A novel persulfate-photo-bioelectrochemical hybrid system promoting the degradation of refractory micropollutants at neutral pH

Bio-electro-Fenton is emerging as an alternative technology for the efficient and cost-effective removal of refractory micropollutants. Though promising, there are still several challenges that limit its wide application, including acidic operating conditions (pH at 2-3), the addition of supporting electrolytes (e.g., Na2SO4), and the issue of iron sludge generation. To address these challenges, a novel hybrid persulfate-photo-bioelectrochemical (PPBEC) system is proposed to remove model micropollutants (carbamazepine and clorfibric acid), from secondary effluent at low persulfate (PS) dosage and neutral pH. The effect of crucial operating parameters on the process was studied, including input voltage, cathodic aeration velocity, and PS dose. Under optimal conditions (0.6 V, 0.005 mL min -1 mL -1 and 1 mM), the PPBEC system achieved approx. 0.56-1.71 times greater micropollutant removal with 93% lower energy consumption when compared to the Jo ur na l P re -p ro of individual processes (UV/PS and PBEC). The improved performance was attributed to a faster production of sulfate radicals by UV irradiation, hydrogen peroxide activation and single-electron reduction, and hydroxyl radicals generated by UV irradiation. Furthermore, the transformation products of carbamazepine and clorfibric acid were identified and the probable pathways are proposed. Finally, the ecotoxicity of the PPBEC treated effluent was assessed by using Vibrio Fischeri, which exhibited a non-toxic effect.


Introduction
In recent years, high levels of residual pharmaceuticals have been detected in municipal sewage due to the increasing global consumption of pharmaceuticals. This in turn brings new challenges to traditional sewage treatment facilities. Current wastewater treatment discharge standards do not typically have threshold limits for these refractory pharmaceuticals, which has resulted in their widespread detection in surface water, groundwater, and even drinking water (Monteil et al., 2019).
As a consequence, long-term exposure to these refractory pharmaceuticals has had measurable J o u r n a l P r e -p r o o f adverse impacts on various organisms and even humans (e.g., spread distribution of antibioticresistant genes) (Pazda et al., 2019).
Among existing treatment methods, AOPs have received widespread attention due to their ability to efficiently eliminate refractory pharmaceuticals in a reliable manner, via the generation of free radicals with strong oxidizing capabilities, such as hydroxyl radicals (•OH). Recently, bioelectrochemical process-based AOPs, in particular bio-electro-Fenton (BEF) process, have become a focal point of AOP research due to several advantages. Firstly, they inherit the advantage of the traditional electro-Fenton (EF) process that can efficiently treat high-concentration (mg L -1 and even to g L -1 ) pharmaceutical-containing wastewater through in situ production of •OH (Eq.(1)).
Secondly, they are highly controllable and require only mild operating conditions. Thirdly, the energy consumption of BEF systems is typically low (Ahmed et al., 2017;Monteil et al., 2019).
Despite being such a promising technology, there are still several challenges that need to be addressed for BEF processes. Firstly, the use of iron as a catalyst in the BEF process can lead to iron sludge problems (Nadais et al., 2018). Additionally, as both EF and BEF processes are optimized at pH 2-3, pH adjustments are thus needed, both before and after treatment, which inevitably increases their operating costs. Furthermore, the addition of supporting electrolytes (e.g. Na 2 SO 4 ) has been required to increase the current flow and the treatment efficiency. This can also increase direct operating costs, as well as requiring further downstream treatment (e.g. reverse osmosis) before discharge, to reduce potential negative effects on receiving water bodies (Nadais et al., 2018). These bottlenecks have so far restricted the wide application of BEF technology.
Furthermore, the variability in quantity and quality of natural organic matter (NOM) and inorganic ions within real water matrices may also adversely affect the treatment due to the non-selectivity of •OH radicals, and their subsequent scavenging effect.

