Removal of Ammonia Using Persulfate during the Nitrate Electro-Reduction Process

NH4+ is often produced during the electro-reduction of NO3−, which results in inadequate total nitrogen (TN) removal during advanced sewage treatment. In this study, the electro-reduction byproduct NH4+ was oxidized and removed using sulfate radical (SO4•−)-based advanced oxidation. Persulfate (PS) was activated by electrocatalysis, using Co/AC0.9-AB0.1 particle electrodes to produce SO4•−. Results showed that when the influent concentration of NO3−-N was 20 mg/L, a PS dosage of 5.0 mM could completely oxidize NH4+ at 0.1 A (nondetectable in effluent) reducing the TN concentration from 9.22 to 0.55 mg/L. The presence of coexisting PO43−, CO32− and humic acid suppressed the oxidation and removal of NH4+. Electron spin resonance (ESR) spectra and quenching experiments revealed SO4•− as the dominant radical in the process of indirect NH4+ oxidation, while •OH radicals only had an assisting role, and the surface accumulated free radicals were responsible for the indirect oxidation of NH4+. Cyclic voltammetry (CV) curves indicated that NO3− was primarily reduced via atomic H*-mediated indirect reduction. Therefore, the activation of PS using Co/AC0.9-AB0.1 particle electrodes might be a promising alternative method for oxidizing byproduct NH4+ in the electro-reduction of NO3− and reduce TN concentration in advanced sewage treatment.


Introduction
The total nitrogen (TN) content of the effluent of most sewage treatment plants can reach as high as 10 mg/L, even after primary and secondary treatment [1]. In sewage wastewater, TN is mainly composed of NO 3 − , with a small amount of NH 3 /NH 4 + [2]. It has been well established that NO 3 − and NH 4 + bear a negative impact on aquatic environments, such as eutrophication and the occurrence of algal blooms [3]. Therefore, advanced treatment methods are required to reduce the TN content of sewage.
Electrochemical reduction is a feasible option for the conversion of NO 3 − to N 2 , with the benefits of simple operational methods, low-environmental impact and economic efficiency [4]. The electro-reduction of NO 3 − occurs via a complex process involving multi-electron transfer between the different valence states of nitrogen and its reaction products, such as N 2 H 4 , NH 3 , NH 2 OH, N 2 , N 2 O, NO, NO 2 − , and NO 2 [5]. Direct reduction and indirect reduction pathways have both been shown to be involved in the electro-reduction of NO 3 − [6]. During the direct reduction process, NO 3 − is first adsorbed onto the electrode and then converted to NO 3 − (ad) . Electron transfer leads to the generation of NO 3 2− (ad) , NO 2 − (ad) and other shortlived intermediates [7]. Subsequently, NO 3 2− (ad) and NO 2 − (ad) are converted (in a stepwise reduction process) to NO (ad) , then finally to N 2 and NH 4 + , sometimes via the intermediates N 2 O, NH 2 OH and N 2 H 4 [5]. The indirect reduction process refers to atomic H*-mediated reactions, where cathode surface-adsorbed atomic H* is produced via the reduction of protons. The products formed during the indirect reduction process usually include NO 2 − , NO, N 2 ,

Experimental Setup
As shown in Figure 1, the mesh plates composed of Ti/RuO 2 (12 × 2.5 cm 2 ) and Ti (12 × 2.5 cm 2 ) in rectangular cells were obtained from Hengli Ti Co., Ltd. (Beijing, China), for use as the anode and cathode, respectively. Then, 15 g of particle electrodes were packed between the two mesh plates (total L × W × H = 2.5 × 2.4 × 12 cm 3 ). Simulated wastewater with PS dosing (namely K 2 S 2 O 8 ) was continuously pumped into the reactor and treated without circulation.
with PS dosing (namely K2S2O8) was continuously pumped into the reactor and treated without circulation. Figure 1. The schematic diagram for ammonia removal using electric activation of persulfate during the nitrate electro-reduction process. Co/AC0.9-AB0.1 particle electrodes are packed between a pair of anode and cathode mesh plates.

Preparation of Particle Electrodes
The preparation of particle electrodes was performed according to a previously reported method [22]. Briefly, 90 g of powdered activated carbon (AC) and 10 g of powdered acetylene black (AB) were impregnated into 300 mL Co(NO3)2 (0.4 mol/L) and 100 mL distilled water, respectively. Then, AC and AB were separated by centrifugation and dried at 105 °C. The dried AC and AB samples were mixed thoroughly using polyvinyl alcohol as an adhesive at a mass ratio of 9:1, then molded into small cylindrical granules (~Φ × H = 4 mm × 5 mm) using a granulator. After drying at 105 °C for 24 h, the Co/AC0.9-AB0.1 particles were calcinated in a muffle furnace at 600 °C for 4 h under an N2 atmosphere.

Experimental Procedure
The simulated wastewater was prepared by dissolving KNO3 in distilled water (NO3 − -N = 20 mg/L), using 10.0 mM Na2SO4 as an electrolyte. After 120 min, adsorption reached saturation point and the electrical current was turned on. During the study of PS dosage, PS doses were applied from 1.5-5.0 mM. The treatments were carried out under successive currents of 0.1, 0.2, 0.3, and 0.4 A for 2 h/each current. Particle electrodes were replaced with new ones after each group treatment from 0.1-0.4 A. To investigate the effects of coexisting substances, the current was set at 0.3 A and the PS dosage remained constant at 5.0 mM. The concentrations of PO4 3− -P and CO3 2− -C ranged from 0.5-3.0 mg/L and 50.0-200.0 mg/L, respectively. The humic acid (HA) was used to simulate the dissolved organic matter. The concentration of HA was expressed by chemical oxygen demand, namely 50.0-120.0 mg COD/L (COD) in the experiments. Therefore, the concentration of HA was expressed by COD, respectively. Particle electrodes were replaced with new ones after treatment for each coexisting substance. The hydraulic retention time (HRT) remained constant at 60 min for all experiments. To assess the comparative effect of Cl − , the dosage of Cl − was varied from 1.5-5.0 mM, while all other conditions remained the same as described for PS experiments. Samples were collected from the outlet at 20 min intervals for the detection of NO3 − -N, NO2 − -N, NH4 + -N, and S2O8 2− . The experiments under initial 20 mg/L NO3 − -N and without PS dosing were used as controls.

