Singlet oxygen generation for selective oxidation of emerging pollutants in a flow-by electrochemical system based on natural air diffusion cathode

The decay of free radicals involved in side reactions is one of the challenges faced by electrochemical degradation of organic pollutants. To this end, a non-radical oxidation system was constructed by a natural air diffusion cathode (ADC) and a Ti-based dimensional stable anode coated by RuO2 (RuO2-Ti anode) for cathodic hydrogen peroxide activation by anodic chlorine evolution. The ADC fabricated by the carbon black of BP2000 produced a stable concentration of hydrogen peroxide of 339.94 mg L−1 (current efficiency of 73.4%) without aeration, which was superior to the cathode made by the XC72 carbon black. The flow-by ADC-RuO2 system consisted of an ADC and a RuO2-Ti anode showed high selectivity to aniline (AN) compared to benzoate (BA) in a NaCl electrolyte, whose degradation efficiencies were 97.72% and 1.3%, respectively. Rapid degradations of a mixture of emerging pollutants and AN were also observed in the ADC-RuO2 system, with pseudo-first-order kinetic constants of 0.51, 1.29, 0.89, and 0.99 min−1 for Bisphenol A (BPA), tetracycline (TC), sulfamethoxazole (SMX) and AN, respectively. Quenching experiments revealed the main reactive oxygen species for the pollutant degradation was singlet oxygen (1O2), which was also identified by the electron spin resonance (ESR) analysis. Finally, the steady-stable content of 1O2 was quantitatively determined to be 6.25 × 10−11 M by the method of furfuryl alcohol (FFA) probe. Our findings provide a fast, low energy consumption and well controlled electrochemical oxidation method for selective degradation of organic pollutants. H2O2 generated on an air diffusion cathode by naturally diffused O2, reacts with ClO− produced from chloride oxidation on the RuO2-Ti anode to form singlet oxygen (1O2). The electrochemical system shows an efficient oxidation to electron-rich emerging pollutants including bisphenol A, tetracycline, sulfamethoxazole and aniline, but a poor performance on the electron-deficient compounds (e.g., benzoate).


Introduction
Non-radical pathways due to their stable oxidative ability receive increasing attention in the field of water and wastewater treatment. Researches show non-radical oxidative pathways exhibit great resistance to certain anions (Duan et al. 2015;Luo et al. 2019). Among them, singlet oxygen ( 1 O 2 ) is the excited state of molecular oxygen, which has higher reactivity than the unexcited state (Stacey and Pope 2013). It is a non-radical and electrophilic oxidant that selectively reacts with electron-rich moieties of organic compounds (Agnez-Lima et al. 2012). Moreover, singlet oxygen shows an ability of anti-interference to other coexisting substrates (such as natural organics, inorganic ions) and the characteristic of producing less halogenated disinfection by-products (Huang et al. 2021).
Oxidation of H 2 O 2 by hypochlorite is known to generate 1 O 2 (Held et al. 1978). Based on this reaction, some electrochemical systems were developed for organic pollutants oxidation (Guo et al. 2021a, b;Lu et al. 2022;Tian et al. 2016). For example, Tian et al. utilized in situ generation of HOCl at an anode of RuO 2 -Ti and H 2 O 2 at a cathode of activated carbon fiber for singlet oxygen dominated cyanide oxidation (Tian et al. 2016). Lu et al. developed an electrochemical disinfection system by a dimensionally stable anode and a graphite-felt cathode with O 2 aeration, where the important role of 1 O 2 was also verified (Lu et al. 2022). Guo et al. designed a flow-through electrochemical system based on a functional carbon nanotubes cathode filter to generate 1 O 2 by in situ formed H 2 O 2 with ClO − electrolyte (Guo and Liu 2020). However, the production of H 2 O 2 in these systems all depends on dissolved oxygen content which is limited by the saturation solubility in the electrolyte (Sirés and Brillas 2021). In addition, extra energy input for the aeration increases the cost of treatment.
