hydroxyapatite were deposited onto the surface of scorodite via supersaturation-controlled heterogeneous deposition (Lagno 2005). Of these materials aluminum phosphate was the easier material to deposit as a coating on scorodite particles (Lagno et al., 2010). At the proper ph and temperature conditions, an acidic sulfate solution containing P(V) and Al(III) was maintained within the metastable zone of aluminum phosphate crystallization that allowed for heterogeneous deposition of this material onto the surface of scorodite (Figure 2.4). After three cycles of aluminum phosphate deposition, the release of arsenic from scorodite was reduced by one order of magnitude under both oxic (ph 8) and anoxic conditions (100 mv, ph 7) (Lagno et al., 2010). Figure Aluminum phosphate coating around a single scorodite particle (Lagno et al. 2010) Preliminary results on the hydroxyapatite (HAP) coating of scorodite were inconclusive (Lagno, 2005) and prompted revisiting of this encapsulation system along with the relating fluoroapatite (FAP) system by Katsarou (Katsarou, 2011). Again HAP and FAP coatings were deposited onto scorodite using the same supersaturation-controlled method. Under oxic 16

27 conditions (ph 9) scorodite coated with hydroxyapatite released 30% less arsenic than uncoated scorodite. Under anoxic conditions (150 mv, ph 8) the coated scorodite released one order of magnitude less arsenic than the uncoated scorodite. The fluoroapatite coating also reduced the release of arsenic but not as successfully as the hydroxyapatite coatings apparently because of difficulties with the directed deposition of FAP on scorodite. Microencapsulation of scorodite through heterogeneous deposition while successful at reducing the arsenic released from scorodite, would be costly and difficult to implement at the industrial level. Moreover the coating films were rather thin and discontinuous (2.5 µm for aluminum phosphate and < 1 µm for HAP/FAP) meaning they may not be robust enough as they appeared to undergo low release of phosphate hence judged unsuitable as a long-term solution. As alternative- Macroencapsulation- may offer a simpler coating process. Leetmaa (2008) explored this approach by producing an aluminum gel (Leetmaa et al., 2013) that could be mechanically blended with atmospheric scorodite to produce a stabilization matrix. Aluminum salt solutions were prepared by dissolving aluminum chloride (AlCl 3 ) or aluminum sulfate (Al 2 (SO 4 ) 3 ) salts in deionized water. Sodium hydroxide (NaOH) was added to the salts whereby partial neutralization resulted in the formation of an aluminum hydroxyl-gel at ph 4 via hydrolysis. The gel was in a viscous semi-solid form (Leetmaa, 2008). Before blending with scorodite, the aluminum gels were washed in order to remove the Na and SO 4 or Cl constituents from the gel. At the industrial level, if the gels are disposed of as waste with scorodite, these by-products will easily leach from the gel and act as contaminants. Using a liquid/solid w/w ratio of 10:1, the gels were washed for 24 hours while being stirred in a beaker using a magnetic stirrer. The gels were then filtered (Leetmaa, 2008). In order to allow for condensation of the gels into a more dense product, the gels were aged in both an open and closed environment at different temperatures (22, 40 and 70 C) (Leetmaa, 2008). Stability testing results, at ph 8, revealed that gels aged at 70 C in a closed environment, released the highest concentration of arsenic followed by the gels aged at 40 C and room temperature. In an open system (exposed to air), gels aged at all temperatures released less arsenic, with the sulfate gels releasing the least amount of arsenic (0.02 mg/l after 44 days). Leetmaa (2008) described the ageing process of the gels (temperature, open/closed system) as an important factor that promotes a hydroxide type of coating to form providing protection to scorodite that goes beyond 17

28 simple adsorption of mobilized arsenate. Leetmaa s work was continued to include different aspects of gel preparation and stability testing under both oxic and anoxic conditions (Demopoulos et al., 2013). For example there was less than 1 mg/l As release under oxic conditions and ~ 5 mg/l of As release under reducing conditions (E h ~200 mv) at ph 9 after 110 days. By comparison arsenic release from unprotected scorodite at ph 9 was ~ two orders of magnitude higher hence providing legitimacy as proof-of-concept to the encapsulation process. Table 2.1- Summary of arsenic release data from previous encapsulation studies under oxic conditions Encapsulating As release ph Time Researcher Material Aluminum phosphate < 1 mmol/l ph 8 10 days F. Lagno (Ph.D thesis 2005 ) Hydroxyapatite 11.2 mg/l ph 8 10 Days F. Lagno (Ph.D thesis 2005) Hydroxyapatite 1 mg/l ph 9 40 days L. Katsarou (M.Eng thesis 2011) Fluoroapatite 7.7 mg/l ph 9 40 days L. Katsarou (M.Eng thesis 2011) Aluminum chloride gel 7.9 mg/l (Al/As=0.1) 4.6 mg/l (Al/As=1) ph Days K. Leetmaa (M.Eng thesis 2008) Aluminum sulfate gel 0.3 mg/l ph Days K. Leetmaa (M.Eng thesis 2008) 0.02 mg/l (open ageing) Aluminum sulfate gel < 1 mg/l ph Days Demopoulos, Becze, and Daenzer (Unpublished Report 2013) Table 2.2- Summary of arsenic release data from previous encapsulation studies under anoxic conditions Encapsulating As release ph and E h Time Researcher Material Aluminum Phosphate ~3 mmol/l ph 7, 100 mv 42 Days F. Lagno (Ph.D thesis 2005 ) Hydroxyapatite 73.5 mg/l ph 8, 150 mv 40 Days L. Katsarou (M.Eng thesis 2011) Fluoroapatite 12.3 mg/l ph 9, 150 mv 40 Days L. Katsarou (M.Eng. thesis 2011) Aluminum sulfate gel ~5 mg/l ph 9, 200 mv 110 Days Demopoulos, Becze, and Daenzer (Unpublished Report 2013) 18

29 2.4.2 Silica-based Encapsulation Methods Silica, or silicon dioxide (SiO 2 ), known as α-quartz in its most common mineral form, is the most abundant mineral in the earth s crust. Quartz is both mechanically and chemically resistant to weathering (Figure 2.5). Figure 2.5- E h -ph diagram of Si-H 2 O at 25 C, Si=10-6 mol/kg, calculated with OLI 3.2 (MSE model) Amorphous silica can be found in nature in the form of volcanic glass or tektites (round glass figures that formed during an impact crater), again known to be resistant for millions of years. Biogenic silica is produced by life forms such as the skeletons of radiolara, diatoms and sponges. In some plants such as rice, the ash from burning the hulls of the seeds can contain 95% wt. silica (Heaney, 1994). Hence silica would be the ultimate encapsulation matrix for hazardous materials if a cost-effective process can be developed for industrial application. In recognition of this potential various silica-based materials have been used to encapsulate certain pollutants or heavy metals in order to prevent soil contamination, acid mine drainage (AMD) and other hazardous waste material problems. One such silica-based material is polysiloxane, a ceramic silica foam (50% vinyl- polydimethyl- siloxane, 20% quartz (SiO 2 ), 5% water) (Randall and Chattopadhyay, 2004). The material sets at room temp (30 C). The polysiloxane foam was evaluated with mercury containing waste and found to lower mercury release to 0.06 mg/l in a TCLP test. However the polysiloxane material was not effective with 19

30 other pollutants such as chromium that was released at levels greater than 500 mg/l (Randall and Chattopadhyay, 2004). The Klean Earth Environmental Company (KEECO) developed the so-called silica micro encapsulation technology (SME). This technology consists of a calcium/silica based powder that becomes activated when exposed to moisture (Mitchell, 2000). The EPA tested this material s ability to encapsulate mercury. Mercury ore (2 kg) was placed in a plastic column and manually mixed with ~200 g of the SME material. The columns were then exposed to an 8-week leaching period. Compared to the control column, the column containing SME retained 88% more mercury (Anonymous, 2004). In another study, Arocha et al. (1996) used a combination of shredded tires and sodium silicate as an encapsulating material to prevent soil contamination by volatile organic compounds (VOCs) like toluene. This technique was studied as possible replacement for slow- setting ordinary Portland cement. Shredded tires were considered a useful additive as they are composed of 31% carbon black, a strong adsorbent. Concentrated sodium silicate was neutralized with acetic acid prior to its use. The final material consisted of 14% silica. This study saw 80% of the total toluene retained and prevented from solubilizing in the soil, compared to soil with no treatment that saw only 3% toluene retention (Arocha et al., 1996). Evangelou (2001) used sodium metasilicate to coat pyrite and prevent its weathering, which results in AMD. Columns of crushed pyrite were treated initially with hydrogen peroxide (H 2 O 2 ), a strong oxidizing agent, and sodium metasilicate solution (0.01 M). What resulted was the formation of a surface layer of oxyhydroxide (FeOOH), favoured at ph 6. The silicate groups in solution then reacted with surface OH groups to form a silica protective polymer film that greatly reduced the dissolution of pyrite (Evangelou, 2001). Bessho et al. (2011) further simplified this process by eliminating H 2 O 2 treatment and simply mixing pyrite with sodium silicate solution (0.01 M) that was previously acidified with hydrochloric acid down (from its natural alkaline level) to ph 7. After filtration the residues were dried (80 C) and filtrate tested for Fe release in water. The solutions containing silica released one order of magnitude less Fe than the control- pyrite (Bessho et al., 2011). While this method was successful at preventing pyrite dissolution (stability test performed for 6 hours at different ph (3-9)), it should be noted that the surface of pyrite needed 20

