Sarah
Balgooyen
a,
Gabrielle
Campagnola
b,
Christina K.
Remucal
*ab and
Matthew
Ginder-Vogel
*ab
aEnvironmental Chemistry and Technology Program, University of Wisconsin–Madison, 660 N. Park St., Madison, WI 53706, USA. E-mail: remucal@wisc.edu; Web: http://@remucal Fax: +1 608 262-0454; Tel: +1 608 262-1820
bDepartment of Civil and Environmental Engineering, 660 N. Park St., Madison, WI 53706, USA. E-mail: mgindervogel@wisc.edu; Web: http://@profmattgv Fax: +1 608 262-0454; Tel: +1 608 262-0768
First published on 27th November 2018
Bisphenol A (BPA) is an endocrine disrupting compound commonly found in natural waters at concentrations that are considered harmful for aquatic life. Manganese(III/IV) oxides are strong oxidants capable of oxidizing organic and inorganic contaminants, including BPA. Here we use δ-MnO2 in stirred flow reactors to determine if higher influent BPA concentrations, or introduction rates, lead to increased polymer production. A major BPA oxidation product, 4-hydroxycumyl alcohol (HCA), is formed through radical coupling, and was therefore used as a metric for polymer production in this study. The influent BPA concentration in stirred flow reactors did not affect HCA yield, suggesting that polymeric production is not strongly dependent on influent concentrations. However, changes in influent BPA concentration affected BPA oxidation rates and the rate of δ-MnO2 reduction. Lower aqueous Mn(II) production was observed in reactors at higher BPA introduction rates, suggesting that single-electron transfer and polymer production are favored under these conditions. However, an examination of Mn(II) sorption during these reactions indicated that the length of the reaction, rather than BPA introduction rate, caused enhanced aqueous Mn(II) production in reactors with low introduction rates and longer reaction times due to increased opportunity for disproportionation and comproportionation. This study demonstrates the importance of investigating both the organic and inorganic reactants in the aqueous and solid phases in this complex reaction.
Environmental significanceManganese oxides are capable of degrading bisphenol A (BPA) and other phenolic compounds via oxidation in both environmental and engineered systems. Although the oxidation mechanism of BPA is complex, the influent concentration of BPA into a manganese oxide reactor does not alter the oxidation mechanism or products. The concurrent reduction of manganese oxide is strongly driven by the introduction rate of BPA, due to the longer reaction time at low introduction rates. BPA oxidation products include polymers, which can couple with organic matter and other compounds found in the environment forming unknown high molecular weight products. The results of this study show that product distribution observed in a controlled setting will reflect that of the environment, despite the 20–160 μM range of influent BPA concentrations used in this study. |
BPA is susceptible to oxidation by manganese oxides.12,18,26,27 Similar to other phenols, BPA undergoes a one-electron transfer with manganese oxides to form a radical species that can form polymeric products through radical coupling or undergo further oxidation through a second one-electron transfer.1–3,10,11,28,29 BPA oxidation is affected by MnO2 concentration, pH, and metal cosolutes, and 11 transformation products have been identified, including 10 phenols and 4 polymers.12 4-Hydroxycumyl alcohol (HCA) is a major product of BPA oxidation and is generated at yields of up to 64% HCA per mole of BPA.30 Note that this calculation is based on direct measurement of HCA and does not consider oxidation of HCA by manganese oxide.30 Since HCA is formed through radical coupling, its production can potentially be used to probe the relative amount of polymeric coupling.
Oxidation rates of organic compounds by manganese oxide are highly dependent on mineral properties. The reaction follows pseudo-first-order kinetics during the initial phase, but the rate of oxidation decreases as the reaction proceeds.11,12,14,31,32 Previous investigations of manganese oxide transformation during organic compound oxidation are limited, but they provide unique insights into changes to the mineral surface. For example, decreased rates of phenol, aniline, and triclosan oxidation by MnOx are associated with decreasing oxidation state and accumulation of reduced manganese species and organic species on the mineral surface.19 Similarly, δ-MnO2 can transform to other phases after accumulating Mn(III) in the presence of high concentrations of fulvic acid.33 Our previous study using Mn(III)-rich MnO2 and BPA shows that changes at the mineral surface are enough to decrease the oxidation rate even without changes in the bulk mineral oxidation state.18
Previous studies typically use batch reactors to characterize the reactivity of manganese oxides.1 These closed systems are experimentally simple and results can be readily compared to previous data. However, stirred flow reactors can provide further benefits, allowing for slow and constant addition of an influent media that can be easily altered. Additionally, batch reactors retain both organic and inorganic reaction products, which can affect the reaction. For example, the addition of Mn(II) can considerably decrease phenol oxidation rate by manganese oxides.12,13,26,27,34 In the environment and in stirred flow reactors, these products are constantly removed. A few studies use stirred flow reactors32,35 or column reactors36,37 to examine manganese oxide reactivity, but none have investigated both the changes in the aqueous and solid phases.