To develop a 1 O 2 -mediated electrochemical system for practical applications, a cathode for efficient H 2 O 2 production is critical since it produces one of the main reactants for 1 O 2 generation. In order to elevate the efficiency of H 2 O 2 electrosynthesis, researchers have developed cathodes of various structures and materials (Brillas 2022a; Luo et al. 2015;Walker et al. 2021;Zhang et al. 2022a, b). The mass transport of O 2 to the cathode is essential for the production of hydrogen peroxide (Sirés and Brillas 2021). Carbonaceous cathodes like carbon sponge, carbon felt, seem the better choices for high mass transport of O 2 and low cost (Brillas 2022b). Zhang et al. prepared a cathode of synthesized agarose catalyst using the air-brushing method to obtain a H 2 O 2 yield of 690.2 mg L −1 (Zhang et al. 2022a, b). Among them, air diffusion cathodes (ADC) are usually reported for highly efficient production of H 2 O 2 under extensive range of current densities (Liao et al. 2022). Márquez developed a 3D-like air-diffusion cathode to favor the H 2 O 2 production for Fenton's reaction (Márquez et al. 2020). The special structure of ADC gives rise to a three-phase interface for O 2 gas, electrolyte and catalyst, which brings fast O 2 mass transfer from air (An et al. 2019). Our previous work developed a flow-by electrochemical system where O 2 used for H 2 O 2 production was sourced from natural diffusion of air . Thus, the energy cost by this design was largely decreased. Nevertheless, coupling the natural air diffusion cathode into the electrochemical system is still rarely utilized for 1 O 2 generation.
In this study, we developed an ADC for highly efficient production of H 2 O 2 without aeration and then constructed a flow-by electrochemical system. The main objective is to explore the generation mechanism of singlet oxygen, leverage the characteristics of 1 O 2 to overcome the interference of chloride ions and effectively degrade emerging pollutants. Firstly, devoted to choosing the catalyst material for the ADC fabrication, the microscopic properties of the cathodes, as well as hydrogen peroxide production were compared with two different carbon black. Secondly, the oxidation selectivity of the electrochemical system was examined by individual pollutants and a mixture of four emerging pollutants. Thirdly, the contribution of singlet oxygen to the pollutant oxidation was elucidated by quenching tests and ESR analyses. Finally, the yields of H 2 O 2 , ClO − , and steady-state concentration of singlet oxygen in the system were quantitatively measured.

Fabrication of air diffusion cathode
Air diffusion cathode (ADC) was made by rolling a mixture of conductive carbon black (CB) and PTFE emulsion. Firstly, CB was dispersed in anhydrous ethanol and stirred well. Then, PTFE emulsion with a solid content of 60% was added dropwise. The above two steps need to be mixed under ultrasonically stirring for 30 min. The mass ratios of carbon black to PTFE in gas diffusion layer (GDL) and catalytic layer (CL) were 1:3 and 2:1, respectively. The suspension obtained was dried in water bath at 80 ℃ to remove the excess ethanol and to form a dough. Then, the dough was pressed into sheet of about 0.2 mm under a roller press. The sheets of GDL and CL were placed on two sides of a stainless-steel mesh (the collector layer) and pressed into a sandwich-like ADC. For each type of carbon black material, three cathodes with the same composition were fabricated to verify the reproducibility of the results.

Experimental approach
Pollutant degradation experiments were performed at a parallel plate electrochemical device. As shown in Fig. S1, the volume of resin electrochemical cell was 8.0 mL, where the central cross section was a square (1.0 × 1.0 cm). The ADC (8.0 × 1.0 cm) was assembled tightly between the cell and air chamber (8.0 × 1.0 cm) with CL facing the electrolyte and GDL facing the air. RuO 2 -Ti anode was placed parallel the ADC with a distance of 1 cm.
Constant current was supplied by a direct current (DC) power supply (GPS-X303/C, Taiwan Goodwill Instrument), and the current intensity was measured by a multimeter (FLUKE F15B +). Simulated wastewater in a beaker of 1 L was pumped into the cell from the bottom (Fig. S1) with hydraulic residence times (HRT) of 2 min at the flow rate of 4 mL min −1 by a peristaltic pump (iPump2s, Baoding Sinuo Fluid Technology Co., Ltd, China). Unless specified, the simulated wastewater contains 0.1 M NaCl, and 0.1 mM contaminants. The initial pH value was adjusted to 10 by adding diluted HCl and NaOH. Samples were taken from the outlet at a predetermined time for concentration analysis. All experiments were performed in duplicate.