31 to be oxidized to form an oxyhydroxide layer before forming a layer of silica polymer. Scorodite is an arsenate mineral meaning the surface of scorodite is already oxidized. While the encapsulation processes discussed by Evangelou (2001) and Bessho et al. (2011) could not be fully re-created for scorodite, the development of a silica coating using sodium silicate solution as a precursor material, is a merited option. In considering silica-based coatings of particles although not used in environmental applications it is worthy to refer to the relating sol-gel process. Such process employed extensively in advanced material applications to coat particles via the core-shell concept relies typically on the use of alcohols and organic silicate precursors like TEOS (tetraethyl orthosilicate). Stöber 1968 was the first to develop a silica coating method based on sol-gel chemistry. Silica is produced in this method through a condensation reaction involving a methanol-butanol alcohol mixture, alkyl silicates, water and ammonia (Stöber et al., 1968). There have been many successes based on the sol-gel process. Deng et al. (2005) used for example the Stöber method to develop a silica coating for magnetite nanoparticles. Silica formed on the surface of the particles through hydrolysis (Reaction 2.1) and condensation reactions as shown below (Reaction 2.2; where Si-OR is the tetraethyl silicate and ROH the ethyl alcohol product): Si-OR + H 2 O Si-OH + ROH Hydrolysis Reaction 2.1 Si-OR + OH-Si Si-O-Si + ROH Condensation Reaction 2.2 (Brinker, 1988; Deng et al., 2005) The use of organic silicate precursors and alcohols limits the usefulness of the Stöber method when it comes to large-scale industrial application. However, it is interesting to note that Lü et al. (2005) was able to substitute sodium silicate, glycol and sulfuric acid for the expensive TEOS precursor in forming a silica shell on SrAl 2 O 4 :Eu 2+, Dy 3+ particles. These particles are used in plastics and ceramic industries for their luminescence properties. As these particles tend to lose their luminescence properties when they come in contact with water it was necessary to encapsulate them with silica. The powders (10 g) were dispersed in glycol (50 g) and the 21

32 suspension was placed in a water tank. Sodium silicate solution (0.25M) was added to the suspension while mildly stirring the suspension. The ph was adjusted using sulfuric acid (0.1 mol/l). The mixture was heated to 80 C. Encapsulation was completed after 3 hours. Scanning electron microscope (SEM) images and energy dispersive x-ray spectroscopy (EDX) point analysis revealed a silica coating around the core particles (Lü, 2005). While glycol allowed for a successful coating, it is an organic solvent that would prove too costly to use at an industrial scale for scorodite encapsulation. Nevertheless the previous studies by Evangelou (2001), Bessho et al. (2011) and Lü (2005) provide evidence that dilute sodium silicate solutions may lead to the formation of a protective coating hence it deserves consideration as possible basis for the encapsulation of scorodite particles Sodium Silicate-derived Gels Sodium silicate has been used in industry for a number of purposes including adhesives and binders. The composition of bulk sodium silicate is made of 29% amorphous silica, 9% sodium oxide and 62% water (Fisher Scientific). As received sodium silicate has a high ph that is incompatible with the stability of scorodite (ph 11.8; refer to Figure 2.2). When sodium silicate solution is progressively acidified by mixing with an acid, such as hydrochloric acid, the silicate groups lose their charge and begin to link into a three dimensional network. At the point where these microgel entities occupy more space than the sol part, the viscosity greatly increases (gel point reached,) and a semi-solid polymeric silica material forms made of O-Si-O groups and Si-OH groups the latter as network of silicic acid, Si(OH) 4 (Iler, 1979). The gel can undergo further transformation through condensation (release of water) to form stronger Si-O-Si groups and networks (Iler, 1979). The transformation reaction involving condensation of the silicate gel through ageing is shown below in Reaction 2.3 (Brinker, 1988; Mantha et al., 2012): Reaction 2.3 (Brinker, 1988) 22

33 By acidification of sodium silicate the ph does drop to a level that coincides with the ph stability zone of scorodite and at the same time leads to gel formation that can advantageously be used to mechanically coat scorodite. The reaction between sodium silicate and acid to form gel can be conducted either by adding acid to sodium silicate solution or the reverse. Kalapathy et al. (2002) tried the latter and found gel to form at a lower ph (ph 7) than the former. Moreover the same authors observed the gels produced by the reverse mixing procedure to render easier sodium ion removal via washing. Removal of sodium is important because at an industrial size process, sodium that remains in the material that has been deposited as waste can be released into the water system and act as a contaminant. The gels were first dried and afterwards crushed and washed with deionized water and centrifuged to remove the excess sodium in the supernatant liquid. The washing step was repeated two more times. The gels were then dried at 80 C for 24 hours to produce pure silica xerogel (Kalapathy et al., 2002). Note that in the present study the aim is not to produce xerogel but a viscous gel for easy coating of scorodite by mechanically mixing as in cement making. Kalapathy et al. (2002) produced gels at both ph 4 and 7 and compared the structures of the gels. It was observed that when the ph of the gel decreases from ph 7 to 4, the major form of silica present in the gel shifts from SiO 2 to Si(OH) 4. A decrease in ph suggests a decrease in siloxane bonds. This also means that at ph 7, the Si-O bonds are covalently bonded, whereas at ph 4, the Si-OH bonds consist of weaker Van der Waals forces and H-bonds. Therefore, a softer less firm gel forms at ph 4 (Kalapathy et al., 2002) Characterization of Sodium Silicate-derived Gels Characterization of amorphous silica materials using vibrational (Fourier Transform Infrared (FTIR) and Raman) and NMR (Nuclear Magnetic Resonance) spectroscopies allow for the identification of key groups and transformation reactions including interaction of the silicatederived gel with scorodite. As these techniques are used in the Results & Discussion chapter it was considered appropriate to prepare Table 2.3 with all characteristic spectroscopic data useful to characterization of silicate species. 23

34 Table 2.3- Summary of spectroscopic characterization data relevant to the analysis of sodium silicate and amorphous silica Source Chemtob al et Vella et al Nordström 2011 Zhuravlev 2000 Nordström 2011 Silica material Hydrous amorphous silica Silica (purified sand) Sodium silicate solution/ gel Amorphous silica Sodium silicate solution/ gel Group Peak position Description/ Observations Si-O-Si 798 cm -1 and 1059 cm -1 (FTIR) 440 cm -1 (Raman) O 3 SiOH 955 cm -1 (FTIR) 480 cm -1 (Raman) Symmetric and asymmetric stretching. -Silica tetrahedral with one hydroxyl group -Asymmetric stretching O phases 1210 cm -1 -Movement of oxygen atoms out of phase -Asymmetric stretching Si-OH Shift from 955 cm -1 to 980 cm -1 (FTIR) 971 cm -1 (Raman) Si-O-Si 798 cm -1 and 1059 cm -1 (FTIR) 796 cm -1 and 1068 cm -1 (Raman) -Corresponds to smaller contribution of Si-OH that are hydrogen bonded to other Si-OH or molecular H 2 O. -Asymmetric stretching Symmetric and asymmetric stretch Si-OH ~3500 to ~3680 cm -1 (FTIR) Band arises from the overlap of a number of silanol bond configurations Si-O 860, 1000 cm -1 The shift to a higher frequency (FTIR) suggests a shift to lower ph. Si-O Increased amount of Si-O groups cm -1 (FTIR) related to the lowering of ph getting closer to gelation. Free OH 3750 cm -1 (FTIR) Apart of water, that may not be part of the silica structure Attached OH 3550 cm -1 (FTIR) Attached to Si Q 0 -Si(OH) 4 Q 1 -Si(OH) 3 Q 2 -Si(OH) 2 Q 3 -Si(OH) Q 4 - SiO 2 At ph 11 Q 0-0.3% Q 1-4% Q 2-25% Q 3-52% Q 4-19% (NMR, Si 29 ) As the ph decreases and moves closer to the gelation point, there is an increase in Q 3 and Q 4 and a decrease in Q 1 and Q 2. Nordström (2011) characterized sodium silicate solution and the effect of acid addition using NMR and FTIR. As sodium silicate is added to acid, an increasing rate of aggregation occurs. At ~ph 5-6, a minimum in colloidal stability is observed. Above this ph, increasing surface charge causes electrostatic repulsion of colloidal particles leading to increased colloidal stability (gel formation). FTIR analysis revealed an increasing amount of Si-O-Si groups as the ph moved towards this gelling point. However it should be noted that the peak positions for 24