J o u r n a l P r e -p r o o f
In addition to •OH, sulfate radicals (•SO 4 -) have also been adopted in AOPs to oxidize organics due to their advantages of having longer reaction lifetimes, selective electron transfer, and ability to work efficiently over a wider pH range (2-8) (Pan et al., 2019;Xu et al., 2017). •SO 4 can be generated by activating persulfate (PS) or permonosulfate (PMS) in a variety of ways, including the use of H 2 O 2 , transition metals, ultraviolet (UV) irradiation, visible light, etc (Wang and Wang, 2018a). Among them, UV irradiation has been shown as one of the most effective ways to activate PS or PMS. So far, the UV/PS process (Eq. (2)) has proven to be both more efficient and economical for the removal of refractory pharmaceuticals, than the UV/PMS process (Deng et al., 2013).
However, the performance of the UV/PS process in the treatment of real wastewater is often limited because the wastewater characteristics including turbidity, chromaticity, and suspended solids could reduce the UV transmittance, thereby reducing the formation of •SO 4 -. Besides, the removal efficiency of the UV/PS process is mainly dependent on the PS dosage. This increases both the treatment cost, as well as increases the production of SO 4 2waste which requires further processing.
In this context, the combination of the UV/PS process and the BEF process could not only greatly improve the removal efficiency but also overcome the shortcomings of the individual UV/PS process and the BEF process mentioned above. Specifically, photocatalysis by UV irradiation to eliminate pharmaceuticals can also be achieved through direct photolysis, or by activation of the cathodic H 2 O 2 generated in situ, to produce •OH (Eq. (3)). These processes can also help to overcome the abovementioned drawbacks in the BEF system, including the requirement for pH adjustment and iron sludge generation. Moreover, the activation of PS by the bioelectrochemical system could be enhanced to overcome the influence of UV transmittance (Durán et al., 2017), as it could also generate •SO 4 by single-electron reduction reaction (Eq. (4)) and the cathodic in situ generated H 2 O 2 activation (Eq. (5) Herein, a novel hybrid PS enhanced photo bioelectrochemical (PPBEC) system was developed, by combining a bioelectrochemical system and a UV/PS process, to be used as a tertiary treatment process for the removal of refractory pharmaceuticals from the real WWTPs secondary effluent at neutral pH. This hybrid system could significantly improve the treatment performance, as well as save costs, over previously developed UV/PS and BEF processes.
Carbamazepine (CBZ) and clofibric acid (CA) have been widely detected in the various water types such as surface water and WWTPs effluent (ranging from ng·L −1 to μg·L −1 ) around the world over recent decades (Calisto et al., 2011;Li et al., 2010). For example, it is reported that the highest CBZ concentration detected in surface water and groundwater were 11561 and 390 ng L -1 , respectively (Moztahida et al., 2019). In addition, the content of CA detected in Canadian surface water ranges from 0.101 to 0.175 μgL -1 (Rebelo et al., 2020). It can be expected that these pollutants will continue to accumulate in the environment to higher concentrations in the long term due to missing of efficient treatment. In addition, by 2016 at least 631 pharmaceutical active compounds have been detected in various water bodies (Ramírez-Morales et al., 2020). The total concentration of these compounds can easily end up to several mg L −1 . In this study, a higher initial concentration (mg L -1 level) of CBZ and CA were selected as the model compounds for refractory pollutants to examine the treatment performance of the novel PPBEC system. Furthermore, these concentration gradients were selected in order to compare the results with the existing studies. In many recent studies on the removal of CBZ, the adopted concentrations are about 5-15 mg L -1 (Lai et al., 2021;Thanekar et al., 2018;Yang et al., 2021;Yu et al., 2021). The impact of operating parameters including input J o u r n a l P r e -p r o o f voltage, cathodic aeration velocity, and PS dosage on the removal of the CBZ and CA were then evaluated. The major transformation products were also investigated and the effluent ecotoxicity was assessed. Furthermore, the applicability of the PPBEC system for the removal of two model pharmaceuticals in different water matrices was also explored. Finally, the PPBEC system was compared to the UV/PS process, the BEF process, and other AOPs such as the EF process, from the perspective of both removal efficiencies and economic cost.

Chemicals
For detailed information on the main chemicals used in this study, see Text S1 (supporting information). As for the detailed water characteristics of the 4 different water bodies, see Table S1.

PPBEC system setup and operation
A laboratory-scale PPBEC reactor was constructed, consisting of two identically-sized chambers where the effective size of each chamber was 8 cm × 5 cm × 5 cm (Fig. S1). Accordingly, the total volume of each chamber was 200 mL, with a working volume of 180 mL. The PPBEC design and materials for individual reactor components including the anode electrode (carbon brush), cation exchange membrane, and Ag/AgCl reference electrode in the cathode chamber, and the corresponding pretreatment methods were identical to those documented within our previous study (Nadais et al., 2018). In addition to these components, a commercial-grade graphite plate (4 cm × 4 cm) coated with carbon black was used as a gas diffusion cathode (GDC), and the specific preparation method was carried out as previously described (Yu et al., 2015). A detailed description of running the PPBEC system was shown in Text S2.