Analysis
The concentration of H 2 O 2 was determined by a titanium sulfate method with a AOE UV-1800 spectrophotometer (Ao Yi Instrument, Shanghai, China) at 405 nm (Liao et al. 2013). Free chlorine concentration was determined by the DPD method at 515 nm.
AN, BPA, TC, and SMX were detected by a FL2200 high performance liquid chromatography (HPLC; Fuli Instruments) equipped with a UV-vis detector and a Shimadzu InertSustain C18 column (5 μm, 4.6 × 150 mm) with a mobile phase of 70% methanol in DI water at a flow rate of 1 mL min −1 . The detection wavelength was 230 nm, 276 nm, 280 nm, and 275 nm for AN, BPA, TC, and SMX, respectively (Ahmed et al. 2012; Kyotani et al. 1996;Sattler et al. 1994;Zhang et al. 2014). BA was detected using the same column in a mobile phase containing methanol and 0.1% trifluoroacetate solution (v/v = 30/70) at 255 nm (Nicoli et al. 2008). FFA was analyzed using the same column in a mobile phase of methanol and DI water (v/v = 50/50) at 220 nm (Kohn and Nelson 2007). The pH value of the electrolyte was measured by a portable pH meter (wtw-350i, Germany). Total organic carbon (TOC) was monitored by a TOC analyzer (Shimadzu, TOC-L CPH, Japan).
Current efficiency (CE) also known as Faraday efficiency was calculated as Eq. (1) (Liu et al. 2015), where n was the number of electrons transferred from oxygen reduction to hydrogen peroxide (2), F was Faraday's constant (96,486 mol L −1 ), C represented the concentration of hydrogen peroxide concentration (mol L −1 ), q was flow rate (L s −1 ), and I was current intensity (A).
The electric energy consumption (EEC, kWh kg −1 ) was calculated by Eq. (2) , where U and I were the applied voltage (V) and current (A), C was the concentration of H 2 O 2 (mg L −1 ), and q was flow rate (L h −1 ). The FFA decay model was assumed to follow the pseudo-first-order kinetics with the rate constant of k obs,FFA . The steady state concentration of singlet oxygen could be calculated by Eq. (3)

Characterization
The morphology of ADC was observed under field emission scanning electron microscope (SEM, Zeiss G300). The N 2 adsorption analysis of the sample was analyzed at 77 K by a specific surface area and pore size analyzer (SSA-4300 Beijing Builder Electronic Technology Co., Ltd.). The specific surface area was measured by the Brunauer-Emmett-Teller (BET) method, and the Barrett-Joyner-Halenda (BJH) method was used to study the porosity and volume distribution of the material. The electronic spinning resonance (ESR) was carried on an electron paramagnetic resonance spectrum (JES-FA200). 2,2,6,6-Tetramethyl-4-piperidone hydrochloride (TEMP) was used as the spin-trapping agent of 1 O 2 . The trapping agent of 60 mM TEMP was added to electrolyte containing 0.1 M NaCl or 0.1 M Na 2 SO 4 before electrolysis.

Performance of the ADC
Our study sets out with a detailed exploration of the suitable carbon materials for fabrication of the the catalytic layers (CL) of the ADC. The structure, composition and morphology were compared for the CL made by two carbon materials (i.e., BP2000 and XC72), which were chosen because of wide use and high performance in electrosynthesis of H 2 O 2 . As shown in Fig. 1, honeycomb-like pores distributed evenly on the flat surface, and black-grey carbon particles were attached to the white polytetrafluoroethylene (PTFE) grid for the catalytic layers (CL) made of BP2000 and XC72. The images also showed massive micropores, mesopores and micropores co-existed, which composed channels for gas and electron transfer. However, the catalytic layer made by BP2000 had a flatter surface with more micropores and mesopores. For XC72, the macropores and cracks appeared significantly, which may be due to the large difference in the particles of the two carbon black materials.
