Factors to Enable Crystallization of Environmentally Stable Bioscorodite from Dilute As ( III )-Contaminated Waters

Applicability of the bioscorodite method (use of the thermo-acidophilic Fe(II)-oxidizing archaeon Acidianus brierleyi for arsenic (As) oxidation and immobilization at 70 ◦C) was tested for synthetic copper refinery wastewaters of a wide range of dilute initial As(III) concentrations ([As(III)]ini = 3.3–20 mM) with varying initial [Fe(II)]/[As(III)] molar ratios ([Fe(II)]ini/[As(III)]ini = 0.8–6.0). Crystallization of scorodite (FeAsO4·2H2O) tends to become increasingly challenging at more dilute As(III) solutions. Optimization of conditions such as initial pH, seed feeding and initial [Fe(II)]/[As(III)] molar ratio was found critical in improving final As removal and product stability: Whilst setting the initial pH at 1.2 resulted in an immediate single-stage precipitation of crystalline bioscorodite, the initial pH 1.5 led to a two-stage As precipitation (generation of brown amorphous precursors followed by whitish crystalline bioscorodite particles) with a greater final As removal. The formation process of bioscorodite precipitates differed significantly depending on the type of seed crystals fed (bioversus chemicalscorodite seeds). Feeding the former was found effective not only in accelerating the reaction, but also in forming more recalcitrant bioscorodite products (0.59 mg/L; Toxicity Characteristic Leaching Procedure (TCLP) test). Under such favorable conditions, 94–99% of As was successfully removed as crystalline bioscorodite at all dilute As(III) concentrations tested by setting [Fe(II)]ini/[As(III)]ini at 1.4–2.0. Providing an excess Fe(II) (closer to [Fe(II)]ini/[As(III)]ini = 2.0) was found beneficial to improve the final As removal (up to 98–99%) especially from more dilute As(III) solutions.


Introduction
Arsenic (As) contamination is a growing problem in metallurgical operations due to the increasing metal demand that necessitates processing of As-bearing low-grade copper sulfide ores such as enargite (Cu 3 AsS 4 ) and tennantite (Cu 12 As 4 S 13 ).As one of the approaches to immobilize soluble As species from wastewaters, formation of scorodite (Fe III As V O 4 •2H 2 O) is considered an ideal form of As disposal due to its high thermodynamic stability, low iron demand and high density [1].Early studies on scorodite synthesis employed hydrothermal processes targeting high As(V) concentrations of 170-460 mM (13,000-35,000 mg/L) [2,3].They were later followed by studies using atmospheric pressure (mostly at around 95 • C), again targeting high As(V) concentration ranges of 130-670 mM (10,000-50,000 mg/L) [4][5][6][7][8][9][10].While these chemical approaches are generally effective for high As(V) concentrations, even after treatments with a good As removal efficiency, residual soluble As(V) must yet meet the effluent standard (e.g., 95% As removal from [As(V)] ini = 10,000 mg/L would still leave 500 mg/L of residual As(V) in solution [7]).Also, since such chemical treatments generally require a pretreatment step to oxidize As(III) to less toxic and less mobile As(V) using strong chemical oxidants prior to the scorodite precipitation step, residual As(III) can still persist in the solutions.There are also cases where dilute As(III) solutions (up to 25 mM) are produced in metallurgical operations, such as from solid-liquid separation followed by Cu recovery after enargite concentrate leaching, wherein chemical scorodite synthesis turns ineffective (JX Nippon Mining & Metals Co., Ltd., Tokyo, Japan; personal communication).
Hence, to realize effective scorodite formation even from dilute As solutions, a microbiological approach using Fe(II)-oxidizing thermophiles (72-80 • C) was proposed by Gonzalez-Contreras et al. (2010, 2012a, 2012b) using 13-37 mM (1000-2800 mg/L) As(V) as starting reactant [11][12][13].Later our studies enabled simultaneous microbial As(III) and Fe(II) oxidation using the thermophilic archaeon Acidianus brierleyi (70 • C), which for the first time realized one-step bioscorodite crystallization from original wastewater solutes of As(III) and Fe(II), without necessitating any As(III) oxidation pretreatments or chemical oxidants [14,15]: So far bioscorodite was successfully produced by Ac. brierleyi from 13 mM (1000 mg/L) of As(III) directly, with 86-99% final As removal depending on initial conditions [14].Whilst scorodite crystallization becomes more difficult at more dilute As concentrations under milder temperature condition, exploiting factors to enable such reaction is of great importance to broaden the applicability of the scorodite method for a wide range of As(III)-contaminated waters.
From studies of conventional abiotic scorodite syntheses, seed feeding (scorodite or other heterogeneous crystals) was shown to be one of the influential factors to promote scorodite formation [4,6,16].Initial pH was also reported to affect the stability and Fe/(As + S) ratio of the resultant scorodite product [17].Initial solution Fe/As ratio is also an affecting factor; in hydrothermal scorodite synthesis at 150 • C, the rate of As precipitation decreased with increasing initial solution Fe(III)/As(V) molar ratios from 1.0 to 1.5 and 2.0 [3].Our previous bioscorodite study also found that the form of As precipitation changes at different initial solution Fe(II)/As(III) ratios [14].
Based on the above, the aim of this study was set to find factors that enable effective bioscorodite crystallization from dilute As(III)-bearing solutions (3.3-6.5 mM; so far undescribed by scorodite studies): In particular, (i) bio-and chemical-scorodite particles were produced to compare their characteristics; (ii) their effectiveness as seed crystals was evaluated for the bioscorodite crystallization reaction from dilute 4.7 mM As(III) solution (wherein initial pHs 1.5 or 1.2 also compared); (iii) stability of the resultant bioscorodite products was evaluated by the toxicity characteristic leaching procedure (TCLP) test, and finally; (iv) a range of As(III) solutions (3.3-20 mM) with varying initial [Fe(II)]/[As(III)] molar ratios (0.8-6.0) was tested for bioscorodite crystallization to grasp the overall applicability of the bioscorodite method.
The resultant bio-and chemical-scorodite particles were collected respectively and freeze-dried overnight.Production of seed crystals did not involve seed feeding.

