Methyl Arsenic Adsorption and Desorption Behavior on Iron Oxides

Environmental Science & Technology 0 (proofing), .... Bong Kyun Kang , Byeong Seok Lim , Yeojoon Yoon , Sung Hoon Kwag , Won Kyu Park , Young Hyun ...
1 downloads 0 Views 611KB Size
Environ. Sci. Technol. 2005, 39, 2120-2127

Methyl Arsenic Adsorption and Desorption Behavior on Iron Oxides B. J. LAFFERTY† AND R. H. LOEPPERT* Department of Soil and Crop Sciences, Texas A&M University, College Station, Texas 77843-2474

In virtually all Earth surface environments, methylated forms of arsenic can be found. Because of the widespread distribution and toxicity of methyl-arsenic compounds, their adsorption by soil minerals is of considerable interest. The objective of this study was to compare the adsorption and desorption behavior of methylarsonic acid (MMAsV), methylarsonous acid (MMAsIII), dimethylarsinic acid (DMAsV), dimethylarsinous acid (DMAsIII), arsenate (iAsV), and arsenite (iAsIII) on iron oxide minerals (goethite and 2-line ferrihydrite) by means of adsorption isotherms, adsorption envelopes, and desorption envelopes (using sulfate and phosphate as desorbing ligands). MMAsIII and DMAsIII were not appreciably retained by goethite or ferrihydrite within the pH range of 3 to 11, while iAsIII was strongly adsorbed to both iron oxides. MMAsV and iAsV were adsorbed in higher amounts than DMAsV on goethite and ferrihydrite at all pH values studied. MMAsV and iAsV exhibited high adsorption affinities on both goethite and ferrihydrite from pH 3 to 10, while DMAsV was adsorbed only at pH values below 8 by ferrihydrite and below 7 by goethite. All arsenic compounds were desorbed more efficiently by phosphate than sulfate. MMAsV, iAsV, and DMAsV each exhibited adsorption characteristics suggesting specific adsorption on both goethite and ferrihydrite. Increased methyl substitution resulted in both decreased adsorbed arsenic at low arsenic concentrations in solution and increased ease of arsenic release from the iron oxide surface.

Introduction Arsenic is widely distributed throughout the environment as a result of both geological processes and anthropogenic activities (1). Applications of arsenic have included use in pesticides and herbicides, wood preservation, and manufacturing. Arsenic occurs as a component of many minerals and can be released into the environment by weathering processes. Arsenic has two predominant oxidation states in most environmental systems: AsIII and AsV. Although these oxidation states are positive, arsenic exists as an oxyanion in most aqueous systems. Redox potential and pH are the two main environmental factors that determine the equilibrium oxidation state of arsenic (2). For the arsenic oxyanions, pH is important in determining the protonation state and mo* Corresponding author: Soil & Crop Sciences Department, Texas A&M University, 2474 TAMU, College Station, TX 77843-2474. Phone: (979) 845-3663. Fax: (979) 845-0456. e-mail: r-loeppert@ tamu.edu † Present address: Department of Plant and Soil Sciences, University of Delaware, Newark, DE 19717-1303. 2120

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 7, 2005

TABLE 1. pKa Values for the Various Arsenic Species As species

pKa1

pKa2

pKa3

H3AsO4°(25) H2AsO3(CH3)° (26) HAsO2(CH3)2°(26) H3AsO3°(25)

