Acid birnessite was synthesized using the acid reduction method from McKenzie (1971) A 42 mL solution of 10 M HCl (certified ACS Plus grade, Fisher Chemical) was added dropwise to a vigorously boiling 0.58 L solution of 0.40 M KMnO₄ (Beantown Chemical, 99%) in ultrapure water (with a resistivity of 18.2 MΩ.cm). After the addition of the acid, the solution was boiled for an additional 10 min and cooled to room temperature. Using vacuum filtration with filter paper (Fisher Brand, P2 fine porosity glass fiber filter paper), the precipitate was isolated and washed with ultrapure water until the pH was 6. Then the precipitate was dried for 72 h in a 60°C oven.

The “c-disordered” H⁺ birnessite was synthesized using the redox methods from Villalobos et al. (2003) A 160 mL solution of 0.20 M KMnO₄ was stirred, while a solution of 180 mL of 0.51 M NaOH (Alfa Aesar, 98%) was slowly added. Next, a 160 mL solution of 0.58 M MnCl₂ (Acros Organics, 99%) was slowly added to the mixture before settling for 4 h. Using the triclinic sodium-birnessite method, the suspension was washed with a solution of 1.0 M NaCl (Fisher Bioreagents, 99.5%) and ultrapure water. Then using NaOH, the pH was adjusted to 8. The precipitate was then dried for 72 h in a 50°C oven.

In order to monitor and quantify the kinetics of the oxidation of SeO₃ ²⁻ to SeO₄ ²⁻, batch experiments were conducted using the different birnessites as a function of concentration level at varying pH in 50 mL centrifuge tubes. Each mineral was suspended in ultrapure water and adjusted to the desired pH before being spiked with a stock solution of SeO₃ ²⁻ (as NaSeO₃, VWR, high purity grade). The pH of the solution was periodically checked and adjusted using HCl or NaOH. At specific time points, the samples were centrifuged for 10 min at 5,000 rpm (ThermoScientific, Sorvall ST16 Centrifuge). For analysis, an aliquot of the supernatant was taken to be analyzed using tandem ion chromatography–inductively coupled plasma–mass spectroscopy (IC-ICP-MS). Sorption control experiments were performed under identical conditions, but spiked with SeO₄ ²⁻ (as NaSeO₄, ThermoFisher Scientific, 99.8+% purity) instead of SeO₃ ²⁻. These experiments were conducted to determine if the mineral will retain the analyte and cause an underestimation of the SeO₄ ²⁻ concentration. Additional control experiments were conducted to determine the volatility of the selenium species under the investigated conditions, where SeO₃ ²⁻ and SeO₄ ²⁻ were spiked into solutions of ultrapure water at the desired pH values.

The two selenium species were analyzed using tandem IC-ICP-MS, with the individual instrument conditions having been described in Szlamkowicz et al. (2022) This tandem method allows for the simultaneous speciation and quantification of selenium at environmentally realistic concentrations (nM range). This occurs through the IC separating the different species of selenium using a selective ion exchange column, the effluent of which is fed into the ICP-MS through a capillary line. Since the ICP-MS has much lower limits of detection than the IC does on its own, it was used as a detector in place of the one within the IC. Thus, tandem IC-ICP-MS combines the separatory capabilities of IC with the low limits of detection of an ICP-MS, all in a single analytical step.

The two birnessite minerals were characterized using X-ray diffraction (XRD, Empyrean, PANanalytical), with analysis parameters having been previously described in Stanberry et al. (2021a) and Stanberry et al. (2021b). Additionally, the manganese average oxidation state (AOS), specific surface area, and pH where the surface of the mineral has a net neutral charge (pH_pzc) was performed, with the results having been described previously in Szlamkowicz et al. (2023).