Toxicity Characteristic Leaching Procedure (TCLP) Test
The TCLP test was conducted by following the EPA method 1311 [19].Original scorodite seeds formed in Section 2.2 and final bioscorodite products formed in Section 2.3 (at [As(III)] ini = 4.7 mM; [Fe(II)] ini = 9.5 mM) were transferred into 25 mL vials containing 10 mL acetate buffer (pH 4.9) at a pulp density of 5% (w/v) and incubated at 25 • C, rotated at 30 rpm for 18 h.Liquid samples were filtered (0.45 µm glass fiber) to measure total soluble Fe and As concentrations.Tests were conducted in duplicates.

Ion Activity Product (IAP) Calculation
The IAP of bioscorodite was calculated as Equation ( 1), based on its congruent dissolution (Equation ( 2)).Ionic activities of Fe(III) and As(V) were calculated from Equations ( 3) and (4), using the measured pH values (a H + = 10 −pH ) and total soluble concentrations of Fe(III) and As(V) (M), respectively.The hydrolysis constants (K i ) at 70 • C were calculated using the van't Hoff equation (Equation (5), Table A1) [11,20].The resultant bio-and chemical-scorodite particles were collected respectively and freeze-dried overnight.Production of seed crystals did not involve seed feeding.

Toxicity Characteristic Leaching Procedure (TCLP) Test
The TCLP test was conducted by following the EPA method 1311 [19].Original scorodite seeds formed in Section 2.2 and final bioscorodite products formed in Section 2.3 (at [As(III)]ini = 4.7 mM; [Fe(II)]ini = 9.5 mM) were transferred into 25 mL vials containing 10 mL acetate buffer (pH 4.9) at a pulp density of 5% (w/v) and incubated at 25 °C, rotated at 30 rpm for 18 h.Liquid samples were filtered (0.45 µm glass fiber) to measure total soluble Fe and As concentrations.Tests were conducted in duplicates.

Solid Analysis
Precipitates were collected at designated times by filtration and freeze-dried overnight for X-ray diffraction (XRD; Ultima IV, Rigaku, Tokyo, Japan; Cu Kα 40 mA, 40 kV), Fourier transform infrared spectroscopy (FT/IR-670Plus, JASCO, Tokyo, Japan; KBr pellet method), thermo gravimetry (TG-DTA2000SA, Bruker AXS, Karlsruhe, Germany; by heating from room temperature to 500 • C at 10 • C/min at 100 mL/min N 2 ) and scanning electron microscope (SEM; ULTRA55, ZEISS, Oberkochen, Germany) analyses.Precipitates were embedded in an epoxy resin (SpeciFix-20, Struers, Ballerup, Denmark), polished using 2000 and 4000 grit emery papers on a polishing machine (Doctor Lap ML-182, Maruto, Tokyo, Japan) and carbon-sputtered (JEC-560, JEOL, Tokyo, Japan) to observe SEM cross-section views.The scorodite particle size was measured using a laser diffraction particle size distribution analyzer (Partica LA-950, HORIBA, Kyoto, Japan).Following the drying pre-treatment at 80 • C for 50 h, BET specific surface area measurements were performed on scorodite seeds (BEL-Max, MicrotracBEL, Osaka, Japan).To analyze the chemical composition of original seed crystals and final bioscorodite products, 50 mg of dried sample was digested in 10 mL of 35% HCl and concentrations of soluble Fe and As were measured by ICP-OES.