2.20 4.19 6.14 9.22

6.97 8.77 12.13

11.53 13.4

lecular symmetry of arsenic. The pKa values of the various species of arsenic are summarized in Table 1. Arsenic also occurs as methylated forms and more complex organic compounds in environmental systems. The methylated monomeric arsenic species are: dimethylarsonic acid (H2AsO3(CH3), MMAsV), methylarsonous acid (H2AsO2CH3, MMAsIII), dimethylarsinic acid (HAsO2(CH3)2, DMAsV), dimethylarsinous acid (HAsO(CH3)2, DMAsIII), trimethylarsine oxide (AsO(CH3)3, TMAsOV), and trimethylarsine (As(CH3)3, TMAsIII). The methyl-arsenic species are similar to the inorganic arsenic species with respect to the relative stabilities of their oxidation states in environmental systems. That is, MMAsV and DMAsV are both stable in oxidized systems, while DMAsIII and MMAsIII are unstable under oxidizing conditions and are readily oxidized (3). Methylation can be carried out by a variety of organisms ranging from bacteria to fungi to mammals (1), and it occurs by alternating steps of arsenic reduction and oxidative methylation of the reduced arsenic species (1, 4). Inorganic arsenic species (i.e., arsenite, iAsIII; arsenate, iAsV) have long been known for their toxic properties. Increased incidences of cancer, along with other negative health effects, have been associated with long-term chronic exposure to arsenic (5). Recent data suggests that some methyl arsenic species (MMAsIII and DMAsIII) can be as toxic or more toxic than inorganic species (iAsIII or iAsV) (6-11). Because of the widespread distribution and toxicity of inorganic arsenic and methyl arsenic in the environment, the understanding of their adsorption behaviors on soil minerals is important in dealing with environmental systems. The arsenic binding sites in many soils are primarily provided by iron oxide minerals (12-15). Studies employing extended X-ray absorption fine-structure spectroscopy (EXAFS) as well as Fourier transform infrared (FTIR) spectroscopy have shown that iAsV and iAsIII both form primarily bi-nuclear bi-dentate complexes with the surface sites of iron oxides (16-19). The interaction between arsenic and iron-oxide surfaces is highly dependent on pH and the species of arsenic (20). Maximum adsorption of iAsV occurs on ferrihydrite within the pH range of 3.5-5.5, while the iAsIII adsorption maximum is between pH 8 and 10. These trends have been attributed to the variable charge characteristics of both the iron-oxide surface and the arsenic species. While considerable research has been conducted with inorganic arsenic and soil minerals, particularly iron oxides, there has been considerably less research on the interactions of methyl-arsenic compounds with soil constituents. MMAsV and DMAsV adsorption have been previously studied with several soil minerals (alumina, hematite, quartz, hydrous ferric oxide) (21-23). In these studies, it was observed that for MMAsV and DMAsV maximum adsorption occurs at low pH and decreases with increasing pH, which is similar to the adsorption behavior of iAsV. Adsorption of iAsV, MMAsV, and DMAsV on alumina and hydrous ferric oxide (HFO) was insensitive to changes in ionic strength, indicating that these arsenic species form inner-sphere complexes with these minerals (21). The desorption of iAsV, iAsIII, MMAsV, and 10.1021/es048701+ CCC: $30.25

 2005 American Chemical Society Published on Web 02/18/2005

DMAsV have been examined using amorphous iron oxide and goethite as adsorbents and phosphate (pH 3 and 7) and hydroxide as desorbing ions (24). It was determined that neither of the desorbates (phosphate or OH-) was effective for quantitatively removing any arsenic species from goethite or amorphous iron oxide. The first objective of the current research was to investigate the effects of chemical speciation, pH, arsenic concentration, and iron-oxide mineralogy on adsorption of arsenic. By comparing the adsorption of iAsV, iAsIII, MMAsV, MMAsIII, DMAsV, and DMAsIII, the effects of both the oxidation state of arsenic and the extent of arsenic methylation were examined. The second objective was to compare the ability of environmental ligands to desorb methyl-arsenic compounds from iron oxides across the pH range of 3 to 11.

