Environ. Sci. Technol. 2002, 36, 493-500

Arsenic(III) Oxidation by Birnessite and Precipitation of Manganese(II) Arsenate CHRISTOPHE TOURNASSAT,† L A U R E N T C H A R L E T , * ,† DIRK BOSBACH,‡ AND ALAIN MANCEAU† LGIT-CNRS/UJF, Universite´ Joseph Fourier (Grenoble), P.O. Box 53, F 38041 Grenoble, France, and Institut fu ¨r Nukleare Entsorgung (INE), Forschungszentrum Karlsruhe, P.O. Box 3640, 76021 Karlsruhe, Germany

Solution chemical techniques were used to investigate the oxidation of As(III) to As(V) in 0.011 M arsenite suspension of well-crystallized hexagonal birnessite (H-birnessite, 2.7 g L-1) at pH 5. Products of the reaction were studied by scanning electron microscopy coupled with energy dispersive spectroscopy (SEM-EDS), atomic force microscopy (AFM), and X-ray absorption near-edge structure spectroscopy (XANES). In the initial stage (first 74 h), chemical results have been interpreted quantitatively, and the reaction is shown to proceed in two steps as suggested by previous authors: 2>MnIVO2 + H3AsO3 + H2O f 2>MnIIIOOH + H2AsO4- + H+ and 2>MnIIIOOH + H3AsO3 + 3H+ f 2Mn2+ + H2AsO4- + 2H2O. The As(III) depletion rate was lower (0.02 h-1) than measured in previous studies because of the high crystallinity of the H-birnessite sample used in this study. The surface reaction sites are likely located on the edges of H-birnessite layers rather than on the basal planes. The ion activity product of Mn(II) and As(V) reached after 74 h reaction time was the solubility product of a protonated manganese arsenate, having a chemical composition close to that of krautite as identified by XANES and EDS. Krautite precipitation reaction can be written as follows: Mn2+ + H2AsO4- + H2O ) MnHAsO4‚ H2O + H+ log Ks ≈ -0.2. Equilibrium was reached after 400 h. The manganese arsenate precipitate formed long fibers that aggregated at the surface of H-birnessite. The oxidation reaction transforms a toxic species, As(III), to a less toxic aqueous species, which further precipitates with Mn2+ as a mixed As-Mn solid characterized by a low solubility product.

Introduction Manganese oxides play a distinctive role in superficial soil or near surface environments due to their narrow Eh-pH stability field. They are found in oxic soils, aquifers, and oceanic and aquatic systems (1, 2). They can be readily reductively dissolved by a variety of inorganic and organic compounds (3-5). Conversely, Mn2+ should be oxidized in most oxic waters on the basis of thermodynamic considerations, but it is not oxidized by O2 within several years (6) unless mineral surfaces (7, 8) or microorganisms (9, 10) catalyze this slow redox reaction. * Corresponding author e-mail: [email protected]; phone: +33 4 76 82 80 18; fax: +33 4 76 82 81 01. † Universite ´ Joseph Fourier (Grenoble). ‡ Forschungszentrum Karlsruhe. 10.1021/es0109500 CCC: $22.00 Published on Web 01/04/2002

 2002 American Chemical Society

FIGURE 1. X-ray powder diffraction pattern for the unreacted H-birnessite and structural model (23, 26). The structural formula is H+0,33Mn3+0,111Mn2+0,055(Mn4+0,722Mn3+0,11100,167)O2‚(H2O)0,50. Black dots represent OH groups around octahedral cavities. Morgan recognized such limitations in the chemical thermodynamics approach to dynamic natural systems (11). His broad contribution to aquatic chemistry includes kinetic studies of both the oxygenation of Mn2+ and the reductive dissolution of manganese oxides (e.g., refs 3 and 12-14). He further created the concept of oxidative capacity (OXC) of natural systems, which in the pE dimension is similar to that of buffer capacity in the pH dimension (15). These two achievements are of course linked together since manganese oxide is an important part of the OXC in natural waters. One of the chemicals that can reductively dissolve manganese oxides is arsenious acid (H3AsO30), a toxic soluble As(III) species that plagues a variety of water-drinking bodies in the world (16). The oxidation of As(III) by manganese oxides has been shown to occur in surface water, after mixing of oxic and anoxic lake waters (17). The reaction could lead to some decontamination of surface and drinking water bodies since the acute toxicity of As(V) species is lower than that of As(III) species (18) and since As(V) can be easily removed from solution by precipitation and adsorption on a variety of minerals (7, 19). However, detailed kinetic and spectroscopic data are needed to optimize the application of MnO2/ mineral mixtures as oxidants and sinks of arsenic in technical systems. Following the former investigation by Oscarson et al. (20), a number of studies have been devoted to the oxidation of As(III) by manganese oxides using either solution chemistry (21, 13) or spectroscopic tools (22), but none have combined the two types of technique. In this study, the kinetics of arsenic transformation by hexagonal birnessite (H-birnessite) is examined. Solution chemistry, microscopic and spectroscopic tools were used, but microscopic and spectroscopic investigations imposed some constraints: high As concentration (∼11 mM) and a great quantity of solid had to be used. This high As concentration is more representative of mine drainage water than of contaminated drinking water. In addition, the thorough chemical and morphological analysis of reaction products required the use of a well-crystallized birnessite whose chemical formula and structure are well-known. In contrast to hexagonal birnessite, natural phyllomanganates are extremely fine-grained and are devoid of three-dimensional ordering. The structural formula of hexagonal birnessite is H+0.33Mn3+0.111Mn2+0.055(Mn4+0.722Mn3+0.11100.167)O2‚ (H2O)0.50, where 0 is a Mn vacancy in the octahedral layer (23). A polyhedral representation of its structure is presented in Figure 1, in which black dots stand for OH groups around octahedral vacancies. These vacancies were shown to be the reactive sorption sites for Zn2+, Cr3+, and Co2+ (5, 24-26). VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