To further analyse the microscopic structure, N 2 adsorption and desorption tests were carried out. Figure 2a-b illustrated type IV isotherms for both GDL and CL, which indicated the formation of mesopores and capillary condensation . Furthermore, H3-type hysteresis loops were observed for two carbon materials suggesting the existence of micropores (Guo et al. 2021a, b). To measure the specific surface area porosity and volume distribution of the material, BET and BJH method were carried out. As shown in Table 1, BET surface area of CL made by BP2000 was 396.04 m 2 /g, which was about 8 times of XC72 (51.54 m 2 /g). Larger surface area provided more Besides, the CL made by BP2000 had bigger pore volume (2.40 cm 3 g −1 ), and smaller average pore size (19.36 nm) than those of XC72, indicating higher content of micropores and mesopores. This was consistent with the huge surface area of CL fabricated by BP2000. More micropores and mesopores were conducive to the liquid-solid-gas three-phase, which may promote the production of hydrogen peroxide (F. Guo et al. 2021a, b).
Finally, we conducted electrolysis experiments to examine the electrocatalytic activity of the ADC. As shown in Fig. 2c-d, the yield of hydrogen peroxide achieved 336.55 mg L −1 for the ADC made by BP2000 after 4 HRT, with a stable current efficiency (CE) of 73.4%, which were much higher than that made by XC72 (production of 275.34 mg L −1 and CE of 62.67%). During the electrolysis process, the ADC made by BP2000 always showed higher yield of hydrogen peroxide than that of XC72. The results were supported by the higher BET surface area and special pore size distribution of BP2000 ADC. Therefore, BP2000 was chosen for the fabrications of ADC in the following experiments.
To get a deeper understanding of the performance of the self-made ADC, the effect of current intensity, flow rate, solution pH value and electrolyte concentration was explored. As shown in Fig. S3a, the stable yield of H 2 O 2 increased from 72.18 to 630.42 mg L −1 , while the CE significantly increased from 50 to 200 mA but decreased from 200 to 500 mA, showing the highest CE of 73.4% at 200 mA. High current intensity was generally thought to be harmful for H 2 O 2 production, owing to side reactions such as H 2 evolution and H 2 O 2 decomposition (An et al. 2019). The production of hydrogen peroxide increased remarkably from 99.38 to 335.27 mg L −1 as the flow rate was decreasing from 16 to 4 mL min −1 but decreased to 298.34 mg L −1 at 1 mL min −1 (Fig. S3b). Low flow rate and long resident time facilitated the H 2 O 2 accumulation, but which was limited by auto decomposition of H 2 O 2 at high concentrations. At the pH range of 4-10, the ADC exhibited less dependence on the solution pH (Fig. S3c). As reported, poor performance would be caused by H 2 evolution and combination of H + under strong acidic condition (Wang et al. 2020). Even so, the ADC exhibited good performance in a wide pH range. The yield of H 2 O 2 increased significantly when the concentration of electrolyte increased to 0.1 M (Fig. S3d). From the economic point of view, the current intensity of 200 mA, flow rate of 4 mL min −1 , and 0.1 M electrolyte was employed for the following experiments.
We presented a detailed comparison of the energy consumption of the ADC system in this work with those reported (Table S1). As shown in Table S1, an average EEC of 6.7 kWh kg −1 was obtained for the ADC made of BP2000. Yu et al. fabricated a gas diffusion cathode of carbon black and PTFE for hydrogen peroxide production with the EEC of 15.9 kWh kg −1 , while the aeration energy consumption (64.1 kWh kg −1 ) was three times higher than the EEC (Yu et al. 2015). In another work, the cathode for H 2 O 2 yielding also showed a high aeration energy consumption up to 86 kWh kg −1 (Cordeiro et al. 2022). The ADC in this work shows outstanding electrocatalytic activity towards H 2 O 2 production together with low energy consumption. Utilization of natural diffusion air for H 2 O 2 electrosynthesis saves a large part of energy consumption on the aeration, outperforming traditional cathode.