Two different forms of the oxidizing agent, birnessite, are initially investigated to determine its influence on the oxidation of selenium under environmental conditions. The two forms of birnessite used are acid birnessite and “c-disordered” H⁺ birnessite, whose main differences in mineral structure has been established previously by Szlamkowicz et al., with the X-ray diffraction (XRD) patterns seen in Supplementary Figure S1 (Szlamkowicz et al., 2023). These two minerals exhibit various degrees of reactivity towards selenite (SeO₃ ²⁻), as seen below in Figure 2. The concentration of SeO₃ ²⁻ in solution decreases more rapidly and extensively with the “c-disordered” H⁺ birnessite as compared to the acid birnessite. The rapid disappearance of Se in the aqueous phase is attributed to the fast and strong adsorption of SeO₃ ²⁻ onto the active sites along the surface of the mineral through strong inner-sphere complexes (Dwivedi et al., 2022). The disappearance of SeO₃ ²⁻ in solution with the “c-disordered” H⁺ birnessite was quite rapid leading to only the data from acid birnessite being used when trying to determine the reaction order for these experiments. It was determined that the reaction between SeO₃ ²⁻ and acid birnessite follows a second order reaction, as seen in Supplementary Figure S2. It was determined that the reaction rate constant was 3.73 × 10⁻⁴ M⁻¹ min⁻¹. This follows the second order reaction observed in a previous study, where the sorption and oxidation of SeO₃ ²⁻ was studied using a birnessite mineral in stirred flow reactors (Balistrieri and Chao, 1990).

In the case of both birnessite minerals at pH 5, there is a significant disappearance of SeO₃ ²⁻ from the solution and no release of selenate (SeO₄ ²⁻) into the aqueous phase. This is attributed to the lower pH of the system exhibiting a weaker electrostatic repulsion between SeO₃ ²⁻ and the surface of the mineral, which results in a greater extent of sorption (Balistrieri and Chao, 1990). This is attributed to the selenite in the system forming inner-sphere complexes, while selenate forms outer-sphere complexes with MnO₂ minerals (Li et al., 2021). Electrostatic interactions cause selenite to form outer-sphere complexes, while inner-sphere complexes require more than purely Coulombic forces (Balistrieri and Chao, 1990). Control samples of SeO₄ ²⁻ were prepared, to determine if there is a sorption mechanism of SeO₄ ²⁻ in the system, which would cause an underestimation of the mass balance of selenium. However, the sorption controls of SeO₄ ²⁻ indicate that the minerals are capable of negligible levels of sorption (Figure 2). This indicates that the disappearance of selenium from the aqueous phase is only a result of the sorption of SeO₃ ²⁻, and not from oxidation. This is due to SeO₃ ²⁻ having a much stronger adsorption to metal oxide surfaces through the formation of strong inner-sphere complexes (Dwivedi et al., 2022). Conversely, SeO₄ ²⁻ adsorbs to metal oxide surfaces through weaker outer-sphere complexes (Dwivedi et al., 2022). This follows XPS data from Li et al. where the post reaction solids only exhibited the reduced form of selenium, SeO₃ ²⁻ (Balistrieri and Chao, 1990).

The percent of SeO₃ ²⁻ oxidized under the varying pH values for both minerals has been summarized below in Table 1. As the pH increased, acid birnessite experienced a decrease in the extent of SeO₃ ²⁻ sorption, from 90% ± 2 at pH 5%–64% ± 4 at pH 8. However, the “c-disordered” H⁺ birnessite was not impacted by the increasing pH (Figure 3). This disparity is attributed to the “c-disordered” H⁺ birnessite having a consistent net surface charge throughout the different investigated pH values. The reason the “c-disordered” H⁺ birnessite was less affected by the increase in pH of the system is due to it having a lower pH_pzc for the vacancy sites than acid birnessite, <3 and 3.5–4.0, respectively (Szlamkowicz et al., 2023; Arias et al., 2019). Acid birnessite has a pH_pzc that is closer to pH 5, causing the vacancy sites to have more positive sites amidst the negative sites along the mineral surface. On the other hand, the surface of “c-disordered” H⁺ birnessite has a mostly negative charge, with few residually positive charged vacancy sites. The mostly negative charge is due to the pH of the system being above the pH_pzc of “c-disordered” H⁺ birnessite. Some residual positive vacancy sites remain on the mineral since the pH of the system is close to the pH_pzc values. The residual positive active sites will have a greater electrostatic attraction to SeO₃ ²⁻ at pH 5, which will decrease as the pH increases to 8 and the sites become negatively charged. With this, “c-disordered” H⁺ birnessite was able to sorb 97% ± 0.2 at pH 5, and 95% ± 0.8 at pH 8.