493

As(III) aqueous species (As(OH)30) may sorb either on these vacancies or on the edges of the octahedral Mn layers. The objective of the present study is to elucidate the reaction mechanism occurring in the oxidative precipitation of arsenic in the presence of birnessite. The morphology and As/Mn stoichiometry of the precipitate were determined by field emission scanning electron microscopy (FE-SEM), and the dimensions of the precipitate particles were obtained by atomic force microscopy (AFM). X-ray absorption near-edge structure (XANES) spectroscopy was used to determine the oxidation state of manganese and arsenic in the solid phase.

Materials and Methods Standards and Reagents. All solutions and suspensions were prepared with boiled Millipore Milli-Q 18 MΩ water, which was cooled under a stream of argon. HNO3 and HCl stock solutions were made from Titrisol solutions. Sodium arsenite salt and HNO3 solution were used to prepare the pH 5 As(III) stock solution. ICP standards for As, Mn, and Na were prepared by dilution of 1000 ppm Alpha and Merck standards. Analytical Method. Aqueous As(III) and As(V) species were separated by an anion-exchange resin (Dowex 1 × 8 anionexchange resin 100-200 mesh) according to Ficklin (27). Total As (i.e., [As(III)] + [As(V)]) and As(III) concentrations were determined analytically, and the As(V) concentration was obtained by difference. Solid sample aliquots were dissolved in 10 mL of hot 4 M HCl solution before analysis. Arsenic, manganese, and sodium concentrations were measured on a Perkin-Elmer Optima 3300 DV inductively coupled plasma atomic emission spectrometer (ICP-AES). Hexagonal Birnessite Preparation. Sodium birnessite was prepared according to Giovanoli’s method (28), modified as follows. Initial solutions of MnCl2 and NaOH were degassed with argon for 1 h. The final suspension obtained from the hydrolysis and oxidation of the Mn solution was stored 48 h at 110 °C in order to increase the crystallinity of birnessite. After the suspension was cooled, the solid was washed with deionized water and centrifuged 10 times to eliminate the excess in Na+ and Cl-. The Na-birnessite was subsequently transformed into H-birnessite at pH 5 according to Silvester et al.’s protocole (26). Equilibrium was reached after 20 days. The H-birnessite was then transferred into a 2.5-L glass reactor. The initial solid to liquid ratio, measured in triplicate, was equal to 2.73 ( 0.06 g L-1. This birnessite content was chosen in order to collect enough material for XANES spectroscopy. Solid Characterization. The unreacted H-birnessite was characterized with a Siemens D5000 X-ray powder diffractometer equipped with a Kevex Si(Li) solid-state detector and Cu KR radiation. The diffraction pattern (Figure 1) was found to be in agreement with that reported by Lanson et al. for the H-birnessite standard (23). Mn K-edge XANES spectra were recorded at the LURE synchrotron radiation laboratory (Orsay, France) on the EXAFS 1 station, and As K-edge XANES spectra were recorded at the European Synchrotron Radiation Facility (ESRF, Grenoble France) on beamline BM32. The spectra were calibrated by taking the maximum of the derivative of the Mn metal foil at 6539 eV and that of erythrite (Co3(AsO4)2‚8H2O) at 11867 eV. FE-SEM experiments were performed at the Institut fur Nukleare Entsorgung (INE, Forschungszentrum Karlsruhe, Germany) using a CamScan CS44FE apparatus. AFM experiments were carried out at INE using a Topometrix Explorer apparatus operating in tapping mode. Experimental Conditions. The experiment was carried out at 25.0 ( 0.5 °C under controlled argon atmosphere. The H-birnessite suspension was allowed to equilibrate at pH 5.0 ( 0.1 in a 0.03 M NaNO3 background ionic medium for 1 week before the introduction of As(III). The suspension pH was monitored with a pH-stat apparatus (Impulsomat 494

