Oxidation of V(IV) by Birnessite: Kinetics and Surface Complexation

Abstract

Vanadium is a redox-active metal that has been added to the EPA’s Contaminant Candidate List with a notification level of 50 μg L⁻¹ due to mounting evidence that V^V exposure can lead to adverse health outcomes. Groundwater V concentration exceeds the notification level in many locations, yet geochemical controls on its mobility are poorly understood. Here, we examined the redox interaction between V^IV and birnessite (MnO₂), a well-characterized oxidant and a scavenger of many trace metals. In our findings, birnessite quickly oxidized sparingly soluble V^IV species such as häggite [V₂O₃(OH)₂] into highly mobile and toxic vanadate (H_nVO₄^(3–n)–) in continuously stirred batch reactors under neutral pH conditions. Synchrotron X-ray absorption spectroscopic (XAS) analysis of in situ and ex situ experiments showed that oxidation of V^IV occurs in two stages, which are both rapid relative to the measured dissolution rate of the V^IV solid. Concomitantly, the reduction of birnessite during V^IV oxidation generated soluble Mn^II, which led to the formation of the Mn^III oxyhydroxide feitknechtite (β-MnOOH) upon back-reaction with birnessite. XAS analysis confirmed a bidentate-mononuclear edge-sharing complex formed between V^V and birnessite, although retention of V^V was minimal relative to the aqueous quantities generated. In summary, we demonstrate that Mn oxides are effective oxidants of V^IV in the environment with the potential to increase dissolved V concentrations in aquifers subject to redox oscillations.

Graphical Abstract

INTRODUCTION

Vanadium (V) is ubiquitous in terrestrial environments and is increasingly recognized as a contaminant in subsurface environments due to both natural processes and anthropogenic activities.^1,2 In aquatic ecosystems, sensitive species are vulnerable to aqueous V exposures as low as 1.2–80 μg L⁻¹.^1,3,4 In either environment, vanadium toxicity is dependent on the concentration, oxidation state, and species of V, with V^V species eliciting the largest degree of toxicity as well as the highest degree of solubility.^5–8 As a result, V was added to the U.S. Environmental Protection Agency Contaminant Candidate List (CCL1) in 1998 with a corresponding notification level of 50 μg L⁻¹ and several U.S. states have since recommended lower notification levels.^9–11

In the environment, V is present in the +II, +III, +IV, and +V oxidation states, but only V^III, V^IV, and V^V are common in soils and sediments.^12–14 V^III is released from mineral phases into the soil solution upon weathering and dissolution of the host material, which quickly oxidizes to V^IV or V^V under suboxic to oxic conditions.^14–16 Despite the prominence of V^III as the most abundant form of V in geologic parent material, its solubility decreases above pH 4.9, limiting its mobility under all but the most reducing and acidic conditions, although precise thermodynamic parameters are not known.¹ As a result, V^IV and V^V are the most common species found in groundwater systems.^14,17

V^IV and V^V exhibit differences in solubility, speciation, and chemical interactions with organic matter and mineral phases.¹ V^IV species are stable under moderately reducing and acidic conditions and have been shown to be the most stable form of V in suboxic to anoxic groundwater, as well as waters with dissolved organic carbon (DOC).^1,18 Oxovanadium^IV (VO²⁺_aq) hydrolyzes above pH 5.7 to form VOOH⁺_aq and VO(OH)_2(s). However, VO(OH)_2(s) precipitation can be inhibited by V^IV complexation with DOC, and as such precipitation occurs primarily under low DOC conditions.^18,19

Under oxic conditions and solution pH above 6, V^V species exist predominantly as the oxyanion vanadate (H_nVO₄^(3–n)–), which is reducible to V^IV by organic ligands or by Fe^II, with a HVO₄²⁻/VO²⁺ standard potential (E⁰) of 1.90 V.^1,20,21 Furthermore, V^V is known to polymerize to form oligomeric polyoxometalate V^V species at concentrations as low as 100 μM.^22,23 While retention of V^V species in the environment is dependent on sorption–desorption and reduction processes,^14,24,25 conditions that lead to changes in redox potential are expected to affect the oxidation state and, therefore, the mobility of V similar to other redox active metal(loid)s such as arsenic and chromium^26,27

Manganese(IV) oxides are strong oxidants and scavengers of metal(loids) like As and Cr in terrestrial systems.^28,29 The ubiquity and reactivity of Mn^IV oxides make them likely candidates to influence the fate of V in these environments. For example, elevated levels of geogenic V (>50 μg L⁻¹) have been detected throughout aquifers in California.^13,17 Many of these aquifers are subject to seasonal redox fluctuations that can lead to the accumulation of Mn^IV oxides in a close spatial proximity to V.^17,29,30 However, previous studies examining the reactivity of V^IV and V^V with soil mineral phases^20,31–37 have not examined the role of Mn^IV oxides in V^IV oxidation or V^V adsorption, despite their importance in controlling the fate of other metal(loid)s. In this study, we investigated the reaction kinetics, sorption processes, and resultant Mn and V solid phase alterations following the reaction of V^IV with birnessite. Accordingly, we combine in situ and ex situ aqueous and solid phase measurements to construct a model that describes the oxidation of V^IV by birnessite.