Identification of Reactive Species and Dominant Radicals
Free radicals were determined and recorded using an electron spin resonance (ESR) spectrometer (EMX Plus, Bruker, Karlsruhe, Germany). The parameter settings included Figure 1. The schematic diagram for ammonia removal using electric activation of persulfate during the nitrate electro-reduction process. Co/AC 0.9 -AB 0.1 particle electrodes are packed between a pair of anode and cathode mesh plates.

Preparation of Particle Electrodes
The preparation of particle electrodes was performed according to a previously reported method [22]. Briefly, 90 g of powdered activated carbon (AC) and 10 g of powdered acetylene black (AB) were impregnated into 300 mL Co(NO 3 ) 2 (0.4 mol/L) and 100 mL distilled water, respectively. Then, AC and AB were separated by centrifugation and dried at 105 • C. The dried AC and AB samples were mixed thoroughly using polyvinyl alcohol as an adhesive at a mass ratio of 9:1, then molded into small cylindrical granules (~Φ × H = 4 mm × 5 mm) using a granulator. After drying at 105 • C for 24 h, the Co/AC 0.9 -AB 0.1 particles were calcinated in a muffle furnace at 600 • C for 4 h under an N 2 atmosphere.

Experimental Procedure
The simulated wastewater was prepared by dissolving KNO 3 in distilled water (NO 3 − -N = 20 mg/L), using 10.0 mM Na 2 SO 4 as an electrolyte. After 120 min, adsorption reached saturation point and the electrical current was turned on. During the study of PS dosage, PS doses were applied from 1.5-5.0 mM. The treatments were carried out under successive currents of 0.1, 0.2, 0.3, and 0.4 A for 2 h/each current. Particle electrodes were replaced with new ones after each group treatment from 0.1-0.4 A. To investigate the effects of coexisting substances, the current was set at 0.3 A and the PS dosage remained constant at 5.0 mM. The concentrations of PO 4 3− -P and CO 3 2− -C ranged from 0.5-3.0 mg/L and 50.0-200.0 mg/L, respectively. The humic acid (HA) was used to simulate the dissolved organic matter. The concentration of HA was expressed by chemical oxygen demand, namely 50.0-120.0 mg COD/L (COD) in the experiments. Therefore, the concentration of HA was expressed by COD, respectively. Particle electrodes were replaced with new ones after treatment for each coexisting substance. The hydraulic retention time (HRT) remained constant at 60 min for all experiments. To assess the comparative effect of Cl − , the dosage of Cl − was varied from 1.5-5.0 mM, while all other conditions remained the same as described for PS experiments. Samples were collected from the outlet at 20 min intervals for the detection of NO 3 − -N, NO 2 − -N, NH 4 + -N, and S 2 O 8 2− . The experiments under initial 20 mg/L NO 3 − -N and without PS dosing were used as controls.

Identification of Reactive Species and Dominant Radicals
Free radicals were determined and recorded using an electron spin resonance (ESR) spectrometer (EMX Plus, Bruker, Karlsruhe, Germany). The parameter settings included a center field of 3513 G, seep width of 100 G, microwave power of 20 mW, and a scan time of 60 s [33]. The test samples were mixed with 50 mM 5,5-dimethyl-1-pyrrolidine N-oxide (DMPO, TCI Development, Tokyo, Japan). Quenching experiments were performed to identify the contribution of different radical species using simulated wastewater, composed of 5.0 mM PS in 6.0 mg/L NH 4 + -N in distilled water, using different molar ratios of [scavenger]/[PS] to screen free radicals and establish their contributions. Molar ratios of 1000:1 tert-butyl alcohol (TBA, ≥99%), 1000:1 and 2000:1 ethanol (EtOH, ≥99%) and 200:1 phenol (≥99%) were separately added to simulated sewage samples prior to treatment. Due to the potential interference caused by organic reagents, samples were filtered through a 0.22 µm membrane prior to NH 4 + determination.

NO 3
− and NO 2 − were determined via ion chromatography (Dionex ICS-3000, Sunnyvale, CA, USA) using a Dionex IonPac AS11-HC analytical column (4 mm × 250 mm) and an AG11-HC guard column (4 mm × 40 mm). Elution was performed using 12.5 mM NaOH at a flow rate of 1.0 mL/min. NH 4 + was analyzed using salicylic acid spectrophotometry (UV2600, Techocomp, Beijing, China) [34]. Total nitrogen (TN) was calculated as the sum of NO 3 − -N, NO 2 − -N and NH 4 + -N. The pH value was determined using a pH-meter (PB-10, BSISL, China). S 2 O 8 2− was determined by UV spectrophotometry [35]. The activation efficiency of PS was calculated according to Equation (1), as follows: where, C 0 and C t are the PS concentration of the influent and effluent, respectively.

Electrochemical Measurements
Cyclic voltammetry (CV) measurements were performed using an electrochemical workstation (CHI 660D, Shanghai CH Instruments, China), with Pt filament and Ag/AgCl electrodes employed as the counter and reference electrodes, respectively. A glassy carbon disk electrode (3.0 mm) was used as the working electrode. Approximately 20.0 µL of pre-dispersed catalyst ink was coated dropwise onto the polished glassy carbon electrode. CV tests were performed with a cycle ranging from −0.39 V to +1.61 V and a scan rate of 0.1 V/s for 10 segments. A solution containing 0.5 M Na 2 SO 4 was used as the supporting electrolyte. All solutions were prepared using ultrapure water and purged with N 2 gas for 20 min prior to measurements. The potentials were established in reference to a reversible hydrogen electrode (RHE).

Statistical Analysis
The one-way analysis of variance (ANOVA) was carried out using IBM SPSS v.20.0 (SPSS Inc., Chicago, IL, USA) software. Unless otherwise stated, results were considered to indicate significant differences if p < 0.05.

