Peroxymonosulfate-Based Electrochemical Advanced Oxidation: Complication by Oxygen Reduction Reaction

Abstract

Peroxymonosulfate (PMS)-based electrochemical advanced oxidation processes (EAOPs) have received widespread attention in recent years, but the precise nature of PMS activation and its impact on the overall process performance remain poorly understood. This study presents the first demonstration of the critical role played by the oxygen reduction reaction in the effective utilization of PMS and the subsequent enhancement of overall pollutant remediation. We observed the concurrent generation of H₂O₂ via oxygen reduction during the cathodic PMS activation by a model nitrogen-doped carbon nanotube catalyst. A complex interplay between H₂O₂ generation and PMS activation, as well as a locally increased pH near the electrode due to the oxygen reduction reaction, resulted in a SO₄^•–/^•OH-mixed oxidation environment that facilitated pollutant degradation. The findings of this study highlight a unique dependency between PMS-driven and H₂O₂-driven EAOPs and a new perspective on a previously unexplored route for further enhancing PMS-based treatment processes.

Short abstract

We report a systematic investigation of how oxygen reduction and H₂O₂ formation promote the electrochemical activation of peroxymonosulfate.

Introduction

Electrochemical advanced oxidation processes (EAOPs) have emerged as a viable alternative to conventional advanced oxidation processes (AOPs) for removing organic pollutants in wastewater.^1,2 In conventional AOPs, peroxides such as hydrogen peroxide (H₂O₂) and peroxymonosulfate (PMS; HSO₅^–) are typically activated within the AOP system to produce reactive radicals such as hydroxyl radical (^•OH) and sulfate radical (SO₄^•–),^3,4 possessing relatively high redox potentials [E°(^•OH/OH^–) = 1.90–2.7 V vs NHE, and E°(SO₄^•–/SO₄^2–) = 2.60–3.10 V vs NHE],^5,6 which facilitate the oxidation of organic pollutants. Given sufficient electrolyte conductivity,^2,7 an EAOP enables an electric current to directly drive redox reactions involving precursors, leading to greater energy efficiency compared to that of conventional AOPs that consume electricity to exert photolysis, sonolysis, and thermolysis of precursors.^8,9 An EAOP can be readily modularized and is, therefore, better suited to small scale, decentralized treatment compared to conventional AOPs employing redox agents (e.g., Fe²⁺/Fe³⁺ for H₂O₂ and Co²⁺/Co³⁺ for PMS) that typically require a larger reactor as well as secondary treatment (e.g., sludge disposal for Fenton-based AOPs).^7,10,11 Compatibility for automated control via current and voltage adjustment is another beneficial property of EAOPs.¹²

Reactions that lead to radical generation in electrochemical cells are complex. Precursors are activated mainly through cathodic reduction as presented in eqs 1 and 2. Various carbonaceous electrodes such as carbon nanotubes,¹³ activated carbon,¹⁴ and carbon nanofibers,¹⁵ often decorated with metal catalysts such as iron and cobalt,¹⁶⁻¹⁹ have been employed. In addition, radicals can be generated from water oxidation (eqs 3 and 4)²⁰ and sulfate oxidation (eq 5)^21,22 on the anode, depending on the property of the electrode (e.g., having a high overpotential for oxygen evolution reaction) and the composition of water (e.g., SO₄^2– concentration). It is worth noting that precursors themselves can be electrochemically generated, for example, via direct oxidation of the sulfate anion (eq 6).²³⁻²⁵ The potential production of H₂O₂ via oxygen reduction reaction (ORR, eq 7) by carbonaceous cathodes often used for precursor activation is another reaction that can complicate matters.^{16,17,26−28}

1

2

3

4

5

6

7

We provide an in-depth examination of the cathodic reactions that occur during the PMS-based EAOP with a particular focus on evaluating how inevitable cathodic H₂O₂ production affects PMS activation and pollutant removal performance. Previous studies of PMS-based EAOPs focused on the activation of PMS but rarely considered the potential complications of accompanying H₂O₂ production and subsequent reactions involving both AOP precursors. We present a complex interplay between a PMS-driven AOP and a H₂O₂-driven AOP as a phenomenon unique to PMS-based EAOPs in marked contrast to conventional AOPs, to guide further development and more robust engineering of this promising technology.

