Total aqueous As in the slightly acidic (pH 5.2 to 6.1) surface and porewaters of the Gola di Lago peatland ranged from 1.8 to 180 μg L-1. Arsenic initially entering the peatland via an intermittent inflow (IF), was mainly arsenate (study 1, Figure 1a). With increasing lateral distance to the IF, the percentage of arsenate decreased while thioarsenates increased (profiles B1L < B2L and B1R < B2R < B3R in Figure 1a, study 1).
The percentage of arsenite remained comparatively low (2-30%) in surface water samples. At greater depth (∼90 cm), arsenite increased to 25-80% of total As (study 1, Figure 1b). Overall, the maximum percentage of thioarsenates reached 93% of total As (profile B3R, 60 cm depth). Concentration and fraction of thioarsenates did not show a correlation with any of the aqueous parameters such as pH, redox potential, or concentration of total As and S in the solid phase. Mostly, the dominant species in all profiles was MTAs(V) with up to 91% of total As followed by significantly lower fractions of TriTAs(V) and DTAs(V) (median 3% and 2%, respectively), whereas TetraTAs(V) was instable at these low pH values.168 It is important to note that the speciation data presented here only reflects a single-point observation from November; a seasonally changing water table or increased microbial activity in the summer half-year may lead to a slightly different As speciation, especially in the most affected shallow peat layers.
The significant fractions of MTAs(V) observed in Gola di Lago versus a complete lack of information on the effect of MTAs(V) formation on overall As mobility in reduced S and organic carbon-rich environments were the basis for several laboratory sorption studies.
In time series experiments with 3 mM sulfide-reacted model peat, total As sorption upon 50 µM MTAs(V) addition increased in the order pH 8.5 < 7.0 < 4.5, whereby sorption at pH 4.5 (32 μmol As mol-1 C) was
∼5 times higher than at pH 8.5 (7 μmol As mol-1 C, study 1, Figure 2). The same trend occurred for arsenite, but the difference between As sorption at pH 4.5 (29 μmol As mol-1 C) and 8.5 (14 μmol As mol-1 C) was only a factor of 2. Compared to arsenite, the initial sorption of total As was slower in experiments where we added MTAs(V), but after 984 h (41 days), the extent of sorption was comparable, at least for pH 4.5 and 7.0. For pH 8.5, the amount of total As adsorbed remained at only∼50% of that observed in the arsenite experiments.
Aqueous As speciation analysis for the MTAs(V) experiments showed, in contrast to the arsenite experiments, significant transformations of MTAs(V) to mainly arsenite (study 1, Figure 3).
Transformation rates decreased with increasing pH from half-lives of ∼35 to∼470 h at pH 4.5 and 7.0, respectively, to almost no transformation at pH 8.5 over the duration of the experiment (study 1, Figure 3a,c,e). In a blank experiment with background electrolyte and NaN3 only minor transformation of MTAs(V) to arsenite at pH 4.5 occurred after 120 h and no transformation at pH 7.0 and 8.5 was observed (study 1, Figure SI-9). Furthermore, no significant increase in S oxidation products (sulfate and thiosulfate) occurred during our sorption experiments confirming anaerobic conditions and indicating no obvious
Results and Discussion
microbial influence. The transformation of MTAs(V) was therefore not driven by pH or ionic strength but clearly triggered by the presence of sulfide-reacted model peat.
As described before, MTAs(V) is in equilibrium with arsenite and S(0)-species. Low pH will drive the equilibrium toward arsenite and S(0)-species,169 but kinetics in solution are usually so slow that any MTAs(V) transformation is not likely to be observed over several days (study 1, Figure SI-9). However, previous reports115,120 and our own experiments (study 1, Figure 2) show that arsenite adsorbs strongly to thiol-rich NOM, particularly at low pH. This is consistent with the observed order of MTAs(V) transformation kinetics. We conclude that arsenite sorption drives the chemical equilibrium toward arsenite and S(0)-species, therefore accelerating MTAs(V) transformation. Sorption of S(0)-species would have the same effect.