Analytical methods and calculations
The detection method for the in situ produced H 2 O 2 was the same as has been previously described (Nadais et al., 2018). The CBZ and CA concentrations and the corresponding transformation products during the PPBEC system were quantified by high-performance liquid chromatography Instruments, Inc., Cleveland, OH, USA), and the corresponding system current can be calculated by Ohm's law.
The removal of four selected model pharmaceuticals was followed by pseudo-first-order kinetics during the BEF process and UV/PS process (Nadais et al., 2018) , where the apparent rate constant (K app ) can be calculated via Eq. (6).
The electrical energy per order (E EO ) was adopted to express electric energy consumption, as it was most conducive for comparison with other AOPs in terms of energy consumption. The calculations of E EO (see Text S4) were conducted as previously described (Pan et al., 2019). Specifically, the bulk of the treatment cost of the PPBEC process originated from two aspects, including electrical energy consumption; and (2) chemical reagent (PS) cost. To further facilitate comparisons, the energy consumption in this paper was divided into four parts including UV lamp consumption, MEC reactor consumption (direct voltage input), aeration, and agitation consumption on the treatment of WWTPs secondary effluent (Lyngby) spiked with CBZ and CA (10 mg L -1 of each).
Calculations were based on the local electricity price (Denmark) and the PS price (0.25 € kWh -1 and 0.62 € kg -1 , respectively) and the corresponding total expenditure was further calculated in terms of cost per order (Cost/O total ).
Finally, one-way analysis of variance (one-way ANOVA) was carried out via SPSS 24.0 Statistics software (IBM Crop, USA) to analyse the significant difference among the obtained results. When the p value was <0.05, the data were considered statistically significant.
J o u r n a l P r e -p r o o f

Eco-toxicity assays
Microtox tests were conducted through a bioluminescence inhibition experiment, using Vibrio fischeri as the toxicity indicator (Zou et al., 2020b). This method was used to assess the ecotoxicity of both the secondary biological treatment effluent and the effluent from the novel PPBEC system. This finding is consistent with previous studies regarding the removal of emerging pharmaceuticals during UV/PS and BEF processes (Bu et al., 2016;Zou et al., 2020a). Furthermore, it was found that CBZ was still not easily removed by direct UV irradiation alone (7% after 8 h). In contrast, CA was better degraded by these methods (65% after 8 h), which was also consistent with previous literature (Ali et al., 2018;Lu et al., 2018). As depicted in Eq. (7) and Eq. (8), the degradation of CBZ and CA under direct UV irradiation can be divided into two steps. Firstly, the aromatic rings and heteroatoms such as nitrogen and chlorine are promoted to an electronically excited state under UV irradiation (254 nm) (Eq. (7)), which then react to form a new product through chemical decomposition (Ali et al., 2018;Carlson et al., 2015;Khan et al., 2017) (Eq. (8)). Unlike CA, CBZ contains an amide group (RCONH 2 ), which makes further degradation difficult (Ali et al., 2018).

Synergistic effect of PBEC with UV/PS
The observed degradation of CA under direct UV photolysis was consistent with previous reports,