Materials and Methods Materials. The inorganic arsenic compounds (iAsV and iAsIII) were obtained from Alpha Aesar (Ward Hill, MA) as oxides (As2O5 and As2O3, respectively). MMAsV was obtained as a sodium salt (monosodium acid methane-arsonate sesquihydrate, CH4AsNaO3), and DMAsV was obtained in the acid form (dimethylarsenic acid, C2H7AsO2), both from Chem Service (West Chester, PA). MMAsIII and DMAsIII were synthesized by Dr. William R. Cullen (University of British Columbia) as iodides (CH3AsI2 and C2H6AsI, respectively) following the procedures outlined by Burrows (25, 26). All other chemicals used in this study were analytical grade, and were obtained from Fisher (Pittsburgh, PA). Goethite and 2-line ferrihydrite (hereafter referred to as ferrihydrite) were prepared according to the methods outlined by Schwertmann and Cornell (27). These methods were slightly modified by using sodium hydroxide (NaOH) instead of potassium hydroxide (KOH) to hydrolyze ferric nitrate (Fe(NO3)3), in the preparation of both goethite and ferrihydrite. Samples for subsequent arsenic adsorption and desorption experiments were dialyzed to remove excess salt, diluted to 1-gFe (as oxide) L-1 final volume, and stored at 2 °C. The mineralogy of separate freeze-dried samples was verified by powder X-ray diffraction (XRD) analysis. Point of zero charge (pzc) was determined from the intercept of batch acid-base titration curves of goethite and ferrihydrite obtained in 0.001, 0.01, and 0.1 M NaCl ionic-strength buffers. Adsorption capacities of goethite and ferrihydrite were determined by means of iAsV adsorption isotherms and adsorption maxima from linear transformations of the Langmuir function (discussed below). Adsorption Isotherms. Adsorption isotherms were obtained as batch experiments. Arsenic concentrations ranged from 0 to 150 mg L-1 for ferrihydrite, and 0-38 mg L-1 for goethite. The concentration of iron as iron oxide was 0.44 gFe L-1 for both adsorbents, constituting As:Fe molar ratios ranging from 0 to 0.05 and 0 to 0.22 for goethite and ferrihydrite, respectively. Reactions were conducted in 0.044 M NaNO3 (final concentration) ionic strength buffer. Individual isotherms were obtained for each arsenic-compound and iron-oxide combination at pH 4 and 7 by adjusting the pH of samples individually by the addition of NaOH and nitric acid (HNO3) and bringing the final volume of each sample to 22.5 mL. The samples were allowed to equilibrate for 12 h on a rotary platform shaker, and then centrifuged at 15000g filtered through a 0.22 µm nominal pore-size membrane filter, and analyzed by flow-injection hydridegeneration flame-atomic-absorption spectrometry (FI-HGFAAS). Adsorption isotherms of all AsV compounds were also transformed using the linear form of the Langmuir equation [C/q ) (1/KLb) + (C/b)], where C is the equilibrium concentration of the particular arsenic species in solution in mmolAs L-1, q is the concentration of arsenic adsorbed by