While aqueous As speciation already implied that arsenite sorption is an important mechanism in the MTAs(V) experiments, it could not answer the question whether sorption observed in the MTAs(V) experiments occurred exclusively as arsenite or also included direct sorption of MTAs(V) to sulfide-reacted model peat. Thus, we analyzed normalized AsK-edge XANES spectra by ITFA in order to follow As solid-phase speciation from selected sampling points of the time series experiments. The three standards “arsenite adsorbed via thiol-S to peat” (As(III)-S-NOM), “arsenite reacted with untreated peat” (As(III)-NOM), and
“arsenate reacted with untreated peat” (As(V)-NOM) were found to be the statistically most plausible components in our system. MTAs(V) was not detected on the solid phase even if present in high aqueous concentrations at pH 8.5. This finding was further confirmed by EXAFS shell-fitting of resin spectra from MTAs(V) solutions reacted with the thiol-rich Ambersep GT74 resin at pH 7.5 and 8.0. MTAs(V) was stable in the aqueous solution at both pH values and no contribution of a double-bound S (typically 2.13-2.18 Å38) was observed on the solid phase. Instead, As was 3-fold coordinated at each pH to S with a distance of 2.26 Å, indicating typical bond lengths of arsenite bound via thiol-S.120
The ITFA calculations revealed an increase of As(III) fractions with decreasing pH in the same manner as MTAs(V) transformation to arsenite in the aqueous phase (study 1, Figures 3 and 4). Hereby, a systematically higher portion of As was complexed via thiol-S to model peat at pH 4.5 compared to pH 7.0.
No thiol complexation higher than the detection limit (5%) was observed at pH 8.5 (study 1, Figure 4). The solid-phase speciation data clearly show that there was no significant MTAs(V) sorption to sulfide-reacted model peat via thiol groups. Probably, the reason for MTAs(V) showing no detectable sorption to S-NOM is that MTAs(V) has pKavalues of 3.30 (pKa1), 7.20 (pKa2), and 11.0 (pKa3) and was therefore single- or double-negatively charged under our experimental conditions (pH 4.5-8.5). The negative charge causes a strong electrostatic repulsion with the negatively charged sulfide-reacted model peat, thus excluding MTAs(V) sorption, regardless of the pH value and S content (study1, Figure 4) and thus renders MTAs(V) mobile in the S-NOM system.
However, arsenate with very similar pKa values26 showed, though to a very low extent, sorption in the
Results and Discussion
presence of sulfide-reacted model peat (study1, Figure 2 and 4). Further, relatively high fractions of arsenite, originally reacted with untreated model peat, were found in the sulfide-reacted peat solid phase.
For both As species, as well as for MTAs(V), extent of sorption and underlying sorption mechanisms to untreated peat (low to no reduced S content) were still elusive and are thus addressed in the following.
Highest sorption to untreated model peat after incubation for 96 h was observed for arsenite, followed by MTAs(V) and arsenate, even though the difference between the extent of MTAs(V) and arsenate sorption was small (study 2, Figure 1). Earlier studies reported higher sorption of arsenate to HA compared to arsenite,147,148,170 but did not consider potential ternary complex formation of arsenate with Fe in their investigations.144 In agreement with earlier studies,147-150,170 sorption of arsenite and arsenate was stronger at pH 7.0 than at pH 4.5. However, for MTAs(V), sorption was stronger at pH 4.5, which can be attributed to the partial transformation of MTAs(V) to arsenite, as revealed by the AsK-edge XANES spectra (study 2, Figure SI-7). This finding suggests that the acid-assisted transformation will be feasible in the presence of any sorbent that adsorbs arsenite more strongly than MTAs(V), not only in presence of thiol groups as shown before.
The calculated logKOC (L kg-1 C) at pH 7.0 were 0.83-1.01 for arsenite, 0.40-0.56 for MTAs(V), and 0.34-0.41 for arsenate; at pH 4.5, corresponding values were 0.73-0.85, 0.44-0.66, and 0.31-0.38 (study 2, Figure SI-3). While the extent of sorption for arsenate and MTAs(V) was considerably low, complexation of arsenite with O-containing functional groups of model peat (log KOC: 0.83-1.01) was comparable with binding through Fe(III)-bridged ternary complexes (logKOC: 1.2-1.5)146, but ranged clearly below As(III)-S-Corg coordination (logKOC: 0-2.9)120 at pH 7.0.