J o u r n a l P r e -p r o o f
and is in line with pseudo-first-order kinetics with a rate constant (K app ) of 0.128 h -1 (R 2 =0.97) (Khan et al., 2014).
CBZ (CA) → CBZ* (CA*) (7) CBZ* (CA*) by-products (8) Since individual PS oxidation or MEC processes could not effectively remove CBZ and CA, and direct UV photolysis was also ineffective towards CBZ, further processes including PBEC (both with and without 50 mM Na 2 SO 4 addition), UV/PS, and PPBEC were also investigated. In the UV/PS process, degradation was significantly enhanced, achieving approximately 84.5% and 100% within 8 h for CBZ and CA, respectively. This can be attributed to the fact that the bond of O-O in PS is broken under UV irradiation due to the production of •SO 4 − containing a lone pair of electrons via Eq.
(2). This then has a strong oxidation ability, and can therefore quickly degrade both CBZ and CA (Gao et al., 2012). The PBEC process also showed a clear improvement in CBZ and CA removal, both with and without the addition of 50 mM Na 2 SO 4 . Here, the pharmaceuticals removal efficiency during the first 8 hrs without 50 mM Na 2 SO 4 addition was approximately 87.5% and 100% for CBZ and CA, respectively. The efficiency was increased to 100% for both compounds within 5 h, with the addition of 50 mM Na 2 SO 4 to the process. Similarly, as the UV/PS process was described above, PBEC can effectively remove CBZ and CA due to the in situ generation of H 2 O 2 .
Under UV irradiation via Eq (2), H 2 O 2 is converted to •OH, which has an even stronger oxidation ability than •SO 4 − (Yuan et al., 2009). The addition of Na 2 SO 4, resulted in the improvement of cathodic conductivity (from 1.45 to 7.81 ms cm -1 ) and subsequently led to the increase of system current (from approximately 2.25 to 5.50 mA, Fig. S2). This ultimately resulted in an increased yield of H 2 O 2 and •OH for the degradation of CBZ and CA Moreira et al., 2017).
The degradation of CBZ and CA during the above processes were found to follow pseudo-firstorder kinetics.

J o u r n a l P r e -p r o o f
h -1 where each value corresponds to UV/PS, PBEC and PBEC (with 50 mM Na 2 SO 4 ) processes, respectively. Although the PBEC process with the addition of 50 mM Na 2 SO 4 achieved the highest removal efficiencies for our target pollutants, in practice the addition of such a high concentration of salts would require further processing before discharge, and hence limit its wide application.
Fortunately, this limitation could be overcome by the addition of PS in place of Na 2 SO 4  demonstrated the stability of the bioanode-based PBEC system, as well as for the derived PPBEC system. Due to the use of bipolar membranes, the pH of the entire treatment process increased slightly (from 7.60 to 8.28, Fig. S2). Unlike the traditional EF and BEF processes, pH adjustment was therefore not required. Overall, the PPBEC process was proven the most effective method for the removal of selected pollutants.
J o u r n a l P r e -p r o o f

Applied voltage
The circuit current is directly related to the rate of H 2 O 2 generation from the cathode (Eq. (1)), which can subsequently affect the formation of •OH (Xu et al., 2016;Zhao et al., 2017). Therefore, The PPBEC process was further investigated at a range of applied voltages (0, 0.2, 0.4, 0.6, and 0.8 V). As shown in Fig. 2 (16) The residual H 2 O 2 concentrations observed under different input voltages could further support the above explanation ( Fig. S3) could support this explanation. From the above results, 0.6 V was considered as the most appropriate input voltage for the PPBEC process.

Cathodic aeration velocity
As illustrated in Eq.
(1), cathodic O 2 supply can be another crucial operational parameter affecting the synthesis of H 2 O 2 and the pharmeceuticals removal. Therefore, the impact of O 2 supply on the performance of the PPBEC process was further studied at varied cathodic aeration velocities ranging from 0 to 4 mL min -1 (corresponding to 0 to 0.02 mL min -1 mL -1 ). As shown in Fig. 3, the removal efficiency of CBZ and CA was initially improved when the aeration velocity was raised from 0 to 0.005 mL min -1 mL -1 , and then subsequently dropped when the aeration velocity was further increased to 0.01 and even 0.02 mL min -1 mL -1 . One-way ANOVA (p<0.05, Table S3) further showed a statistically significant difference in aeration rates between the four selected groups, suggesting that the contribution of O 2 was significant. This pattern was likely due to the introduction of proper cathodic aeration, which may maintain the dissolved O 2 concentration in the catholyte, and thus maintain the synthesis of H 2 O 2 for better removal efficiency than that without aeration. However, when the aeration velocity was further increased, the amount of bubbles also increased, and this, in turn could increase the internal resistance of the system. The current J o u r n a l P r e -p r o o f decreased when the aeration velocity was increased to 0.01 and further to 0.02 mL min -1 mL -1 (Fig.   S4), which is further evidence for this hypothesis (Luo et al., 2015;Zhou et al., 2013). Furthermore, the bubble sizes generated at faster aeration velocities were larger, which may hinder binding to the active site of the cathode (Liu et al., 2007). A faster aeration velocity may also lead to a decrease in UV radiation, which in turn reduces the production of •SO 4 − and •OH for pharmaceuticals removal (Moreira et al., 2017). As a result of the above investigations, an aeration velocity of 0.005 mL min -1 mL -1 was deemed most appropriate and was therefore chosen for subsequent experiments.