the iron oxide in mmolAs mmolFe1-, b is the calculated adsorption maximum in mmolAs mmolFe1-, and KL is the calculated Langmuir adsorption constant in L mmolAs1-. The parameters of this equation were used to further evaluate adsorption behavior. Adsorption Envelopes. Adsorption envelopes were obtained in batch experiments by varying pH (pH 3 -10) while keeping the ionic strength (0.044 M NaNO3) and the concentrations of arsenic (0.67 mgAs L-1 in goethite systems and 13.3 mgAs L-1 in ferrihydrite systems) and iron (0.44 gFe L-1) as iron oxide constant. The pH of each sample was adjusted by the addition of NaOH and HNO3 and brought to a uniform final volume of 22.5 mL. Samples were allowed to react for 12 h on a rotary shaker, and then centrifuged. The pH of each sample was then measured, and samples were subsequently filtered through a 0.22 µm nominal poresize membrane filter, and analyzed by FI-HG-FAAS. Desorption Envelopes. Desorption envelopes were used to evaluate the influence of phosphate and sulfate on desorption of arsenic across a range of pH values. The concentrations of As and Fe in these experiments were selected to match the As: Fe molar ratio used in adsorption envelope experiments. Experiments were conducted in two stages: the adsorption stage and the desorption stage. The adsorption stage was conducted in bulk at pH 4.5, with As concentrations of 30 mgAs L-1 for ferrihydrite (0.02 As: Fe molar ratio) and 1.5 mgAs L-1 for goethite (0.001 As: Fe molar ratio), constant ionic strength (0.044 M NaNO3), and Fe oxide concentration of 1.0 gFe L-1. Each arsenic species was allowed to react with goethite or ferrihydrite for 12 h during the adsorption phase of the experiment. After the initial 12 h reaction, arsenic in solution was measured by FI-HG-FAAS to verify quantitative adsorption of each species. The desorption stage was accomplished as a batch experiment by adding either sodium phosphate or sodium sulfate solutions (0.1 M final concentration) to an aliquot from the adsorption stage. The addition of phosphate or sulfate resulted in As:S(or P):Fe molar ratios of 0.022:11.2:1.0 (for ferrihydrite) and 0.0011:11.2:1.0 (for goethite). These ratios correspond to total arsenic concentrations of 15 mgAs L-1 (for ferrihydrite systems) and 0.75 mgAs L-1 (for goethite systems), and iron concentrations (as iron oxide) of 0.5 gFe L-1 (for both goethite and ferrihydrite). The pH was varied between pH 3 and 10 at increments of approximately 0.4 pH unit by preadjustment of the phosphate buffer solutions (with NaOH or HNO3) or by adjusting the pH of samples individually (in the case of sulfate desorption) by the addition of NaOH and HNO3. Samples were brought to a uniform final volume (20 mL), allowed to react for 12 h on a rotary shaker for desorption of arsenic, and then centrifuged. The pH of each sample was then measured, and samples were subsequently filtered through a 0.22 µm nominal pore-size membrane filter and analyzed by FI-HG-FAAS. Arsenic Analysis. Arsenic analysis was conducted using flow-injection hydride-generation flame-atomic-absorption spectroscopy (FI-HG-FAAS) with sodium borohydride (NaBH4) as the reductant and hydrochloric acid (HCl) as the eluent. Maximum absorbance of MMAsV and DMAsV by AAS was obtained using 1.5 M HCl as the eluent, whereas for iAsV, iAsIII, MMAsIII, and DMAsIII maximum absorbance was obtained using 5 M HCl as the eluent.

Results and Discussion Mineral Phases. The goethite and ferrihydrite mineralogy was confirmed by powder XRD. Points of zero charge for the goethite and ferrihydrite samples were 8.0 and 8.3, respectively. The adsorption maxima as determined by iAsV adsorption at pH 7.0 were 0.0067 and 0.10 molAs molFe. These values correspond to adsorption maxima of 1.17 and 0.070 VOL. 39, NO. 7, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2121

FIGURE 1. Adsorption isotherms for iAsv, MMAsv, and DMAsv on 2-line ferrihydrite (0.44 gFe L-1) at pH 4 and 7 in 0.044 M NaNO3. The insert (A.) is an expanded representation at low equilibrium dissolved arsenic concentration.

FIGURE 2. Adsorption isotherms for iAsv, MMAsv, and DMAsv on goethite (0.44 gFe L-1) at pH 4 and 7 in 0.044 M NaNO3. The insert (A.) is an expanded representation at low equilibrium dissolved arsenic concentration. mol kg-1 for ferrihydrite and goethite from structural formula of Fe5HO8‚H2O and FeOOH, respectively. Adsorption Isotherms. Methyl substitution influenced the quantity of arsenic adsorbed by ferrihydrite and goethite. MMAsV adsorption maxima were similar to those of its inorganic analogue (iAsV) at high solution concentrations, but at low solution concentrations, the amount of iAsV adsorbed at any given arsenic concentration in solution was always more than in the case of MMAsV (Figures 1 and 2). That is, MMAsV exhibited slightly less adsorption affinity than iAsV. Spectroscopic studies have shown that iAsV forms predominantly bi-dentate bridging complexes with surfaces of iron oxides (16, 17, 19). Although the size of iAsV and MMAsV are different, these two compounds are similar in that they both possess at least two oxygen atoms that are available for complexation with the oxide surface. Thus, MMAsV could have allowed the formation of bi-nuclear bridging complexes, as is the case with iAsV. If MMAsV forms a bi-nuclear bridging complex as iAsV does, the differences in adsorption could have been influenced by the electron-donating characteristics of the methyl group in MMAsV. Compared to iAsV and MMAsV, DMAsV was adsorbed in much smaller amounts on goethite and ferrihydrite (Figures 1 and 2). These results suggest that the adsorption mechanism for DMAsV with the iron-oxide surfaces was probably different than that of iAsV and MMAsV. DMAsV was not quantitatively adsorbed under any of the experimental conditions in this 2122