Nitrogen Conversion and Removal under Different PS Dosage and Current Conditions
As shown in Figure 2, the Co/AC 0.9 -AB 0.1 particle electrodes exhibited excellent catalytic activity for the electro-reduction of NO 3 − . For samples without PS dosing, the percentage of NO 3 − conversion reached 80.5-90.2%, with the percentage conversion increasing in accordance with the applied current. This phenomenon occurred due to Co on the surface of particle electrodes having a strong catalytic activity for the production of atomic H*, which was beneficial to the indirect reduction of NO 3 − [22]. Following the addition of PS at dosages of 1.5-5.0 mM, NO 3 − reduction was further increased by 9.4-16.7%. Although PS activation occurred at the cathodic side of the particle electrode [36], the PS activation process did not interfere with the reduction of NO 3 − . PS is frequently used in advanced oxidation processes, but it can also facilitate the formation of Co-H* on electrode surfaces [36]. Therefore, PS promotes the atomic H*-mediated reduction of NO 3 − . Compared with the control samples, the addition of 1.5 mM PS significantly enhanced the reduction of NO 3 − at 0.2-0.4 A (p < 0.05), although this effect was not observed at 0.1 A. When the PS dosage was increased further to 3.0 and 5.0 mM, the enhancement effect was significantly greater than that of 1.5 mM PS (p < 0.05). However, there were no significant differences found between 3.0 and 5.0 mM (p > 0.05), implying that increases in the dosage of PS beyond 3.0 mM, did not further improve NO 3 − reduction.
advanced oxidation processes, but it can also facilitate the formation of Co-H* on electrode surfaces [36]. Therefore, PS promotes the atomic H*-mediated reduction of NO3 − . Compared with the control samples, the addition of 1.5 mM PS significantly enhanced the reduction of NO3 − at 0.2-0.4 A (p < 0.05), although this effect was not observed at 0.1 A. When the PS dosage was increased further to 3.0 and 5.0 mM, the enhancement effect was significantly greater than that of 1.5 mM PS (p < 0.05). However, there were no significant differences found between 3.0 and 5.0 mM (p > 0.05), implying that increases in the dosage of PS beyond 3.0 mM, did not further improve NO3 − reduction. During the electro-reduction process at 0.1-0.4 A, approximately 10.2-28.2% of NO3 − was converted into NH4 + without PS dosing ( Figure 3). Therefore, when the influent NO3 − -N concentration was ~20.0 mg/L, 2.1-5.5 mg/L of NH4 + -N was discharged into the effluent. It is known that NH4 + -N concentrations over 2.0 mg/L can cause toxicity to aquatic life, such as fish [37]. However, compared with controls, the NH4 + -N concentration was significantly reduced after PS dosing at all assessed currents (p < 0.05). Even low current conditions of 0.1 A, and the addition of 5.0 mM PS resulted in NH4 + being undetectable in the effluent. In contrast, when 5.0 mM of Cl − was added to the system, the concentration of During the electro-reduction process at 0.  Figure 3). Therefore, when the influent NO 3 − -N concentration was~20.0 mg/L, 2.1-5.5 mg/L of NH 4 + -N was discharged into the effluent. It is known that NH 4 + -N concentrations over 2.0 mg/L can cause toxicity to aquatic life, such as fish [37]. However, compared with controls, the NH 4 + -N concentration was significantly reduced after PS dosing at all assessed currents (p < 0.05). Even low current conditions of 0.1 A, and the addition of 5.0 mM PS resulted in NH 4 + being undetectable in the effluent. In contrast, when 5.0 mM of Cl − was added to the system, the concentration of NH 4 + in the effluent continually accumulated throughout the whole experimental process, reaching 4.7 mg/L. As can be seen from Figure 4, the PS activation efficiency reached >94.1% at 0.1 A, suggesting that PS can easily be activated by Co/AC 0.9 -AB 0.1 particle electrodes. Due to the effective activation of PS, the oxidation of NH 4 + was positively correlated with PS dosage, with higher PS dosages achieving a better NH 4 + oxidation efficiency. Furthermore, increases in current could overcome the effects of low PS dosage, with a current of 0.3 A resulting in NH 4 + being undetectable in the effluent at a PS dose of 3.0 mM. When 3.0 mM Cl − was applied, the effluent still contained 2.8 mg/L NH 4 + -N. Although active chlorine species have the capacity to oxidize NH 4 + , they cannot oxidize and remove NH 4 + as efficiently as PS at an equivalent dosage.
NH4 + in the effluent continually accumulated throughout the whole experimental process, reaching 4.7 mg/L. As can be seen from Figure 4, the PS activation efficiency reached >94.1% at 0.1 A, suggesting that PS can easily be activated by Co/AC0.9-AB0.1 particle electrodes. Due to the effective activation of PS, the oxidation of NH4 + was positively correlated with PS dosage, with higher PS dosages achieving a better NH4 + oxidation efficiency. Furthermore, increases in current could overcome the effects of low PS dosage, with a current of 0.3 A resulting in NH4 + being undetectable in the effluent at a PS dose of 3.0 mM. When 3.0 mM Cl − was applied, the effluent still contained 2.8 mg/L NH4 + -N. Although active chlorine species have the capacity to oxidize NH4 + , they cannot oxidize and remove NH4 + as efficiently as PS at an equivalent dosage.   During the treatment processes, all samples dosed with PS achieved lower TN concentrations than the relevant controls ( Figure 5). In order to investigate the NH4 + oxidation products, samples were prepared by dissolving (NH4)2SO4 (6.0 mg/L NH4 + -N) into 10.0 mM Na2SO4, followed by treatment for 120 min at 0.3 A with the addition of 5.0 mM PS. Results showed that only 0.06 mg/L NO3 − -N was generated, while NO2 − -N was undetectable ( Figure 6). These results suggest that the oxidation products of NH4 + mainly consisted of nitrogenous gases. The release of gaseous substances from aqueous solution resulted in a decrease in TN concentration. When a PS dosage of 5.0 mM was applied, a TN removal efficiency of >99% could be achieved, even at 0.1 and 0.2 A. In contrast, the maximum TN removal efficiency achieved with the addition of 5.0 mM Cl − , was only 89% at 0.4 A. Therefore, due to the high removal effects achieved under low currents, the oxidation of NH4 + by PS not only improves the denitrification efficiency of the system, but also reduces the energy consumption requirements. During the treatment processes, all samples dosed with PS achieved lower TN concentrations than the relevant controls ( Figure 5). In order to investigate the NH 4 + oxidation products, samples were prepared by dissolving (NH 4 ) 2 SO 4 (6.0 mg/L NH 4 + -N) into 10.0 mM Na 2 SO 4 , followed by treatment for 120 min at 0.3 A with the addition of 5.0 mM PS. Results showed that only 0.06 mg/L NO 3 − -N was generated, while NO 2 − -N was undetectable ( Figure 6). These results suggest that the oxidation products of NH 4 + mainly consisted of nitrogenous gases. The release of gaseous substances from aqueous solution resulted in a decrease in TN concentration. When a PS dosage of 5.0 mM was applied, a TN removal efficiency of >99% could be achieved, even at 0.1 and 0.2 A. In contrast, the maximum TN removal efficiency achieved with the addition of 5.0 mM Cl − , was only 89% at 0.4 A. Therefore, due to the high removal effects achieved under low currents, the oxidation of NH 4 + by PS not only improves the denitrification efficiency of the system, but also reduces the energy consumption requirements.   Initial NH 4 + -N was 6.0 mg/L. The 10.0 mM Na 2 SO 4 was used as electrolyte.