PS dose in the cathode
Previous sections have discussed the impact of operating parameters related to the in situ production of H 2 O 2 in the PPBEC process. The dosage of PS will clearly affect the generation of •SO 4 − , and thus, affect the overall treatment process. As such, the PPBEC system was further investigated with PS concentrations between 0 and 1 mM, as these concentration levels are commonly used in the individual PS process (Lu et al., 2018;Zhang et al., 2015). The removal efficiency of CBZ and CA during the PPBEC process increased as expected with the increase of PS dose from 0 to 0.75 mM (Fig. 4), whereafter only a slight improvement was achieved at 1 mM.
Similar to the above, statistical analysis showed that it had a significant effect on CBZ and CA removal rates within the PS range studied (p<0.05,

J o u r n a l P r e -p r o o f
Along with the increase of PS dose, the conductivity of the catholyte and the system current both increased (Fig. S5) (20) Taking into account that further increasing the PS dosage will not only enhance the abovementioned suppression effect but also further increase the cost of chemicals, 1 mM was chosen as the optimal dose for the system.

Initial CBZ and CA concentration
Since the initial concentration of pollutants is regarded as one of the main influencing factors that determine the treatment capacity of AOPs, three concentration gradients of CBZ and CA (5, 10 and 15 mg L -1 ) were selected to further evaluate the PPBEC process (Ali et al., 2018). Our results show that both CBZ and CA were completely removed within 6 h by the PPBEC process, within the selected concentration range (Fig. 5)

Impact of the water matrix
Due to the complexity and variability of real water and wastewater matrices, the composition of the matrix can have a significant impact on the effectiveness of AOPs like BEF and UV/PS processes (Nadais et al., 2018;Xiao et al., 2016). Therefore, in addition to the WWTP secondary effluent (Lyngby), three additional types of water matrices including tap water, WWTP influent (Lyngby) and secondary effluent from another WWTP (Herning) were chosen and their impact on the PPBEC process was investigated. As presented in Fig.6, the removal rates of CBZ and CA in different water matrixes were in the following order: tap water > WWTPs secondary effluent (Lyngby) > WWTPs secondary effluent (Herning) > WWTPs influent (Lyngby). The results of one-way ANOVA further confirmed the significant impact of water quality on the degradation of CBZ and CA in the PPBEC process (p<0.05,  Although tap water typically contains far less NOM and inorganic ions (e.g., Cl − ) than the other tested water matrices, the removal efficiency of CBZ and CA wasn't significantly higher than that of influent and secondary effluents of WWTPs. This could be attributed to the high alkalinity of tap water (Table S1), as carbonate and bicarbonate can consume •OH and •SO 4 − (Lu et al., 2018). As expected, the influent from Lyngby WWTPs with a higher concentration of NOM and inorganic ions (e.g., Cl − ) led to a stronger inhibitory effect and the lowest pollutants removal rate amongst the water matrices tested.

Economic analysis
In general, the E EO and corresponding Cost/O total value of PPBEC process were much lower than that of UV/PS and PBEC processes ( Table 1). The results indicate that the hybrid PPBEC process could reduce treatment costs by more than 93% when compared to the PBEC and UV/PS processes.
Notably, the energy consumption of the UV lamp and pump for agitation accounted for 73.76% and 24.59% of the overall energy consumption, respectively. This means that both the reactor configuration and UV lamp type can be adjusted and optimized to improve the energy efficiency of UV light and further reduce energy consumption (Miklos et al., 2018). Moreover, previous reports on the UV/H 2 O 2 process for wastewater treatment have verified that the scale of such treatment technology can significantly reduce energy consumption, when scaling up the process from labscale to pilot-scale and even further to the full-scale application (Miklos et al., 2018).
The treatment cost of the PPBEC process was further compared with some analogous technologies that have been developed for the treatment of pharmaceutical wastewaters. As shown in Table 2, ultrasound-activated PS or PMS methods consumed much more energy than the PPBEC process when treating secondary wastewater effluent (Yin et al., 2018;Zhou et al., 2016). Furthermore, individual UV, UV-activated PS, UV-activated PMS, or UV-activated H 2 O 2 processes have much higher energy consumption than the PPBEC process, even at lower pharmaceutical concentrations J o u r n a l P r e -p r o o f (Deng et al., 2013;Zheng et al., 2019). In addition, the PPBEC process had both significantly lower treatment cost and higher removal efficiency than the BEF process (in both lab-scale and upscaled forms), without the need to adjust pH before and after treatment nor to deal with iron sludge issues (Nadais et al., 2018). Although the treatment cost of the PPBEC process was slightly higher than that of UV based photocatalysis using a PVDF membrane with immobilized TiO 2 , it was still within the same order of magnitude (33-58 vs. 88.10 kWh m -3 ), with the benefit of a faster removal rate (Paredes et al., 2019). Biological processes are well-known to be amongst the more cost-effective treatment methods, but are also typically far less efficient for pharmaceutical and trace chemicals removal (Monteil et al., 2019). Overall, the PPBEC process shows huge potential for the removal of residual pharmaceuticals in wastewater, not only due to its high efficiency, but also due to its costeffective nature.