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

FIGURE 2. Experimental setup. At the beginning of the experiment, the volume of the suspension V ) 2.16 L, [As(III)] ) 0.0106 mol L-1, and content of H-birnessite ) 2.73 g L-1. At the end of the experiment, V ) 2.15 L.

FIGURE 3. Aqueous amount of As(III), As(V), Mn(II), H+, and As in the solid as a function of time. (A) Time interval: 0-600 h. (B) Time interval: 0-50 h. (Metrohm 614) and Dosimat (Metrohm 665)). Proton addition (HNO3 0.1 M) was controlled by a computer (Figure 2). At t ) 0, 2.28 × 10-2 mol of As(III) was added to the suspension, corresponding to an As/Mn molar ratio of 0.44 and to a 0.011 M arsenite concentration in water. Samples were collected with a plastic syringe and filtered through a 0.1-µm filter membrane (Durapore 0.1 µm VVLP). The first milliliter of the filtered supernatant was discarded to limit losses due to sorption on the filter membrane. This filtration step lasted less than 1 min. The filtered solution was then diluted prior to analysis. The solid collected on the filter was either dissolved for chemical analysis or lyophilized for spectroscopic analysis.

Results and Discussion Concentration of Dissolved Species versus Time. Figure 3 reports the changes in moles of aqueous As(III), As(V), and Mn(II) in the reactor and moles of added protons with time, consecutively to the addition of As(III) to the H-birnessite suspension. The depletion of As(III) from solution was quite

TABLE 1. As(III) Depletion Rates Reported in the Literature and Obtained in This Study as a Function of Initial Solution Parameters

this experimenta ref 21a MnAs1b (29) MnAs2b (29) MnAs9b (29) MnAs10b (29) ref 20a

kobs (h-1)

half-life (h)

pH

Astot/Mn

T (°C)

-0.02 -0.57 -4.2 -2.1 -2.3 -2.0 -2.1

35 1.2 0.17 0.33 0.30 0.35 0.33

5.0 7.5 4.0 4.0 5.85 6.8 7.0

0.44 0.48 0.07 0.11 0.04 0.08 0.08

25 25 25 25 25 25 25

a Rate calculated at the beginning of the reaction. the Scott thesis (29).

b

MnAsX refers to

slow, and As(III) was not completely oxidized even after 600 h of reaction time. The decrease in As(III)(aq) content occurred in three distinct steps, which are discussed below. The first 74 h were characterized by a rapid decrease in moles of As(III)(aq), which paralleled the increase in moles of As(V)(aq). From 74 to 400 h, As(III)(aq) depletion was slower, and arsenic started to be incorporated in the solid phase. Beyond 400 h, As(V)(aq) concentration remained approximately constant, while As(III)(aq) concentration progressively approached zero. Initial Reaction Step (0-74 h) and Particle Size Effect. In the first 74 h of reaction time, aqueous As(III) species were transformed to aqueous As(V) species. Since the only oxidant present in solution was H-birnessite and since at any time a very little amount of As was present in the solid phase, As(III) must have been oxidized by H-birnessite in a multistep oxidation reaction in which the release of As(V) in solution was faster than both the As(III) sorption and the electron-transfer reaction step at the H-birnessite surface. One mole of Mn2+ and 1 mol of H2AsO4- (the predominant As(V) species at pH 5, according to the Minteq database) were produced for every consumed mole of H3AsO30 (the predominant As(III) species at pH 5) and for nearly every mole of proton removed from solution. Therefore, the chemical reaction can be written as follows:

>MnIVO2 + H3AsO30 + H+ f Mn2+ + H2AsO4- + H2O (1) where >MnO2 is a Mn(IV) reactive H-birnessite surface functional group. The apparent first-order rate equation for the depletion of As(III) can be written as

-d[As(III)] ) kobs[As(III)] dt or after integration:

(

ln

[As(III)]t

)