Experimental Section

Materials

Peroxymonosulfate (Oxone), an aqueous H₂O₂ solution (30 wt %), para-chlorobenzoic acid (pCBA), 2-hydroxyterephthalic acid (hTPA), tert-butyl alcohol (TBA), isopropyl alcohol (IPA), and cobalt(II) chloride were all ACS reagent grade or higher and purchased from Sigma-Aldrich. Disodium terephthalic acid (TPA, HPLC grade, ≥99.0%) and acetonitrile (HPLC grade, ≥99.9%) were obtained from TCI and J. T. Baker, respectively. 5,5-Dimethyl-1-pyrroline N-oxide (DMPO, GC grade, ≥99%) was purchased from DOJINDO. All solutions and suspensions were prepared using deionized water (>18.2 MΩ) from a Milli-Q system.

Nitrogen-Doped Carbon Nanotube Electrode Synthesis

Multiwalled carbon nanotubes (MWCNTs, ≥98.0% C, outside diameter of 6–13 nm, length of 2.5–20 μm, Sigma-Aldrich) and 1,10-phenanthroline monohydrate (1,10-phen, reagent grade, Sigma-Aldrich) at a 10:1 weight ratio were mixed with ethanol (EtOH, ≥99.5%) under vigorous agitation overnight. The suspension was dried at 80 °C, and the resulting powder was pyrolyzed using a tube furnace (STF1200 Tube Furnace, Across International) under an Ar atmosphere (UHP, 100%, Airgas) at 600 °C for 2 h at an initial heating rate of 5 °C/min to reach 600 °C.²⁹ N-CNT powder with ∼1.4 wt % N doping was obtained after cooling to room temperature (see Texts S1 and S2 and Figures S1–S3 for characterization). N-CNT ink was prepared by mixing 5 mg of N-CNT with 2 mL of isopropyl alcohol (≥99.5%) and 20 μL of a Nafion 117 solution (5%, Sigma-Aldrich) and subsequently sonicating for 2 h. The N-CNT electrode was fabricated by drop-casting 0.2 mL of a homogeneous ink suspension onto a 1 cm × 1 cm geometric area of a carbon paper (Fuel Cell Store). The electrode was fully dried under an infrared lamp prior to use.

Electrochemical Characterization

All electrochemical experiments were performed using a Bio-Logic VSP potentiostat under ambient conditions and conducted within a conventional glass H-cell separated by a Nafion 115 membrane (Fuel Cell Store). Unless mentioned, the N-CNT electrode fabricated following the aforementioned procedure and an Ag/AgCl electrode (CH111, CH Instruments) were employed as working and reference electrodes, respectively, for the electrochemical experiments; they were sectioned into one chamber of the cell. A Pt plate (99.95%, Chemistry Cabinet) was employed as the counter electrode in the other chamber. The electrodes employed in this study were clamped with stainless steel clamps first, and the current feeders were then clipped onto the clamps to maintain a stable and consistent electrochemical cell setup. Unless specified, all cyclic voltammetry (CV) and linear sweep voltammetry (LSV) experiments were conducted at a scan rate of 50 mV s^–1. Based on the LSV characterizations, we performed chronoamperometric (CA) experiments at −0.5 V vs Ag/AgCl using 20 mL of Na₂SO₄ (50 mM) as the supporting electrolyte (see Figure S4a for the cell setup). Before commencing the electrolysis, we confirmed that the initial pH of the electrolyte containing PMS was observed to be 4.0 ± 0.2 without further adjustment. For experiments under anoxic conditions, the electrolyte was purged with N₂ (99.996%, Airgas) for 30 min prior to electrolysis, and the glass cell was sealed tightly using a septum stopper and parafilm throughout the experimental duration. For characterizations of the N-CNT and CNT catalysts, each material was loaded onto a rotating ring-disk electrode (RRDE) assembly composed of a glassy carbon disk and Pt ring (AFE6R1PTPK, Pine Instruments) (see Figure S4b for the cell setup). The working electrode was prepared via drop-casting catalyst ink onto the glassy carbon surface; ink solutions for both materials were prepared following the same process described above. Next, 10 μL of the ink solution was pipetted onto the glassy carbon disk (0.196 cm²) to obtain a loading of 0.128 mg cm^–2. The H₂O₂ selectivity was calculated using the molar fraction relation

where I_disk, I_ring, and N refer to the disk current, ring current, and collection efficiency (the percentage of material collected at the ring interface), respectively. The collection efficiency was obtained experimentally using the [Fe(CN)₆]^3–/4– redox system at rotation rates between 400 and 2025 rpm and calibrated to be 23.8% (close to the theoretical value of 25.0%). The Pt ring was fixed at 0.59 V vs Ag/AgCl across all measurements. After all experiments, the Pt ring was cleaned by immersing in H₂SO₄ and then cycled under N₂-saturated Na₂SO₄ between −1.0 and 0.5 V vs Ag/AgCl until a stable voltammogram was attained.

