Synthesis and thermodynamic properties of arsenate and sulfate-arsenate ettringite structure phases

Arsenic is a toxic and carcinogenic contaminant of potential concern. Ettringite [Ca6Al2(SO4)3(OH)12·26H2O] has the ability to incorporate oxyanions as a solid solution with SO42−, which could lower the soluble oxyanion concentrations. Therefore, ettringite containing SO42− and AsO43− has been synthesized. Results indicated that AsO43− could substitute for SO42− inside the channels of ettringite in the form of HAsO42−, and a linear correlation existed between Xinitial solution and Xsolid. The thermodynamic characterization of the solid samples was investigated by means of Visual MINTEQ, a freeware chemical equilibrium model, and the solubility product logK of -48.4 ± 0.4 was calculated for HAsO4–ettringite at 25°C. The Lippmann phase diagram and XHAsO4–XHAsO4,aq plot showed that the solid solution series containing arsenate has HAsO4-poor aqueous solutions in equilibrium. These findings can be helpful to arsenate solidification and arsenate leaching modeling projects.


Introduction
Arsenic (As) is known to be toxic, carcinogenic, and possibly teratogenic to humans [1][2][3]. In nature, As is released to the environment through volcanism and weathering. As is also produced by anthropogenic activities, such as mineral processing and melting, coal combustion, and extensive use of As-containing compounds, such as wood preservatives, desiccants, and herbicides, resulting in high concentrations of As in water and soil [4].
However, to date there are no investigations focusing on the solid composition analysis and thermodynamic data of AsO 4 /SO 4 -ettringite solid solutions. The purpose of the present study is to investigate the changes in solid-phase characteristics, solid composition, and solubility resulting from AsO 4 3− substitutions in the ettringite structure. The results of this investigation will be helpful in modeling the potential for ettringite to control AsO 4 2− concentrations.

Materials
All chemicals used in this study were at least of pro-analytical grade. The following substances were used: CaCO 3 powder, NaAlO 2 powder, Na 2 SO 4 , Na 3 AsO 4 Á12H 2 O, sucrose, and HNO 3 .
Deionized water used in this study was boiled, followed by cooling under soda lime in a N 2 (g)-filled glove box to eliminate CO 2 (aq). All polyethylene bottles, tubes, and glassware were soaked in acid solution (5% HNO 3 ) for at least 24 h and rinsed using ultrapure water three times prior to each experiment. All handling of materials, the solid solution synthesis, the sample filtration, the sample drying, and the pH measurements were conducted in a glovebox filled with N 2 to prevent possible CO 2 contamination.

Solid solution synthesis and experiments
The solid solution was synthetized on the basis of Hassett and McCarthy's "modified saccharate method" [17]. Initially, a solution containing a mixture of NaAlO 2 , Na 2 SO 4 , and Na 3 A-sO 4 Á12H 2 O was prepared with a range of arsenate/sulfate ratios (the total number of moles of SO 4 and AsO 4 was constant). A soluble calcium complex prepared by dissolving CaO in a 10% sucrose solution was added slowly (4 mL/min to 7 mL/min) to the mixed solution while stirring. A twofold excess of NaAlO 2 was used to prevent the precipitation of calcium arsenate or sulfate, which would be difficult to separate from the ettringite solid solution. The liquid-tosolid ratio was 20 mL/g. The sample was stirred for 12 h and equilibrated in a thermostatic oscillator for 48 h at a constant temperature of 25˚C. Afterward, the sample was centrifuged for 15 min at 4,500 rpm. The supernatant was filtered using 0.45 μm nylon membrane filters following pH measurement. The solution was stored at 4˚C for analyzing for elemental composition after being acidified with concentrated HNO 3 . The solid phase was washed with acetone after being filtered, and was dried in a desiccator.

Characterization of the solid phase
After drying, part of the solid phase was ground with an agate mortar to <60 μm for analysis by X-ray diffraction (XRD), infrared (IR) analysis, and chemical analysis. XRD analysis was used to determine the purity and crystallinity of the phases, and data were collected on an X'Pert PRO Polycrystalline X-ray Diffractometer from PANalytical Company using Cu Ka radiation. The diffraction scans ranged from 5˚to 60˚2θ with a step interval of 0.0263˚2θ and a counting time of 4 s/step. The IR spectra were recorded on a Thermo Nicolet Nexus Series using potassium bromide pellets in the range of 4,000 cm −1 to 400 cm −1 with a resolution of 2 cm −1 to confirm arsenate in the samples. The morphology of the samples was determined by scanning electron microscopy (SEM) using a ZEISS SIGMA 500 equipped with a Bruker Quantax EDS detector, which can also provide the information on the surface composition.