35 different groups within sodium silicate solution depends on the sodium silicate solution concentration and SiO 2 : Na 2 O ratio (Nordström et al., 2011). Another important factor with silicate gel formation and transformation is the presence of free water versus water that is part of the structure. Two peaks close to the OH peak in FTIR can differentiate between the two types of OH. Free OH is located at 3750 cm -1, whereas vicinal OH (which is part of the structure and located near the surface) is located at 3550 cm -1. The degree of surface hydroxylation of the silanol groups explains the hydrophilic properties of amorphous silica. The electron density is delocalized from the O-H bond to neighboring Si-O bonds. These silanol groups can then form strong hydrogen bonds with water molecules. Therefore, the amount of OH groups within the structure is an important factor in determining the hydrophilic properties of silica (Zhuravlev, 2000). These hydrophilic properties could in turn affect the coating properties of silicate gel. Using thermogravimetric analysis (TGA), Satu el at. (2001) reports the removal of all weakly bonded water after an increase of temperature over the range from C, depending on the type of silica. Therefore thermal treatment of silicate gel may be necessary in order to change its hydrophilic properties. It is very important to use such characterization techniques in order to better understand the properties and composition of the silicate gel and its encapsulation coating/interaction with scorodite. 2.5 Summary Arsenic is a major environmental problem for the mining industry. In the case of arsenicrich waste material (flue dust, residue), further processing is required for safe long-term disposal. Crystalline scorodite produced by atmospheric pressure precipitation involving supersaturation control can stabilize the arsenic waste, however scorodite is only stable in a certain ph range (ph 3 to 7) and under oxic conditions. Encapsulation via the use of an inert material may further enhance the stability of scorodite under a wider range of conditions. In this literature review, the chemistry and previous work related to converting sodium silicate to silicate gel was reviewed, but no specific application to arsenic fixation was found. 25

36 3. CHAPTER 3 EXPERIMENTAL METHODS In this chapter, the experimental procedure is discussed including chemicals used, the method used for scorodite and silicate gel production and the characterization techniques used to analyze final products. 3.1 Chemicals Table 3.1 displays the chemicals used throughout this study. Table 3.1- Chemicals used during this study Group Chemical name Formula Assay/Grade Supplier Salt Ferric sulfate Fe 2 (SO 4 ) 3 xh 2 O 21.6% Fe(III)/ 97% Sigma Aldrich hydrate Oxide Arsenic (V) As 2 O 5 xh 2 O 60.7% As(V)/ 98% Sigma Aldrich pentoxide Silicate Sodium silicate Na 2 SiO 3 xh 2 O 29.2% silica amorphous/ Fisher Scientific solution Technical Bé Hydroxide Calcium hydroxide Ca(OH) 2 Technical Fisher Scientific Acids Sulfuric acid H 2 SO % / ACS-Pur Fisher Scientific Nitric acid HNO 3 Trace metal grade Fisher Scientific Sulfide Sodium sulfide nonahydrate Na 2 S 9H 2 O 98% Fisher Scientific 3.2 Procedures Production of Scorodite Atmospheric scorodite was produced and coated with silicate gel in order to determine whether this coating increases the stability range of scorodite. The scorodite production method has been developed over 20 years by the McGill HydroMET group supervised by Professor G.P. Demopoulos. Atmospheric scorodite was produced in a batch stirred reactor (Applikon Reactor). First, the arsenic reagent (arsenic pentoxide) was dissolved in water (40 g/l) and constantly stirred with a magnetic stirrer for 24 hours until the solution became clear. The iron sulfate 26

37 reagent (ferric sulfate) was then slowly added to the beaker and continuously stirred until a clear dark olive green solution formed. Water was added to reach 1 L. The arsenic to iron molar ratio was 1:1. The ph of the solution was 0.9. The solution was filtered using a pressure filter and 0.22 µm pore filter paper to remove any undissolved material. The filtered solution was then added to the batch reactor and stirred at 750 rpm. The oil bath that circulates silicone oil around the outer glass container of the double-walled reactor was heated to 95 C (Figure 3.1 and Table 3.2). When the temperature inside the reactor reached ~65 C and the ph had dropped to ph 0.45, 5 g of hydrothermally produced scorodite was added to the reactor as seed. The internal temperature of the reactor was increased to 85 C and was maintained at this temperature for 24 hours. The solution was filtered, the filtrate collected and the pressure filter cake washed with 1 L of deionized (DI) water. After that, the scorodite cake was repulped in the reactor with a new 1 L batch of DI water. The scorodite was washed for 24 hours. This process was repeated for ~8 cycles of repulp washing. X-ray diffraction (XRD) was performed to verify the production of crystalline scorodite and inductively coupled plasma- optical emission spectroscopy (ICP-OES) was performed to determine the amount of arsenic removed during the 8-stage washing sequence (Figure 3.2A and B). 27

38 Figure 3.1- Applikon reactor used to produce atmospheric scorodite Table 3.2- List of equipment in reactor setup used to produce atmospheric scorodite (Figure 3.1) Number Equipment Description corresponding to Figure Applikon Reactor vessel 2L glass vessel with discharge points at height= 1, 21 cm attached to Masterflex tubing. Outer glass covering circulates silicone oil from lower discharge point to higher discharge point. 2 Axial impeller 3 baffles and impeller, d= 4.5 cm 3 Temperature probe Temperature adjusted by hot oil bath controller 4 ph electrode ph recorded by controller 5 Hot oil bath and temperature controller Silicone oil cycled through tubing and into outer glass vessel of reactor 6 Masterflex oil transfer tubing Rubber tubing resistant to 110 C 7 Applikon Controller Adjusts impeller stir speed and records temperature and ph 28

39 A) B) Figure 3.2- Atmospheric scorodite synthesis results with A) XRD results and reference spectrum displaying a crystalline scorodite spectrum and B) Total dissolved arsenic released (mg/l) over 5 24 hour washing cycles. 29

40 3.2.2 Production of Silicate Gel Sodium silicate solution, Na 2 SiO 3, was used to produce silicate gel. The initial composition of the sodium silicate solution was: 29.2%, 9.1% and 61.7% amorphous silica (SiO 2 ), sodium oxide (Na 2 O) and water respectively (Fisher Scientific). Sodium silicate gel was produced through a neutralization reaction. Two different mixing sequences were employed. Initially 1M sulfuric acid, H 2 SO 4, was added dropwise from a burret to a beaker containing 1M sodium silicate solution being stirred with a magnetic stirrer. The initial ph of the sodium silicate solution was ~ 12. Once the ph lowered to ~ph 10, a viscous gel formed (Figure 3.3). Figure 3.3- Silicate gel (from 1M sodium silicate solution) Since the above mixing sequence resulted in a high ph gel that was deemed aggressive for scorodite a reverse titration procedure was employed whereby 1M sodium silicate solution was added slowly into 1M sulfuric acid while stirring. By starting at low ph with sulfuric acid, and adding the base (sodium silicate solution) to the acid to slowly increase the ph, the ph was 30

41 lower at the point of gelation. Upon gelation ph ~6.5 was reached. Using this method, gels using 5M and 0.25M sodium silicate solution were also produced Washing and Blending A number of experiments investigated the effect of washing, blending and ageing parameters on the scorodite-gel (hereafter referred to as Scorogel) mixture. To remove sodium and sulfate from the gel, a washing procedure was developed. Initially a 10:1 w/w ratio of DI water to gel was used and agitated using a shaker table for 14 hours. Subsequent to that, washing experiments used less water (2:1 w/w ratio of DI water and gel) and included more vigorous mixing using a Roto-Shaker for shorter amounts of time (3 cycles of 2 hours each, water replaced each time). After the washing stage the gel was filtered and manually blended with scorodite using a metal spatula. Different molar Si/As ratios were used for different series of experiments (1:0.1, 1:1, 0.5:1, and 0.1:1 Si/As molar ratios) Ageing Ageing of the silicate gels was done for the purpose of transformation and condensation of the silicate gels into encapsulating silica. Different ways were tried involving open air or sealed, in-aqueous solution or in oven drying or hydrothermal conditions. One series of experiments looked at the effect of fully drying out the Scorogel. The method consisted of placing the blend in an oven heated to 40 C for 24 hours. The second method involved air drying the Scorogels at room temperature for 2 weeks. Finally, a series of experiments tested the effect of hydrothermal treatment on the transformation and coating capability of silicate gel. Gels were placed in an autoclave (acid digestion vessel, Parr Instrument Company-Figure 3.4), and then heated in an oven for 8 hours at 110 C and 160 C. Both 5M and 1M gels were hydrothermally (HT) treated. 31