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 7, 2005

FIGURE 3. Adsorption isotherms for iAsIII, MMAsIII, and DMAsIII on 2-line ferrihydrite (0.44 gFe L-1) at pH 4 and 7 in 0.044 M NaNO3. study. The smaller amount of adsorption could have been the result of the additional methyl group (compared to MMAsV) in a position that was active in the surface complexation of the other AsV compounds (i.e., iAsV and MMAsV). The lower adsorption of DMAsV might also have been influenced by its molecular geometry and possible reduction in spatial compatibility with surface adsorption sites, as influenced by its two methyl groups. The overall effect of pH on adsorption was that each of the AsV species was adsorbed in higher concentrations on goethite and ferrihydrite at pH 4 than at pH 7. Similar results were previously obtained with iAsV on ferrihydrite and goethite (20, 22, 23). The adsorption of each of the AsV compounds was at least 10 times greater on ferrihydrite than on goethite (Figures 1 and 2), primarily due to the large differences in particle size and concentration of surface sites (on an equal mass basis) of these two minerals (27). The adsorption of MMAsIII and DMAsIII was negligible on goethite (data not shown) and ferrihydrite (Figure 3) at both pH 4 an pH 7. This lack of any appreciable adsorption of MMAsIII and DMAsIII by iron oxides indicates that the presence of methyl groups drastically altered the bonding characteristics of AsIII. Adsorption of iAsIII was greater at pH 7 than at pH 4 and did not appear to reach a maximum under the conditions of the experiment (Figure 3). These adsorption patterns of iAsV and iAsIII follow the same patterns as previously published data which also showed that iAsV is adsorbed in greater amounts at low pH values and iAsIII at high pH values (20). Langmuir Adsorption Model. Table 2 lists the constants obtained by linear regression analysis of the adsorption isotherm data collected for all AsV compounds in this study, using the linear form of the Langmuir function as defined by Sparks (28). In the current study each of the calculated linear functions exhibited a good coefficient of correlation (r2), with the smallest value being 0.982. The linear Langmuir function results indicate that adsorption maxima (b) of AsV species were greater at pH 4 than at pH 7 (Table 2), which agrees with the observed patterns of adsorption isotherms (Figures 1 and 2). The calculated relative adsorption maxima (b) increased in the order of iAsV ≈ MMAsV > DMAsV at both pH 4 and 7 (Table 2). The calculated KL parameters of the Langmuir function followed the following relative trend: iAsV g MMAsV > DMAsV (Table 2). In the case of iAsV adsorption on ferrihydrite at pH 4, the calculated KL was 10,800, while the KL values for other arsenic species adsorbed on ferrihydrite at pH 4 were below 700. Also, for each species and for each pH value, the calculated KL parameter was greater for goethite than for ferrihydrite. Though the empirical KL parameter has little or

TABLE 2. Correlation Coefficients (r2), Calculated Adsorption Maxima (molAs molFe-1) (b), and Binding Constants (KL), as Derived from the Linear Form of the Langmuir Equation for Arsenic Compounds Adsorbed on Goethite and Ferrihydrite ferrihydrite As species

pH

r2

goethite b

(mmolAs mmolFe-1) iAsV iAsV MMAsV MMAsV DMAsV DMAsV

4 7 4 7 4 7

0.9999 0.9959 0.9979 0.9968 0.9960 0.9850

0.1548 0.1045 0.1738 0.0943 0.0752 0.0467

KL

As species

pH

r2

-1)

(L mmolAs 10800 372 619 312 144 37.5

KL

b -1)

iAsV iAsV MMAsV MMAsV DMAsV DMAsV

4 7 4 7 4 7

0.9997 0.9995 0.9996 0.9994 0.9915 0.9823

(mmolAs mmolFe 0.0077 0.0067 0.0080 0.0068 0.0027 0.0004

(L mmolAs-1) 20800 7685 15300 8070 167 497

FIGURE 4. Adsorption envelops for iAsV, MMAsV, and DMAsV on 2-line ferrihydrite (0.44 gFe L-1), with initial dissolved arsenic concentration of 13.3 mmolAs L-1 (equivalent to 0.022 molAs molFe1-) and 0.044 M NaNO3 ionic strength buffer.