Analytical Methods

The concentration of pCBA, TPA, and hTPA was measured using a high-performance liquid chromatograph (HPLC, Agilent Technologies 1260 Infinity) equipped with an Eclipse XDB-C18 column (5 μm, 4.6 mm × 150 mm) under an isocratic mode. pCBA was detected at 230 nm with a mobile phase of acetonitrile (50%) and a 0.1% phosphoric acid solution (50%) with a flow rate of 1.5 mL/min. For TPA and hTPA, we detected them at 315 nm with a mobile phase of acetonitrile (15%) and a 0.1% phosphoric acid solution (85%) at a flow rate of 2.0 mL/min. The concentration of PMS and H₂O₂ were quantified via a spectrophotometric titration method using potassium iodide (KI) and cerium sulfate [Ce(SO₄)₂], respectively.^30,31 Detailed experimental methods are described in Texts S3−S5 and Figure S5. We conducted in situ electron paramagnetic resonance (EPR) analysis using a Bruker EMXplus-6/1 instrument to monitor the DMPO adducts with sulfate and hydroxyl radical.

Results and Discussion

Influence of ORR on PMS Activation

We first confirmed that the N-CNT cathode effectively activated PMS; nearly 70% of PMS was activated within 20 min under a constant applied potential of −0.5 V vs Ag/AgCl (Figure 1a). The PMS activation was accompanied by the nearly complete removal of a model pollutant, pCBA (k_{pCBA/SO4^•–} = 3.6 × 10⁸ M^–1 s^–1 at pH 7; k_pCBA/^•OH = 5.0 × 10⁹ M^–1 s^–1 at pH 6–9.4)³²⁻³⁴ within 10 min of EAOP treatment (Figure 1b). Neither PMS nor pCBA decayed without the current, which verified that these reactions were electrochemically driven. When a pristine CNT-loaded cathode was used instead of the N-CNT cathode, PMS activation and pCBA degradation were both minimal (Figure S6), highlighting the importance of employing a proper cathode material to enable the PMS-based EAOPs. The high performance of N-CNT toward PMS activation is consistent with past studies that employed N doping to enhance the adsorption of oxygen functionalities.³⁵⁻³⁸ Pyridinic (38.7%) and graphitic (center, 15.3%; valley, 20.1%) nitrogen atoms on the N-CNT (Text S2, Tables S1−S2, and Figure S1f) are known to be effective catalytic sites for PMS activation.³⁸⁻⁴¹ In addition, EIS analysis suggests that N-doping also contributes to faster charge transfer compared with that of pristine CNT (Figure S3).

Figure 1.

(a) PMS decomposition and (b) pCBA degradation in the N-CNT cathode system. Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 2.5 mM PMS, ORR eliminated by N₂ purging of the electrolyte (with N₂ purging for the absence of ORR and without N₂ purging for the presence of ORR), applied potential of −0.5 V, temperature of 25 ± 1 °C. LSV scans with varying amounts of PMS on the N-CNT cathode (c) with N₂ purging (absence of ORR) and (d) with O₂ purging (presence of ORR). LSV scans with varying amounts of PMS on the CNT cathode (e) with N₂ purging (absence of ORR) and (f) with O₂ purging (presence of ORR). Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 0, 1, 4, and 8 mM PMS, temperature of 25 ± 1 °C. Note that the ohmic drop was negligible in this system on the basis of the observation depicted in Figure S7.

We performed the same chronoamperometric experiment except with purging of the electrolyte with N₂ to remove dissolved O₂ and exclude any ORR contributions at the cathode. This test was motivated by past reports labeling ORR as a potential parasitic reaction that competes with PMS reduction on the cathode surface.^14,15,42 Interestingly, we observed a drastic deterioration of the performance in the absence of ORR (i.e., with N₂ purging); <20% of PMS was activated (Figure 1a), and only 23% of pCBA was degraded (Figure 1b). These results suggest that the ORR significantly contributes to PMS activation and pCBA degradation.