Effects of Coexisting Substances
Various coexisting anions and organic compounds are present in actual municipal sewage. Therefore, the effects of several commonly coexisting species, including PO 4 3− , CO 3 2− and HA, were investigated. As shown in Figures 7-9, the coexistence of PO 4 3− , CO 3 2− and HA had a negative impact on the reduction of NO 3 − , especially in terms of the oxidation of NH 4 + . PO 4 3− and CO 3 2− have been shown to adsorb the surface of electrodes, resulting in competition for active sites and interfering with the adsorption and conversion of NO 3 − [38,39]. HA is an amphoteric substance that can be directly or indirectly reduced on the surface of Co/AC 0.9 -AB 0.1 particle electrodes, with both systems interfering with the reduction of NO 3 − [22]. During the oxidation of NH 4 + , HPO 4 2− (hydrolyzed by PO 4 3− ) and CO 3 2− often act as scavengers of the free radicals SO 4 •− and •OH, therefore reacting with SO 4 •− and •OH to yield weak oxidants, such as HPO 4 •− and CO 3 •− (Equations (2)- (5)) [39].

Effects of Coexisting Substances
Various coexisting anions and organic compounds are present in actual municipal sewage. Therefore, the effects of several commonly coexisting species, including PO4 3− , CO3 2− and HA, were investigated. As shown in Figures 7-9, the coexistence of PO4 3− , CO3 2− and HA had a negative impact on the reduction of NO3 − , especially in terms of the oxidation of NH4 + . PO4 3− and CO3 2− have been shown to adsorb the surface of electrodes, resulting in competition for active sites and interfering with the adsorption and conversion of NO3 − [38,39]. HA is an amphoteric substance that can be directly or indirectly reduced on the surface of Co/AC0.9-AB0.1 particle electrodes, with both systems interfering with the reduction of NO3 − [22]. During the oxidation of NH4 + , HPO4 2− (hydrolyzed by PO4 3− ) and CO3 2− often act as scavengers of the free radicals SO4 •− and •OH, therefore reacting with SO4 •− and •OH to yield weak oxidants, such as HPO4 •− and CO3 •− (Equations (2)-(5)) [39].   Similarly, HA can also be oxidized by consuming SO 4 •− and •OH [36]. Therefore, the coexistence of these substances reduced the oxidation performance of the radical-based system. The oxidation of NH 4 + was greatly affected and the efficiency of NO 3 − and NH 4 + removal was reduced, negatively affecting TN removal. However, after actual sewage treatment, the TP content of the effluent typically ranges from 0.5-1.0 mg/L, while the concentration of HA, expressed as COD, is generally less than 50.0 mg/L [40] and therefore, the concentration of NH 4 + -N would be below 2.0 mg/L under this coexisting substance concentration, which would not adversely affect the water environment. Even if the concentration of CO 3 2− -C reached 50.0 mg/L, a concentration of NH 4 + -N in the effluent would measure 2.8 mg/L, which would only have a slight influence on the water environment. However, the negative impact can be mitigated by increasing the PS concentration accordingly. Similarly, HA can also be oxidized by consuming SO4 •− and •OH coexistence of these substances reduced the oxidation performance system. The oxidation of NH4 + was greatly affected and the efficien removal was reduced, negatively affecting TN removal. However, treatment, the TP content of the effluent typically ranges from 0.5concentration of HA, expressed as COD, is generally less than 50.0 fore, the concentration of NH4 + -N would be below 2.0 mg/L under stance concentration, which would not adversely affect the water e the concentration of CO3 2− -C reached 50.0 mg/L, a concentration of N would measure 2.8 mg/L, which would only have a slight influence o ment. However, the negative impact can be mitigated by increasing accordingly.

Reactive Species Identification
In order to confirm the active radical species in the PS-Co/AC experiments were carried out under different conditions. As show were no free radical signals detected in the 2D system. After the add a weak signal was observed for the DMPO-SO4 adduct (1:1:1:1:1:1, a aN = 1.4 G, aH = 0.8 G), whereas the DMPO-OH adduct signal (1:2:2:1, much stronger. This suggests that DMPO-SO4 and DMPO-OH were the activation of PS, while the •OH radical was yielded from SO4 •− rea as shown in Equations (6)