[As(III)]0

) -kt

where [As(III)]t is the concentration of As(III) remaining in solution at time t, and kobs is the apparent first-order rate constant. The measured kobs value, 0.02 h-1, is 30-200 times lower than those reported previously in the literature (Table 1). This difference could arise from a difference in pH and As to Mn ratio. The influence of these parameters was studied by Oscarson et al. (20), Moore et al. (21), Scott (29), and Scott and Morgan (13). An increase in pH led to a decrease in As(III) depletion rate: kobs was found to decrease from 4.2 to 1.96 h-1 as pH was increased from 4.0 to 6.8 (29) (Table 1, experiments MnAs1 and MnAs10, respectively). Moore et

al.’s (21) experiment was conducted at the same As to Mn ratio and temperature as in the present study, but their kobs value was much larger (0.57 h-1) than ours, although they worked at pH 7.5. Therefore, differences in experimental conditions cannot account for our low kobs value, and it will be argued below that this difference is due to the higher crystallinity of our birnessite sample. In previous studies (13, 20, 21, 29), birnessite was prepared following the method of McKenzie (30), which yields a disordered birnessite. For example, the X-ray diffraction pattern of the birnessite synthesized by Scott (29) displays hk bands characteristic of a turbostratic stack of layers (31). In contrast, Mn layers are hexagonally stacked in our birnessite sample as attested by the presence of hkl reflections in its diffraction pattern (Figure 1). The better crystallinity of the H-birnessite in the c* direction is also apparent from the full width at half-maximum (fwhm) of the (001) and (002) reflections, which are equal to fwhm001 ) 0.3° and fwhm002 ) 0.2° in H-birnessite, compared to fwhm001 ≈ 2.5° and fwhm002 ≈ 2.5° in the Scott sample (29) and to fwhm001 ≈ 2° and fwhm002 ≈ 2° in the Moore et al. sample (21). Although the lateral dimensions of the birnessite particles synthesized by the McKenzie’s method (30) are unknown, we can logically infer from their intrinsic disorder in the c* direction that these particles also had a smaller lateral size than H-birnessite. Higher crystallinity usually implies a lower density of reactive sites. According to reaction 1, the kinetic law should be expressed as

d[As(III)]t ) k[As(III)]t[H+][sites] ) kobs[As(III)]t dt with kobs ) k[H+][sites] Accordingly, the 2 orders of magnitude difference between the kobs values obtained in this study and by Scott and Morgan (13) suggests that the density of As reactive sites is about 100 times less in H-birnessite. If As(III) sorbed in the interlayer space of birnessite layers, one would expect little dependency of the As(III) oxidation rate with crystallinity because the density of vacancies is independent of the lateral particle size. The reverse is true for edge sites because their density tremendously increases when the lateral dimension of birnessite platelets decreases. Therefore, the observed difference in kobs values suggests that arsenic does not sorb in the interlayer region of birnessite, as do Zn, Cr(III), and Co(II) (24-26), but rather sorb on the edges of birnessite layers. Experimental support for this finding comes from FE-SEM observations, which showed that the size and the edge sharpness of birnessite platelets decreased from 0 to 6 h of reaction time (Figure 4). Therefore, the reductive dissolution of birnessite particles likely took place on particle edges and did not proceed by pits formation on the basal surfaces. It will be shown now that this structural interpretation is also qualitatively consistent with differences in measured kobs values. The ratio of H-birnessite to McKenzie’s birnessite edge surface areas (SHBi/SMcK) can be estimated from the length of particles. If birnessite particles are modeled by as L × L × H parallelepipeds, the edge surface area ratio (Sedge1/ Sedge2) for two birnessite samples of identical weight is equal to their basal length ratio, L2/L1. In the present study, H-birnessite particles were 1-2 µm long (Figure 4a), while those of Oscarson et al. (20) were less than 0.1 µm long and formed aggregates of about 0.2-0.5 µm. Aggregates of similar size were observed by Scott (29). Therefore, LHBi/LMcK can be roughly estimated to 10-20 and, hence, SHBi/SMcK ∼ 0.050.1. This reduction of the edge surface area in H-birnessite is commensurable with the reduction in kobs value. Initial Step (0-74 h) and Oxidation Mechanism. To quantify the transfer between species (Sp) in solution (Sp ) VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

495

FIGURE 4. SEM images of the As(III)-reacted H-birnessite particles. (A) 1-2 µm H-birnessite particles just after adding As(III) in suspension. (B) Particles after 6 h of reaction. Particles are smaller and edges are smoother than initially. (C) Particles after 162 h of reaction with krautite-like fibers. H-birnessite particles and krautite-like fibers are aggregated. (D) Isolated fiber on the edge of an aggregate. (E) Krautite-like fibers at the end of the experiment. (F) AFM image of fibers. (G) Topographic transect from the previous image. Width and height values are in agreement with SEM scale. Mn2+, As(III), or As(V)), in the solid (Sp ) Mn(II), Mn(III), Mn(IV) or As), together with the amount of added H+, these species are expressed in moles rather than in concentration. 496