We further examined the PMS reduction reaction and ORR on the N-CNT cathode by performing LSV experiments. We first ruled out the effects of hydrogen evolution reaction (HER) on the basis of the minimal currents observed in the absence of both PMS and O₂ [0 mM PMS with N₂ purging (Figure 1c)]. Consequently, we fixed the boundary of the reductive window to −0.8 V for subsequent experiments. As we increased the PMS concentration under N₂ purging, we observed small yet definitive increases in reductive current for the N-CNT electrode (Figure 1c). This reductive current became substantially larger when O₂ was introduced (Figure 1d). While the voltammograms did not change at relatively low PMS concentrations (i.e., 1 mM), the reductive currents were substantially larger at higher PMS concentrations (i.e., 8 mM). This result asserts a synergistic effect between PMS reduction and ORR. Also consistent with the results presented above, PMS negligibly impacted the currents for unmodified CNT in the absence (Figure 1e) and presence (Figure 1f) of O₂.

H₂O₂ Generation via Oxygen Reduction

We further confirmed that H₂O₂ was generated through two-electron ORR (eq 7) under these reductive conditions using an RRDE setup. In the absence of PMS and pCBA under O₂-saturated conditions, we observed approximately 20% H₂O₂ selectivity at −0.5 V vs Ag/AgCl for both CNT and N-CNT (Figure 2a,b). ORR currents were higher for the N-CNT, as the nitrogen sites have been previously reported to promote this pathway.^43,44 We then confirmed such trends in the batch cell; under the same experimental conditions, the measured H₂O₂ concentration was ∼0.76 mM after electrolysis for ∼1 h (Figure 2c). When PMS was added to the system, we noticed that the accumulation of H₂O₂ in the electrolyte was depleted (Figure 2c). In contrast, the impact of the addition of pCBA on this depletion was minimal. This observation suggests that H₂O₂ produced from ORR would react with PMS, which prevents H₂O₂ from accumulating in the electrolyte.

Figure 2.

(a) LSV ring and disk currents of bare glassy carbon, CNT, and N-CNT-deposited glassy carbon and (b) calculated H₂O₂ selectivity (%) from the results of panel a. (c) Electrochemically generated H₂O₂ concentration without N₂ purging (presence of ORR) with the N-CNT cathode. Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 2.5 mM PMS, applied potential of −0.5 V, temperature of 25 ± 1 °C.

Now having evidence for H₂O₂ generation via ORR, we first postulate that cathodic H₂O₂ activation (eq 2) is responsible for the aforementioned pCBA degradation enhancement by H₂O₂ (Figure 1b). If this assumption is correct, we should be able to observe some pCBA degradation by externally adding H₂O₂ (i.e., instead of generating via ORR), even in the absence of PMS. We therefore added 5 mM H₂O₂ to a N₂-purged electrolyte containing pCBA but no PMS. Surprisingly, we found that pCBA did not degrade at all (Figure S8a). Even though ∼15% of the H₂O₂ was decomposed within 30 min of electrolysis (Figure S8b), the oxidative environment was not sufficient to induce measurable pCBA degradation. When H₂O₂ was generated via ORR (∼0.4 mM over 30 min) in the absence of PMS, we did not observe any pCBA degradation. In addition, we could not detect any ROS such as ^•OH, SO₄^•–, and ¹O₂ via in situ EPR analysis (Figure S9). These results collectively suggest that the decreases in H₂O₂ concentration in the cathodic chamber (Figure S8b) are likely due to futile reduction and disproportionation into water (leading to the overall reaction as represented by eq 8),⁴⁵ not cathodic activation into ^•OH. Instead, pCBA degradation enhancement by H₂O₂ appears to require the presence of PMS.

8

Influence of pH

Another consequence of ORR would be the increase in pH during the course of PMS-based EAOPs. The two-electron ORR that produces H₂O₂ (eq 7) and the 4e^– ORR to H₂O (eq 9) consume protons and increase the pH. We indeed observed that the pH of the electrolyte increased from 4.1 to 10.5 (Figure 3a) during a 20 min electrolysis without N₂ purging. However, when we performed the same electrolysis after N₂ purging, the pH did not change at all, confirming the decisive role of the ORR in the pH increase.

9

Figure 3.

(a) pH of the electrolyte during electrolysis with and without N₂ purging. Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 2.5 mM PMS, applied potential of −0.5 V, temperature of 25 ± 1 °C. (b) pCBA degradation and (c) PMS decomposition at different pH values of the electrolyte in the presence of ORR. (d) H₂O₂ concentration resulting from ORR with and without PMS. Conditions: 50 mM phosphate buffer (electrolyte) with different pH values (3, 5, 7, 9, and 11), 25 μM pCBA, 2.5 mM PMS, applied potential of −0.5 V, temperature of 25 ± 1 °C.