42 Figure 3.4- Description of autoclave (acid digestion vessel- Image by Parr Instrument Company) Stability Testing The Scorogel (scorodite blended with silicate gel) containing 5 g of scorodite was placed in Erlenmeyer flasks with DI water at 40:1 w/w water to scorodite ratio. The samples were placed on a shaker table, running at 50 rotations min -1. Stability testing was done under harsh conditions in order to better assess the efficacy of the encapsulation capability of the silicate gels. Two conditions were employed: (a) Oxic involving equilibration at ph 9 using 0.5 M calcium hydroxide, Ca(OH) 2, as base; and Anoxic involving the use of a reducing agent, M sodium sulfide nonahydrate, Na 2 S 9H 2 O, to maintain near 0 mv E h at ph 9. Samples were taken using a 10 ml syringe and filtered with a 0.2 µm syringe filter. Samples were acidified using a 4% nitric acid solution. 32

43 3.3 Analysis and Characterization Elemental Analysis To determine the composition of the silicate gel produced and measure the concentration of arsenic in solution released from stability tested Scorogel samples, Thermo Jarell Ash Trace Scan inductively coupled plasma- optical emission spectroscopy (ICP-OES) was used FTIR and Raman Spectroscopy Normal practice for sample preparation for FTIR Analysis: To observe the chemical groups present in the silicate gel, the attenuated total reflectance (ATR) technique using Fourier Transform Infrared (FTIR) spectroscopy was used. The equipment was from Perkin Elmer, model Spectrum BX. The FTIR machine contained components from Pike Technology, model Miracle, mounted with a single bounce diamond crystal. Eight scans were taken of each sample. Novel sample preparation for FTIR mapping was performed on a different machine set-up, the UMA 600 ATR system. The detector image size was 32x32 pixels. Each pixel was 1.1 µm. This translates to a 35x35 µm size mapping image. At each pixel a full range ( cm -1 ) FTIR spectrum was measured. Each final spectrum at each pixel consisted of 256 co-added scans/acquisitions. Sample preparation consisted of placing the Scorogel (dried) powder onto a calcium fluoride glass window. Using this method, the infrared laser was able to penetrate the entire sample and develop spectra from the acquired scans. Normal practice for sample preparation for Raman Analysis: The Renishaw InVia Raman microscope system was used to analyze the groups present in the Scorogel (dried). The 785 nm near infrared laser and a 50x short working distance was used. Both regular scans with acquisition times of 40 to 60 seconds were taken as well as Raman mapping analysis of a single cross section of a coated scorodite particle. A line was drawn across the thin sectioned particle, and scans were acquired at equally spaced points along the line. The mapping function employed the arithmetic function of the WIRE 2.2 Renishaw software. The arithmetic function used a known and reliable spectrum from scorodite (Gomez et al., 2010a) and developed a ratio 33

44 between this strong signal and a weaker signal from silica. This ratio ensured that the silica signal did in fact exist at each point of analysis. Novel sample preparation for Raman mapping: The Scorogel was air dried and a small amount of Scorogel powder was placed on a piece of parafilm. The parafilm was then sealed with the powder inside. The parafilm with the sample was then placed between two stainless steel blocks and compacted. When parafilm with Scorogel powder was exposed at the top, between the two steel blocks, a stainless steel razor was used to cut this exposed part off, revealing a thin section of the sample SEM, TGA and XRD Analysis Scanning electron microscopy using the Hitachi S-4700 FE-SEM, equipped with an Everhart Thornley secondary electron detector, was used to take images of the Scorogel (after drying) at 5 KeV beam strength and 100,000X magnification. Electron dispersive X-ray (EDS) spectroscopy was used to perform point and mapping chemical analysis at 20 KeV. To determine the water content in the silicate gel, the TA instrument, model Q500 was used to complete thermogravimetric analysis of the gels with a heating rate of 20 C/min up to 800 C. X-Ray Diffraction was used to confirm the production of crystalline scorodite using a Philips PW 1710 diffractometer consisting of a copper target (Cu Kα radiation, λ= Å), a crystal graphite monochromator and scintillation detector. The analysis was performed at a scan rate of 0.1 degrees every 3 seconds X-Ray Photoelectron Spectroscopy (XPS) K-Alpha X-Ray Photoelectron Spectroscopy (XPS) equipment from ThermoFisher Scientific was used to perform a depth profile analysis on the Scorogel (dry sample). Sample preparation was very important. The Scorogel samples had to be dried in an oven for 24 hours at 110 C to remove essentially all of the moisture in the sample. If there is too much moisture in 34

45 the sample, the pressure in the analysis chamber of the XPS will be too high and the machine will not function. The depth profile analysis consisted of 6 levels of etching, 60 seconds long, with 3 KeV beam strength. After each etching cycle, an 8 second scan was taken of the exposed surface of the particles using 1 KeV beam strength. 3.4 Process Flow Chart Figure 3.5 provides a flowchart summarizing the sequence of procedural steps and range of conditions considered in investigating the effectiveness of the silicate gel-based encapsulation system. Figure 3.5- Flow chart, summarizing the various procedural steps followed from gel formation, washing, blending and ageing to stability testing of Scorogel material. Note: Different concentrations of sodium silicate solutions were prepared (0.25, 1 and 5M) however 1M is used in the flow chart as this was the concentration used for most series of work 35

46 4. CHAPTER 4 RESULTS AND DISCUSSION This chapter describes the various experiments performed along with the obtained results and discusses the effectiveness or ineffectiveness of silicate gels as scorodite encapsulation materials. The adopted silicate gel encapsulation sequence is summarized in the form of a flowchart in Figure 3.5. Before the silicate gel results are presented the stability of scorodite produced by the atmospheric precipitation process is considered in order to establish a benchmark for comparing silicate gel coated scorodite (termed hereafter Scorogel ) with uncoated scorodite. Next the silicate gel formation is described followed by its investigation as a scorodite encapsulant. The effectiveness of the encapsulation product (scorodite coated with silicate gel or Scorogel) is evaluated under harsh (for scorodite) conditions (ph 9, oxic, ~600 mv E h and sub-oxic/anoxic conditions, ~0 mv E h ) by employing an EPA Toxicity Characteristic Leaching Procedure (TCLP)- like modified stability testing protocol. Because of persisting instability issues the ageing/transformation of silicate gel via elevated temperature (drying in an oven or hydrothermally in an autoclave) processing was sought. Finally to provide insight as to the underlying factors responsible for the release of arsenic from Scorogel, the interaction between silicate, iron and arsenate groups is discussed on the basis of different analytical and characterization studies. 4.1 Stability of Atmospheric Scorodite Atmospheric scorodite was produced as described in Chapter 3 following the method discussed in the theses of Leetmaa, 2008 and Katsarou, The method favours the growth of crystalline scorodite on seed particles (Figure 4.1) by operating within the established metastable zone as described in Chapter 2 (refer to Figure 2.1). Minor changes were made including the temperature at which to add the seed (67 C, ph 0.4), and the temperature/length of time at which precipitation took place (24 hours at 85 C). In a single batch (40 g/l As(V) solution; under the applied conditions) ~ 100 g of scorodite was produced having the physical appearance indicated in Figure 4.2. Positive identification, done by XRD, was included in Chapter 3. 36

47 Figure 4.1- Hydrothermally produced scorodite that was used as seed in atmospheric scorodite production (Leetmaa, 2008) Figure 4.2- Images of scorodite produced by atmospheric precipitation at 85 C- (from left to right): Scorodite after filtration (image by C.Verdugo, HydroMET Laboratory) and SEM images of scorodite particles in silicate gel matrix (0.5:1 Si/As molar ratio) at different magnification The produced-by-atmospheric precipitation scorodite was of excellent quality not only in terms of particle growth characteristics (Figure 4.2) but also in terms of stability. This is exemplified with the data of Figure 4.4 where its stability is compared to that of hydrothermally produced - at 160 C - scorodite (and used in this work as seed-fig. 4.1). The stability testing involved exposure of scorodite samples (L/S= 40 at 22 C) over 30 days to ph 9 adjusted solution. The hydrothermal scorodite data are from a previous HydroMET laboratory study carried out by Bluteau (2004 and Bluteau and Demopoulos 2007). By Day 30, the hydrothermal scorodite (sample 22-9) had released ~140 mg/l As, whereas the atmospheric scorodite had released ~8 mg/l As (Figure 4.3). This drastic difference could be explained by two factors, the better crystallinity of the atmospheric scorodite produced via supersaturation-controlled surface nucleation and particle growth as opposed to fast homogeneous nucleation of the smaller 37