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

∆Sp is the difference between the moles of Sp in the reactor at t and at t ) 0. A detailed examination of Figure 3B shows a discrepancy between experimental and predicted ∆H+ and

∆As(V) calculated according to reaction 1, specifically in the very beginning of the reaction. Indeed, whereas ∆Mn2+/∆t and ∆As(V)/∆t are equal within experimental error, ∆H+/∆t and ∆Mn2+/∆t are not (Figure 3B). Scott and Morgan (13) found also that Mn2+ was released in solution with some delay relative to As(V). A preliminary interpretation of this phenomenon was warranted by Nesbitt et al. (22) using X-ray photoelectron spectroscopy (XPS). They suggested a twostep reaction mechanism:

2>MnIVO2 + H3AsO3 + H2O f

2>MnIIIOOH + H2AsO4- + H+ (2)

2>MnIIIOOH + H3AsO3 + 3H+ f

2Mn2+ + H2AsO4- + 2H2O (3)

where >MnIIIOOH is a Mn(III) H-birnessite reactive site. On the basis of our results, this reaction scheme can be quantitatively evaluated in the following way. The net release and consumption of Mn2+ and H+ will be assumed to be only due to reactions 2 and 3. Arsenic present in the solid phase will be neglected during the first 74 h as it has been shown to account for less than 1% of the total arsenic mass balance. Net changes in aqueous moles of As(III) and As(V), denoted as ∆As(III) and ∆As(V), can then be calculated according to

∆Mn2+ ) 2∆As(V)(3) ) -2/3∆H+(3) ) -2∆As(III)(3) ) -∆>MnIIIOOH(3) (A) ∆H+(2) ) ∆As(V)(2) ) 1/2∆>MnIIIOOH(2) ) -∆As(III)(2) ) 1

IV

- /2∆>Mn O2 (B) with

∆H+ ) ∆H+(2) + ∆H+(3)

(C)

∆As(III) ) ∆As(III)(2) + ∆As(III)(3)

(D)

∆As(V) ) ∆As(V)(2) + ∆As(V)(3)

(E)

and

where the (2) and (3) subscripts refer to concentration changes due to the advancement of reactions 2 and 3, respectively. Combining eqs A-C, one obtains

∆As(V) ) 2∆Mn2+ + ∆H+

(F)

∆As(III) ) -2∆Mn2+ - ∆H+

(G)

Concentrations of As(III) and As(V) were computed according to eqs F and G on the basis of ∆H+ and ∆Mn2+. The computed and experimental ∆As(III) and ∆As(V) values now agree within experimental errors (Figure 5A). The model also allows obtaining the >MnIVO2 and >MnIIIOOH proportions in the solid phase:

∆>MnIIIOOH ) 2∆Mn2+ + 2∆H+

(H)

∆>MnIVO2 ) -3∆Mn2+ - 2∆H+

(I)

According to eqs H and I, the Mn(III)/Mn(IV) ratio in the solid phase increases rapidly in the first 2 h of the reaction (from 31% to 37%) and is then stable. Besides, the Mn(III)/ Mn(IV) slope parallels the As(V) slope (Figure 5A). This

FIGURE 5. Experimental amounts of aqueous species (points + error bars) compared with predicted amounts (solid lines, dotted lines ) incertitude on the calculation) based on the mass balance model described in the text. Mn(III), Mn(IV), and Mn(III)/Mn(IV) molar ratio in the solid phase were calculated with the same model. (A) Time interval: 0-100 h. (B) Time interval: 0-600 h. suggests that reaction 3, i.e., the oxidation of As(III) by Mn(III), is the rate-limiting step of reaction 1. This finding agrees with XPS results by Nesbitt et al. (22), who showed (reaction 2) the appearance of intermediate Mn(III) and hydroxyl groups. Intermediate Nucleation Step (74-162 h). After 100 h of reaction time, no more protons had to be added to maintain the pH constant. Mn2+ and As(V) concentrations reached their maximum at 162 h and then decreased gradually later on. In light of the data obtained during the 162-583-h period, the 74-162-h interval corresponds to the nucleation of a new solid. Manganese Arsenate Precipitation (162-583 h). From 162 to 583 h, the As(III) content in solution decreased with no net production of aqueous As(V) and Mn(II). Instead, As(V) and Mn(II) contents decreased at the same rate. These ions either coprecipitated or were uptaken by birnessite. The position in energy of XANES spectra and the shape of the edge maximum indicated that As was predominantly pentavalent in the solid phase (Figure 6). The As average oxidation state in the solid could not be determined accurately because of the lack of reference products with intermediate As oxidation states. Semiquantitative estimate indicated that the As(V) to total As ratio was greater than 90%. This value was confirmed by EXAFS spectroscopy (Tournassat et al., manuscript in preparation). Therefore, the precipitated solid was a manganese arsenate, and the As(III) removal from solution resulted from an oxidative precipitation reaction. Numerous manganese arsenate compounds have been reported in the literature with different As:Mn molar ratio, but Mn3(AsO4)2‚8H2O(s) is the only manganese arsenate compound for which a thermodynamic solubility product is available. It will be shown below that the precipitation of this manganese(II) arsenate was plausible on the basis of solubility thermodynamic considerations but did not fit the stoichiometry of the produced species. VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