To examine the effects of pH, we performed a set of experiments in which we fixed the electrolyte pH to various values using a phosphate buffer (50 mM). Our assumption about the interaction of PMS and H₂O₂ contributing to pCBA degradation is also consistent with the observation that pCBA degraded faster with an increase in pH, but this was true only up to pH 9 (Figure 3b). The pCBA degradation was the fastest at neutral to weakly basic pH (7–9), reaching approximately 98% removal within 10 min. At this pH, similar to the results depicted in Figure 1b, the pCBA degradation efficiency also drastically decreased with N₂ purging (Figure S10), once again highlighting the involvement of ORR. We also observed that PMS decayed faster as the pH increased (Figure 3c), consistent with greater H₂O₂ production at a higher pH (Figure 3d, in the absence of PMS, data for electrolysis for ≤20 min). At low pHs (3 and 5), pCBA degradation was slow, likely due to an inefficient ORR and the low availability of H₂O₂. H₂O₂ selectivity would be low under this condition; the accumulation of protons near the electrode surface may further reduce nearby H₂O₂ molecules back to H₂O (eq 10).⁴⁶

10

At a higher pH of 11, the kinetics of pCBA degradation slowed despite faster PMS decay and sufficient H₂O₂ production. We attribute this to the occurrence of a few parasitic reactions involving deprotonated species (HSO₅^– ↔ H⁺ + SO₅^2–, pK_a = 9.3; H₂O₂ ↔ H⁺ + HO₂^–, pK_a = 11.6 at 25 °C).⁴⁷⁻⁴⁹ First, PMS can disproportionate to non-reactive sulfate according to the reaction presented in eq 11.⁵⁰ Second, a similar disproportionation reaction may also possibly occur with H₂O₂ at a high pH (eq 12), once again decreasing the oxidation power of H₂O₂. Finally, PMS can also self-decompose to form ¹O₂ (eq 13). Formation of ¹O₂ may decrease the rate of pCBA degradation, because ¹O₂ is a much weaker oxidant and less reactive with pCBA [E°(¹O₂ /O₂^•–) = 0.81 V vs NHE; k_pCBA/¹O2 = 1.4 × 10⁷ M^–1 s^–1]^51,52 compared to SO₄^•– and ^•OH.

11

12

13

To further probe the pH dependency of the interaction between PMS and H₂O₂, we spiked PMS into the H₂O₂ accumulated electrolyte after PMS-free electrolysis for 20 min as discussed above (Figure 3d). At pH 3, the ORR occurred continuously regardless of PMS addition, resulting in a continuous increase in the level of H₂O₂. At pH 5, after spiking PMS, the rate of H₂O₂ generation plateaued, which indicates that the rate of H₂O₂ production became comparable to the rate of PMS consumption. At neutral to alkaline pHs, we observed a significant decrease in the H₂O₂ concentration with PMS addition. Therefore, we conclude that the interaction between PMS and H₂O₂ would also be more favorable at high pH.

These results collectively suggest that pH plays a critical role in determining the efficiency of PMS-based EAOPs. The most effective pCBA degradation at pH 7–9 with buffer in fact explains our observation made with the unbuffered system. We note an initial lag phase in pCBA decay followed by rapid degradation in Figure 1b, mirroring the rapid increase in pH after an initially low pH in Figure 3a. The initial lag in pCBA decay is due to low beginning pH with the addition of acidic PMS (acidity of OXONE = 2.93).⁵³ Subsequent faster pCBA degradation is achieved as the pH increases due to ORR and reaches a neutral to slightly alkaline pH range, which has also been reported to be optimum for PMS-based EAOPs.^54,55

Identification of Reactive Species

We have so far demonstrated that the ORR has a significant influence on the PMS-based EAOP by generating a secondary oxidant, H₂O₂, and adjusting the pH to slightly alkaline conditions. Regardless of the way in which PMS and H₂O₂ interact (which will be discussed below), the radical species that result from the activation of PMS and H₂O₂ are primarily responsible for the oxidative degradation of pCBA, because we observed efficient pCBA degradation only when PMS was present and ORR occurred.