Oxidation capacity of the ADC-RuO 2 system
In order to verify the oxidation ability of the ADC-RuO 2 system, two model pollutants, BA and AN, were selected for the degradation experiments. As shown in Fig. 3a, AN was removed by 97.72% when the reaction reached equilibrium after 1 HRT. However, BA was hardly oxidized after 2 HRT with degradation efficiencies of around 1.3%. This finding was consistent with earlier observations that the oxidation system based on singlet oxygen exhibited poor oxidation of BA compared to other pollutants ). These results suggested that the oxidation in the system might not be induced by hydroxyl radical but more likely by a selective oxidant. The high selective oxidation of benzyl alcohol to benzaldehyde by photocatalysis was also observed in other studies (Tayyab et al. 2022). Although AN was efficiently removed in ADC-RuO 2 system, the TOC concentration only decreased by 6.90% (Fig. S4). This observation suggested that the ADC-RuO 2 system showed a poor mineralization ability. The selective removal of aniline suggests a non-radical oxidation pathway for the ADC-RuO 2 system. Studies have shown that organic substrates with electron-rich groups (such as anilines) can act as electron donors to reactive oxides, such pollutants were more prone to be oxidized by electrophilic reactions through non-radical processes (Lee et al. 2016(Lee et al. , 2015Zhou et al. 2015). BA was an organic substrate with electron-deficient groups, which it was difficult to be oxidized. Additional to the single pollutant of AN, a mixture of AN with several emerging contaminants (i.e., bisphenol A (BPA), tetracycline (TC) and sulfamethoxazole (SMX)) were used to evaluate the oxidation ability of the system. These emerging pollutants were chosen as they are stable in water, but cause harm to the human body even in tiny concentration (Horikoshi et al. 2008;Kwon et al. 2018;Liu et al. 2002;Zhang et al. 2017). As shown in Fig. 3b, the four kinds of pollutants were simultaneously removed with high efficiency of > 90% in 2 HRT degradation by the ADC-RuO 2 system. Among the four pollutants, the removal rates followed a sequence of TC > AN > SMX > BPA. Figure 3c showed that the degradation of the four pollutants (BPA, TC, SMX and AN) basically conformed to the first-order kinetics (k obs = 0.51, 1.29, 0.89, and 0.99 min −1 , respectively). The results indicated that the reactive oxides preferentially reacted with TC, followed by AN and SMX, and the last was BPA. The weakest ability to oxidize BPA may be owing to BPA that existed partly in their neutral form (pKa values of 9.7-10.2) at pH = 10, which showed higher resistance to oxidation by the ROS (Bokare and Choi 2015). However, SMX, TC, and AN all mainly existed as the negative ions or protonated complex as their pK a or pK b values were much lower than 10.0 (Hou et al. 2022;Lin et al. 1997;Wang et al. 2011;Zhang et al. 2022a, b).
These experiments confirmed that the ADC-RuO 2 system exhibited excellent performance for the degradation of antibiotics and industrial raw materials in the wastewater containing high concentrations of chloride ions. In contrast, radicals-based oxidation processes were weakened by the side reactions with some coexisting organic or inorganic substrates (such as chloride ions, etc.) because of their strong oxidizing and non-selective properties (Duan et al. 2015;Huang et al. 2021;Luo et al. 2019). However, the reactive oxidants established in this system exhibited strong antiinterference ability and selective degradation ability. The results suggested that the system may degrade these pollutants through a non-radical process (Zheng et al. 2021). In addition, no chemical agents were added during the treatment by the ADC-RuO 2 system, which make it a green, compact and convenient method for distributed wastewater treatment applications .
Wastewater from different source might show varied ion compositions, especially the concentration of chloride ions, which was reported to have a great impact on the oxidation of pollutants. As shown in Fig. S5, the rate of degradation raised from 52.88 to 97.72% and the pseudo-first-order kinetic constants (k obs ) increased from 0.17 to 1.05 min −1 as the concentration of NaCl increased from 0.01 to 0.10 M. However, there was negligible improvement of the degradation rate when the NaCl concentration increased from 0.1 to 0.15 M. Lower voltages were obtained in higher Fig. 3 Degradation of (a) AN and BA, (b) a mixture of emerging pollutants, (c) plot of ln(C/C 0 ) versus time and (d) pseudo-first-order rate constants (k obs ) for the emerging pollutants. Experiment conditions: initial pH = 10, concentration of pollutants = 0.1 mM, [NaCl] = 0.1 M, electrolysis using a RuO 2 -Ti anode and ADC cathode, current intensity (I) = 200 mA, HRT = 2 min. Error bars represent the standard deviations of duplicate measurements concentration of electrolyte, otherwise high voltage would be harmful for H 2 O 2 production owing to side reactions that spring up such as H 2 evolution and H 2 O 2 decomposition. Moreover, the increase of Cl − concentration may be beneficial to generate reactive oxidation generation on Ti-RuO 2 anode. These results confirmed that the proposed ADC-RuO 2 system had a high tolerance to the wastewater containing high concentrations of Cl − , which might be an efficient treatment method for hypersaline industrial wastewaters.