Shell-by-shell fitting of the AsK-edge EXAFS spectra revealed a significant As-O-Corg coordination for all three As species (study 2, Figure 2 and Table 1). In the arsenite-treated model peat samples, the average As-O distance was 1.79±0.01 Å for both pH values, consistent with reported As-O bond lengths in the AsO3
pyramid.53,171 The As-C distances were 2.73±0.01 Å and lay between As-C distances of 2.77-2.86 Å172,173 (study 2, Table SI-2), representative for hydroxyl coordination, and 2.58-2.68 Å,171 representative for carboxyl coordination, suggesting a mixed coordination between both functionalities. Because of the higher oxidation state of arsenate, the average As-O distance was considerably shorter (1.69±0.01 Å) compared to that of arsenite.144 The fitted As-C paths for both pH values were 2.83±0.01 Å and therefore longer than that in the arsenite-treated model peat. Guénetet al.151 determined comparable As-C distances (2.85-2.86 Å) for a mixture of arsenate and arsenite during oxidation of reduced organic-rich floodplain soil, similar with our results pointing to complexation with organic hydroxyl groups. In the MTAs(V)-treated model peat at pH 7.0, the fitted As-O path with 1.72±0.01 Å was longer than the As-O distance in arsenate- but shorter than in arsenite-treated model peat. The As-S distance was determined as 2.10±0.02 Å, indicating that S was double-bound to As.38 Despite the same oxidation state of As, the longer As-O in the AsSO3 compared to that in AsO4 can be attributed to the less positive partial charge on the central As atom, because of replacement of the double-bonded O with less electronegative double-bonded S. The fitted As-C distances
Results and Discussion
were 2.80±0.02 Å and thus between arsenite- and arsenate-treated model peat.
The assumed underlying mechanism of these chemical reactions is the nucleophilic attack by carboxylic and/or hydroxyl groups to the partially positively charged As atom.147 The partial positive charge of the central As atom is highest in arsenate, followed by MTAs(V) and arsenite, because of the double-bonded O of arsenate compared to arsenite and the replacement of O with the less electronegative S in MTAs(V).
Thus, the expected order of complexation would be arsenate > MTAs(V) > arsenite. However, the order was exactly the opposite (study 2, Figure 1), which may be explained by the electrostatic repulsion between negatively charged, deprotonated carboxyl and hydroxyl groups of peat (study 2, Figure SI-1) and the As species. While arsenite (pKa1 = 9.17) is neutrally charged at pH 4.5-7.0 and adsorbed most, arsenate (pKa1
= 2.30, pKa2 = 6.99) and MTAs(V) (pKa1= 3.30, pKa2= 7.20) are negatively charged, and electrostatic repulsion probably predominated over the effect of higher partial positive charge of the As center in these two species. Despite electrostatic repulsion, the driving force for complexation of As species to NOM is attributed to the energetic stability gained by donation of the negative charge of the carboxylate (R-COO-) and hydroxyl (R-O-) ions to the partially positively charged As atom and/or additional chelation and H bonding with nearby functional groups.147 We hypothesize that the energy gain forming covalent bonds together with the increased nucleophilicity of hydroxyl groups at higher pH 7.0 overcame the effect of electrostatic repulsion between arsenate or MTAs(V) and peat functional groups, however, resulting in a lower extent of sorption and relatively longer As-C bonds compared to neutral arsenite (study 2, Table 1).
Thus, our results demonstrate that compared to theKOCvalues of ternary As(III)-O-Fe-Corg complexation and in addition to the reported As(III)-O-Corg contributions of other studies146,151 and our study 1, O-containing functional groups can be an important alternative or additional binding mechanism for arsenite in anoxic, organic-rich systems. The low complexation of arsenate and MTAs(V) with O-containing groups and the lack of MTAs(V) binding in systems with high thiol-group-contents, underline the high mobilization potential of arsenate and especially MTAs(V), in reduced S and organic-rich systems.
3.2 Influence of reduced S on the fate of As during sulfurization of As-bearing model peat