We verified the dominant role of radicals by observing that pCBA degradation was completely inhibited by adding excess EtOH (50 mM), an effective scavenger for both SO₄^•– (k_{EtOH/SO4^•–} = 1.6–7.7 × 10⁷ M^–1 s^–1) and ^•OH (k_EtOH/^•OH = 1.2–2.8 × 10⁹ M^–1 s^–1) (Figure 4a).⁵⁶⁻⁵⁸ When we added excess TBA (50 mM), however, the pCBA degradation was initially similar (i.e., little scavenging) but later significantly inhibited. Overall, only ∼50% of pCBA removal was achieved over 20 min. TBA is an effective scavenger for ^•OH (k_TBA/^•OH = 3.8–7.6 × 10⁹ M^–1 s^–1) but not for SO₄^•– (k_{TBA/SO4^•–} = 4.0–9.1 × 10⁵ M^–1 s^–1).⁵⁷⁻⁵⁹ Furthermore, the observed abatement of pCBA removal was indeed attributed to scavenging reactions rather than other factors such as electrode damage (Figures S11 and S12). Consequently, we concluded that SO₄^•– played a dominant role during the early phase of pCBA degradation and ^•OH became a dominant oxidation species during the later phase.

Figure 4.

pCBA degradation with EtOH (50 mM) and TBA (50 mM) as radical scavengers in an electrolyte (a) in the presence of ORR without N₂ purging and (b) in the absence of ORR with N₂ purging. Without scavenger plots is the duplication of data in Figure 1b. (c) Generation of hTPA from the reaction of ^•OH and TPA in electrolysis with and without N₂ purging. Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 2.5 mM PMS, 0.5 mM TPA, applied potential of −0.5 V, temperature of 25 ± 1 °C. (d) In situ EPR analysis of SO₄^•– and ^•OH in an electrolyte in the presence of ORR. Conditions: microwave frequency of 9.83 GHz, microwave power of 6.325 mW, modulation frequency of 100 kHz, modulation amplitude of 1.0 G, sweep time of 30 s, 50 mM Na₂SO₄ (electrolyte), 2.5 mM PMS, 100 mM DMPO, applied potential of −0.5 V, temperature of 25 ± 1 °C.

We further found that, with N₂ purging, pCBA degradation was not significantly inhibited by excess TBA (Figure 4b); i.e., under this condition, the contribution of ^•OH was less important than that of SO₄^•–. On the basis of the extent of pCBA removal rate retardation by scavengers, we calculated the contributions of SO₄^•– and ^•OH to pCBA degradation (Figure S13) under the assumption that those two radicals were the dominant reactive species in this EAOP system. We employed another probe compound, TPA, which produces hTPA from the reaction with ^•OH (k_TPA/^•OH = 4.4 × 10⁹ M^–1 s^–1)⁶⁰ to quantify the contribution of ^•OH-driven oxidation versus SO₄^•–-driven oxidation (Figure 4c). These observations were further confirmed by conducting EPR analysis of the electrolyte in the presence of ORR (Figure 4d and Table S3). The overall level of generation of both radicals increased during the electrolysis, with the initial observation of SO₄^•–, followed by ^•OH enhancement subsequently. Results shown in Figure 4 and Figure S13 suggest that when we eliminated the ORR, a smaller fraction of pCBA removal resulted from ^•OH-driven oxidation.

We have discussed above the fact that the pH of the electrolyte increased over time; even when the bulk pH reached only up to neutral to slightly alkaline, we suspected that the local pH would be higher due to surface-mediated ORR. Alkaline PMS activation with HO₂^–, which results from disproportionation of H₂O₂ at higher pH values, can be expected as described in eq 14.^61,62 Due to this additional PMS activation, the overall amount of SO₄^•– production increases in the presence of ORR. In addition to alkaline PMS activation, we expect conversion of SO₄^•– to ^•OH particularly near the electrode surface due to the reaction (eq 15) (k = 6.5 × 10⁷ M^–1 s^–1).⁶³⁻⁶⁵ Results so far suggest that both SO₄^•– and ^•OH were the main reactive species leading to pCBA oxidation in this system. Their relative contribution appears to change as the pH increases due to ORR, consistent with past findings that SO₄^•–-dominant oxidation coverts to ^•OH-dominant oxidation above approximately pH 8.5–9.0 in PMS-based AOPs.⁶⁶

14

15

Interaction between PMS and H₂O₂

While we identified a complex interplay between PMS and H₂O₂ as a determining factor for the efficiency of the PMS-based EAOP, the question of how exactly H₂O₂ interacts with PMS remains. We first exclude direct interaction between PMS and H₂O₂. In a simple mixture (i.e., without an electrode), pCBA degradation was minimal within the experimental time scale (Figure 5a). We recall the previous conclusions that (1) H₂O₂ alone does not lead to pCBA degradation with or without current and (2) PMS alone (i.e., in the absence of ORR) leads to pCBA degradation but at a much slower rate than in the presence of H₂O₂ produced from ORR. These conclusions combined leave only one possible scenario for radical generation; electrochemical PMS activation initiates radical production and H₂O₂ subsequently enhances such production.