Transformation products and probable pathways
The transformation products from the degradation of CBZ and CA through the PPBEC process were investigated by using HPLC-MS/MS with the m/z ranged from 60 to 300 in ESI positive and negative full scan modes, respectively. The system operating parameters for these trials were adopted according to the results obtained in section 3.2.

CBZ
Total ion chromatography and MS2 spectra showed that ten major transformation products were produced during the PPBEC treatment process ( Fig.S8 and Fig.S9 (Bo et al., 2017;Jelic et al., 2013;Wang and Wang, 2017;2018b). Acridine-9-carboxaldehyde has been previously identified in the degradation of CBZ by AOPs (e.g. ozonation and photocatalytic processes), with a m/z 208 (Azaïs et al., 2017;Bo et al., 2017). Interestingly, this transformation product was not found by standard product verification within our study (Fig.S10). The transformation product of m/z 208 was found to be further converted into intermediates m/z 226 (possible structures TP-D1 and D2) and m/z 180 (possible structure TP-E) through •OH addition followed by ring contraction (Bo et al., 2017;Jelic et al., 2013;Wang and Wang, 2017;2018b). The products of TP-C1, TP-C2 and TP-E were further hydroxylated to intermediate m/z 196, with four possible structures highlighted by literature (Azaïs et al., 2017;Bo et al., 2017;Jelic et al., 2013). Finally, •SO 4 − and •OH radicals caused cleavage of the benzene ring to produce various small molecular transformation products (TP-G to TP-K) which were eventually converted into CO 2 and H 2 O (Bo et al., 2017;Jelic et al., 2013). Figure S11 also shows the trend of the abundance of various transformation products of CBZ versus time. Notably, the results indicate that all transformation products first increased and but then decreased in abundance as the reaction proceeds, and were finally completely removed.

CA
In addition to CBZ, the degradation products of CA were also investigated to infer its possible degradation pathways. Since the structure of CA was less complex than CBZ, it was easier to remove and mineralize accordingly. Therefore, unlike the CBZ degradation process which detected many products, only a few were detected during CA degradation ( Fig.S12 and Fig.S13).
Specifically, three major transformation products were detected, with elution times at 3.754-4.173 min (m/z 168.9, 212.9 and 222.9), 5.790-6.157 min (m/z 168.9 and 212.9), and 6.192-6.602 min (m/z 168.9 and 212.9). The first transformation product corresponded to m/z 222.9, and the full proposed transformation pathway is presented in Fig.7b. More specifically, CA can be transformed J o u r n a l P r e -p r o o f into m/z 168.9 (possible structures TP-A1 and A2) by decarboxylation, or CA was dechlorinated by substitution of •OH and then further attacked by •SO 4 − and •OH radicals to produce transformation product m/z 212.9 (possible structures TP-B) (Sun et al., 2015;Zhu et al., 2019a). Subsequently, the adjacent carbon on the benzene ring of TP-A1 was oxidized and the benzene ring was broken to produce product m/z 222.9 (possible structures TP-C1 and C2). A similar phenomenon has been observed in a previous study on the removal of CA by UV/O 3 and FeOOH based Fenton-like processes (Sable et al., 2015;Wang et al., 2019). Finally, TP-B,C1, and C2 were oxidized to CO 2 and H 2 O. The abundance trend observed for CA degradation products was consistent to that observed for CBZ degradation (Fig.S14). This indicates that the PPBEC system has a relatively higher mineralization ability for our target pollutants than the ones that have been shown for other AOPs, such as vacuum-UV oxidation and UV/Cl 2 Zhu et al., 2019b).