Characterization of Scorodite Seeds
Scorodite seed crystals were prepared by either biological or chemical methods and identified by XRD (Figure 1a).According to TG-DTA analysis, weight losses (at 125-250 • C) of bioscorodite and chemical scorodite seeds were 12.5% and 15.0%, respectively (Figure 1b), suggesting that their structural water contents were FeAsO 4 •1.55H 2 O and FeAsO 4 •1.91H 2 O, respectively (cf. the theoretical structural water content is 15.6% for scorodite (FeAsO 4 •2H 2 O)).A gradual and continuous weight loss observed with bioscorodite (but not with chemical scorodite) was likely from the result of decomposition of organic matters deriving from cells, as was detected by FT-IR (Figure 1c).Different XRD peak sharpness between bio-and chemical-scorodite seeds (Figure 1a) may, therefore, reflect difference in their crystallinity due to encrustation of cells within bioscorodite particles (Figure 1c) as well as in their structural water content (Figure 1b).
Morphological differences were clearly seen by SEM observation.Chemical scorodite seeds were aggregates of orthorhombic crystals (Figure 2d,e), with the average particle size and specific surface area of 98 µm and 0.7 m 2 /g, respectively (Table 1).Meanwhile, bioscorodite seeds were found as spherical aggregates (Figure 2a,b) with the average particle size and specific surface area of 36 µm and 2.0 m 2 /g, respectively (Table 1).There was a significant difference in crystal surface roughness and small almond-shaped grains covered the bioscorodite surface (Figure 2b).SEM cross-section views found hollow bioscorodite particles (Figure 2c) whereas chemical scorodite particles were completely filled (Figure 2f).Other than the presence/absence of cells, this morphology difference may be partly attributed to Fe and As ionic species available during the scorodite crystallization process: In the chemical process, As(V) is initially fed where Fe(III) is gradually provided from Fe(II) oxidation by dissolved oxygen.In the biological process, on the other hand, microbial Fe(II) oxidation to Fe(III) is rapidly completed, after which As(V) is gradually provided by microbial As(III) oxidation.Although the microbial oxidation speed of As(III) and Fe(II) was not affected in all cases, a clear difference in As and Fe precipitation behaviors was found at pH 1.5 (Figure 3a,b).The first-stage As removal observed at day 0-2 (brown-colored amorphous precursor formation [15]) was an almost common phenomenon, but the second-stage As removal observed at day 14-21 (whitish green-colored scorodite formation [15]) was promoted on bioscorodite seeds only, resulting in 98% As removal at day 21 (Figure 3a,b).
The supersaturation of scorodite was evaluated based on the ion activity product (IAP) (Figure 3c) calculated from Fe(III) and As(V) ion activities (Figure 3d,e).The log Ksp for scorodite was reported to be −25.8[24].The greater first-stage As-removal on bioscorodite seeds (than on chemical scorodite seeds; Figure 3a) led to the lower and stable supersaturation level during day 2-14
Although the microbial oxidation speed of As(III) and Fe(II) was not affected in all cases, a clear difference in As and Fe precipitation behaviors was found at pH 1.5 (Figure 3a,b).The first-stage As removal observed at day 0-2 (brown-colored amorphous precursor formation [15]) was an almost common phenomenon, but the second-stage As removal observed at day 14-21 (whitish green-colored scorodite formation [15]) was promoted on bioscorodite seeds only, resulting in 98% As removal at day 21 (Figure 3a,b).
The supersaturation of scorodite was evaluated based on the ion activity product (IAP) (Figure 3c) calculated from Fe(III) and As(V) ion activities (Figure 3d,e).The log K sp for scorodite was reported to be −25.8[24].The greater first-stage As-removal on bioscorodite seeds (than on chemical scorodite seeds; Figure 3a) led to the lower and stable supersaturation level during day 2-14 (Figure 3c), before a sudden second-stage As removal was triggered to produce crystalline scorodite (Figure 3a).
(Figure 3c), before a sudden second-stage As removal was triggered to produce crystalline scorodite (Figure 3a).SEM cross section views revealed contrasting bioscorodite crystallization patterns on bio-(Figure 4(a1-a6) and chemical-scorodite seeds (Figure 4b).Hollow bioscorodite seed particles were found increasingly filled with newly formed scorodite (Figure 4(a1-a6)), whilst solid chemical seeds induced their surface to be thoroughly coated with new scorodite precipitates (Figure 4(b1-b6)).SEM cross section views revealed contrasting bioscorodite crystallization patterns on bio-(Figure 4(a1-a6) and chemical-scorodite seeds (Figure 4b).Hollow bioscorodite seed particles were found increasingly filled with newly formed scorodite (Figure 4(a1-a6)), whilst solid chemical seeds induced their surface to be thoroughly coated with new scorodite precipitates (Figure 4(b1-b6)).Other abiotic studies under atmospheric pressure condition using As(V) and Fe(III) as starting species also have reported a positive effect of seeds with high specific surface area, e.g., the better scorodite precipitation kinetics were observed by using finer seeds such as hematite and hydrothermal scorodite [6].Caetano et al. (2009) reported that a specific surface area higher than 270 m 2 /g was required to achieve 85% As removal [16].
The fact that the specific surface area of bioscorodite (2.0 m 2 /g; Table 1) was nearly three-times greater than that of chemical scorodite (0.7 m 2 /g; Table 1) likely allowed the former to show better seeding effect.Other abiotic studies under atmospheric pressure condition using As(V) and Fe(III) as starting species also have reported a positive effect of seeds with high specific surface area, e.g., the better scorodite precipitation kinetics were observed by using finer seeds such as hematite and hydrothermal scorodite [6].Caetano et al. (2009) reported that a specific surface area higher than 270 m 2 /g was required to achieve 85% As removal [16].
The fact that the specific surface area of bioscorodite (2.0 m 2 /g; Table 1) was nearly three-times greater than that of chemical scorodite (0.7 m 2 /g; Table 1) likely allowed the former to show better seeding effect.
Chemical scorodite syntheses are generally conducted at pH 1.0 or lower [2,6,8,9,25].Fujita et al. (2009) chemically synthesized stable scorodite particles at the pH range of 0.3-1.0,but higher pH levels (pH > 1.2) negatively affected scorodite stability [17].Therefore, the effect of lowering pH was also tested here.Although As(III) and Fe(II) oxidation tendencies by Ac. brierleyi were not affected, lowering pH from 1.5 to 1.2 (using bioscorodite seeds) resulted in a different As removal behavior: A continuous formation of whitish bioscorodite particles was observed from the beginning at pH 1.2, instead of displaying two-stage As removal at pH 1.5 (Figure 3a).However, As removal remained relatively incomplete at pH 1.2 (91%), compared to at pH 1.5 (98%) (at day 30; Figure 3a).Overall lower supersaturation levels (Figure 3c) owing to lower As(V) ion activities at pH 1.2 (Figure 3d) as well as continuous As precipitation likely promoted steady and continuous bioscorodite crystal growth, as was reported under abiotic atmospheric pressure conditions [4,5,16].