497

FIGURE 7. Normalized Mn K-edge XANES spectra of the As(III)reacted H-birnessite for various times. FIGURE 6. Normalized As K-edge XANES spectra of the As(III)reacted H-birnessite for various times. NaAsO2 and MnAsO4 are the references for As(III) and for As(V), respectively.

TABLE 2. Log of Ion Activity Product (IAP) for Precipitation of Mn3(AsO4)2‚8H2O and Krautite as a Function of Sampling Timea ref 21

this study

t (h) 2 6 8 25 33 74 162 391 583

ref 29

log(IAP) Mn3(AsO4)2‚ log(IAP) t log(IAP) 8H2O krautite (h) krautite

t (h)

-2.0 -1.3 -1.1 -0.50 -0.36 -0.08 0.07 -0.14 -0.21

0.08 0.17 0.33 0.50 0.75 1.00 1.50 2.00 2.50

6.9 8.7 9.3 10.6 11.0 11.7 12.1 11.6 11.4

1 2 4 8 16 24 32 48 64

-0.91 -0.16 0.08 0.29 0.13 0.04 -0.07 -0.17 -0.13

MnAs1b MnAs10b log(IAP) log(IAP) krautite krautite -7.8 -7.0 -6.6 -6.5 -6.3 -6.3 -6.3 -6.3 -

-5.9 -4.9 -8.7 -7.8 -8.3 -8.0 -7.8 -7.6 -7.4

For comparison log(Ksadiq) ) 9.5 and log(Kminteq) ) 12.5 for the precipitation of Mn3(AsO4)2‚8H2O. No data are available for krautite. b MnAs1 and MnAs10 refer to the Scott thesis (29). a

A large range of solubility products have been reported for the manganese(II) arsenate solid. For the dissolution reaction

Mn3(AsO4)2‚8H2O(s) + 6H+ ) 3Mn2+ + 2H3AsO40 + 8H2O (5) Sadiq (19) reported an equilibrium constant of KsSadiq ) 109.5, whereas the Minteq and Wateqf4 thermodynamic databases yielded Ksdatabase ) 1012.5. The possible precipitation of Mn3(AsO4)2‚8H2O(s) during our experiment was assessed by computing with phreeqc2 the ion activity product IAP ) (Mn2+)3(H3AsO4)2/(H+)6 at each sampling time (Table 2). Log IAP increased from 0 to 12.1 between 0 and 162 h and then leveled off to 11.4 at 583 h. The concentration of solution species was thermodynamically consistent with the precipitation of Mn3(AsO4)2‚8H2O(s) according to Sadiq (19) as log IAP became greater than log(Ksadiq) at t ) 25 h. The final log IAP value (at 583 h) was reached only at t ) 74 h, meaning that the beginning of the precipitation may have occurred 498

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

actually from 74 h and not from 25 h. In this case, Mn(II) concentration in the solid phase is expected to have increased after the 74th hour. This assumption was confirmed by XANES spectroscopy. Figure 7 shows Mn K-edge spectra as function of reaction time together with the position of the edge maximum for Mn(IV) (6553 eV), Mn(III) (∼ 6558 eV), and Mn(II) (6562 eV) reported in the literature (10). A shift of the edge position toward lower energy and a concomitant apparition of a shoulder at 6562 eV were observed after 74 h, attesting for the apparition of Mn(II). The divalent manganese present in the solid phase necessarily originated from the solution as the solid phase was devoid of Mn(II) before 74 h. The possible Mn3(AsO4)2‚8H2O formation was further checked by mass balance computations according to