Figure 5.

pCBA degradation with PMS and varying H₂O₂ (a) without the electrode and (b) with current in the absence of ORR (with N₂ purging). Conditions: 50 mM Na₂SO₄ (electrolyte), 25 μM pCBA, 2.5 mM PMS, 1, 2.5, 5, and 10 mM H₂O₂, 5 mM for H₂O₂ only condition, applied potential of −0.5 V, temperature of 25 ± 1 °C. (c) H₂O₂ decomposition in the mixture with PMS and Co ion. Conditions: background solution of 50 mM Na₂SO₄, 2.5 mM PMS, 1 mM H₂O₂, 50 μM Co²⁺, initial pH adjusted to 7.

We found that externally added H₂O₂ exerted the same effect. Upon addition of H₂O₂ to the PMS-based EAOP with N₂ purging (i.e., no ORR and no H₂O₂ generation), pCBA degradation was enhanced (Figure 5b), and the kinetic enhancement was greater with a larger amount of H₂O₂ added. We thus conclude that SO₄^•– originates from the electrochemical PMS reduction (eq 1). PMS reduction can also produce ^•OH via HSO₅^– + e^– → SO₄^2– + ^•OH. Because we have already provided evidence that H₂O₂ is not a direct precursor for ^•OH (i.e., no pCBA degradation in the absence of PMS), the occurrence by ^•OH is likely the result of H₂O₂ formation from ORR and subsequent interaction with SO₄^•– originated from PMS activation to produce additional radicals (i.e., ^•OH and ^•OOH) (k_{H2O2/SO4^•–} = 1.2 × 10⁷ M^–1 s^–1).^67,68

We question whether a similar interaction between PMS and H₂O₂ can be realized by activating PMS via a different route, i.e., using Co²⁺, a well-known transition metal catalyst, to activate PMS into SO₄^•–.⁶⁹ We observed 50% decomposition of H₂O₂ in 20 min with the addition of Co²⁺ (50 μM) to the solution containing both PMS and H₂O₂ (Figure 5c). In contrast, the H₂O₂ concentration did not change with PMS alone (without Co²⁺) or with Co²⁺ alone (without PMS). Results shown in Figure 5c suggest that H₂O₂ consumption occurred only when PMS was activated. Considering that we produced SO₄^•– dominantly from PMS activation through Co²⁺, this result reaffirms that PMS activation and SO₄^•– formation are a prerequisite for H₂O₂ activation; collectively, they cooperatively enhance the EAOP performance.

A reaction pathway constructed for the PMS-based EAOP system that we examined is schematically illustrated in Figure 6. In the absence of O₂, PMS is reductively activated to produce SO₄^•– as the main reactive radical species. As is well-known in the literature, depending on pH, SO₄^•– further reacts with water to produce ^•OH to create a mixed-radical environment.^55,58 This represents the typical oxidation environment of the conventional PMS-based AOPs. In case of the PMS-based EAOP, the same cathode used to activate PMS produces H₂O₂, another AOP precursor, through ORR. Initial PMS activation via cathodic reduction to produce SO₄^•– is followed by further production of ^•OH from reaction of SO₄^•– with H₂O₂, reaction of SO₄^•– with OH^– (due to an increase in the local pH by ORR), and possibly other reactions for which we might not have accounted. This transition of the SO₄^•–-dominant oxidation environment to the SO₄^•–/^•OH-mixed oxidation environment is concurrently affected by the ORR and the resulting change in pH. This conjecture explains how H₂O₂ produced from ORR significantly enhances the PMS-based EAOP via a previously unexplored pathway.

Figure 6.

Schematic of the interaction pathway and reactive species in the PMS-based EAOP in the absence and presence of ORR.