Stability of Final Bioscorodite Products
TCLP tests were conducted for original scorodite seeds (bio-and chemical seeds) as well as for the resultant bioscorodite products (Table 2).
Original seed crystals: Despite their smaller particle size and larger specific surface area (Table 1), bioscorodite seeds showed greater stability (As; 0.33 ± 0.08 mg/L; Table 2) compared to chemical scorodite seeds (As; 5.49 ± 0.03 mg/L; Table 2).Although the Fe/As molar ratios of all scorodite seeds and products were 1.0-1.1 and 1.3-1.4,respectively, the amount of Fe leached generally remained only below one-tenth of that of As (Table 2), due to selective re-precipitation of Fe(III) at pH 4.9.Other chemical scorodite studies reported TCLP leachabilities ranging 0.1-13.6As-mg/L [7,10,16], generally often higher than those of bioscorodite (Table A2).Such stability difference between bioscorodite seeds and chemical seeds may be attributed to factors such as involvement of cell encrustation, structural water content and morphological difference, and crystal maturity (speed of crystallization), but the exact reasons remain to be unclear.
Final bioscorodite products: Stability of final bioscorodite products was improved (As; from 5.49 ± 0.03 to 1.86 ± 0.05 mg/L; Table 2) when chemical seed surface became coated with fresh bioscorodite precipitates, whilst As leachability remained at a low level both before (As; 0.33 ± 0.08 mg/L) and after (As; 0.59 ± 0.08 mg/L) the bioscorodite crystallization reaction when biological seeds were used (Table 2).Lowering initial pH to 1.2 using bioscorodite seeds slightly improved the product stability (As; 0.20 ± 0.01 mg/L), compared to that at pH 1.5 (As; 0.59 ± 0.20 mg/L; Table 2), although the final As removal was lower at pH 1.2 (Figure 3a).Overall, by comparing results at pH 1.5, it can be said that the use of bioscorodite seeds is advantageous from the following viewpoints; (i) significantly improved final As removal (Figure 3a) (at least partly due to their larger specific surface area; Table 1); (ii) higher product stability (lower TCLP leachability of As); and (iii) hollows of bioscorodite particles can be filled by fresh precipitates.
3.2.Optimal Fe(II)/As(III) Molar Ratios for As Removal as Bioscorodite from a Range of Dilute As(III)-Bearing Solutions Following the above findings that feeding bioscorodite seeds and setting initial pH to 1.5 have favorable effects to improve bioscorodite formation from dilute As(III) solution (4.7 mM), a range of initial As(III) concentrations (3.3-20 mM) were tested to find their respective optimal Fe(II)/As(III) molar ratio under these conditions.Feeding (even at 0.15%) of bioscorodite seeds was found significantly effective especially when targeting more dilute [As(III)] ini of 3.3 mM (95% or 11% As removal with or without seeds, respectively at [Fe(II)] ini /[As(III)] ini = 1.4;Table 3), and of 4.7 mM (94% or 15% As removal with or without seeds, respectively at [Fe(II)] ini /[As(III)] ini = 1.4;Table 3).The optimal [Fe(II)] ini /[As(III)] ini ratio was generally 1.4-2.0 at all [As(III)] ini tested, enabling 94-99% As removal as scorodite (although starting at lower [As(III)] ini required longer incubation time), of which, providing an excess Fe(II) (closer to [Fe(II)] ini /[As(III)] ini = 2.0) was found beneficial to improve the final As removal (up to 98-99%) especially from more dilute As(III) solutions (Table 3).Setting [Fe(II)] ini /[As(III)] ini = 1.0 (theoretical Fe/As molar ratio of scorodite) produced scorodite but with a decreased final As removal.On the other hand, setting [Fe(II)] ini /[As(III)] ini ≥ 2.5-3.0 led to formation of unwanted amorphous precipitates or jarosite at all initial As(III) concentrations tested (Table 3).In abiotic hydrothermal/atmospheric pressure methods starting with high concentrations of oxidized species, the presence of excess Fe(III) (e.g., [Fe(III)] ini /[As(V)] ini > 1.5) generally negatively affected the precipitation kinetics of scorodite and, thus, the ratios of 1.0-1.5 are used [3,7,25].However, when treating these dilute, reduced species of As(III) and Fe(II) as starting materials microbiologically, it was found that Fe(III) should be added to an slight excess of 1.4 ≤ [Fe(III)] ini /[As(V)] ini ≤ 2.0 for the best performance.Chemical composition of the resultant bioscorodite products ([Fe] im /[As] im ratio) generally became closer to 1.0 (the theoretical ratio for scorodite) at higher initial As(III) concentrations (Table 3).Slightly higher [Fe] im /[As] im ratios (1.2-1.5)observed at lower initial As(III) concentrations may be due to incorporation of SO 4 2-, as was reported in abiotic hydrothermal [25][26][27][28] and atmospheric scorodite synthesis [7].