∆As(V)s(5) ) 2/3∆Mn(II)s(5) ) 1/3∆H+(5) ) -∆As(V)(5) ) -2/3∆Mn2+(5) (J) The inferred Mn/As stoichiometry is unrealistic because of the following: (i) The calculated amount of Mn(III) consumed exceeded largely the quantity of Mn(III) available in the solid (i.e., the sum of initial structural Mn(III) and Mn(III) produced by reaction 2). (ii) The calculated amount of Mn(IV) in the solid would be expected to increase with reaction time, whereas XANES spectroscopy showed a decrease in average Mn oxidation state. Other manganese(II) arsenate compounds with Mn/As ratio comprised between 1 (krautite; 32) and 3.5 (allactite; 33) may have precipitated. No solubility product is available for any of these compounds (K. Nordstrom, personal communication). The possible precipitation of one of these compounds was tested only on the basis of stoichiometry considerations using eqs A-J because it could not be tested on the basis of solubility considerations. The best fit to our experimental data was obtained assuming the formation of krautite (MnHAsO4‚H2O) (Figure 5B). However, 0.2H+ per As had to be added to the solid to obtain a good fit to experimental data. For this Mn/As stoichiometry, the Mn(IV) content in the solid should decrease and the Mn(III) content should remain constant in agreement with XPS (22). Further support for the precipitation of a krautite-like solid was obtained by SEM-EDS and AFM (Figure 4). Large 0.05-

surface functional groups, further decreasing the As(III) depletion rate.

Acknowledgments This research was partly funded by the Indo-French Center for the Promotion of Advanced Research (Project 2200-W1), by EGIDE Procope (Project 99151), and by the ACI “Eau et Environnement” from the CNRS. The LURE and ESRF are acknowledged for the provision of beamtime. FE-SEM and AFM facilities at the Institut fu ¨ r Nukleare Entsorgung of the Forschungszentrum Karlsruhe was used. Special thanks to M. Plaschke (INE) for his help with the AFM. FIGURE 8. EDS Spectrum of the isolated fiber presented in Figure 4D (arrow). The measured molar As/Mn ratio is 0.9 after ZAF correction. 0.2 µm wide fibers and agglomerated at the surface of birnessite were observed after 162 h. These fibers are longer than the initial platelets of birnessite, and the platelets are still recognizable (Figure 4). Therefore, one can reasonably reject the hypothesis that these fibers are the result of the dissolution of birnessite platelets. In addition, the EDS analysis (Figure 8) of one isolated fiber shown in Figure 4d indicated that the precipitate had a Mn to As ratio close to 1, in agreement with krautite chemical composition. The krautite precipitation/dissolution reaction can be written as

MnHAsO4‚H2O + H+ ) Mn2+ + H2AsO4- + H2O Kskrautite )

(Mn2+)(H2AsO4-) (H+)

(6)

The ion activity product for this reaction (IAPkrautite) has been calculated with phreeqc2 at each step in our experiment, as well as in Moore et al.’s (21) and Scott’s (29) experiments (Table 2). For our experiment, log IAPkrautite reached a maximum (0.07) at 162 h and then leveled off to -0.21 at the end of the experiment. Therefore, one can consider this value to be the solubility product of krautite (Kskrautite). Earlier, i.e., after 74 h, log IAPkrautite was greater than -0.21, and krautite nucleation may have occurred. In the experiment of Moore et al. (21), log IAPkrautite became larger than -0.21 after only 2 h because of a faster kinetics and a higher pH. In this case, precipitation may have occurred already after 1 h. After 4 h, they observed the formation of a gel having a Mn/As molar ratio of 1.05 ( 0.05, in agreement with the present study. Although Oscarson et al. (20) did not measure [Mn(II)]aq, they probably precipitated a krautite-like solid because their initial As(III) concentration and pH value were similar to those of Moore et al. (21). In contrast, Scott (29) did not observe any precipitate as expected from the low log IAPkrautite values reported in Table 2. They used a lower pH value in their MnAs1 experiment and a lower As concentration and solid to solution ratio in their MnAs10 experiment. The precipitation or not of a Mn(II)-As(V) solid phase provides a clue to the discrepancies in As(III) oxidation rates reported in the literature. Moore et al. (21) and Oscarson et al. (20) observed a two-step reaction kinetics. The first step, characterized by a fast As(III) depletion, lasted less than 1 h (i.e., up to the point where IAP reached the Kskrautite value). In the second step, the As(III) depletion was limited and lasted up to the disappearance of As(III) from solution. In contrast, no low As(III) depletion rate was observed by Scott and Morgan (13). Therefore, the existence of a slow As(III) depletion step can be correlated to the onset of a manganese(II) arsenate precipitation. If precipitation occurred on the edge of the birnessite platelets, it would obstruct the reactive