Chemical analysis
Part of powder sample was dissolved in a 1% solution of HNO 3 . Then, solid stoichiometry was determined using ICP-OES for calcium, sodium, aluminum, and arsenic and using ion chromatography for sulfur.

Solubility study and geochemical model
Finely ground synthetic solid samples were mixed with distilled deionized water (1:10 (w/v)) and equilibrated in a shaker bath for 100 days at 25 ± 1˚C. The supernatant was isolated by centrifugation and passage through a 0.45 μm nylon membrane filter, and the concentrations of calcium, sodium, aluminum, arsenic, and sulfur in the filtrate were determined. Ionic species and their activities were calculated from the experimental values of ionic concentrations and pH values using Visual MINTEQ (Version 3.1).
Visual MINTEQ is a freeware chemical equilibrium model, which was developed from the DOS program MINTEQA2 and originally coded by the US EPA. Visual MINTEQ can calculate the speciation of inorganic ions and complexes in water [18]. And the databases used by Visual MINTEQ (Version 3.1) included pertinent and updated thermodynamic data from available literature (Table 1).

Results and discussion
Solid phases of the solid solution series Table 2 lists the chemical composition of the solid solution series, which shows that the solids had relatively constant Ca and Al ratios at variable SO 4 :AsO 4 ratios in the initial solution. The ideal stoichiometry of ettringite is 6Ca:2Al:3SO 4 . However, small deviations from the ideal stoichiometry were observed, which might have resulted from the synthesis method and the amount of water present [25]. Nevertheless, the Ca/Al ratio in the solid with no sulfur was 4.7, and far greater than the ideal value 3, which occurred in despite of the two-fold excess of soluble Al in the synthesis solution. This could be the result of the production of Ca phases (such as portlandite, calcite or calcium-arsenic compound). Furthermore, the molar ratio of AsO 4 to (AsO 4 + SO 4 ) in the initial solution (which is represented by X initial solution ) was not obtained in the solid. At low molar ratio (X initial solution < 0.3), the solids were more enriched in AsO 4 than the original solution, whereas the solids had more SO 4 at high molar ratio (X initial solution > 0.3). Linear regression analysis of the data of the molar ratio of AsO 4 to (AsO 4 + SO 4 ) in solids (which is represented by X solid ) and the initial solution led to the following equation: X solid = 0.7858X initial solution + 0.0409 with a coefficient of correlation of 0.994. This equation can be used to approximately estimate the solution mix ratio of SO 4 to AsO 4 required to prepare a particular solid solution composition.
Sharp peaks in the X-ray diffractograms (Fig 1) of the synthesis products indicated good crystallinity, and ettringite was the only stable crystalline phase except the sample h. Although the low intensities of the diffraction peaks are obtained in the pure As-sample (Fig 1h), the ettringite phase clearly can be recognized. The peaks showed slight shifts to higher angles (smaller d-spacing) with substitution of AsO 4 for SO 4 (including the sample h). This decrease in basal spacing implied that the interchannel for sulfate and water molecules was compressed after AsO 4 uptake, which can be attributed to the intercalation of an arsenate anion in exchange for sulfate and the displacement of the ordered water molecules [26,27]. The solids with high arsenic content (X solid > 0.5) evidently had lower intensities of the diffraction peaks, which indicated lower crystallinity of samples. The peak intensity (around 9.9˚) decreased with increasing As content in the solid solution series, and disappear when there was no sulfur.
That was due to the elevated electron density in the structure, which resulted from the substitution of SO 4 by AsO 4 . A new peak (around 12.6˚) appeared when As contents in the samples were higher than 0.65. This could be attributed to CaAl 2 O 4 Á10H 2 O. And there was another new peak at 30.4˚in sample h, which was due to Ca 3 (AsO 4 ) 2 Á10H 2 O. The peak intensity around 37.3˚rapidly decreased with As increasing in solid samples, which may be attributed to As substitute for partial H 2 O in the channel of ettringite. And the rapid decrease of the peak (37.3˚) caused a false appearance of peak split in f, g and h.
The poorer crystalline solid with increasing As content in the samples could also be observed through the SEM micrographs of the samples (Fig 2). The club-shaped ettringite particles were approximately 5 μm to 12 μm in length when arsenic was absent. However, the grain length of ettringite decreased to <1 μm with increases in solid-phase arsenic concentration, and the  length-to-diameter ratio was lower. Typical ettringite rods dominated when As contents in samples were lown(<0.3), while some irregular particles attached to the rods with short length at higher As content. The irregular particles also contained the same elements with the prisms, which were showed from EDS mapping (Fig 3). Oxyanion speciation could be confirmed by Fourier transform infrared (FTIR) spectroscopy. Fig 4 shows the spectra of the solid solution series of AsO 4 -SO 4 -ettringite. The assignments of vibration are shown in Table 3. As indicated by the peak intensities, the SO 4   concentration in the solid solutions decreased with the increase in the solid phase AsO 4 concentration. The FTIR spectra of solid solutions exhibited an As-O stretching peak at 855 cm −1 , which was higher than 810 cm −1 reported by Siebert H. [28]. AsO 4 3was protonated to form HAsO 4 2− , which caused the As-OH symmetric stretch to shift to higher wavenumbers [29]. Thus, AsO 4 3− substituted SO 4 2− inside the channels in the form of HAsO 4 2− , which was also   [30,31], which could indicate carbonate contamination, but in less than 5%, because no compound containing carbonate was observed from the X-ray diffractograms.