Here calculated HCA yield, aqueous Mn(II) production, and solid phase characterization are used to detect differences in BPA oxidation mechanism by δ-MnO2 in stirred flow reactors as a function of influent BPA concentration. This method is used to test the hypothesis that higher influent BPA concentrations lead to greater polymer production and therefore less overall electron transfer. Radicals formed via single-electron transfer are of concern as they can couple with dissolved organic matter or other compounds in the environment, forming unknown high molecular weight products.38–40 HCA is used as an indicator of conditions that favor single-electron transfer (i.e., polymer production), rather than two sequential electron transfers (i.e., benzoquinone production). Aqueous Mn(II) is produced by reductive dissolution and is used along with solid-phase Mn speciation to quantitatively compare electron transfer across different solution conditions. We use this data, along with measurements from solid phase characterization using X-ray absorption near edge structure (XANES) spectroscopy and X-ray diffraction (XRD), to make inferences about the mechanism of the redox reaction and how the δ-MnO2 structure changes throughout the reaction. By performing these reactions in a stirred flow reactor, this study provides novel insights into the effects of contaminant loading over long time periods, which is more representative of contaminants in a flow-through treatment system.36,41
Stirred flow reactor effluent was analyzed by HPLC and ICP-OES to quantify BPA, HCA, and aqueous manganese. These samples were not filtered or quenched using excess ascorbic acid since the reaction stops upon exiting the δ-MnO2 reactor. Media with varying BPA concentrations (20–160 μM, corresponding to a BPA introduction rate of 20–160 nmol min−1) were used in stirred flow experiments to determine mechanistic differences due to influent BPA concentrations. The length of the experiment was adjusted so that 20–25 μmol BPA was added to each reactor in total.
Solids were recovered at the end of the reaction. This material was washed in methanol to remove organics, dried at room temperature, and ground before analysis. Average manganese oxidation number (AMON) was determined using XANES spectra collected at beamline 10-BM at the Advanced Photon Source at Argonne National Laboratory (Section S4†). Samples were prepared by diluting 3 mg of manganese oxide into 8 mg of polyvinylpyrrolidone, grinding until homogenous, and pressing into a 7 mm pellet. XANES data was analyzed for AMON using the Combo method.48 XRD patterns were collected (Rigaku Rapid II, Mo Kα source; λ = 0.7093 Å) to determine changes in the order and crystallinity of the mineral.
![]() | (1) |
In stirred flow reactors, a steady-state assumption was used to estimate the HCA yield once the reactions reach a plateau in BPA and HCA concentrations using the following equation:
![]() | (2) |
![]() | (3) |
![]() | ||
Fig. 1 Measured BPA concentrations and measured and theoretical HCA concentrations over time in a batch reactor containing 80 μM BPA and 0.33 g L−1 δ-MnO2 in a pH 5 acetate buffer. |
HCA is also susceptible to oxidation by δ-MnO2, albeit at a slower rate than BPA, which is consistent with previous studies.18,30 For example, the pseudo-first-order oxidation rate constants of BPA and HCA at pH 5 are 0.228 min−1 and 0.029 min−1, respectively (Fig. S6; Table S1†). As shown in Fig. 1, HCA concentration increases with time in BPA oxidation reactions, but then reaches a maximum and slowly decreases. Using relative BPA and HCA initial oxidation rate constants determined in separate batch reactors, theoretical HCA yields can be calculated for BPA oxidation reactions in both batch reactors and stirred flow reactors using eqn (1) and (2). For example, the HCA yield in a batch reactor with 0.33 g L−1 δ-MnO2 and 80 μM BPA at pH 5 is 44% (least-squares fit line in Fig. 1).