Identification of reactive oxidizing intermediates
To elucidate the oxidation mechanism and identify the reactive oxidizing intermediates, ESR and radical scavenging experiments were carried out. TBA, NaHCO 3 , and p-BQ were adopted as exclusive scavengers for Hydroxyl radicals (•OH), free chlorine, and superoxide radicals (HO 2 •/O 2 • − ), respectively (Cai et al. 2022;Luna-Trujillo et al. 2020;Yang et al. 2018). Figure 4 showed that the addition of TBA or p-BQ barely inhibited the AN degradation. With the presence of 10 mM TBA or p-BQ, the removal efficiency of AN decreased insignificantly from 97.72 to 96.25% or 97.72 to 97.36%, respectively (Fig. 4a). And the k obs decreased from 1.07 to 0.93 min −1 and 1.05 min −1 respectively, which means negligible contributions of the (•OH) and HO 2 •/O 2 • − to AN oxidation. While in the presence of 10 mM NaHCO 3 the removal efficiency of AN decreased from 97.72 to 86.42% and the k obs reduced from 1.07 to 0.51 min −1 , which suggested a small contribution of free chlorine (Fig. 4b). In general, it could be concluded that the removal of AN in the ADC-RuO 2 system was not mainly induced by a radical process. To verify the contribution of singlet oxygen that produced from the reactions of hypochlorous ions and hydrogen peroxide (Guo and Liu 2020;Lu et al. 2022), FFA was used as a probe compound (k 1O2 , FFA = 1.2 × 10 8 M −1 s −1 ) (Cheng et al. 2017). Figure 4a showed that the degradation efficiency of AN obviously decreased from the 97.75 to 71.47%, and the k obs was significantly reduced from 1.07 to 0.30 min −1 with the presence of 10 mM FFA. When the concentration of FFA was further increased to 50 mM and 100 mM, the degradation efficiency of AN decreased sharply to 53.9 and 26.97% respectively (Fig. 5a). According to these data, we can infer that 1 O 2 may be the dominant reactive oxides for the degradation of AN in the system.
In order to confirm the assumption, the electron spin resonance (ESR) spectroscopy was conducted. As shown in Fig. 4d, there was no obvious signal when ADC electrodes or RuO 2 electrodes were employed alone. However, in the ADC-RuO 2 system operated at 100 mA, intense threeline signals (1:1:1) were observed, which can be assigned to tetramethylpiperidine-oxide (TEMPO) generated by the oxidation of 1 O 2 (An et al. 2019). Moreover, as increasing the current to 200 mA, the intensity of TEMPO signal was significantly enhanced, indicating that elevating the current

Quantitative analysis of singlet oxygen generation
Although a singlet oxygen-mediated oxidation pathway was confirmed, a quantitative analysis was still in need to elucidate the oxidation mechanism in depth. Considering the 1 O 2 was generated from the reactions of hypochlorite (ClO − ) and H 2 O 2 in the RuO 2 -ADC system , we firstly measured the production rates of ClO −− and H 2 O 2 under different current density in batch reactors (the experimental facilities were shown in Fig. S6). Figure 6 illustrated that production of ClO −− and H 2 O 2 both increased lineally with time. The yield rates of H 2 O 2 and free chlorine increased from 0.05 to 0.31 mM min −1 and from 0.05 to 0.35 mM min −1 with the current intensity increasing from 50 to 300 mA, respectively ( Fig. 6a-b). As shown in Fig. 6c, the relationship of H 2 O 2 and free chlorine was following monotonic function with current intensity (Eqs. (4)- (5) These results suggested that the generation of H 2 O 2 and free chlorine could be modulated by the current applied. At 200 mA, according to the fitted equation, the instantaneous production of hydrogen peroxide and free chlorine was 0.21 mM min −1 and 0.23 mM min −1 , respectively. According to the reaction formula (Eq. (6)) , the theoretical production of 1 O 2 can be preliminarily calculated to be 0.21 mM min −1 at a current intensity of 200 mA.