Solid solution solubility products
The ion concentrations in solution of the dissolution experiments were shown in Table 4. And the Ca:Al:SO 4 :AsO 4 ratios in solution were different from those of the solid, suggesting nonstoichiometric dissolution. This was expected, since the presence of secondary phases (Ca-As compound, gypsum and Al-hydroxide) leaded to reduction of dissolved Ca, Al, SO4 and AsO4 concentrations in solution.
FTIR spectroscopy analysis of the solid samples and chemical analysis of the liquid phase revealed that AsO 4 3− substituted for SO 4 2− inside the channels in the form of HAsO 4 2− . Thus, HAsO 4 2− will be used for the thermodynamic study of the solid solution series. The solubility products of the solid solution series was calculated according to the following reaction: The curly brackets {} denote aqueous activities. The Ksp calculated are shown in Fig 5. The calculated AsO 4 -ettringite and SO 4 -ettringite solubility products change as a function of X HAsO4 for mixed phases, with a linear correlation between logKsp and X HAsO4 . This finding indicates that solid solutions exist. Solubility calculations of all precipitation and dissolution experiments resulted in a mean logK As-ettringite = -48.4 ± 0.4 and K SO4-ettringite = -43.9 ± 0.6. The solubility of SO 4 -ettringite in the current study was higher than that determined by Barbara and Thomas (-44.9) [14,32], which could be due to a small CO 2 amount in the system that lowers the pH. Certainly other factors, such as the choice of the activity coefficient model, analytical errors, and the presence of other complexes not included in the activity calculations, may affect estimates based on minimizing the variance in Ksp.
The aqueous solubility of the binary solid solution system HAsO 4 -ettringite and SO 4ettringite could be predicted by plotting the Lippmann's solidus and solutus relationships on the ordinate against two superimposed scales, i.e., X HAsO4 and X HAsO4,aq , on the abscissa, which could provide the solid-phase and aqueous-phase compositions for a series of possible thermodynamic equilibrium states [33][34][35]. In this case, the solidus and solutus equations can be expressed as follows: and SP eq ¼ 1 where the curly brackets {} denote aqueous activities; X HAsO4 and X SO4 are the mole fractions of HAsO 4 and SO 4 (X SO4 + X HAsO4 = 1) in the solid, respectively; X HAsO4,aq and X SO4,aq are the activity fractions of HAsO 4 2− and SO 4 2− ions in the aqueous solution, respectively; K HAsO4 and Properties of arsenate and sulfate-arsenate ettringite structure phases K SO4 are the solubility products of pure HAsO 4 -ettringite and SO 4 -ettringite, respectively; and γ HAsO4 and γ SO4 are solid activity coefficients. The solid-phase activity coefficients determined from the modified Guggenheim regular excess free energy model [36]can be expressed as follows: Regular nonideal solid solutions only need one Guggenheim fitting parameter (a 0 ), which was determined by the MBSSAS code [33,37]. A value of a 0 = 1.69 was obtained. Thus, Eqs 5 and 6 can be simplified as follows: The Lippmann diagram for this system at 25˚C is shown in Fig 6. The calculated SP of the solid solution series fits best to the nonideal model. The pure HAsO 4 -ettringite and SO 4 -ettringite endmember solubility products in this system differ by four orders of magnitude. As a result, a strong preferential distribution of the less soluble endmember toward the solid phase was observed. This expression (Eq 9) can be used to construct a X HAsO4 -X HAsO4,aq plot (Fig 7), which describes the coexisting compositions of solid and aqueous solutions under equilibrium conditions. X HAsO 4 ¼ K SO 4 X SO 4 g HAsO 4;aq ðK SO 4 g SO 4 À K HAsO 4 g HAsO 4 ÞX HAsO 4;aq þ K SO4 g SO 4 ð9Þ Fig 7 shows that the X HAsO4 -X HAsO4,aq curve approximates to two straight lines forming a right angle, which implies that HAsO 4 -poor aqueous solutions are in equilibrium with HAsO 4 -rich solid phases in a wide X HAsO4 range.