Aqueous Mn(II) is a product of Mn(III/IV) reduction by phenols and is commonly used to quantify reductive dissolution.12–14,19,26,31,32,34 Here, Mn(II) appears in the reactor effluent after the first hour (Fig. 2), showing that reductive dissolution of the mineral is occurring.2,3,50 After six hours, dissolved Mn(II) reaches a maximum and returns to below detection limit as δ-MnO2 stops reacting with BPA and its transformation products. The BPA concentration experiments described below focus on the early stages of this reaction, where 70–100% of BPA is oxidized by δ-MnO2 during the retention time, similar to previous studies investigating As(III)35,51 and aniline32 oxidation by MnO2.
The BPA oxidation rate decreases with exposure of δ-MnO2 to BPA throughout the shorter time-scale of these experiments, as was observed in the 140 hour experiment (Fig. 2). The BPA concentration in the effluent reaches an apparent plateau in which 5–30% of the BPA is oxidized, depending on the concentration of the influent solution (Fig. 3a). At the lowest BPA introduction rate (20 nmol min−1), BPA appears in the effluent after 11 μmol of BPA are introduced and reaches a plateau almost immediately (i.e., after 13 μmol of BPA are introduced). This trend is also followed for the 40 and 80 nmol min−1 reactors. The plateau that is reached is modeled using a steady-state approximation to calculate BPA oxidation rate constants and HCA yields. At the highest BPA introduction rate (160 nmol min−1), BPA also appears in the effluent after 11 μmol of BPA are introduced but no plateau is observed during the reaction period (25 μmol of BPA introduced).
HCA concentration in the effluent varies with influent BPA concentration, with higher concentrations of BPA leading to higher concentrations of HCA in the effluent and vice versa (Fig. 3b). HCA is found in the effluent earlier in the reaction with lower BPA introduction rates (e.g., after 8 μmol of BPA introduced for 20 nmol min−1 reactor and after 13 μmol of BPA for 160 nmol min−1 reactor). Furthermore, a plateau of HCA concentration is reached sooner in reactors with lower BPA introduction rate (i.e., after 11 μmol of BPA for 20 nmol min−1 reactor and after 25 μmol of BPA for 160 nmol min−1 reactor). However, when plotted as a fraction of BPA consumed in the reactor (Fig. 3c), all reactors produce the same ratio of moles of HCA in effluent per moles of BPA consumed (30.0–33.1%) by the end of the reaction, excluding the 160 nmol min−1 reactor, which does not fully reach a plateau.
The steady-state approximation (Section S5†) and relative initial rates of BPA and HCA oxidation in batch reactors (Table S1; Fig. S6†) are used to calculate BPA oxidation rate constants and HCA yields in stirred flow reactors when the reaction reaches a plateau (e.g., after 12 μmol of BPA introduced for the 20 nmol min−1 reactor). Although the system is not truly at steady-state because δ-MnO2 reactivity changes gradually over extended reaction times (Fig. 2), the steady-state approximation is valid because the BPA and HCA concentrations are not changing from one time point to the next within this shorter timeframe. However, it is not possible to calculate HCA yield with the 160 nmol min−1 introduction rate because this reactor does not reach a distinct plateau by the end of the reaction time. Calculated BPA oxidation rate constants steadily increase with BPA introduction rate in reactors with 20, 40, and 60 nmol min−1 introduction rates (Fig. 4). This observation is in agreement with previous batch reactor studies that show an increase in oxidation rate with increases in either phenol concentration or MnO2 concentration.5,12
![]() | ||
Fig. 4 Calculated BPA oxidation rate constants (kBPA) and HCA yields (see eqn (1–3) in stirred flow reactors containing 1.58 g L−1 δ-MnO2 exposed to 20 μmol of BPA in 10 mM acetate buffer (pH 5) at an introduction rate of 20–160 μmol min−1 BPA. Reaction times range from 2.5–20 hours. |
HCA yields provide insight into changes in BPA reaction mechanism as a function of BPA concentration. Reactors with 20, 40, and 80 nmol min−1 BPA introduction rates have nearly identical HCA yields between 38–40% at the reaction plateau (Fig. 4). As a major oxidation product formed through radical coupling,30 the fraction of HCA production is theoretically proportional to the fraction of one-electron transfer reactions in this system. Since these polymeric products are more likely to form when there are high concentrations of BPA, we hypothesized that HCA yield would be higher when δ-MnO2 is exposed at a higher BPA introduction rate. However, the data indicates that there is no difference in HCA yield within this BPA concentration range.