48 hydrothermally produced scorodite particles (compare Figures 4.1 to 4.2); and the base that was used to adjust the ph during stability testing. During stability testing of hydrothermal scorodite, NaOH was used to maintain the solution at ph 9. For stability tests within the scope of this thesis research, Ca(OH) 2, or lime, was used to maintain ph 9. NaOH is known to be a strong base releasing instantaneously its alkalinity in a localized manner that at all likelihood results in further scorodite attack and dissolution. In contrast ph adjustment by lime results in smoother release of alkalinity. Similar observations were made by Leetmaa (2008). Figure 4.3- Comparison of scorodite stability for two differently produced crops: Atmospheric vs. hydrothermal. Total dissolved arsenic (mg/l) released over 4 weeks of stability testing at ph 9 under oxic conditions from atmospheric scorodite (this study) and hydrothermal scorodite (Bluteau, 2004); ph adjustment done with Ca(OH) 2 in this study and NaOH in Bluteau study. 4.2 Silicate Gel Formation Reaction of Sodium Silicate with Sulfuric Acid Sodium silicate (Na 2 SiO 3 ) solution is the precursor material used to produce silicate gel. The as-received solution composed of soluble silica (29.2%), sodium oxide (9.1%) and water (61.7%) (Fisher Scientific) corresponds to 6.4M. The structure of sodium silicate is depicted in Figure 4.4 where the silicate tetrahedra are visible along with the sodium that is not bonded into the structure and therefore can be washed out of the silicate gel that forms. 38

49 Figure 4.4- Silicate tetrahedron (left) and molecular structure of sodium silicate (right). Grey spheres= silicon atoms, red spheres= oxygen atoms, purple spheres= sodium atoms. H 2 O groups are not shown. (McDonald and Cruickshank, 1967; Mills, 2008) Sodium silicate is very alkaline in nature having a ph ~12. Prior to its use sodium silicate was diluted at various initial concentrations from 0.25 to 5.0M with 1.0M being the most common. As described in Chapter 2, acidification of the diluted sodium silicate solution via a reaction like the one below results in the formation of a semi solid network- silicate gel as shown in Reaction 4.1: n(na 2 SiO 3 + (1-x)H 2 O + H 2 SO 4 ) nna 2 SO 4 + [SiO x (OH) 4-2x ] n Reaction 4.1 Sulfuric acid (1M) was titrated into previously diluted sodium silicate solution that was under constant ph monitoring and stirring. Initially silicic acid is assumed to form, Si(OH) 4. With further acid addition that took only a few extra minutes, the ph dropped from approximately ph 12 to ph 10. At this point the semi-solid silicate gel formed, [SiO x (OH) 4-2x ] n, having the consistency shown in Figure 4.5. The silicate gel consists of a network of polymerized Si-O-Si-OH groups that traps water via solvation and hydrogen bonding forces (Iler, 1979). TGA analysis revealed 90% of the gel mass (produced from 1M sodium silicate solution and 1M sulfuric acid) to be water (Figure 4.6A). After the gel was air dried for 7 days TGA analysis revealed that 5% water remained as part of the silicate gel structure (Figure 4.6B). Finally it should be mentioned that the gel contained as well (confirmed by ICP analysis) 21.5% sulfate, 10.5% sodium, and 61% silicon dioxide. This prompted the investigation of its washing described in the next section. 39

50 Figure 4.5- Image of silicate gel prepared by titration of 1M sulfuric acid (30 ml) into 1M sodium silicate solution (60 ml) A B Figure 4.6- TGA analysis of 1M sodium silicate gel, A) Not dried B) Air dried for 7 days Silicate Gel Washing - Removal of Sodium and Sulfate If silicate gel were to be used to coat scorodite at a large industrial scale and then be deposited into a tailings impoundment, sodium and sulfate from the silicate gel could leach into the water system and act as contaminants as well. An important aspect of this part of the work was to determine 1) the effectiveness of washing in removing the bulk of sodium and sulfate from the silicate gel and 2) whether the presence of sodium and sulfate within the silicate gel adversely or positively affects its consistency and coating ability. Initially, silicate gel was washed for 14 hours on a shaker table using a water-to-gel w/w ratio of 10:1. This method was effective in removing essentially all Na + and SO 2-4 from the silicate gel (nearly 100% Na and 93% SO 4 removed after 12 hours, Figure 4.7) without affecting the gel s consistency (see Figure 40

51 4.8). Equally washed and unwashed silicate gels were found not to differ in terms of their encapsulation capacity. The relevant data are presented in Appendix A.1. Hence it is concluded that Na + and SO 2-4 do not participate in the formation of the gel. In the investigations that follow unless otherwise stated the silicate gels are used after they have been washed. Figure 4.7- Percent total sulfate (left) and sodium (right) removed after 14 hours of washing (water-to-gel w/w ratio of 10:1) Figure 4.8- Appearance of silicate gel before and after washing In an effort to reduce the use of water and washing time an alternative method was tested whereby 3 cycles of washing was employed, 2 hours each, using a 2:1 w/w ratio of water to gel. More vigorous washing was employed by using a Roto-shaker. With this method, less Na + and SO 2-4 were removed (55% Na +, 53% SO 2-4 removed). This suggests that Na + and SO 2-4 removal is most likely governed by a slow diffusion process that may prove practically a limiting factor for such system to find large scale application further considering that Na 2 SO 4 will have to be recovered ultimately from the wash water. 41

52 4.2.3 Silicate Gel Ageing As discussed in Chapter 2, the silicate gels upon ageing are supposed to undergo condensation by releasing part of the water and replacing hydroxyl (OH - ) with oxo (O 2- ) groups hence forming a more densely packed network of siloxane groups, Si-O-Si (Reaction 4.2 (Brinker, 1988)): Si-OH + OH-Si Si-O-Si + H 2 O Reaction 4.2 (Brinker, 1988) As standard ageing procedure a simple holding of the Scorogel (scorodite-gel blend) in a closed environment (sealed flasks) at room temperature for 5 days was adopted. All results that follow in this section were obtained with this ageing method. Towards the end of the thesis (section 4.4) the effect of ageing is revisited by considering drying and hydrothermal treatment as alternative ageing methods. 4.3 Silicate Gel as Encapsulant for Scorodite General Initial encapsulation tests involved testing the 1M sodium silicate gel by blending with scorodite at different Si/As molar ratios (0.1:1, 0.5:1, 1:1 and 1:0.1). From a more practical aspect taking that the as- received sodium silicate has 6.4 M concentration, a 1:1 Si/As molar ratio is equivalent to 28.5 g of as-received sodium silicate per 100 g of scorodite. Figure 4.9 displays the appearance of silicate gel manually blended with scorodite (Scorogel). Figure 4.9- Scorodite manually blended with silicate gel at 0.5:1 Si/As molar ratio (Scorogel) 42

53 SEM images of scorodite particles coated with silicate gel after open air drying for 2 days are shown in Figure The scorodite particles are seen to be dispersed in a matrix of silicate gel (1:0.1 Si/As molar ratio). Figure 4.10B displays an individual Scorogel particle. EDS mapping (Figure 4.10D) clearly revealed the engulfing of the scorodite particle by the silicate coating material. This coating however is not uniform. Figure 4.10C displays a hole in the silicate coating that exposes the substrate scorodite particle. It must be noted that the images in Figure 4.10 correspond to silicate gel-coated scorodite using a large excess of silicate gel (Si/As molar ratio=10). In the sections that follow lower ratios are investigated. Based on these SEM images and X-ray analysis, it was concluded that silicate gel can coat scorodite and further investigation into the effectiveness of these coatings as encapsulating material was undertaken. A B C D Figure SEM images and EDS mapping of silicate gel coated scorodite particles (Scorogel) (1:0.1 Si/As molar ratio) A) Thin section image of scorodite coated particles B) Silicate gel coated particle C) Hole observed in silicate gel coating, D) EDS mapping of Scorogel particle 43