Literature Cited (1) ) McKenzie, R. M. In Minerals in Soil Environment; Dixon, J. B., Weed, S. B., Eds.; Soil Science Society of America: Madison, WI, 1977; pp 181-193. (2) Wehrli, B.; Friedl, G.; Manceau, A. In Aquatic Chemistry: Interfacial and Interspecies Processes; Huang, C. P., O’Melia, C., Morgan, J. J., Eds.; Advances in Chemistry 244; American Chemical Society: Washington, DC, 1995; pp 111-134. (3) Stone, A. T.; Morgan, J. J. Environ. Sci. Technol. 1984, 18, 450456. (4) Hering, J. G.; Stumm, W. In Mineral-Water Interface Geochemistry; Hochella, M. F., White, A. F., Eds.; Reviews in Mineralogy 23; Mineraological Society of America: Washington, DC, 1990; pp 427-465. (5) Silvester, E.; Charlet, L.; Manceau, A. J. Phys. Chem. 1995, 99, 16662-16669. (6) Diem, D.; Stumm, W. Geochim. Cosmochim. Acta 1984, 48, 15711573. (7) Davis, J. A.; Fuller, C. C.; Rea, B. A.; Claypool-Frey, R. G. WaterRock Interact. 1989, 187-189. (8) Junta, J. L.; Hochella, M., Jr. Geochim. Cosmochim. Acta 1994, 58, 4985-4999. (9) Tebo, B. M.; Ghiorse, W. C.; van Waasbergen, L. G.; Siering, P. L.; Caspi, R. Geomicrobiology: Interactions Between Microbes and Minerals; Banfield, J. F., Nealson, K. H., Eds.; Reviews in Mineralogy 35; Mineralogical Society of America: Washington, DC, 1997; pp 225-266. (10) Bargar, J. R.; Tebo, B. M.; Villinski, J. E. Geochim. Cosmochim. Acta 2000, 64, 2775-2778. (11) Morgan, J. J. In Equilibrium Concepts in Natural Water Systems; Stumm, W., Ed.; American Chemical Society: Washington, DC, 1967. (12) Stone, A. T.; Morgan, J. J. Environ. Sci. Technol. 1984, 18, 617624. (13) Scott, M. J.; Morgan, J. Environ. Sci. Technol. 1995, 29, 18981905. (14) Davies, S. H. R.; Morgan, J. J. J. Colloid Interface Sci. 1989, 129, 63-77. (15) Scott, M. J.; Morgan, J. J. Energetics and Conservative Properties of Redox Systems; Melchior, Basset, R. L., Eds.; ACS Symposium Series 419; American Chemical Society: Washington, DC, 1990. (16) Hering, J. G.; Elimelech, M. Presented at the American Water Works Association Annual Conference, 1995. (17) Kuhn, A.; Sigg, L. Limnol. Oceanogr. 1993, 38, 1052-1059. (18) Coddington, K. Toxicol. Environ. Chem. 1986, 11, 281-290. (19) Sadiq, M. Water Air Soil Pollut. 1995, 93, 117-136. (20) Oscarson, D. W.; Huang, P. M.; Liaw, W. K.; Hammer, U. T. Soil Sci. Soc. Am. J. 1983, 47, 644-648. (21) Moore, J.; Walker, J.; Hayes, T. Clays Clay Miner. 1990, 38, 549555. (22) Nesbitt, H. W.; Canning, G. W.; Bancroft, G. M. Geochim. Cosmochim. Acta 1998, 62, 2097-2110. (23) Lanson, B.; Drits, V. A.; Silvester, E. J.; Manceau, A. Am. Mineral. 2000, 85, 826-838. (24) Manceau, A.; Charlet, L. J. Colloid Interface Sci. 1992, 148, 443458. (25) Manceau, A.; Drits, V. A.; Silvester, E.; Bartoli, C.; Lanson, B. Am. Mineral. 1997, 82, 1150-1175. (26) Silvester, E.; Manceau, A.; Drits, V. A. Am. Mineral. 1997, 82, 962-978. (27) Ficklin, W. H. Talanta 1983, 30, 371-373. (28) Giovanoli, R.; Sta¨hli, E.; Feitknecht, W. Helv. Chim. Acta 1970, 53, 209-220. (29) Scott, M. J. In Kinetics of Adsorption and Redox Processes on Iron and Manganese Oxides: Reactions of As(III) and Se(IV) at VOL. 36, NO. 3, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

499

Goethite and Birnessite Surfaces; California Institute of Technology: Pasadena, CA, 1991. (30) McKenzie, R. M. Mineral. Mag. 1971, 38, 493-502. (31) Drits, V. A.; Silvester, E.; Gorskhov, A. I.; Manceau, A. Am. Mineral. 1997, 82, 946-961. (32) Catti, M.; Franchini-Angela, M. Am. Mineral. 1979, 64, 12481254.

500

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 3, 2002

(33) Moore, P. B. Am. Mineral. 1968, 53, 733-741.

Received for review May 8, 2001. Revised manuscript received September 24, 2001. Accepted October 2, 2001. ES0109500