FIGURE 5. Adsorption envelops for iAsV, MMAsV, and DMAsV on goethite (0.44 gFe L-1), with initial dissolved arsenic concentration of 0.67 mmolAs L-1 (equivalent to 0.022 molAs molFe1-) and 0.044 M NaNO3 ionic strength buffer.

no significance as a theoretical chemical binding constant, this parameter is very useful for evaluation of comparative adsorption behavior between treatments. The value of KL is strongly influenced by the data points that represent low equilibrium dissolved adsorbate concentrations. In the current study, the high KL values indicate a relatively high retention capacity of arsenic at low dissolved arsenic concentration, i.e., a relatively high arsenic bonding strength by the iron oxide. The results of the current study suggest that arsenic bonding strength decreased in the following order: iAsV g MMAsV > DMAsV, and that in all cases bonding was stronger to goethite than to ferrihydrite. Dixit and Hering (29) similarly observed stronger bonding of iAsV to goethite than to ferrihydrite. In the figure insert of Figure 1, which represents adsorption at low equilibrium arsenic concentrations, it is evident that MMAsV was not adsorbed as readily as iAsV by ferrihydrite. The higher KL for iAsV compared to MMAsV is consistent with its stronger retention at low dissolved arsenic concentrations. Adsorption Envelopes. Adsorption of AsV Species. The adsorption of iAsV and MMAsV followed similar trends in adsorption envelope experiments (Figures 4 and 5), indicating that there could be similar adsorption mechanisms for these species on both goethite and ferrihydrite. Under the conditions of this experiment, MMAsV and iAsV approached quantitative adsorption below pH 8.5 on both goethite and ferrihydrite (Figures 4 and 5). Adsorption of both MMAsV and iAsV began to decrease at pH > 8.5, but more rapidly in the case of MMAsV. The decrease in adsorption of iAsV above pH 9 has been reported previously (20), and is probably a result of increasing repulsive potentials between the negatively charged arsenic species and the negatively charged iron-oxide surface (above the pzc of the iron oxide) and increasing competition by OH- with increasing pH. Also, at pH > 9, the increasing solubility of iron oxide and resulting

desorption of arsenic might also impact the net adsorption of arsenic species (30). The more pronounced decrease in adsorption of MMAsV compared to iAsV with increasing pH could not have been caused by repulsive forces resulting from electrostatic charge, because of the higher pKa2 value of MMAsV compared to iAsV. The decrease in adsorption of MMAsV above pH 9 might have been influenced by the electron donating characteristics of the methyl group. The methyl group could have influenced the weakening of the Fe- -O-As bond, thereby causing MMAsV adsorption to decrease more rapidly than that of iAsV at high pH values. The presence of the additional methyl group on DMAsV compared to MMAsV influenced its adsorption behavior. DMAsV was adsorbed in smaller quantities than iAsV and MMAsV at all pH values (Figures 4 and 5). Under the conditions of this experiment, DMAsV was not quantitatively adsorbed to ferrihydrite at any pH. Approximately 92% of DMAsV added (0.022 molAs molFe1-) was adsorbed by ferrihydrite at pH < 7, but above pH 7 the adsorption of DMAsV decreased sharply until it reached zero at approximately pH 9 (Figure 4). The pH of the rapid decrease in adsorption of DMAsV was close to the pzc of ferrihydrite (pH 8.3) (31). DMAsV adsorption by goethite was also pH dependent. Adsorption of DMAsV by goethite increased slightly from pH 3 to 4, had a broad adsorption maximum from pH 4 to pH 5.5, and then decreased with increasing pH (Figure 5). Adsorption of DMAsV by goethite was not observed above pH 7.5 (Figure 5). The differences in the adsorption behavior of DMAsV by goethite and ferrihydrite are partially attributable to differences in the proportion of arsenic to surface adsorption sites in these two suspensions. With the goethite adsorption envelopes, the amount of DMAsV used represented 48% of the adsorption maximum (from the adsorption isotherm) at pH 4 and 260% of the adsorption maximum at pH 7. While in ferrihydrite adsorption envelopes, only 27% VOL. 39, NO. 7, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2123