54 4.3.2 Initial Testing- High ph Silicate Gel The initial method for gel formation involved the titration of sulfuric acid (1M) into sodium silicate solution (1M; initial ph 11.8). Gel formation occurred almost immediately (within 30 seconds) at ph 10. The resultant 1M silicate gel was blended with scorodite at different Si/As molar ratios; 2:1, 0.5:1 and 0.1:1. Initial short term stability testing (at ph 9) revealed this type of gel not to be protective but the opposite to cause even higher arsenic release than scorodite alone. The amount of arsenic release was found to be proportional to the amount of silicate gel used. Thus the 2:1 Si/As Scorogel released 35 mg/l As after Day 1. Correspondingly Scorogels with 0.5:1 and 0.1:1 Si/As molar ratio released 10 mg/l and 5 mg/l As after Day 3, higher than scorodite alone that released 4.4 mg/l As. The adverse effect the silicate gel had on scorodite stability is attributed to the harsh caustic microenvironment created by the initial high ph of the silicate gel (~10). In other words despite the fact that the stability test was performed at ph 9 (bulk solution measurement) it is hypothesized that locally at the gel/scorodite interface a much higher ph was operating and as such made scorodite less stable given its high solubility in the alkaline ph range (Figure 2.2, (Bluteau and Demopoulos, 2007)). To avoid this undesirable effect silicate gel was produced via a reverse titration procedure Reverse Titration- Low ph Silicate Gel Through reverse titration whereby sodium silicate solution was titrated into sulfuric acid, gel formation occurred at a lower ph than the original titration procedure. Typical reverse curves and the point of gelation for silicate gels produced from different concentrations of sodium silicate solution are shown in Figure The final gelation point this time varied between 5 and 6.5 (ph 5.14 for 5M, ph 5.52 for 0.25M, and ph 6.5 for 1M) as opposed to >10 for the direct titration method. Silicate gel formation occurred in the same pattern with all three sodium silicate solutions. As the sodium silicate was slowly being added to the sulfuric acid, the ph slowly climbed until reaching the gelation point signified by an immediate steep jump in ph. 44

55 A B C Figure Reverse titration curves (sodium silicate sulfuric acid solution) showing gelation ph point for A) 0.25M B) 1M and C) 5M sodium silicate solutions Oxic and Anoxic Stability Protocols To test the new reverse titration method of gel formation, 1M silicate gel was prepared (final ph 6.5) and blended with scorodite at a 1:1 Si/As molar ratio. The Scorogel as per standard procedure stated earlier after washing and ageing for 5 days was subjected to stability testing (Figure 4.12). Figure Stability testing of Scorogel in an Erlenmeyer flask filled with DI water, 40:1 w/w ratio water to scorodite Stability testing was performed on Scorogel at ph 9, under oxic and anoxic conditions. The ph adjustment was done with slaked lime (calcium hydroxide (Ca(OH) 2 ). On average, on the initial day of stability testing where the first ph adjustments were made to increase the ph from approximately ph 4 to ph 9, 2 ml of 0.5 M Ca(OH) 2 was required. An averaged total of 45

56 approximately 6 ml of 0.5 M Ca(OH) 2 was required to adjust one sample to ph 9 for 15 days. Less Ca(OH) 2 was required further into stability testing as the ph drift decreased and less Ca(OH) 2 was required for adjustment. For the anoxic stability tests different reducing chemicals were tested in their ability to adjust the E h of the solution, namely sodium sulfite (Na 2 SO 3 ), sodium bisulfide (NaHS) and sodium sulfide (Na 2 S). Sodium sulfide (Na 2 S) proved to be the most manageable in terms of speed of E h adjustment and retention to the target level of 0 mv. A 0.125M Na 2 S solution was selected as providing good strength with minimum dilution and less occurrence of local hot spots (in terms of overcharging/over reducing). On the initial day of stability testing, on average 1 ml of M Na 2 S was required to reach approximately 0 mv E h. Over 15 days, on average a total of 6.5 ml 0.125M Na 2 S was required for E h adjustment within 200 ml of solution. While the E h drift did decrease over time, a steady amount of Na 2 S was required for adjustment (approximately 0.5 ml per day) Oxic and Anoxic Stability Response Figure 4.13 compares Scorogel to scorodite alone over 8 days of stability testing under oxic conditions. Scorogel did release less arsenic than scorodite over 8 days, however by Day 8, the release of arsenic from Scorogel reached 10 mg/l. On Day 8, scorodite alone released 15.3 mg/l As. While Scorogel released less As than scorodite alone, this concentration still far exceeds the Metal Mining Effluent Regulations maximum allowed level of 1 mg/l arsenic in a grab sample (Anonymous, 2012). After 6 days of ph adjustment to ph 9, both the Scorogel and scorodite ph showed to stabilize (less ph drift) (Day 6- ph 7.21 (Scorogel), 7.3 (scorodite), Day 7- ph 7.34, 7.56, Day 8- ph 7.67, 7.94), however neither system reached equilibrium, because the ph continued to drop after each adjustment suggesting As release would continue. While the ph adjustment for both samples was similar, the pattern of As release was not. While a gradual increase over time was observed for Scorogel, there was a greater initial rate of release of As for scorodite, followed by an apparent reduction in As concentration after Day 4. 46

57 Figure Short term stability testing of Scorogel compared to scorodite alone- Total As (mg/l) released over 8 days (ph 9, oxic conditions- 600 mv E h ; 1:1 Si/As molar ratio, 1M silicate gel prepared using reverse titration method) The registered drop could have been due to experimental error. However Bluteau et al. (2009) found that between ph 5 and 9, the presence of calcium results in less arsenic release from scorodite. In this stability test Ca(OH) 2 was used to adjust to ph 9. Bluteau et al. also observed a decrease in the Ca/S molar ratio over time. At ph 9, the Ca/S ratio decreased to 0.1:1. Hence it is possible a similar interaction between lime and scorodite caused the drop in As concentration in the present work. Returning to the relatively high concentration of released arsenic from the Scorogel, it was again thought that the large amount of silicate gel used (1M sodium silicate, 1:1 Si/As molar ratio) led to the creation of a caustic environment at the interface of the scorodite particle surface and the silicate gel coating. Upon review of previous investigations involving sodium silicate solution it was noticed that the latter had been used at much higher dilution level than 1M used here. For example, Evangelou (2001) used a 0.01M sodium silicate solution to coat the oxidized surface of pyrite, Bessho et al. (2011) used a 0.01M sodium silicate solution to coat the surface of pyrite again, while Lü (2005) used a 0.25M sodium silicate solution to coat SrAl 2 O 4 :Eu 2+,Dy 3+ particles. Their successes with low concentrations of sodium silicate solution was interpreted that a low concentration of sodium silicate solution could produce an effective silicate gel coating without experiencing the seemingly caustic effects of the silicate gel. In an attempt to verify that the microenvironment alkaline ph of the gel may be related to the instability of Scorogel, a series of experiments was done using a less concentrated sodium 47

58 silicate solution. A 0.25M sodium silicate solution and 1M sulfuric acid solution were used to form a silicate gel, with a final ph of The silicate gel was washed and blended with scorodite at a 0.1:1 Si/As molar ratio. Figure 4.14 displays total dissolved As released during stability testing of the Scorogel and scorodite alone at ph 9 under both oxic and anoxic conditions. After 15 days, the Scorogel under oxic conditions had released less arsenic than scorodite alone. However, the As release of the Scorogel was increasing at a greater rate than scorodite over those 15 days. During this time, while the ph did stabilize after each daily adjustment to ph 9, by Day 15, the ph had dropped to ph 7.75 for Scorogel and 7.86 for scorodite. This suggests that equilibrium had not been reached for either and the release of As would continue. After Day 15, As release from Scorogel surpassed the As release from scorodite alone (29 and 22 mg/l). Under anoxic conditions, on Day 10 Scorogel had released equivalent arsenic as scorodite (15 and 14.5 mg/l). The ph or E h by Day 15 had not stabilized and continued to drift for Scorogel (ph 7.78, 205 mv) after adjustment to ph 9 and 0 mv again suggesting further release of As. It is concluded from these results that the ph microenvironment that was hypothesized to be responsible for the increased arsenic release by Scorogel not to be the culprit hence another factor was investigated- That of ageing. A B Figure Medium term stability testing of Scorogel compared to scorodite alone for the case of low concentration silicate gel- Total As (mg/l) released over 30 days: A) Oxic conditions (600 mv E h, ph 9) and B) Anoxic conditions (0 mv E h, ph 9); Scorogel produced from 0.25 M sodium silicate solution by reverse titration (0.1:1 Si/As molar ratio) 48

59 4.4 Effect of Modified Ageing on Silicate Gel Transformation and Scorogel Stability By reducing the ph of the gel via the adopted reverse titration method and the concentration of the sodium silicate solution, arsenic release from the Scorogel was reduced but still remained above that of scorodite alone. As such, other silicate gel factors must be responsible for aggravating the release of arsenic from scorodite. As can be seen from the SEM images presented in Figure 4.10, the silicate gel was able to physically coat the scorodite particles (Figure 4.15). Such coating remained intact even after 30 days of stability testing (not shown). This suggests that the problem is not physical/mechanical but rather of chemical nature. Since the SEM pictures depicting the presence of coating were collected after the Scorogel was first dried it was thought that altering the ageing process including trying drying may lead to a more effective encapsulating layer. In other words by altering the ageing conditions it was sought to promote silicate gel condensation and transformation into a stronger network of Si-O- Si siloxane bonds that could produce a silicate gel matrix around the scorodite particles with better sealant properties. Figure Silicate gel coated scorodite particles (Scorogel) at 1:1 molar ratio. Washing and ageing occurred under standard conditions (3 cycles of 2 hours, 2:1 water: solid w/w ratio, closed sealed 22 C ageing) Two alternative ageing methods tried: (a) Slow drying at ambient temperature or accelerated drying at 40 C; and (b) Hydrothermal ageing above the boiling point of water. The latter was inspired from the use of sodium silicate (also called liquid glass ) as sealant in automotive radiators where apparently heating at ~105 C causes its conversion into a solidified 49