MATERIAL AND METHODS

Birnessite Synthesis.

Birnessite was prepared as described by McKenzie et al.³⁸ Briefly, 63 g of KMnO₄ (Fisher) was dissolved into 1 L of ultrapure water (18.2 MΩ·cm, Millipore) and heated to 90 °C. Over the course of 10 min, 66 mL of concentrated trace metal-grade HCl was added to the KMnO₄ solution. The mixture was kept at 90 °C for 10 min before being cooled to 21 °C. Manganese oxides in the solution were collected using vacuum filtration and repeatedly rinsed with ultrapure water to remove excess KMnO₄ and then dried and crushed with a mortar and pestle. Birnessite was confirmed by X-ray diffraction (XRD) using Cu Kα radiation on a Siemens D500 X-ray diffractometer (Figure S5) and a specific surface area was 37.8 m²g⁻¹ (r² = 0.999) using the BET surface area analysis (Quantachrome Nova 2000e).

In Situ and Ex Situ Experiments.

Two sets of stirred batch reactors were carried out under anoxic conditions to determine the reaction kinetics of V^IV oxidation by birnessite using chemicals that were ACS grade or higher. For both sets of experiments, anoxic 25 mM NaCl was used as a background electrolyte and 50 mM PIPES was used to buffer the solutions at pH 7. One set of experiments (hereafter referred to as in situ) was performed at beamline 2–2 at the Stanford Synchrotron Radiation Lightsource (SSRL) using the semiquick (~90 s per scan) X-ray absorption near-edge structure (XANES) spectroscopy at the V K-edge. The progression of the reaction was tracked in real time by monitoring the increase in the pre-edge peak intensity to determine the V^IV:V^V ratio. These experiments were initiated by suspending birnessite at final loadings of 0.5, 1, 3, or 10 g L⁻¹ in an anoxic buffer at 75% of the final volume. V^IV was added to the remaining 25% of the separately aliquoted buffer solution as the hydrated salt VOSO₄·5H₂O. However, the pH of the buffer caused it to immediately precipitate into a variety of solid V^IV oxyhydroxides, the identity of which is further discussed in the Supporting Information Section 3. The final concentration of V^IV was 3.2 mM, with the final Mn:V ratios of 2.2, 3.7, 11.9, and 47.1, which are similar to concentrations of V and Mn that have been observed in California’s aquifers.^17,39

The second set of experiments (hereafter referred to as ex situ) was conducted in the laboratory in an identical manner with the solids loading of birnessite fixed at 1 g L⁻¹ and the concentration of V^IV varied between 1 and 3.5 mM for final Mn:V ratios of 11.1 and 3.2, respectively. While elevated, these concentrations were chosen to get adequate X-ray fluorescence signals based on the single-channel silicon drift detector that was used in the collection of both in situ and ex situ V X-ray absorption spectroscopic (XAS) data collection, and it is a reason for the similarly elevated concentrations used in prior studies.⁴⁰ Additionally, the chosen V concentrations were necessary to reach our targeted Mn:V ratios based on the amount of birnessite that was used. Further details of the reactors and analysis methods are available in Section 1 of the Supporting Information. Samples were collected from the ex situ experiments for both aqueous and solid phases analysis. Briefly, total aqueous V and aqueous V^V speciation was quantified using inductively coupled plasma optical emission spectrometry (ICP-OES) and nuclear magnetic resonance (⁵¹V-NMR); solid-phase characterization and V speciation included X-ray diffraction, XANES spectroscopy, and extended X-ray absorption fine structure (EXAFS) analyses.

Kinetic Modeling.