Feasibility Assessment

To explore the feasibility of this system, we further examined the oxidation of varying pollutants [i.e., sulfamethoxazole (SMX), bisphenol A (BPA), phenol (PN), and 4-chlorophenol (4-CP)], total organic carbon (TOC) mineralization of pCBA, and the stability of the N-CNT electrode. The results, as shown in Figure S14, demonstrated similar degradation trends for the four model organic pollutants in the PMS-based EAOP in the presence of ORR, consistent with the observations made in this study. Furthermore, we observed significantly higher levels of TOC mineralization (Figure S15) in the presence of ORR (∼41.7%) than in the absence of ORR (∼3.6%), corresponding to the oxidation results of pCBA. In addition, we evaluated the stability of the N-CNT electrode by subjecting it to 10 consecutive cycles under the same experimental conditions. By the end of the 10th cycle, we achieved complete pCBA removal with approximately ∼58% PMS activation (Figure S16).

Considering the significant role of applied potential in determining both PMS activation and ORR, we further investigated the influence of varying applied potentials (i.e., −1.0, −0.75, 0, and 0.25 V vs Ag/AgCl) for broader application of this system. As illustrated in Figure S17, we observed a faster pCBA removal at a more reductive potential (i.e., −1.0, −0.75, and −0.5 V vs Ag/AgCl), accompanied by more effective PMS activation and H₂O₂ generation, compared to those at less reductive potentials (i.e., 0 and 0.25 V vs Ag/AgCl). These findings indicate that an appropriate applied potential window is crucial for achieving successful PMS activation and H₂O₂ generation via ORR, thereby establishing an efficient PMS-based EAOP system.

Environmental Implications

Peroxymonosulfate-driven AOPs, including PMS-based EAOPs, have been extensively pursued to exploit the strong oxidizing power of SO₄^•–. In many application scenarios, mixed oxidants were found to be instrumental, e.g., ^•OH overcoming the selective nature of SO₄^•– and broadening target pollutants.^3,55 Although H₂O₂ is the major reactive species in H₂O₂-driven AOPs, few past studies have discussed PMS-based AOPs in the same context as H₂O₂-based AOPs; instead, PMS-based AOPs were more frequently presented as a distinct alternative to H₂O₂-based AOPs. From this perspective, the results of our study point toward an interesting intersection where PMS-based AOPs can benefit from H₂O₂-based AOPs, e.g., enhancing the PMS-based (E)AOP by collaborating with H₂O₂. This is attributed to a parasitic supplement that eliminates the need for adding H₂O₂ chemically, illustrating a simplified approach to enhancing the EAOP system by leveraging the complex interactions commonly observed in EAOPs. Furthermore, our results offer an advantage over conventional AOPs and general electro-Fenton systems for water treatment that have limited working pH values (e.g., slightly acidic for a H₂O₂-driven AOP) and rely on heterogeneous transition metal catalysts (e.g., metal leaching, physical separation or recovery of catalysts, and generation of sludge waste). We believe that the insights obtained from this study can be extended to encompass a broader range of electrocatalytic materials beyond CNTs by applying them to the practical and scalable electrode materials capable of activating PMS and facilitating ORR, as shown in Figure S18. However, caution must be shown to avoid excess H₂O₂ that could quench radicals (e.g., decrease in pCBA decomposition kinetics with 10 mM H₂O₂ in Figure 5b),⁷⁰⁻⁷² as it is facilely generated by the carbonaceous cathode.

Note that we here focused on cathodic reactions only. The overall performance of the PMS-based EAOP, often performed using a membrane-less cell, can be more complex due to the occurrence of reactions such as the direct electron transfer oxidation of pollutants and active chlorine evolution on the anode.^9,73 More importantly, ^•OH can be generated by water oxidation (eqs 3 and 4)²⁰ and SO₄^•– by sulfate oxidation (eq 5),^21,22 which will directly impact the chemistry discussed in our study. We also suspect significant variation of the mechanism depending on the type of electrode material, which determines the kinetics of the surface redox reactions (e.g., ORR and HER). Future studies need to consider these additional factors with further efforts to identify strategies to maximize the synergism created by seemingly disparate, yet highly correlated, AOP schemes involving different precursors.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c06156.

• Materials and methods, N-CNT characterization results, schematic of the RRDE experimental setup and H cell setup for electrolysis experiments, assessment of PMS and H₂O₂ quantification and EPR analysis in the ORR only system, pCBA degradation and PMS activation of the CNT electrode, assessment of the influence of scavengers on the electrode, pCBA removal rate depending on the scavenger concentration, supplementary results of electrolysis with N-CNT in the absence and presence of ORR, electrode stability and influence of applied potential, and supplementary results of electrolysis with nitrogen-doped carbon black (N-CB) cathode in the presence of ORR (PDF)

The authors declare no competing financial interest.