60 material (Barks, 1987). This property of sodium silicate was also exploited during the recent Fukushima nuclear reactor disaster in Japan to stem a hot radioactive water leak by solidifying the contaminated soil. Solidification of sodium silicate was promoted by the heat of the hot radioactive water (Anonymous, 2011) Ageing Scorogel by Drying at 22 C and 40 C Scorodite was blended with silicate gel (0.1:1 Si/As molar ratio) then aged in the oven at 40 C for 24 hours or air dried for 2 weeks at 22 C. Stability testing results under oxic and anoxic conditions show a steep increase in arsenic release for all samples (Figure 4.16). However after Day 12, the Scorogel aged at 40 C showed a decreased rate of arsenic release. The Scorogel aged at room temperature (22 C) continued to climb in terms of arsenic release for the entire 15 days (RT Scorogel, Day 4= 12 mg/l As, Day 15= 19 mg/l As, 40 C Scorogel, Day 4= 11.3 mg/l As, Day 15= 13.1 mg/l As). This pattern was exhibited for the anoxic samples as well (RT Scorogel, Day 4= 13.3 mg/l As, Day 15= 21.6 mg/l, 40 C Scorogel, Day 4= 11 mg/l, Day 15= 13.7 mg/l). Both Scorogel samples under oxic conditions did release more As than scorodite alone after Day 10 and 12 (RT= 15.9 mg/l As, 40 C= 15.1 mg/l As, scorodite= 14.8 mg/l As). Under anoxic conditions both Scorogel samples released more As than scorodite alone almost from the beginning of stability testing (Day 0 and Day 4 for RT and 40 C Scorogel). Figure 4.17 displays the ph and E h (for anoxic samples only) before and after adjustment during stability testing. Scorodite under anoxic conditions showed larger E h drift throughout the 15 days, compared to the Scorogel samples. There was little difference in E h drift between the two Scorogel samples (Figure 4.17B). The ph drift was less for Scorogel than scorodite alone for both the oxic and anoxic groups (Figure 4.17A). 50

61 A B Figure Effect of drying on silicate gel ageing and Scorogel stability- Total As (mg/l) released over 15 days: Scorogel air dried at room temperature (22 C) for 2 weeks or for 24 hours at 40 C (using washed 1M silicate gel, 0.1:1 Si/As molar ratio) compared to scorodite alone under A) Oxic conditions and B) Anoxic conditions. ph and E h levels are displayed for Day 15 (before adjustment to ph 9 and 0 mv E h (anoxic samples only)) During the 15 days of stability testing the ph of the Scorogel aged at 40 C showed less drift from ph 9 compared to the RT aged Scorogel. This could be a sign of a reduced rate of arsenic release for the 40 C Scorogel (RT= ph 8.21, 40 C= 8.61, Figures 4.16 and 4.17A) in agreement with the data of Figure The apparent increased stability of the Scorogel aged at 40 C suggests that ageing under higher temperatures can improve the stability of Scorogel, after an initially higher release of As. 51

62 A B Figure ph (A) and E h (anoxic samples) (B) profiles during stability testing for scorodite and Scorogel samples aged at 40 C for 24 hours and at RT (22 C) for 2 weeks (measured and adjusted values shown) Ageing by Hydrothermal Treatment of Silicate Gel and Scorogel (110 and 160 C) Transformation of silicate gel: Here both silicate gel alone and blended with scorodite (Scorogel) were hydrothermally tested in an autoclave as an alternative ageing process. First, 1M silicate gel was treated at 110 C for 8 hours. The 8-hour treatment time was selected following preliminary tests at 2 hours and analysis by FTIR characterization that showed 8 hours was needed to achieve silicate gel transformation. FTIR spectra of the 1M gel before and after this 8- hour hydrothermal treatment at 110 C are shown in Figure The main difference between the two spectra is observed at approximately 3400 cm -1 where a decrease in intensity of the Si- OH peak is observed for the silicate gel that underwent HT treatment, suggesting a more condensed network of O-Si-O groupings have formed (Figure 4.18A, (Vella et al., 2011)). This can also be seen in Figure 4.18B, where the peak representing the presence of H 2 O in the structure has reduced in intensity for the silicate gel that underwent HT treatment. The two peaks on the far right of Figure 4.18 (C and D) represent Si based groupings. The slight shift to the left observed in the HT treated silicate gel peak in box C and the greater intensity of the peak in box D may suggest a more extensive polymerization of the silicate gel network has occurred (Chemtob et al., 2012). 52

63 Figure FTIR spectra of washed 1M silicate gel before and after hydrothermal treatment in an autoclave at 110 C for 8 hours. A) Si-OH group; B) H 2 O group; C) and D) Possible Si-based groups Stability of HT Aged Scorogel: Having determined that hydrothermal treatment (HT) does lead to silicate gel transformation next Scorogel blends were HT treated at either 110 C or 160 C for 8 hours and subjected to stability testing. Hydrothermal treatment either involved the silicate gel alone (that was after blended with scorodite) or the Scorogel blend HT treated. The obtained results are plotted in Figure The accompanying ph and E h profiles are provided in Appendix A.4. Overall, once more, both of the two Scorogel blends led to higher release of As than scorodite alone. This implies again some sort of destabilizing chemical interaction between scorodite and silicate gel. Apparently this destabilizing chemical interaction was promoted with temperature elevation. For example the Scorogel HT treated at 160 C released more arsenic (45 mg/l and 55 mg/l As under oxic and anoxic conditions) than that Scorogel HT treated at 110 C (28 mg/l and 33 mg/l As under oxic and anoxic conditions). By contrast the arsenic release from Scorogel prepared by separately HT treating the gel and then blending it with scorodite was significantly lower. Thus the corresponding numbers (after 15 days of testing) are: ~17 mg/l As for both 110 C and 160 C HT gels under oxic conditions; and 17 vs. 10 mg/l As for 110 C vs. 160 C HT gels under anoxic conditions. It is interesting to note in this case that the HT temperature had no effect on the oxic test while in the anoxic test, the higher HT temperature resulted in a gel releasing less arsenic. This may be interpreted that the release of arsenic is not 53

64 governed by the ph or Eh of the stability medium (oxic vs. anoxic) but by the temperature and time of the silicate gel-scorodite contact. By examining further the arsenic release curves of Figure 4.19, it is noticeable that like the stability testing results for Scorogel aged for 24 hours at 40 C (Figure 4.16), both oxic and anoxic Scorogel samples appear to reach a plateau after Day 10. In other words it looks as if the maximum (within the limitations of the short term test) of releasable arsenic appears to have been reached. It is possible that the so-labeled releasable arsenic has in fact been mobilized (released) during the hydrothermal treatment, and because the Scorogel was not washed after HT treatment, it ended up in the stability test liquor. This issue is further investigated in the next section. A B Figure Effect of hydrothermal treatment on ageing and Scorogel stability- Total As (mg/l) released over 15 days: Scorogel (using washed 1M silicate gel, 0.1:1 Si/As molar ratio) compared to scorodite alone under A) Oxic conditions and B) Anoxic conditions. Silicate gel and Scorogel were hydrothermally (HT) treated at 110 C and 160 C for 8 hours. ph and E h levels are listed for Day 15 (before adjustment to ph 9 and 0 mv E h ) 54

65 4.5 How Does Silicate Affect the Release of Arsenic from Scorodite? It is apparent from the stability testing of the Scorogel prepared and aged under different methods, that the use of silicate gel did not prevent but actually increased the release of arsenic from scorodite. How is this possible? As already stated earlier all evidence pointed out to certain destabilizing chemical interactions between the silicate gel and scorodite. A possible interaction may have involved a certain ion-exchange or metathetic reaction by which silicic acid groups (Si(OH) 4 ) exchanged with arsenic acid (H 3 AsO 4 ) groups on the surface of scorodite. To consider the feasibility of such interaction the stability constants of Si(OH) 4 and arsenate complexes with Fe 3+ were compared (data listed in Table 4.1). The complex that forms between silicic acid and Fe 3+ has a higher stability constant (Log K) than the complexes that form between arsenate and Fe 3+. This suggests that the iron-silicate complex is stronger hence it is favoured to form at the expense of the arsenate complexes (at least 7 orders of magnitude difference). This implies that it is possible that silicate reacted with iron (III) on the surface of scorodite taking the place of arsenate, thereby causing the increased release of arsenic. Such competition between arsenate and silicate for ferric sites was also observed by Singh et al. (2005) who studied the effect of soluble silicates on the efficiency of arsenate removal through adsorption onto ferrihydrite. In the presence of silicate, more As(V) remained in solution as its adsorption via surface complexation on hydrous ferric oxide/ferrihydrite (Fe 2 O 3 0.5H 2 O) was hindered by competing adsorption of silicate (Dzombak and Morel, 1990). Similar observations were made by Meng et al. (2000). Table 4.1- Stability constants of Fe 3+ silicate and Fe 3+ arsenate complexes Complex + 2+ FeSiO(OH) 3 0 FeAsO 4 2+ FeH 2 AsO 4 + FeHAsO 4 Log K (Stability Constant) Temperature ( C) Source C Smith and Martell 1976 Weber and Stumm C Langmuir et al C Langmuir et al C Langmuir et al