Reactive Species Identification
In order to confirm the active radical species in the PS-Co/AC 0.9 -AB 0.1 system, ESR experiments were carried out under different conditions. As shown in Figure 10, there were no free radical signals detected in the 2D system. After the addition of PS (2D + PS), a weak signal was observed for the DMPO-SO 4 adduct (1:1:1:1:1:1, a N = 13.2 G, a H = 9.5 G, a N = 1.4 G, a H = 0.8 G), whereas the DMPO-OH adduct signal (1:2:2:1, a N = a H = 14.9 G) was much stronger. This suggests that DMPO-SO 4 and DMPO-OH were both generated from the activation of PS, while the •OH radical was yielded from SO 4 •− reacting with H 2 O/OH − , as shown in Equations (6) and (7) [39]: 3460 3480 3500 3520 3540 3560 For the 3D system, a weak signal was observed for the DMPO-H adduct (1:1:2:1:2:1:2:1:1), with hyperfine coupling constants of aN = 15.5 G and aH = 20.6 G. This indicated that Co/AC0.9-AB0.1 particle electrodes had performed hydro-reduction of NO3 − . Atomic H* was generated by the electrolysis of H2O with catalytically active Co on the particle electrode [22]. After the addition of PS (3D + PS), a new adduct signal (1:2:2:2:2:2:2:2:1) was observed, which was ascribed to the original signal of DMPO-H overlapped with that of DMPO-OH [36]. However, the signal for DMPO-OH was weaker than that of the 2D + PS system, which was due to neutralization of the generated H* and •OH. Moreover, the DMPO-SO4 signal was too weak to be detected in the 3D + PS system, which can be ascribed to the much higher reaction rate constant between •OH and DMPO compared to between SO4 •− and DMPO [41].
As seen in Figure 10, after NO3 − was added to the 3D system (3D + NO3 − ), rather than disappearing, the DMPO-H signal was slightly strengthened. This implies that the generation rate and the amount of atomic H* in the 3D system was large enough to reduce NO3 − . When NH4 + was added to the 3D system (3D + NH4 + ), the DMPO-H signal disappeared and a signal for DMPO-OH emerged. As for the 3D + PS + NH4 + system, the DMPO-OH signal weakened slightly thereafter, although it remained consistently strong. Although DMPO-OH was generated in both 3D + NH4 + and 3D + PS + NH4 + systems, the NH4 + removal efficiencies of these two systems were apparently different. NH4 + could not be effectively oxidized and removed in the 3D system, while in the 3D + PS system NH4 + was completely oxidized and removed. Although the DMPO-SO4 signal was not observed in the 3D + PS system, SO4 •− and •OH can interconvert, with •OH being transferred to SO4 •− via the assistance of SO4 2− and HSO4 − (Equations (8) and (9)) [31,42]. Furthermore, the slow reaction rate constant for •OH and NH3/NH4 + led to the low level of NH3/NH4 + oxidation by •OH [26,43]. Therefore, it can be inferred that the oxidation of NH4 + occurred mainly through its reaction with SO4 •− in the 3D + PS system, resulting in the NH4 + -N concentration measuring below the detection limit (0.04 mg/L).
•OH + SO 4 2-→ SO 4 •-+ OH - •OH + HSO 4 -→ SO 4 •-+ H 2 O As shown in Figure 10, nearly no free radical signals could be detected in the 3D + PS + NO3 − system. The disappearance of the DMPO-H signal indicated hydro-reduction of For the 3D system, a weak signal was observed for the DMPO-H adduct (1:1:2:1:2:1:2:1:1), with hyperfine coupling constants of a N = 15.5 G and a H = 20.6 G. This indicated that Co/AC 0.9 -AB 0.1 particle electrodes had performed hydro-reduction of NO 3 − . Atomic H* was generated by the electrolysis of H 2 O with catalytically active Co on the particle electrode [22]. After the addition of PS (3D + PS), a new adduct signal (1:2:2:2:2:2:2:2:1) was observed, which was ascribed to the original signal of DMPO-H overlapped with that of DMPO-OH [36]. However, the signal for DMPO-OH was weaker than that of the 2D + PS system, which was due to neutralization of the generated H* and •OH. Moreover, the DMPO-SO 4 signal was too weak to be detected in the 3D + PS system, which can be ascribed to the much higher reaction rate constant between •OH and DMPO compared to between SO 4 •− and DMPO [41]. As seen in Figure 10, after NO 3 − was added to the 3D system (3D + NO 3 − ), rather than disappearing, the DMPO-H signal was slightly strengthened. This implies that the generation rate and the amount of atomic H* in the 3D system was large enough to reduce NO 3 − . When NH 4 + was added to the 3D system (3D + NH 4 + ), the DMPO-H signal disappeared and a signal for DMPO-OH emerged. As for the 3D + PS + NH 4 + system, the DMPO-OH signal weakened slightly thereafter, although it remained consistently strong. Although DMPO-OH was generated in both 3D + NH 4 + and 3D + PS + NH 4 + systems, the NH 4 + removal efficiencies of these two systems were apparently different. NH 4 + could not be effectively oxidized and removed in the 3D system, while in the 3D + PS system NH 4 + was completely oxidized and removed. Although the DMPO-SO 4 signal was not observed in the 3D + PS system, SO 4 •− and •OH can interconvert, with •OH being transferred to SO 4 •− via the assistance of SO 4 2− and HSO 4 − (Equations (8) and (9)) [31,42]. Furthermore, the slow reaction rate constant for •OH and NH 3 /NH 4 + led to the low level of NH 3 /NH 4 + oxidation by •OH [26,43]. Therefore, it can be inferred that the oxidation of NH 4 + occurred mainly through its reaction with SO 4 •− in the 3D + PS system, resulting in the NH 4 + -N concentration measuring below the detection limit (0.04 mg/L).
As shown in Figure 10, nearly no free radical signals could be detected in the 3D + PS + NO 3 − system. The disappearance of the DMPO-H signal indicated hydro-reduction of NO 3 − , while disappearance of the signals for DMPO-OH and DMPO-SO 4 occurred as a result of consumption of NH 4 + produced from the electro-reduction of NO 3 − . In addition, due to the reaction of SO 4 •− with NH 4 + , more •OH was converted into SO 4 •− , further weakening the DMPO-OH signal.

Identification of Dominant Radical
In order to establish the relative contributions from SO 4 •− and •OH in the oxidation of NH 4 + , free radical quenching experiments were conducted. TBA is typically used as a scavenger of •OH, as the second reaction rate (k) of k •OH ((3.8-7.6) × 10 8 M −1 s −1 ) is about 1000-fold greater than that of k SO •−  [42,45]. As shown in Figure 11, the NH 4 + removal efficiency reached 93.3% without the addition of a quenching agent. When the molar ratio of TBA/PS was 1000:1, NH 4 + removal was inhibited and reduced to 80.7%, indicating •OH had participated in the oxidation of NH 4 + . When the molar ratio of phenol to PS was 200:1, NH 4 + removal decreased sharply to 45.3%. The number of radicals necessary for TBA to quench oxidation is approximately consistent with phenol [36]. However, the inhibition of NH 4 + oxidation caused by the addition of a 200:1 molar ratio of phenol was greater than that of 1000:1 molar ratio of TBA, indicating that the contribution of SO 4 •− was much greater than that of •OH. Therefore, it can be concluded that SO 4 •− played a dominant role in the oxidation of NH 4 + , while the contribution from •OH was less.