Additional experiments were conducted at different BPA influent concentrations to further test whether HCA yields change under different conditions. First, influent BPA concentrations below 20 μM were preliminarily examined but were inconclusive. For example, a trial using 5 nmol min−1 BPA, which corresponds to an initial concentration of 5 μM BPA for 83 hours, shows that BPA is entirely consumed in the reaction (Fig. S9†). Therefore, HCA yields could not be determined at lower BPA influent concentrations due to complete BPA oxidation. Second, a separate experiment using longer reaction times compares HCA yields of 20 nmol min−1 and 160 nmol min−1 introduction rates after they have both reached a plateau (Fig. S10†). The observed yields of 44% and 40% respectively indicate that the higher introduction rate does not yield more HCA than lower introduction rate, further disproving our hypothesis that more polymeric products are produced at higher influent BPA concentrations.
Although the organic data does not indicate a shift in BPA oxidation mechanism, the inorganic data shows a strong trend among the reactors. Aqueous manganese concentrations in stirred flow reactor effluent show that there is a relationship between influent BPA concentration and Mn(II) production (Fig. 3d). At lower BPA introduction rates, Mn(II) is produced earlier in the reaction and in larger quantities than at higher BPA introduction rates. For example, the 20 nmol min−1 introduction rate results in 19.8 μmol total Mn(II) beginning after 11.4 μmol BPA is introduced, while the 160 nmol min−1 introduction rate results in 6.3 μmol total beginning after 18.7 μmol BPA is introduced. Despite the difference in Mn(II) produced, minimal bulk mineralogical changes are observed in the XRD patterns or XANES data. Fig. 5a shows that commonly observed changes, such as reduced tailing of the hkl diffraction band at 37° and the appearance of a dip at ∼47°, only noticeably appear in the 20 nmol min−1 reactor. Analysis of XANES data using the Combo method fits the raw data and provides a calculated AMON for the sample. The AMON of the starting material is 3.85 v.u. (90% Mn(IV), 5% Mn(III), 5% Mn(II)). In samples recovered from stirred flow reactors, the AMON decreased due to reduction, but was the same value of 3.67 ± 0.01 v.u. for all BPA introduction rates (Fig. 5b; Table S2†). Unsurprisingly, no mineral phase changes occurred due to the low pH and relatively low accumulation of reduced Mn. This is consistent with previous work that shows changes in Mn(III)-rich δ-MnO2 reactivity can occur due to mineral transformation at the surface, such as increased interlayer Mn(II/III), and not necessarily changes to the bulk structure.18
![]() | ||
Fig. 5 (a) XRD patterns of solids from each reactor in red overlaid by the starting material in black and (b) fitted XANES data of solids recovered from each reactor and the starting material. XANES data was analyzed using the Combo method.48 |
The differences in Mn(II) production, despite lack of changes in the bulk average manganese oxidation number, indicate that there are more overall electron transfer reactions occurring at lower BPA introduction rates, resulting in increased reductive dissolution of δ-MnO2. This is further shown by the estimated net electron transfer from organic compounds to δ-MnO2 calculated for each reactor (Table S3†), where electron transfer in 20, 40, 80, and 160 nmol min−1 reactors are estimated to be 79.7, 72.0, 64.9, and 50.7 μmol, respectively. There are several potential explanations for this observation. First, it is possible that two sequential single-electron transfers are favored in reactors with lower BPA introduction rates, forming more Mn(II) through a second single-electron transfer, rather than just one single-electron transfer reaction to produce Mn(III). However, if HCA is used to determine the relative amount of polymeric products formed, the absence of a trend in HCA yield over the 20–80 nmol min−1 reactors indicates that the proportion of single-electron transfer reactions, and therefore polymer production, is consistent over the range of BPA introduction rates. A second explanation is that the differences in reaction lengths allow for more redox reaction to take place, including oxidation of BPA oxidation products (e.g., HCA) and transfer of electrons within the manganese oxide mineral (i.e., disproportionation and comproportionation). This is supported by both data sets, but does not follow the generally accepted concept that more radical coupling and polymeric production will occur at higher concentrations of the target organic compound.52 A third possibility is that HCA does not accurately quantify polymeric production since other polymers can be formed by BPA oxidation.12
One way to narrow down these possibilities is to determine how accurately Mn(II) production quantifies the total number of electrons transferred to δ-MnO2. When Mn(II), Mn(III), and Mn(IV) are present in the same system, the mineral is susceptible to disproportionation and comproportionation (eqn (4)).53–56 A Mn(II) center and Mn(IV) center can exchange electrons, or comproportionate, to form two Mn(III) centers. Conversely, two Mn(III) centers can disproportionate to form Mn(II) and Mn(IV) centers. Due to these reactions, the fraction of reduced Mn(II) species formed in a reaction is not necessarily the same as the reduced species measured in the aqueous phase.