Rate constants were initially calculated by fitting both first- and second-order rate models to the V and Mn data. However, the interaction of V^IV and V^V with birnessite may proceed through a variety of pathways that include oxidation, adsorption, desorption, and surface passivation. We therefore developed a model that accounts for mixed-order Langmuir adsorption kinetics, surface passivation, and pseudo-second-order oxidation, which is presented as eq 1.

$$\begin{array}{l}\text{ln}\frac{{\left[{\text{V}}^{\text{IV}}\right]}_i-{\left[{\text{V}}^{\text{V}}\right]}_{\text{aq},t}-\left(q_{\text{eq}}\left(\frac{1-{\text{e}}^{-k_{s}t}}{1-f_{\text{eq}}{\text{e}}^{-k_{s}t}}\right){[\text{Birnessite}]}_{t,(\text{g}{\text{L}}^{-1})}\right)}{{[\text{Birnessite}]}_t}\\ =t\left(K_0{\text{e}}^{-bt}\right)\left({\left[{\text{V}}^{\text{IV}}\right]}_i-2{[\text{Birnessite}]}_i\right)+\text{ln}\frac{{\left[{\text{V}}^{\text{IV}}\right]}_i}{{[\text{Birnessite}]}_i},\end{array}$$ (1)

where q_eq is the amount of vanadate adsorbed by birnessite at equilibrium (mmole g⁻¹). k_s is the first-order adsorption rate constant (min⁻¹). f_eq is the batch equilibrium factor and represents the level of second-order contribution to the adsorption of V^V to birnessite. F_eq falls between [0 ≤ F_eq ≤ 1], with 0 yielding a pure first-order and 1 yielding a pure second-order form (dimensionless). K₀ is the initial pseudo-second-order rate constant (mM⁻¹ min⁻¹). b is the pseudo-first-order birnessite surface passivation constant (min⁻¹). i denotes initial concentrations. t denotes concentrations at time t. Note that the concentration of birnessite is reported in g L⁻¹ in the numerator on the left side of the model and in molar units otherwise.

All parameters from the pseudo-second-order surface oxidation (PSOSO) model (Table 1) were determined via inverse modeling with residual sum of squares (RSS) minimization. The reaction orders for V and birnessite were determined by standard natural-log transformations of the initial concentration and the rates of reaction from both the in situ and ex situ experiments followed by linear regression, with the slope being equal to the reaction order.^40,41 The adsorption–desorption parameters (q_eq, k_s, and f_eq) were determined by fitting data from kinetic adsorption and equilibrium adsorption experiments (Supporting Information Section 1). K₀ and b were determined from regressing the model in eq 1 on the data from each experiment using the initial rate method (Phase 1 only) or the entire time series. Additional details on the model construction can be found in the Supporting Information, and values are reported in Table 1.

Table 1.

Kinetic Parameters Derived from Modeling the In Situ and Ex Situ Experiments

experiment	birnessite loading (g L−1)	V concentration (mM)	Mn concentration (mM)	rate (mM min−1) from pre-edge analysis	phase 1 rate (mM V min−1)	phase 2 rate (mM V min−1)	Mn:V ratio
in situ	0.5	2.633	5.747	0.042			2.183
in situ	1	3.112	11.494	0.074			3.693
in situ	3	2.890	34.483	0.116			11.934
in situ	10	2.440	114.943	0.232			47.110
ex situ	1	3.513	11.123		0.079	0.022	3.167
ex situ	1	0.998	11.128		0.029	0.004	11.149

reaction order	V	R 2	Birnessite	R 2
phase 1	0.701	0.913	0.529	0.969
phase 2	1.371	0.992

entire time series (×10−3)		phase 1 only (×10−3
passivation rate constant b (min−1)	9.039	passivation rate constant b (min−1)	0.000
K0 (mM−1 min−1)	2.755	K0 (mM−1 min−1)	3.529

RESULTS AND DISCUSSION

Kinetics of V^IV Oxidation by Birnessite.

Ex situ batch experiments showed that the reaction of 3.5 mM V^IV with birnessite (1 g L⁻¹) under anoxic conditions led to the production of 3.3 mM dissolved V^V and 1 mM dissolved Mn^II within 2 h, whereas the 1 mM initial V^IV experiments led to the production of 0.9 mM dissolved V^V and ~0.05 mM Mn^II_aq (Figure 1). After accounting for the adsorbed V^V fraction, all V^IV were found to be oxidized within 2 h (Supporting Information Section 2 eq 11). V_aq (total aqueous V concentration) increased before plateauing after approximately 101 min, which coincided with a decrease in Mn_aq for both concentrations of V^IV tested (Figure 1).

Figure 1.

Aqueous V and Mn during the reaction of 3.5 mM (left) and 1 mM (right) V^IV with 1 g L⁻¹ birnessite at pH 7 in the ex situ experiments. Note the x-axis is split at 250 min for the 3.5 mM V^IV treatment (left). The solid gray vertical line denotes the end of Phase 1 and the dashed vertical line denotes the end of Phase 2 of reaction.