66 Another interesting feature observed by Singh et al. (2005) was that when an increased concentration of silicate was added into solution, polymerization of Si-O groups attached to the Fe sites took place. Ferrihydrite appeared to act as a template for Si-O-Si polymerization, where silicate linked directly to Fe sites via Fe-O-Si bonds, while other Si-O-Si groups were bonded to the adsorbed Fe-O-Si groups as shown in Figure 4.20 (Swedlund and Webster, 1999). If the possibility of Fe-O-Si bond formation and polymerization of Si-O-Si groups on the surface of ferrihydrite is translated to this study, there could be a possibility of an iron silicate (Fe x (SiO 3 ) x ) phase forming upon blending the silicate gel and scorodite and being promoted with temperature elevation. In order to verify this scenario further work was undertaken involving surfacesensitive characterization techniques as described in the next section. Figure The polymerization of silicic acid (Si(OH) 4 ) adsorbed onto Fe(III) surface bonding sites of ferrihydrite, A) Adsorption of Si(OH) 4 monomer, B) Polymerization, C) Siloxane links (Si-O-Si) between Si(OH) 4 monomers (Swedlund and Webster, 1999) 4.6 Molecular and Surface Characterization of Scorogel Before the surface characterization results are presented SEM pictures of silicate gelcoated scorodite particles that had been hydrothermally treated at 110 C for 8 hours and subjected to stability testing for 30 days are shown in Figure Again as it was seen in Figures 4.10 and 4.15 for Scorogel samples prior to stability testing, it is clear that physically the gel has deposited in a cement-like matrix engulfing and at least partially covering the scorodite particle surface. 56

67 Figure SEM images of HT treated (8 hours at 110 C) Scorogel particles (5M silicate gel, 0.5:1 Si/As molar ratio) and after 30 days of stability testing at ph 9 under oxic conditions Si solution analysis during stability testing was undertaken. By comparing the solution concentration profile of Si during stability testing with that of As (shown in Figure 4.22), a possible cause-effect relationship between Si and As behaviour may be detected. As it can be seen in Figure 4.22 after Day 6, the Si concentration decreased and may be interpreted as possible silicate anchoring onto Fe sites on the surface of scorodite in analogy to the silicateferrihydrite system depicted in Figure 4.20 (Meng et al., 2000; Swedlund and Webster, 1999). Figure Si and As concentration in solution during stability testing (at ph 9) of hydrothermally (HT) treated Scorogel at 110 and 160 C (1M sodium silicate solution and 1M sulfuric acid, 0.1:1 Si/As molar ratio) 57

68 Surface characterization of Scorogel samples involved XPS depth profile analysis and FTIR and Raman mapping techniques to qualitatively look at the distribution of As and Si for the case of Scorogel particles aged at 40 C for 24 hours that then underwent 15 days of stability testing at ph 9. XPS depth profile analysis involved progressive laser (4 kv power) etching of the surface layers of the Scorogel particles over 60 second intervals. Each laser etching cycle based on data of another study involving an amorphous silica standard (60 seconds of etching at 4 kv) removed 8 nm of surface material (Mantha et al., 2012). Figure 4.23 displays the peak intensity and atomic % of As and Si for etch level 0 (no etching) and etch level 6 (final etch level). What is interesting to see is that the presence of As is lower at the surface (exhibited by a lower peak and atomic %) of the Scorogel particle compared to the deeper surface layers (etch level 6). Secondly, the presence of Si while higher on the outer surface of the Scorogel particle, is still present within the deeper surface zone. In other words, silicon appears to be present not just on the surface of the Scorogel as a coating, but also deeper at about 50 nm from the surface if each etch cycle is taken to represent 8 nm. This observation provides indirect evidence of possible bonding of silicate to iron. Figure XPS depth profile analysis (As and Si peaks) of Scorogel particle before and after sixth etching cycles (Scorogel aged at 40 C for 24 hours) 58

69 FTIR mapping was used to determine the distribution of arsenic (As) within the cross section of a Scorogel particle following 15 days of stability testing (Figure 4.24). The Scorogel was aged for 24 hours at 40 C. One of the issues with the use of the specific FTIR mapping technique is the loss of peaks located in the lower end of the wavelength spectrum. Thus it was difficult to use peaks from lower wavelengths (symmetric stretching at 800, 380 and 340 cm -1, anti-symmetric stretching at 483, 450 and 420 cm ) that specifically represented the AsO 4 anion. Instead a broader peak ( cm -1 ) that represents hydrogen bonding between a water group and an arsenate group (Gomez et al., 2010a) was utilized as an indirect marker of the presence of arsenic within the Scorogel particle depicted in Figure As it can be seen in Figure 4.24 the presence of As dissipates moving from the center of the particle towards the surface that was coated with the silicate gel (see also Figure 4.21). This can be interpreted as evidence of loss of arsenic from the outer layers of the Scorogel particle. Figure FTIR mapping displaying a cross section of a Scorogel particle (aged at 40 C for 24 hours, and subjected to 15 days stability testing) and the associated intensity of arsenic depicted in color (red represents high As presence and blue no presence). Peak identified at cm -1 ) 59

70 Raman mapping was used to determine the presence of Si along the cross section of a Scorogel particle aged at 40 C for 24 hours (Figure 4.25). Amorphous silicate gel is characterized by weak peaks that are difficult to identify. However it was possible to use a small peak at 1000 cm -1 that is known to be characteristic of a silica group (Chemtob et al., 2012). A sequence of spectra at that wavelength were taken in a linear fashion along the cross sectioned Scorogel particle as shown in Figure Based on the white color intensity scale representing the relative abundance of the silica group, it can be seen the presence of Si to be strongest on the rim of the Scorogel particle; but what is interesting, is the continued presence of Si even in the interior, albeit at lower occurrence, of the Scorogel particle. This is an indication that Si-O As-O anion exchange might have taken place within the scorodite structure at least in the zone near the surface. Matching the Raman mapping information (Figure 4.25) with what has been obtained from solution analysis (Figure 4.22), XPS depth profile analysis (Figure 4.23), FTIR mapping (Figure 4.24), and stability constant analysis (Table 4.1), a clear pattern emerges pointing to the loss of arsenic from the outer layers of Scorogel particles via arsenate silicate exchange. 60

71 Figure Raman mapping displaying the presence of Si along a cross sectioned Scorogel particle (aged at 40 C for 24 hours, and underwent 15 days stability testing) and the associated intensity of silicon depicted in color (white represents high Si presence). Si peak identified at 1000 cm -1 ) 4.7 What Can be Learned from this Study and the Way Forward Based on the characterization and stability analysis presented, the silicate gel-coated scorodite (Scorogel) is seen to be releasing arsenic via a surface chemical reaction involving silicate-iron bonding with some similarity to the silicate-ferrihydrite system (Meng et al and Singh et al. 2005). Such interaction was promoted with temperature elevation (characteristic of a chemical reaction) as was clearly illustrated in the two hydrothermal treatment tests at 110 and 160 C. As mentioned above, after Day 6 of stability testing of the HT treated Scorogel, the rate of release of arsenic decreased and reached a plateau after Day 10. What could that mean for the Scorogel system s stability? Is it possible that after a certain amount of arsenic was released from the outer layers of the scorodite particle, a resistant iron silicate layer was able to form that prevented the further release of arsenic? This is an issue that is proposed for future 61

72 investigation. If indeed a limited surface reaction leads to the formation of an iron silicate layer as depicted in Figure 4.26 that proves to have good sealant properties for preventing further arsenic release, a robust encapsulation system may become feasible. Considering that in the worst case a total of 45 mg/l of arsenic was released (refer to Figure 4.19) out of 5 g/l scorodite content this may represent a minor sacrificial release (<3%) of total arsenic. No doubt more work is required to determine whether this resistant layer is forming and preventing the release of arsenic. Figure Depiction of a Scorogel particle (scorodite coated with silicate gel), with the possible formation of an iron-silicate phase 62