Mn(II) + Mn(IV) ⇌ 2Mn(III) | (4) |
The desorption experiments indicate that Mn(III) disproportionation occurs under our experimental conditions. In Reactor A, exposure of δ-MnO2 to 20 μM BPA (buffered to pH 5 in 10 mM acetate) for 20 hours produces a total of 6.75 μmol aqueous Mn(II). The addition of 25 mM Ca2+ for three hours desorbs an additional 1.54 μmol of Mn(II) that was generated during BPA oxidation (Fig. 6a). We then added 10 mM pH 5 acetate with no BPA for 7 hours, which does not react with δ-MnO2 (Fig. S1†), but allows time for disproportionation of Mn(III) to occur; minimal Mn(II) is produced during this time (i.e., a total of 0.06 μmol). A second introduction of 25 mM Ca2+ for two hours yielded 0.37 μmol additional Mn(II), indicating that Mn(II) production via disproportionation occurs during the time between Ca2+ introductions.
In Reactor B, δ-MnO2 is exposed to a 20 μM BPA solution (buffered to pH 5 in 10 mM acetate) for 20 hours, 10 mM acetate buffer for 9 hours, and a final 25 mM Ca2+ solution for two hours. Mn(II) is produced during BPA oxidation as described above, but does not desorb during exposure to 10 mM acetate (Fig. 6b). When the Ca2+ solution is introduced, Mn(II) desorbs from the mineral and is found in the effluent. Reactor A produces 1.54 μmol Mn(II), while Reactor B produces a total of 0.3 μmol Mn(II). Since both reactors were exposed to the same amount of reductant (i.e., 24 μmol BPA), this indicates that the Mn(II) produced during BPA oxidation is able to undergo comproportionation with Mn(IV) to form Mn(III) if it is not desorbed from the mineral in a timely fashion.
Three conclusions can be drawn from the desorption experiments. First, Mn(II) does not continue to desorb from the mineral after oxidation with BPA, indicating that desorption of Mn(II) occurs after the mineral reaches saturation either by Mn(II) or organics and is driven by exposure to organics and further production of Mn(II). Second, disproportionation, which is highly dependent on pH, occurs in the system within the timescale of the reaction (i.e., on the order of 7 hours at pH 5). Finally, the amount of Mn(II) desorbed by the first exposure to 25 mM Ca2+ in each reactor (i.e., 1.54 μmol in Reactor A with immediate exposure to Ca2+, and 0.3 μmol in Reactor B with delayed exposure) indicates that comproportionation is occurring in Reactor B. Overall, these desorption experiments show that both disproportionation and comproportionation are possible within the reaction times of our stirred flow experiments. This indicates that the differences in aqueous Mn(II) production (Fig. 3d) are due to the differences in reaction time rather than the BPA introduction rate, as increasing reaction time will lead to a proportional amount of electron transfer between manganese centers.
Footnote |
† Electronic supplementary information (ESI) available: Additional experimental details, Tables S1–S3, and Fig. S1–S10. See DOI: 10.1039/c8em00451j |
This journal is © The Royal Society of Chemistry 2019 |