Although a high concentration of singlet oxygen was predicted in the ADC-RuO 2 system, the steady state concentration would be a minute amount as a reactive oxidant could not exist stably. We utilized FFA as a probe for the detection of singlet oxygen (k FFA = 1.2 × 10 8 M −1 s −1 ) (Cheng et al. 2017). 1 O 2 could be completely scavenged by FFA, thus the reduction of FFA concentration was proportional to the amount of singlet oxygen generated (Cheng et al. 2017). As shown in Fig. 6d, the stable residual rate of FFA was 40.77% and the pseudo-first-order rate constants (k obs,FFA ) of 0.45 min −1 was obtained. According to Eq. (3), the steady state concentration of singlet oxygen was calculated to be 6.25 × 10 −11 M, which was quite less than the theoretical value of 0.2150 mM min −1 . A possible explanation for this might be the short half-life of singlet oxygen (0.2-2.0 μs)  (Salokhiddinov et al. 1981). In addition, because of other substances that existed in the measured free chlorine, such as perchlorate, ClO − yield might be overestimated (Shinohara et al. 2006). Moreover, the reaction rate between H 2 O 2 and ClO − was affected by the relative molar ratio of each other. When the initial concentration of H 2 O 2 and ClO − was 10 mg L −1 , it took 2 min to react completely (Agnez-Lima et al. 2012). However, as a reactive intermediate, 1 O 2 still exhibited significant impact on the pollutant degradation as showed in previous sections.

Treatment of industrial wastewater
To further verify the selective oxidation capacity of the ADC-RuO 2 system, an industrial wastewater was sampled from the balance tank of a sewage plant in Hubei Province (China) for experiment, where a mixture wastewater from pharmaceutical factory, paper mill and insecticide factory were collected. The basic properties of the wastewater are presented in Table S2. High concentrations of Cl − (5000-7000 mg L −1 ) and COD (9000-13,000 mg L −1 ) in the wastewater were refractory to treatments by biological method and Fenton oxidation. As shown in Fig. S7, AN was removed by 30% at an initial concentration of 1 g L −1 in the ADC-RuO 2 system under a flow rate of 4 mL min −1 and current intensity of 200 mA. This finding suggests that the ADC-RuO 2 system was highly selective to the targeted pollutants removal in a complex wastewater environment.

Conclusion
A non-radical oxidation process was constructed by a natural air diffusion cathode (ADC) and RuO 2 -Ti anode for cathodic hydrogen peroxide activation by anodic chlorine evolution. The ADC exhibited great catalytic performance on hydrogen peroxide production on account of abundant micropores and mesopores in the catalysis layer made of BP2000. The self-breathing cathode in a flowby reactor avoids oxygen supplement by aeration, which greatly reduced the energy consumption and improved the efficiency. Different from the free radical process, the system showed higher selectivity for organic pollutants with electron-rich groups (e.g., SMX, AN, TC and BPA). The ADC-RuO 2 system had a high tolerance to the wastewater containing high concentrations of Cl − , which might be an efficient treatment method for hypersaline industrial wastewaters. Quenching experiments and ESR analysis further demonstrated that 1 O 2 was the dominant ROS. The steady-stable content of 1 O 2 was quantitatively determined to be 6.25 × 10 −11 M by the FFA probe. The proposed nonradical oxidation pathway makes the novel system to be highly selective to the targeted pollutants removal in a complex wastewater environment. Overall, our findings provided a highly efficient, low energy consumption and well-controlled electrochemical oxidation system aiming to provide a new idea for future distributed water treatment technology. However, on account of a poor mineralization