V_aq production occurred in two phases during the ex situ expeirments: (1) rapid initial production from 0 to 11 min (3.5 mM V^IV) or 15 min (1 mM V^IV) followed by (2) slower production until 101 min. After these phases, V_aq production plateaued with no V^IV remaining. In the 3.5 mM V^IV experiments, an average rate of 0.079 ± 0.01 mM V^V min⁻¹ was achieved during the first phase followed by a rate of 0.022 ± 0.002 mM V min⁻¹ during the second phase (calculations provided in Supporting Information). In the 1 mM V ex situ experiments, an average rate of 0.029 ± 0.003 mM V min⁻¹ was observed during the first phase, which decreased to 0.0039 ± 0.0005 mM V min⁻¹ in the second phase. This represents a rate decrease of 72 and 87%, respectively, for the two sets of experiments. By combining data from ex situ and in situ measurements, a reaction order for vanadium of 0.70 (R² = 0.912) was calculated for the first phase, which increased to 1.37 (R² = 0.992) in the second phase. The V reaction order for phase 1 approximates the two-thirds reaction order observed in geometric contraction kinetic models, indicating that the reduction of birnessite by V^IV results in a change to the birnessite particle size, likely as Mn^IV is reduced to Mn^II_aq at the particle edges.⁴² Aqueous speciation of dissolved V using ⁵¹V-NMR confirmed the production of H₂VO₄⁻ within the first 2 min followed by the production of H₂V₂O₇²⁻ and V₄O₁₂⁴⁻ (Figure S4a–c).^43–45

Aqueous Mn was also observed in the ex situ experiments which increased linearly until the end of the first phase (Figure 1). This suggests that side reactions like the formation of Mn^III hydroxides were not significant during the first phase, affirming our use of the initial rate method over this time frame. This is further supported by the surface passivation rate constant and Mn XANES analyses discussed below. For the 1 mM experiments, the concentration of Mn_aq plateaued during the second phase before decreasing after 101 min. Likewise, the rate of Mn_aq production slowed in the 3.5 mM experiments during the second phase, with the concentration of Mn_aq gradually decreasing after 101 min. The decrease in Mn_aq in both systems after the consumption of V^IV by the end of the second phase is likely due to comproportionation reactions between the Mn^II_aq and birnessite, resulting in Mn^IIIOOH phases further described below. Maximum Mn_aq produced by the 1 and 3.5 mM experiments were 0.05 and 0.96 mM, respectively, or 5 and 27% of the total V^IV added to each set of systems. Since two moles of V^V are produced per mole of Mn^II released and both spectroscopic and kinetic data suggest that all V^IV were oxidized to V^V (Figure 2b), the remaining reduced Mn must therefore be associated with the solid phase, which is supported by the Mn XAS and XRD findings. The differences in Mn^II_aq between the high and low concentration ex situ experiments, particularly during phase 2, are likely a result of the lowered likelihood of two molar equivalents of V^IV transferring their electrons to the same structural Mn atom at lower V^IV:Birnessite ratios. This would result in the formation of more structural Mn^III than Mn^II_aq, especially as Mn^II_aq is removed from the solution through back reactions with additional equivalents of structural Mn^IV as seen by the decrease in Mn^II_aq after phase 2 and the formation of β-MnOOH.

Figure 2.

In situ V K-edge XANES time series. (a) XANES spectra collected during oxidation of V^IV by 0.5 g L⁻¹ birnessite. Oxidation of V^IV to V^V is primarily observed as an increase in the intensity of the pre-edge feature at 5469 eV over time (shaded area). (b) Increase in the oxidation number of V across the in situ experiments calculated from the analysis of the pre-edge feature.

Measurements of the vanadium average oxidation state (AOS) were obtained using pre-edge peak fitting of XANES spectra collected during the in situ experiments, which accounts for V in both aqueous and solid phases (eq S12). Reaction rates and initial concentrations are presented in Table 1 and Figure 2. In all treatments, the pre-edge analysis revealed a monotonic V^V increase followed by a plateau with an average V AOS increase of 1.19 ± 0.04 valence units. Comparison of the rates across experiments revealed an increase in the observed V^IV oxidation rate of 0.0039 mM V min⁻¹ per unit increase in the Mn:V ratio (R² = 0.966). The reaction order for birnessite was calculated to be 0.55 (R² = 0.988). Since only a single phase of V^V production was observed in the time frame over which the in situ experiments were conducted, any increase in the reaction order for birnessite at later stages of the reaction was poorly defined. Regardless, when combined with the V reaction order, the overall reaction order increased from 1.25 to 1.92 between phase 1 and phase 2, approximating a designation as a second order overall.