Identification of Dominant Radical
In order to establish the relative contributions from SO4 •− and •OH in the oxidation of NH4 + , free radical quenching experiments were conducted. TBA is typically used as a scavenger of •OH, as the second reaction rate (k) of k•OH ((3.8-7.6) × 10 8 [42,45]. As shown in Figure 11, the NH4 + removal efficiency reached 93.3% without the addition of a quenching agent. When the molar ratio of TBA/PS was 1000:1, NH4 + removal was inhibited and reduced to 80.7%, indicating •OH had participated in the oxidation of NH4 + . When the molar ratio of phenol to PS was 200:1, NH4 + removal decreased sharply to 45.3%. The number of radicals necessary for TBA to quench oxidation is approximately consistent with phenol [36]. However, the inhibition of NH4 + oxidation caused by the addition of a 200:1 molar ratio of phenol was greater than that of 1000:1 molar ratio of TBA, indicating that the contribution of SO4 •− was much greater than that of •OH. Therefore, it can be concluded that SO4 •− played a dominant role in the oxidation of NH4 + , while the contribution from •OH was less. Although EtOH can effectively scavenge both SO4 •− and •OH, the addition of EtOH (molar ratio to PS of 1000:1) caused only a 6.0% reduction in NH4 + oxidation ( Figure 11). Increasing the EtOH/PS molar ratio to 2000:1 increased the suppression of NH4 + oxidation to 23.6%. The varied inhibitory effect of the three quenching agents was related to their physicochemical properties and the formation sites of the free radicals [46]. Phenol, TBA and EtOH have varying dielectric constants of 9.78, 12.47 and 28.40, respectively [47]. Dielectric constants reflect the polarity of a substance. Since the polarity of a substance usually reflects its water solubility, the hydrophobic quality of the quenching agents can be ranked in the descending order of phenol > TBA > EtOH. Generally, hydrophobic quenching agents easily react with free radicals present on the surface of catalytic materials. Although EtOH can effectively scavenge both SO 4 •− and •OH, the addition of EtOH (molar ratio to PS of 1000:1) caused only a 6.0% reduction in NH 4 + oxidation ( Figure 11). Increasing the EtOH/PS molar ratio to 2000:1 increased the suppression of NH 4 + oxidation to 23.6%. The varied inhibitory effect of the three quenching agents was related to their physicochemical properties and the formation sites of the free radicals [46]. Phenol, TBA and EtOH have varying dielectric constants of 9.78, 12.47 and 28.40, respectively [47]. Dielectric constants reflect the polarity of a substance. Since the polarity of a substance usually reflects its water solubility, the hydrophobic quality of the quenching agents can be ranked in the descending order of phenol > TBA > EtOH. Generally, hydrophobic quenching agents easily react with free radicals present on the surface of catalytic materials.
Powder activated carbon has the property of highly selective adsorption of hydrophobic organic compounds [48]. Therefore, the hydrophobic phenol can more easily approach the particle electrode surface and react with surface-bound free radicals, while hydrophilic TBA and EtOH prefer to compete for •OH and SO 4 •− in the liquid phase [47,49]. The greater inhibitory effect of phenol compared to TBA and EtOH, indicates that a majority of •OH and SO 4 •− were accumulated on the surface of particle electrodes, with the free radical reaction with NH 4 + mainly occurring in the boundary layer on the surface of Co/AB 0.9 -AC 0.1 .

Electrocatalytic
Performance of PS-Co/AC 0.9 -AB 0.1 The PS-Co/AC 0.9 -AB 0.1 redox process of electrocatalytic denitrification was investigated via CV analysis, with all tests based on the Na 2 SO 4 electrolyte. As shown in Figure 12a, only one oxidation peak was observed in the Na 2 SO 4 electrolyte at a potential (E p ) of 1.44 V, which was ascribed to the oxidation of Co 0 to Co 2+ /Co 3+ [50]. However, no peak was observed for the reduction of Co 2+ /Co 3+ to Co 0 during CV cycles, suggesting that Co 2+ /Co 3+ might receive electrons transferred from the power source to the cathode surface through a circuit, resulting in conversion of the Co valence state [22,51]. CV curves are often used to reflect heterogeneous charge transfer from an electrode to an electroactive species, although it cannot determine charge transfer and valency changes inside the electrode [9,52]. Therefore, the conversion of Co 2+ /Co 3+ to Co 0 mainly serves as an electron shuttle. When NO 3 − was added to the electrolyte solution, no peak of NO 3 − direct reduction appeared (Figure 12b), indicating that the direct electro-reduction of NO 3 − was difficult, with most NO 3 − reduction occurring via a reaction with atomic H* adsorbed on the surface of the Co/AC 0.9 -AB 0.1 electrode. Powder activated carbon has the property of highly selective adsorption of hydrophobic organic compounds [48]. Therefore, the hydrophobic phenol can more easily approach the particle electrode surface and react with surface-bound free radicals, while hydrophilic TBA and EtOH prefer to compete for •OH and SO4 •− in the liquid phase [47,49]. The greater inhibitory effect of phenol compared to TBA and EtOH, indicates that a majority of •OH and SO4 •− were accumulated on the surface of particle electrodes, with the free radical reaction with NH4 + mainly occurring in the boundary layer on the surface of Co/AB0.9-AC0.1.