FIGURE 6. Adsorption envelops for iAsIII, MMAsIII, and DMAsIII on 2-line ferrihydrite (0.44 gFe L-1), with initial dissolved arsenic concentration of 13.3 mmolAs L-1 (equivalent to 0.022 molAs molFe1-) and 0.044 M NaNO3 ionic strength buffer. and 49% of the DMAsV adsorption maxima were used at pH 4 and 7, respectively. The adsorption envelope data for DMAV indicated that the fully protonated, neutral species (pKa1 for DMAsV ) 6.14) was adsorbed in the presence of a nonspecifically adsorbed ion (ionic strength buffer ) 0.044 M NaNO3), which was present in concentration much greater than the DMAsV. There was little electrostatic attraction between the neutral DMAsV and goethite or ferrihydrite below pH 5. The appreciable adsorption of the neutral DMAsV species at pH < 6 indicates that adsorption was due to a specific interaction between DMAsV and the iron-oxide surface. Despite this evidence for the formation of a specific bond between DMAsV and the surfaces of goethite and ferrihydrite, DMAsV was never adsorbed in any appreciable amount above the pzc of goethite or ferrihydrite. This relationship is especially evident in the case of ferrihydrite, where the pH of decrease in DMAsV adsorption was very close to where the pzc of ferrihydrite should occur. In the case of DMAsV, the decrease in adsorption on goethite occurred at a lower pH (approximately pH 5.5). The difference in adsorption behavior indicates that DMAsV probably had a different predominant adsorption mechanism than iAsV and MMAsV, possibly inner-sphere monodentate or outer-sphere complexation versus predominantly bidentate bonding as in the case of iAsV. This possibility would have to be verified by means of spectroscopic experiments. Adsorption of AsIII Species. Both MMAsIII and DMAsIII were negligibly adsorbed by goethite (data not shown) and ferrihydrite (Figure 6), compared to an equal concentration of iAsIII which approached quantitative adsorption across the pH range of 3.5 to 11. MMAsIII adsorption by ferrihydrite slightly increased above pH 8 (Figure 6), which could be attributable to oxidation of MMAsIII to MMAsV and subsequent adsorption of MMAsV. The negligible adsorption exhibited by MMAsIII and DMAsIII under all experimental conditions indicates that these two species were not specifically adsorbed at the goethite or ferrihydrite surface. Desorption Envelopes. Effect of PO4 as a Desorbing Ion. The methyl As III compounds were not studied in desorption experiments because of the poor adsorption of MMAsIII and DMAsIII. In desorption envelopes, the amount of AsV desorbed generally increased as the number of methyl groups increased (i.e., iAsV < MMAsV < DMAsV). Desorption of MMAsV by phosphate from ferrihydrite increased with increasing pH, as did desorption of iAsV (Figure 7). The shapes of these two desorption envelopes were similar (Figure 7), but MMAsV was desorbed in greater quantities than iAsV at any given pH. Upon the treatment of ferrihydrite with phosphate following preadsorption of DMAsV, almost quantitative desorption of DMAsV occurred across the entire 2124

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 7, 2005

FIGURE 7. Desorption envelopes for iAsv, MMAsv, and DMAsv on 2-line ferrihydrite by 0.1 M sodium phosphate, at Fe concentration (as ferrihydrite) of 0.5 g L-1 and As concentration of 15 mg L-1.