Several trends emerge upon application of the PSOSO model to the ex situ data. First, the values of the initial pseudo-second-order rate constant (K₀) and the pseudo-first-order surface passivation rate constant (b) are dependent on the time frame over which the model is applied. Thus, we will discuss values obtained from the application of the model to phase 1 only, as well as those obtained via the application of the model over the whole experimental time frame. When applied to only the first phase of the ex situ data and the single phase of V^V production in the in situ experiments, a K₀ of 0.0035 (mM⁻¹ min⁻¹) was obtained. Over this time frame, the calculated value of b was 0.000 min⁻¹, supporting our conclusion that side and backreactions are minimized over this period. However, when applied to the entire time series, K₀ decreased to 0.0028 (mM⁻¹ min⁻¹), and a value of 0.009 min⁻¹ was obtained for b indicating the increased role of surface passivation with time.

The rate of solid-phase V^IV dissolution was found to be 0.0017 mM V min⁻¹ (R² = 0.952) over the first phase (11 min) of the control experiment in the absence of birnessite (Figure S3). We speculate that the difference between this dissolution rate and the observed oxidation rates may be a result of solid–solid interactions due to particle collisions in the stirred reactor, implying that electron transfer between V^IV oxyhydroxides and birnessite takes place. Additionally, although the V^IV oxyhydroxides are predominantly in the solid phase when introduced into the birnessite slurry, the presence of the oxidant rapidly removes the low concentration of V^IV present before it can back-react with the V^IV solids, spurring the state of the V^IV_aq–V^IV_solid system in the presence of birnessite toward dissolution.

Using the results from the V^IV dissolution control experiment, we calculated a pseudo-first-order dissolution constant of 0.115 min⁻¹ for the first phase. However, a steady increase of V^IV_aq was not observed after this phase. Instead, V^IV_aq reached a maximum concentration of 27 μM before equilibrating to 22 ± 4 μM for 23 h, followed by a slow decrease to 7 μM by the end at 125 h (Figure S3a). The slow V^IV_aq decrease is attributed to re-uptake by the V^IV precipitate, leading to an increase in its crystallinity (additional details provided in Section 3 of the Supporting Information), possibly via Ostwald ripening. Both dissolution and re-uptake processes were modeled using the second-order rate law for mineral dissolution, yielding desorption and resorption constants 0.021 and 3.9 × 10⁻⁷ μM⁻¹ min⁻¹, respectively.⁴⁶

Manganese Transformations.

Vanadium(IV) oxidation is coupled with the reduction of Mn^IV to structural and aqueous Mn^II/III. Understanding the fate of reduced Mn products provides insights into the underlying controls on the observed rate, extent, and products of V(IV) oxidation by birnessite. Manganese XANES analysis of the unreacted and reacted birnessite shows a shift in edge position by −1.7 eV, with E₀ (taken as the peak of the first derivative) decreasing from 6553.16 to 6549.68 (Figure S5b,c). Since the concentrations of V^V_aq and Mn^II_aq are not stoichiometric (Figure 1), the remaining reduced-Mn equivalents must remain associated with the solid phase. The AOS of Mn in the unreacted birnessite was calculated as 3.75 using the Combo method of Manceau et al.⁴⁷ (Supporting Information Section 1). Given the complete oxidation of the V^IV and the concentrations presented in Table 1, a final Mn AOS of 3.45 and 3.67 is expected in the 3.5 and 1 mM ex situ experiments, respectively. The solid phase associated with reduced Mn equivalents were then calculated at each time point using the values of Mn^II_aq and V^V_total (Supporting Information eq 11; Figure 3), yielding AOS values for the 1 and 3.5 mM ex situ batch experiments of 3.66 and 3.53, respectively. These values were validated by applying the Combo method to the Mn XANES of the reacted birnessite from the 3.5 mM ex situ batch experiment, for which an AOS of 3.52 was obtained.

Figure 3.

Percentage increase in reduced structural Mn in birnessite over time (with reduced Mn presented as either Mn^II_sorbed or Mn^III_structural) for ex situ experiments. The increase in reduced structural Mn was determined by calculating the amount of structural Mn^II or Mn^III from the V^V_total and Mn^II_aq at each time step.

The accumulation of Mn^III_structural in the birnessite was observed in both the Mn EXAFS and XRD as well. The presence of corner-sharing Mn^II/III octahedra at interlayer vacancy sites is exhibited by the appearance of a shoulder at 6.3 Å⁻¹ in Mn EXAFS spectra at the end of the reaction (Figure S5), along with an increase in the intensity of the peak at 3.1 Å R + δR in the radial structure function (RSF) (Figure S5, Table S5).^48–52 In addition, a triple corner-sharing Mn octahedra capping a birnessite vacancy site is observed at a distance of 3.50 Å, comparable to previously published distances.⁵³ The intensity of the Mn–Mn single scattering peak at 5.1 Å R + δR increased over the reaction suggesting that these vacancies were filled by disordered arrangements of Mn^II/III as the reaction progressed.^49,53,54 We then used XRD to see if these changes affected the long-range structure of the Mn oxide.