Electrocatalytic Performance of PS-Co/AC0.9-AB0.1
The PS-Co/AC0.9-AB0.1 redox process of electrocatalytic denitrification was investigated via CV analysis, with all tests based on the Na2SO4 electrolyte. As shown in Figure  12a, only one oxidation peak was observed in the Na2SO4 electrolyte at a potential (Ep) of 1.44 V, which was ascribed to the oxidation of Co 0 to Co 2+ /Co 3+ [50]. However, no peak was observed for the reduction of Co 2+ /Co 3+ to Co 0 during CV cycles, suggesting that Co 2+ /Co 3+ might receive electrons transferred from the power source to the cathode surface through a circuit, resulting in conversion of the Co valence state [22,51]. CV curves are often used to reflect heterogeneous charge transfer from an electrode to an electroactive species, although it cannot determine charge transfer and valency changes inside the electrode [9,52]. Therefore, the conversion of Co 2+ /Co 3+ to Co 0 mainly serves as an electron shuttle. When NO3 − was added to the electrolyte solution, no peak of NO3 − direct reduction appeared (Figure 12b), indicating that the direct electro-reduction of NO3 − was difficult, with most NO3 − reduction occurring via a reaction with atomic H* adsorbed on the surface of the Co/AC0.9-AB0.1 electrode. When PS was added to the electrolyte solution (Figure 12c), the current response decreased in anodic sweeps. In contrast, the current response increased in cathodic sweeps, indicating the occurrence of strong electron exchange at the electrode, with a weak reduction peak appearing at 0.76 V. Similarly, a reduction peak was observed at 0.83 V in the When PS was added to the electrolyte solution (Figure 12c), the current response decreased in anodic sweeps. In contrast, the current response increased in cathodic sweeps, indicating the occurrence of strong electron exchange at the electrode, with a weak reduction peak appearing at 0.76 V. Similarly, a reduction peak was observed at 0.83 V in the NaNO 3 + PS system (Figure 12d), with the current also increasing in cathodic sweeps. These peaks were attributed to breakage of the O−O bond of PS [42,53]. The addition of PS promoted the production of atomic H* via the electrolysis of water at the cathode, resulting in an increase in electrode current. In order to establish the PS activation process, CV tests were conducted using different scan rates (v) ( Figure 13). As demonstrated in Figure 13a,b, the reduction peak current gradually increased in accordance with the sweep rate, while E p shifted negatively. E p was found to be proportional to ln v (R 2 = 0.991), indicating that PS activation was irreversible [54]. The electron transfer number (n) was calculated using the Laviron method (Equation (10)) [55,56]. Based on the CV test results (Figure 13), n value of 1.00 implies that PS activation occurred via a single electron transfer process.
where, α represents the electron transfer coefficient; n represents the electron transfer number; k s is the diffusion coefficient; R is the gas constant (8.315 J/(K·mol); F is the Faraday constant (96,500 C/mol); T is the temperature; and v is the sweep rate.  (Figure 12d), with the current also increasing in cathodic sweeps. These peaks were attributed to breakage of the O−O bond of PS [42,53]. The addition of PS promoted the production of atomic H* via the electrolysis of water at the cathode, resulting in an increase in electrode current. In order to establish the PS activation process, CV tests were conducted using different scan rates (v) ( Figure 13). As demonstrated in Figure 13a,b, the reduction peak current gradually increased in accordance with the sweep rate, while Ep shifted negatively. Ep was found to be proportional to ln v (R 2 = 0.991), indicating that PS activation was irreversible [54]. The electron transfer number (n) was calculated using the Laviron method (Equation (10)) [55,56]. Based on the CV test results (Figure 13), n value of 1.00 implies that PS activation occurred via a single electron transfer process.
where, α represents the electron transfer coefficient; n represents the electron transfer number; ks is the diffusion coefficient; R is the gas constant (8.315 J/(K•mol); F is the Faraday constant (96500 C/mol); T is the temperature; and v is the sweep rate. As shown in Figure 12f, when (NH4)2SO4 was added to the electrolyte solution, two oxidation peaks were observed in the anodic sweeps at 0.42 V (Ep1) and 1.18 V (Ep2), with electron transfer numbers of 2.86 and 1.00, respectively ( Figure 14). As shown in Equations (11)-(13), Ep1 was considered as the oxidation of NH3/NH4 + to N2, which involved a threeelectron transfer reaction. When N2 was not rapidly separated from liquid, it was adsorbed by the electrode, causing a single-electron reduction reaction at 0.82 V (Ep3) in the cathodic sweep and generating NH2OH (NH3OH + ) as an intermediate (Equation (14)) [7]. NH2OH can also be oxidized into N2 via single-electron transfer (Equation (15)), resulting in the Ep2 oxidation peak. When PS was added to the (NH4)2SO4 solution (Figure 12e), the peak currents ip1 and ip2 decreased from 2.18 and 3.09 mA to 1.03 and 1.74 mA, respectively. This occurred due to the generation of SO4 •− and •OH accelerating the NH4 + oxidation rate near the electrode surface, resulting in a decrease of NH4 + -N absorbed on the electrode surface. However, the NH4 + in solution could not diffuse to the electrode surface fast enough, weakening the direct oxidation of NH4 + and NH2OH on the electrode and therefore, reducing the current response. As shown in Figure 12f, when (NH 4 ) 2 SO 4 was added to the electrolyte solution, two oxidation peaks were observed in the anodic sweeps at 0.42 V (E p1 ) and 1.18 V (E p2 ), with electron transfer numbers of 2.86 and 1.00, respectively ( Figure 14). As shown in Equations (11)-(13), E p1 was considered as the oxidation of NH 3 /NH 4 + to N 2 , which involved a three-electron transfer reaction. When N 2 was not rapidly separated from liquid, it was adsorbed by the electrode, causing a single-electron reduction reaction at 0.82 V (E p3 ) in the cathodic sweep and generating NH 2 OH (NH 3 OH + ) as an intermediate (Equation (14)) [7]. NH 2 OH can also be oxidized into N 2 via single-electron transfer (Equation (15)), resulting in the E p2 oxidation peak. When PS was added to the (NH 4 ) 2 SO 4 solution (Figure 12e), the peak currents i p1 and i p2 decreased from 2.18 and 3.09 mA to 1.03 and 1.74 mA, respectively. This occurred due to the generation of SO 4 •− and •OH accelerating the NH 4 + oxidation rate near the electrode surface, resulting in a decrease of NH 4 + -N absorbed on the electrode surface. However, the NH 4 + in solution could not diffuse to the electrode surface fast enough, weakening the direct oxidation of NH 4 + and NH 2 OH on the electrode and therefore, reducing the current response.