FIGURE 8. Desorption envelopes for iAsv, MMAsv, and DMAsv on goethite by 0.1 M sodium phosphate, at Fe concentration (as goethite) of 0.5 g L-1 and As concentration of 0.75 mg L-1. pH range (Figure 7). The relative desorption of iAsV, MMAsV, and DMAsV by phosphate indicates a clear difference in their relative binding affinity to the ferrihydrite surface. Desorption trends for iAsV, MMAsV, and DMAsV from goethite were different from those observed for ferrihydrite (Figure 7 vs Figure 8). Both iAsV and MMAsV exhibited minimum desorption between approximately pH 8 and 9 and increasing desorption with both increasing and decreasing pH (Figure 8). However, iAsV was desorbed in larger quantities below pH 8, compared to greater MMAsV desorption above pH 8 (Figure 8). The desorption curves for these systems crossed close to pH 8.0. Comparatively, DMAsV was 70-90% desorbed from goethite from pH 4 to 7.5, but was totally desorbed at pH values outside of this range. The pH range of minimum desorption by phosphate corresponded to the pH range of maximum DMAsV adsorption on goethite in adsorption envelope experiments (Figure 5). Each of the AsV compounds was more readily desorbed from ferrihydrite than from goethite. Methyl substitution impacted the desorption of arsenic by phosphate, but no uniform desorption trend was observed for goethite and ferrihydrite. Of the arsenic compounds studied, none was quantitatively desorbed by phosphate across all pH values, except for DMAsV from ferrihydrite. For iAsV and MMAsV, as pH was increased desorption generally increased from ferrihydrite. However, desorption from goethite was more complex. MMAsV and iAsV showed minimum desorption at a pH range close to maximum stability of goethite (close to the pzc). Minimum desorption of DMAsV was at a

FIGURE 9. Desorption envelopes for iAsv, MMAsv, and DMAsv on 2-line ferrihydrite by 0.1 M sodium sulfate, at Fe concentration (as ferrihydrite) of 0.5 g L-1 and As concentration of 15 mg L-1. lower pH, at which the goethite surface was positively charged. Effect of SO4 as a Desorbing Ion. In sulfate systems at pH values > 8, desorption of AsV compounds from ferrihydrite (Figure 9) increased with increasing methyl substitution, i.e., iAsV < MMAsV < DMAsV. Below pH 7.5, only DMAsV was desorbed. Above pH 9, MMAsV desorption increased with increasing pH much more rapidly than that of iAsV. The pH region that showed an increase in desorption of MMAsV and iAsV corresponded to the pH region of increased iron oxide solubility (27) and increased repulsive potential between the charged AsV species and the charged oxide surface. The greater increase in MMAsV desorption compared to that of iAsV at pH>9 cannot be attributed to repulsive potentials, since the pKa2 is lower for iAsV than for MMAsV. That is, MMAsV is less negatively charged than iAsV at any given pH, yet is more readily desorbed in the presence of sulfate. The greater desorption of MMAsV must be attributable to the influence of the methyl group. More than 95% of the DMAsV was desorbed from ferrihydrite by sulfate above pH 9, with desorption decreasing to a minimum at pH 6.5. The high amount of arsenic in solution above pH 7 could not be attributed to ligand exchange of DMAsV by sulfate, since DMAsV was not absorbed above pH 7 in adsorption isotherm experiments. In the case of DMAsV, desorption from ferrihydrite increased below pH 6. In a comparison of the desorption envelope of DMAsV (by sulfate) with the inverse of the adsorption envelope of DMAsV (i.e., the percent of DMAsV remaining in solution after adsorption) on ferrihydrite, the only appreciable differences were below pH 6 (Figure 10). Desorption of DMAsV in the presence of sulfate is an indication that sulfate can partially compete with specifically adsorbed DMAsV at low pH. This desorption reaction cannot be attributed primarily to oxide dissolution since the DMAsV was largely retained during the adsorption reaction. The trends in desorption of iAsV and MMAsV were similar with goethite (Figure 11) versus ferrihydrite (Figure 9). However, the trends in DMAsV desorption from goethite versus ferrihydrite differed at pH