Sorption and incorporation of Mn^II and Mn^III into birnessite were observed in the XRD patterns as well. A dip at 45° and a broad feature between 37 and 66° 2θ were found in the reacted birnessite (Figure S5), which are associated with Mn^II/III sorption at interlayer vacancy sites at levels close to saturation.^54–58 The lack of the dip at 45° in the unreacted birnessite indicates a low amount of Mn^II/III initially present.⁵⁶ We did not detect a symmetry change from hexagonal to triclinic in the birnessite after the reaction, which has been shown to occur when Mn^II_aq reacts with hexagonal birnessite above a pH of 8.0.⁵⁰ Rather, comproportionation between the adsorbed Mn^II and birnessite resulted in the formation of the Mn^III oxide phase feitknechtite, a product that has been observed in similar studies.^50,59,60 The feitknechtite was observed via the reflection of the [100] plane at 18° 2θ (Figure S5). In contrast to previous studies that investigated the reduction of birnessite by Mn^II, no manganite (γ-MnOOH) was observed in this study.^50,59,60 We attribute the absence of manganite primarily to relatively short experimental time, whereas longer reaction times are necessary for the feitknechtite-to-manganite aging process to occur.^59–61 Although competition between divalent cations and Mn^II for vacancy sites has also been shown to lead to the preferential formation of feitknechtite over manganite,^61,62 V EXAFS results suggest that V^V preferentially interacts with edge sites rather than vacancy sites; thus, manganite formation would be expected if the aging process continued. Furthermore, we expect aqueous V^IV species to preferentially react at birnessite edge sites over surface vacancy sites due to the increased reactivity of undersaturated edge oxygens present at the particle edge, similar to the mechanism of the oxidation of As^III by birnessite.⁵⁶

Reductive manganese oxide transformation initiated by V^IV oxidation and sustained by reaction with V^IV and Mn^II (and likely Mn^III) result in a complex reaction cycle that has implications for controlling both V and Mn concentrations in groundwater. The reaction can lead to the accumulation of secondary Mn phases at the surface of birnessite which is expected to add complexity to the rate of oxidation and adsorption of V^IV/V, which is reflected in part by the V EXAFS described below.

Vanadium Solid Phase Dynamics.

The presence of an oxidant influences the development of the V^IV hydroxide at early stages of its formation. In the presence of birnessite, the V^IV precipitate differs considerably from the V^IV oxyhydroxide observed in the V^IV dissolution control experiment when samples taken at 1-minute postinitiation are compared. Specifically, V^IV was present as an octahedrally coordinated semicrystalline solid primarily composed of häggite in the V^IV dissolution control experiment (Supporting Information Section 3), whereas V^IV solids appeared more amorphous in the presence of birnessite. In the presence of birnessite, V^IV exhibited a trigonal planar coordination while retaining some characteristics of häggite, such as the linear axial arrangement of V^IV octahedra responsible for producing a strongly focused multiple scattering contribution in the EXAFS signal at 3.7 Å (Figure 4, Table S6). In both the control and oxidative experiments, equatorially neighboring V^IV are present at distances similar to häggite and Gain’s hydrate.⁶³ The remaining oxygens constituting the octahedral coordination sphere of V^IV in the presence of birnessite are likely contributed by transient waters.⁶⁴ Although trigonal V^IV surface complexes have been observed on Fe^II/III (oxy)-hydroxides, the inclusion of V^IV–Mn backscattering atoms did not statistically improve the fit as evaluated by the Hamilton test.²⁰ In particular, we considered the formation of a ²C V^IV–Mn complex given the apparent V–V distance at 3.32 Å (Table S6) at the early stages of the reaction, and the absence of that scattering contribution in the early stages of the control experiment, as well as its similarity to the V^IV–Fe^II distance in the ²C complexes reported by Vessey and Lindsay.²⁰ However, the coordination number when the path is fit as Mn instead of V does not rise above 0.8, which is over a unit less than is required by the formation of a ²C complex. Given this, we assigned this path to a V^IV–V^IV backscattering interaction, which takes into account the findings of Besnardiere et al. for a V^IV–V^IV backscattering interaction at 3.37 Å in a häggite sample with a coordination number of one.⁶³ Likewise, although V^V–Mn^II phases exist, they were not detected in the experiments through EXAFS or XANES analyses.⁶⁵

Figure 4.