•
Contrasting bioscorodite crystallization processes on bio-and chemical-scorodite seeds were found.Hollow bioscorodite seed particles became increasingly filled with newly formed scorodite, whilst solid chemical seeds induced their surface to be thoroughly coated with new scorodite precipitates.

•
Lowering the initial pH from 1.5 to 1.2 (using bioscorodite seeds) led to a steady and continuous formation of bioscorodite particles (instead of two-stage As precipitation at pH 1.5).However, As removal remained relatively incomplete at pH 1.2 (91%) compared to at pH 1.5 (98%).

•
Optimal conditions found for the bioscorodite method enabling treatment of dilute As(III) concentrations did not necessarily match those found for chemical scorodite syntheses targeting high As(V) concentrations.

•
For possible practical application using the continuous process (supposedly composed of an aeration tank followed by a settling tank), recycling microbiologically active bioscorodite sludge could maintain both high cell density and seed pulp density to support steady As(III) and Fe(II) oxidation and the resultant scorodite crystallization.

Table 1 . 2 .
Particle size and specific surface area of bio-and chemical-scorodite seed crystals.Effect of Seed Feeding Seeding effects of bio-and chemical-scorodite particles were compared at [As(III)]ini = 4.7 mM, [Fe(II)]ini = 9.5 mM, and pH 1.5.

Table 1 .
Particle size and specific surface area of bio-and chemical-scorodite seed crystals.

Table 2 .
TCLP test results for original scorodite seed crystals (bio-and chemical-) and final bioscorodite products.