Ecotoxicity assessment
Evaluating the overall effectiveness and eco-friendliness of new proposed technology should be documented according to not only the removal efficiency of target pollutants, but also by a toxicity assessment of the degradation products. For example, some AOPs (e.g., gamma irradiation and ozonation) have been found to produce highly toxic intermediate products during the treatment of pharmaceutical-containing wastewater (Stalter et al., 2010) which can limit their applications.
Therefore, the ecotoxicity assessment for samples before and after treatment through the PPBEC process was carried out. This was done by comparing the widely-used luminescence inhibition effect of Vibrio fischeri, with assay times of 10 and 20 mins. As shown in Fig.8, the percentage of bacterial luminescence inhibition at assay time of 10 and 20 mins fluctuated around 10% throughout the whole treatment process, showing a slight increase first and then a continuous decrease and finally lower than the initial value before treatment. The ecotoxicity of compounds like CBZ was found in previous studies (Deng et al., 2017;Lai et al., 2021). Thus, this indicates that the ecotoxicity of water samples did not increase significantly during the treatment process and J o u r n a l P r e -p r o o f was eventually reduced. A similar phenomenon has been observed on CBZ degradation by the catalytic activation of PMS (Deng et al., 2017). It has been previously found that the ecotoxicity of CBZ and CA removal by direct UV photolysis, photocatalysis or ozonation process tended to be be monotonically increased because some of the resulting transformation products (e.g., acridine and hydroxyl-CA) were more toxic than the parent compounds (Donner et al., 2013;Zhu et al., 2019b).
It is important to note that since the ecotoxicity assessment in this work only involved Vibrio fischer, the specific ecotoxicity results should only be used as an indicator rather than simply extending the results to other species. Nevertheless, the results still indicate that the PPBEC process could be considered as an efficient and environment-friendly water treatment technology.

Implications
This work for the first time verified the feasibility and applicability of using this novel PPBEC system to effectively to remove residual refractory micropollutants from WWTPs secondary effluent. The PPBEC system has several advantages over traditional UV based AOPs (e.g., UV/PS and UV/H 2 O 2 /PS) and the recently proposed BEF and MEUC systems, in terms of treatment efficiency, resilience to high chromaticity and turbidity and feasibility of treatment at neutral pH.
Though promising, the following challenges should be addressed before field application. Firstly, membrane fouling could be occurred after long-term operation. Secondly, high capital cost (e.g., electrodes and membranes) for the reactor construction. Thirdly, the reactor configuration should be further improved to improve the efficiency of UV utilization together with further reducing energy consumption required by aeration. To meet the above challenges, future efforts should be devoted to the development of new membrane and electrode materials that are inexpensive and have strong anti-pollution capabilities (Asghar et al., 2015). Moreover, some of the EF reactor design (e.g., flow-through and photo EF reactor configurations) could be further developed and employed for PPBEC process .
J o u r n a l P r e -p r o o f

Conclusions
An innovative PPBEC system was developed and successfully operated for the efficient and costeffective removal of model refractory micropollutants CBZ and CA . This was achieved at a low PS dosage of 1 mM, input voltage of 0.6 V, cathodic aeration velocity of 0.005 mL min mL -1 , UV intensity of 10 mW cm -2 , and at neutral pH throughout the process. Our results show good synergy between the bioelectrochemical-based PBEC system and the UV/PS process, whereby the synergy factor of the PPBEC system was calculated to be as high as 1.71 for CBZ and 0.56 for CA.
Moreover, the PPBEC system was able to remove CBZ and CA in different water matrices.
Furthermore, the transformation products of CBZ and CA were identified and the probable degradation pathways were identified. The low number of degradation products and their abundance further indicated that the PPBC process had a high mineralization capability. Ecotoxicity tests further indicated that the system was environmentally friendly and therefore has broad prospects for recalcitrant wastewater treatment. The initial CBZ and CA concentrations of 10 mg L -1 were spiked into the selected water types, and other parameters were determined based on the earlier described tests, including PS dosage of 1 mM, applied voltage of 0.6 V, cathodic aeration velocity of 0.005 mL min mL -1 , UV intensity of 10