Comparison of V speciation of solids collected at start (1 min) and end (140 h) of the reaction between V^IV and birnessite with 3.5 mM V^IV initial concentration and 1 g L⁻¹ birnessite. (a) k³-weighted EXAFS spectra showing the transition from V^IV octahedra at 1 min to V^V tetrahedra at 140 h. (b) RSF plot of panel (a). (c) XANES spectra showing oxidation of V^IV to V^V indicated by a shift in edge position and the development of absorption features at 5490 and 5507 eV corresponding to 1s to 4_Pz and 4_Pxy transitions. (d) Structural models of V^V adsorbed on birnessite as determined by shell-fitting of EXAFS. Red dashed lines represent the fit, and black solid lines represent the data.

The difference in oxygen coordination between the initial forms of V^IV in the presence and absence of birnessite is likely due to the relative rates of crystallization versus oxidation when in the presence of birnessite. In the control experiment, V^IV precipitated and ripened in solution before being dried under an anoxic atmosphere prior to EXAFS analysis, whereas in the presence of birnessite, the V^IV recrystallization pathway was in direct competition with the oxidation pathway provided by contact with the birnessite. A comparison of the second-order rate constants presented in Table 1 and Table S4 for the oxidation by birnessite and V^IV re-uptake by the V^IV solid (a proxy for the recrystallization process) reveals that oxidation is more than 7000 times faster; thus, aging of the initial V^IV precipitate is inhibited in the presence of the Mn oxide.

Tetrahedrally coordinated vanadate H₂VO₄⁻ was the predominate end-product in the 1 and 3.5 mM ex situ experiments, also forming a surface complex with the birnessite. However, some V^V polymerization occurred by the end of the reaction. Equilibrium speciation modeling predicted approximately 20% of aqueous V^V to be pyrovanadate (H₂V₂O₇²⁻) and 15–47% (1 mM V^V and 3.5 mM V^V, respectively) to be present as tetravanadate (V₄O₁₂⁴⁻). The proportions of pyrovanadate and tetravandate determined by ⁵¹V-NMR were found to be less than predicted, reaching 7.7 and 4.2% of the total V, respectively (Figure S4). Several possible reasons for this discrepancy include the formation of surface-bound pyrovanadate–birnessite complexes and paramagnetic interreference in the NMR spectra by the elevated concentrations of Mn^II_aq. V^V surface complexation by birnessite was dominated by a vanadate bidentate mononuclear edge-sharing complex (²E), likely localized to birnessite edge-sites (Figure 4). The EXAFS signal for this complex included a significant contribution from vanadate intratetrahedral multiple scattering, a phenomenon observed for vanadate adsorbed on Fe oxides.^20,33 When a Mn–O length of 1.90 Å is applied (Table S5), a ∠V–O–Mn angle of 102.8° is obtained which is comparable to the ²E complex of vanadate adsorption on ferrihydrite.³³ Although EXAFS spectra with greater k ranges are needed to definitively identify whether pyrovanadate was also complexed to the birnessite surface, the addition of a neighboring V backscattering atom at the V–V pyrovanadate distance of 3.43 Å leads to an improvement in the overall fit and decreased uncertainties calculated for the floated variables (Table S6, I_r = 0.53). Formation of bidentate corner sharing (²C) V–Mn complexes is also possible, though inclusion of these paths did not statistically enhance the EXAFS fit (I_r = 0.71). Larsson et al.³³ previously noted that the EXAFS contribution from intratetrahedral multiple scattering obscured the ability to resolve any ²C complexes between V^V and ferrihydrite. Likewise, the presence of several types of intratetrahedral multiple scattering with a half-path length of 3.4–3.5 ų³ as well as a V–O–Mn multiple scattering path at 3.36 Å would make the determination of a pyrovanadate V–V distance at 3.4 Å difficult without higher resolution data. A summary of the oxidation of V^IV by birnessite and interactions of the products are illustrated in Figure S6.

Comparison to Other Metal(Loid) Contaminants.