The Mechanism of N Transformation
On the basis of these results, a possible mechanism was proposed for NO3 − reduction and NH4 + oxidation in the PS-Co/AC0.9-AB0.1 electrocatalytic system ( Figure 15). The results of CV test showed that PS activation occurred at the cathode of particle electrodes, generating SO4 •− and subsequently causing •OH to be produced from SO4 •− . The process of PS activation occurred via a single electron transfer process, with an activation efficiency of >94.1%. During the electro-reduction process, in our previous study [22], we discovered that when the current was not applied, the nitrate concentration of the effluent was almost unchanged after 60 min of treatment. This implied that the adsorption did not play the main role in the removal of nitrate ions. When the current was applied, NO3 − was mainly reduced by atomic H*-mediated indirect reduction, with atomic H* produced by electrolysis of H2O/H + via Co catalysis at the particle electrodes. The addition of PS did not interfere with the reduction of NO3 − , instead promoting the generation of atomic H*, which further promoted the indirect reduction of NO3 − . The NO3 − reduction products consisted of nitrogenous gas, NH4 + and residual NO3 − . During the process of electro-oxidation, NH3/NH4 + was directly oxidized via a three-electron reaction at the electrode, generating N2 as a product. A portion of N2 was reduced to NH2OH, which continued to undergo single-electron oxidation. The indirect oxidation of NH4 + occurred mainly via SO4 •− , while •OH functioned only in assistance. Although the DMPO-OH signal was stronger than that of DMPO-SO4, the slow reaction rate constant between •OH and NH4 + resulted in a poor NH4 + removal efficiency. The consumption of SO4 •− could be compensated for

The Mechanism of N Transformation
On the basis of these results, a possible mechanism was proposed for NO 3 − reduction and NH 4 + oxidation in the PS-Co/AC 0.9 -AB 0.1 electrocatalytic system ( Figure 15). The results of CV test showed that PS activation occurred at the cathode of particle electrodes, generating SO 4 •− and subsequently causing •OH to be produced from SO 4 •− . The process of PS activation occurred via a single electron transfer process, with an activation efficiency of >94.1%. During the electro-reduction process, in our previous study [22], we discovered that when the current was not applied, the nitrate concentration of the effluent was almost unchanged after 60 min of treatment. This implied that the adsorption did not play the main role in the removal of nitrate ions. When the current was applied, NO 3 − was mainly reduced by atomic H*-mediated indirect reduction, with atomic H* produced by electrolysis of H 2 O/H + via Co catalysis at the particle electrodes. The addition of PS did not interfere with the reduction of NO 3 − , instead promoting the generation of atomic H*, which further promoted the indirect reduction of NO 3 − . The NO 3 − reduction products consisted of nitrogenous gas, NH 4 + and residual NO 3 − . During the process of electro-oxidation, NH 3 /NH 4 + was directly oxidized via a three-electron reaction at the electrode, generating N 2 as a product. A portion of N 2 was reduced to NH 2 OH, which continued to undergo single-electron oxidation. The indirect oxidation of NH 4 + occurred mainly via SO 4 •− , while •OH functioned only in assistance. Although the DMPO-OH signal was stronger than that of DMPO-SO 4 , the slow reaction rate constant between •OH and NH 4 + resulted in a poor NH 4 + removal efficiency. The consumption of SO 4 •− could be compensated for by the interconversion between SO 4 •− and •OH. The oxidation of NH 4 + occurred mainly due to radicals accumulated on the surface of particle electrodes, with the products mainly consisting of nitrogenous gas and a small amount of NO 3 − .
by the interconversion between SO4 •− and •OH. The oxidation of NH4 + occurred mainly due to radicals accumulated on the surface of particle electrodes, with the products mainly consisting of nitrogenous gas and a small amount of NO3 − .

Conclusions
Oxidation of the byproduct NH4 + -N can be achieved using SO4 •− -based advanced oxidation under low current conditions, resulting in NH4 + being undetectable in the effluent, while also greatly reducing the TN concentration. Compared with active chlorine species, the oxidation of NH4 + via SO4 •− species can improve the denitrification efficiency and reduce energy consumption of the treatment system. SO4 •− can be efficiently and easily produced via the single-electron transfer process of PS on the cathodic side of Co/AC0.9-AB0.1. The PS activation process was found to promote the reduction of NO3 − , rather than interfere with it. The direct reduction of NO3 − on the surface of Co/AC0.9-AB0.1 did not occur easily, with atomic H*-mediated indirect reduction representing the primary pathway of NO3 − reduction. However, the direct oxidation of NH4 + on the Co/AC0.9-AB0.1 surface can be achieved via a three-electron transfer process, generating N2 as a product. During the indirect oxidation of NH4 + , SO4 •− played the dominant role while •OH acted only in assistance. Furthermore, the indirect oxidation of NH4 + was primarily accomplished by radicals that had accumulated on the surface of particle electrodes. The NH4 + oxidation products mainly consisted of nitrogenous gases, with very small amounts of NO3 − -N and undetectable levels of NO2 − -N present in the effluent. Therefore, the activation of PS using Co/AC0.9-AB0.1 particle electrodes might be a promising alternative method for oxidizing the byproduct NH4 + in the electro-reduction of NO3 − and reduce TN concentration in advanced sewage treatment.

Conclusions
Oxidation of the byproduct NH 4 + -N can be achieved using SO 4 •− -based advanced oxidation under low current conditions, resulting in NH 4 + being undetectable in the effluent, while also greatly reducing the TN concentration. Compared with active chlorine species, the oxidation of NH 4 + via SO 4 •− species can improve the denitrification efficiency and reduce energy consumption of the treatment system. SO 4 •− can be efficiently and easily produced via the single-electron transfer process of PS on the cathodic side of Co/AC 0.9 -AB 0.1 . The PS activation process was found to promote the reduction of NO 3 − , rather than interfere with it. The direct reduction of NO 3 − on the surface of Co/AC 0.9 -AB 0.1 did not occur easily, with atomic H*-mediated indirect reduction representing the primary pathway of NO 3 − reduction. However, the direct oxidation of NH 4 + on the Co/AC 0.9 -AB 0.1 surface can be achieved via a three-electron transfer process, generating N 2 as a product. During the indirect oxidation of NH 4 + , SO 4 •− played the dominant role while •OH acted only in assistance. Furthermore, the indirect oxidation of NH 4 + was primarily accomplished by radicals that had accumulated on the surface of particle electrodes. The NH 4 + oxidation products mainly consisted of nitrogenous gases, with very small amounts of NO 3 − -N and undetectable levels of NO 2 − -N present in the effluent. Therefore, the activation of PS using Co/AC 0.9 -AB 0.1 particle electrodes might be a promising alternative method for oxidizing the byproduct NH 4 + in the electro-reduction of NO 3 − and reduce TN concentration in advanced sewage treatment.