The greater extent of disappearance of SeO₃ ²⁻ in the experiments using the “c-disordered” H⁺ birnessite can be attributed to it having a higher specific surface area (111 m²g⁻¹) than acid birnessite (35 m²g⁻¹) (Szlamkowicz et al., 2023). A greater specific surface area indicates a larger number of surface sites for SeO₃ ²⁻ sorption.

Common environmental cations were sorbed onto the two birnessite minerals, to determine if there is an effect of surface passivation. If the active sites on the birnessite mineral were preoccupied by the presence of the sorbed cations, there is the possibility of a decrease in the sorption of selenium in the system. The cations investigated were Ca²⁺ (CaCl₂, Fisher Bioreagents), Mn²⁺ (MnCl₂, 99+%, Thermo Scientific), and Na⁺ (NaCl, Certified ACS, Fisher Chemical). To prepare the sorbed birnessites for use in selenium experiments, the minerals were suspended in a 50 mg L⁻¹ solution of each cation. These mineral suspensions were allowed to spin for 3 weeks, prior to separating the supernatant and analyzing using ICP-MS. The minerals were rinsed using 10 mL of ultra-pure water, before sequestering the supernatant for analysis, to monitor for any cation loss during the washing phase.

Significant extents of sorption was observed for all cations on both birnessite minerals, as summarized in Supplementary Table S1. However, there was no statistical difference in the sorption of SeO₃ ²⁻ between the cation sorbed and plain birnessite minerals, as summarized in Table 2 and in Supplementary Figures S3, S4. This is attributed to there being two different active sites on the surface of birnessite minerals, each with a different localized charge. Having a positively charged edge site allows SeO₃ ²⁻ to readily sorb. On the other hand, the Ca²⁺, Mn²⁺, and Na⁺ are sorbed onto the negatively charged vacancy sites of the mineral.

Artificially elevated concentrations of SeO₃ ²⁻ were also used to investigate the concentration dependency of SeO₃ ²⁻ when reacted with both acid and “c-disordered” H⁺ birnessites. Similar to previous experiments conducted at realistic concentrations of SeO₃ ²⁻, it was seen that as the pH increased, the extent of sorption decreased with acid birnessite while remaining unchanged with the “c-disordered” H⁺ birnessite (Figure 4). The percent of SeO₃ ²⁻ sorbed under the varying pH values for both minerals at elevated concentrations has been summarized below in Table 3. Even throughout the experiments with an elevated initial SeO₃ ²⁻ concentration, there was minimal observed oxidation to SeO₄ ²⁻. The only exception to this trend was with acid birnessite at pH 5, where there was a 12% ± 1 formation and release of SeO₄ ²⁻ during the reaction.

However, it is important to note that the formation of SeO₄ ²⁻ does not begin to appear until a hundred hours into the experiment at artificially elevated concentrations (Figure 4), depending on both the mineral and the pH. Conversely, the experiments using environmentally realistic concentrations (Figure 3) reach equilibrium within 5 h. This illustrates the elevated concentration reaction being a slow sorptive and oxidative process. This can be further seen in Supplementary Figures S2, S5, where the SeO₃ ²⁻ reaction rates are depicted for the disappearance of SeO₃ ²⁻. In these figures, it is observed that the rate of sorption is much slower when using artificially elevated concentrations of SeO₃ ²⁻. This difference of time frame and reactivity is important to understand when devising remediation methods for selenium contamination in the environment.