Although no studies have examined V surface complexes formed on Mn oxides, several studies have characterized V^IV and V^V adsorption on Fe oxides.^20,33,34,66 Vessey and Lindsay²⁰ reported the formation of V^IV and V^V adsorption complexes in ²C and ²E configurations on the surface of mixed-valence Fe^II/Fe^III oxides. Larsson et al. found that V^V did not form a ²C complex on ferrihydrite, which they attributed to the longer V–O bond of V relative to the Me–O distance of an element like As which predominantly forms a ²C complex on the mineral.³³ Our results indicate that V^V primarily forms a ²E complex at birnessite edge sites, which is in agreement with the work by Larsson et al.³³ The V–Mn distance identified in this study (2.79–2.82 Å) is comparable to the ²E V–Fe distance (2.789 Å) on ferrihydrite at neutral pH, but approximately 0.12 Å greater than the ferrihydrite V–Fe²E distances published by Vessey and Lindsay²⁰ likely due to structural differences in ferrihydrite arising from the synthesis techniques techniques used, but is consistent with the ab initio calculations of Peacock and Sherman.³⁴ The V-2E distance was and approximately 0.1 Å greater than the ²E complex formed by As^V on what was hypothesized to be γ-MnOOH-passivated birnessite.^20,56 This confirms that the sorption behavior of V^V is more similar to that of other tetrahedrally coordinated metal(loid)s than octahedrally coordinated metals like Cd or Ni, which have Cd–Mn and Ni–Mn distances of 3.66 and 3.49 Å, respectively.^61,67 Likewise, our predicted ∠V–O–Mn angles range from 101.3° to 102.8°, which are approximately 4° greater than for the V–O–Fe angle with ferrihydrite. This difference indicates that the formation of the ²E complex induces a greater level of Jahn–Teller distortion into the Mn octahedra than when adsorbed onto Fe^III octahedra.³³ The greater distortion could be caused by a shorter average Mn–O distance (1.90 Å) in birnessite than the Fe–O distances of ferrihydrite (1.91, 1.92, or 2.04 Å depending on the Fe center).⁶⁸

Many studies have examined metal anion and cation adsorption on birnessite surfaces that provide insight into other possible V complexes that may have formed. Terminal hydroxyl groups of birnessite can carry a positive charge when doubly protonated and are therefore the most likely sites where anion adsorption is expected to occur (e.g., As, Mo, and Sb).^69–72 Accordingly, we expect V^V to be primarily bound at edge sites. However, V^IV is cationic and could electrostatically attract to negatively charged surface vacancy sites in a manner similar to Mn^II, Zn^II, and Ni^II. This would be physically feasible given a hydrated diameter of <4.2 Å for aqueous V^IV and an interlayer d₁₀₀ spacing of 7.2 Å for birnessite that yields a 5.2 Å cation gap.^61,62,64 However, it likely reacts at the edge sites, given the reactivity of these sites toward other oxyanions.^71,72 We were unable to detect any V^IV sorption complexes on birnessite, which may have been due to relatively fast V^IV oxidation by birnessite compared to the speed of XAS data acquisition. A previous study reported that the fast rate of As^III oxidation by birnessite precluded the observation of any As^III–Mn complex using EXAFS.⁵⁶ However, the role of edge sites in the oxidation of V^IV may be observable in the aqueous ex situ data (Figure 1) via the presence of the two phases discussed above. Rapid accumulation of structural Mn^III at particle edge sites during the first phase of the reaction would lead to the decreased reaction rate observed in the second phase, given the lower potential of Mn^III to oxidize V^IV. The accumulation of Mn^III at birnessite particle edges in response electron transfer from aqueous reductants is well documented and has likewise been shown to influence the adsorption and oxidation of other environmental contaminants such as Sb and Ni.^71,73

The oxidation of metal oxides has been shown to be limited by the solubility of one or both members of the redox couple.²⁹ For example, the initial rate of V^IV oxidation (K₀) was 80 to 200 times slower than the published rates of As^III oxidation, which highlights how V^IV oxyhydroxide oxidation is likely limited by solubility.^74,75 The decreasing solubility of the V^IV oxyhydroxide as it ages from an amorphous precipitate into a more crystalline solid solution of V^IV hydroxides imposes constraints on its subsequent oxidation similar to the constraints imposed on the oxidation of Cr^III by birnessite due to the exceptionally low solubility of Cr^III.^26,76 The observed dissolution rate of the amorphous V^IV precipitate is more than three orders of magnitude greater than the dissolution rate of Cr^III from minerals such as chromite, although the solubility of the V^IV precipitate is reduced as it ages (Figure S3).⁷⁶

Environmental Implications.

In the environment, the concentration of aqueous V^IV is controlled by the solubility of V^IV oxyhydroxides,¹ pH,⁷⁷ pO₂,¹⁸ and the reactivity of other soil mineral phases.^31,35,78 The heterogenous oxidation of sparingly soluble V^IV by birnessite increased aqueous V^V concentrations as well as promoted the reductive release of Mn^II, an underrecognized groundwater contaminant into solution.^79,80 Although birnessite adsorbs a portion of aqueous V^V and Mn^II, back-reactions and the formation of new mineralogical phases like Mn^III hydroxides may alter the reactivity and retention capacity of birnessite. Our findings suggest that in environments subject to redox cycling, such as sediments in aquifers subject to seasonal draw-down and recharge, V can be abiotically oxidized via interactions with Mn oxides, which are also known to accumulate in such environments.^13,17,29,30 We hope that the information presented here can be used to help parameterize predictive models to better predict the extent of groundwater V contamination in areas where Mn oxide content is well defined.