Comparison of UV/H2O2, UV/PMS, and UV/PDS in Destruction of Different Reactivity Compounds and Formation of Bromate and Chlorate

In this study, we compared the decontamination kinetics of various target compounds and the oxidation by-products (bromate and chlorate) of PMS, PDS, and H2O2 under UV irradiation (UV/PMS, UV/PDS, UV/H2O2). Probes of different reactivity with hydroxyl and sulfate radicals, such as benzoic acid (BA), nitrobenzene (NB), and trichloromethane (TCM), were selected to compare the decontamination efficiency of the three oxidation systems. Experiments were performed under acidic, neutral, and alkaline pH conditions to obtain a full-scale comparison of UV/peroxides. Furthermore, the decontamination efficiency was also compared in the presence of common radical scavengers in water bodies [bicarbonate, carbonate, and natural organic matter (NOM)]. Finally, the formation of oxidation by-products, bromate, and chlorate, was also monitored in comparison in pure water and tap water. Results showed that UV/H2O2 showed higher decontamination efficiency than UV/PDS and UV/PMS for BA degradation while UV/H2O2 and UV/PMS showed better decontamination performance than UV/PDS for NB degradation under acidic and neutral conditions. UV/PMS was the most efficient among the three processes for BA and NB degradation under alkaline conditions, while UV/PDS was the most efficient for TCM degradation under all pH conditions. In pure water, both bromate and chlorate were formed in UV/PDS, small amounts of bromate and rare chlorate were observed in UV/PMS, and no detectable bromate and chlorate were formed in UV/H2O2. In tap water, no bromate and chlorate were detectable for all three systems.


INTRODUCTION
Sulfate radical (SO ·− 4 )-based advanced oxidation process has attracted increasing attention as an alternative for traditional hydroxyl radical (HO · )-based advanced oxidation process, due to its high oxidation ability (redox potential of 2.5-3.1 V) (Neta et al., 1988) and adjustability to generating HO · via pH manipulation (Guan et al., 2011). SO ·− 4 was generated through activation of peroxymonosulfate (PMS) and peroxodisulfate (PDS) by UV irradiation, electrolysis, base, heat, quinones, ozone, homogeneous, and heterogeneous transition metals (Anipsitakis and Dionysiou, 2004;Furman et al., 2010;Guan et al., 2013;Cong et al., 2015;Zrinyi and Pham, 2017;Chi et al., 2019;Ding et al., 2020;Li et al., 2020;Liu et al., 2020). PMS, PDS, and hydrogen peroxide (H 2 O 2 ) all have similar -O-O-bond and were usually investigated in comparison. A comparison of UV/PDS and UV/H 2 O 2 was made on decontamination efficiencies. UV/PDS showed a better performance than UV/H 2 O 2 on the removal of carbamazepine (CBZ), 2,4-bromophenol, ofloxacin (OFX), ibuprofen, and cylindrospermopsin (CYN) (He et al., 2013;Yang et al., 2017;Sun et al., 2019;Luo et al., 2019;Xiao et al., 2020a). In studying the degradation of beta amide antibiotics, TOC removal by UV/PDS was slightly better than that by UV/H 2 O 2 (He et al., 2014). The better performance of UV/PDS on decontamination than UV/H 2 O 2 was mainly ascribed to two factors: (1) the higher quantum yield of PDS ( = 0.7 mol Einstein −1 ) than that of H 2 O 2 ( = 0.5 mol Einstein −1 ) and (2) lower steadystate concentration of HO · than SO ·− 4 (Yang et al., 2017). The comparison of three UV/peroxide processes (UV/PMS, UV/PDS, and UV/H 2 O 2 ) was carried out on destruction of atrazine (ATZ), and it was found that the degradation of UV/PDS on ATZ was more efficient than that of UV/PMS and UV/H 2 O 2 under the same conditions. It was attributed to the fact that the molar extinction coefficient and quantum efficiency of PDS at 254 nm are higher than those of UV/H 2 O 2 and UV/PMS (Luo et al., 2015). However, UV/H 2 O 2 exhibited better performance than UV/PDS on clonidine (CLD) removal that initial degradation rate of CLD was 0.68 and 0.46 µM min −1 for UV/H 2 O 2 and UV/PDS. The removal efficiencies were 86.5 and 78.7% in UV/H 2 O 2 and UV/PS by the end of experiments, respectively (Xiao et al., 2020b). When removing imidacloprid, UV/PDS showed higher removal efficiency than UV/PMS. This phenomenon was explained by calculating the rate of radical generation and the radical generation rate of UV/PS is higher than that of UV/PMS (Wang Q. F. et al., 2020), while for the removal of tetracycline, degradation efficiency in UV/PMS was higher than that in UV/PDS (Hu et al., 2019). For the three UV/peroxide processes, the superior process varied as target compound changed. The reactivity of target compound would make a sound besides molar extinction coefficient and quantum efficiency. The comparison of UV/peroxide processes need to be performed on target compounds with different reactivity.
When Br − -or Cl − -containing water was treated by advanced oxidation process, HO · and SO ·− 4 could react with them to form Br · or Cl · . Br · or Cl · could react with Br − or Cl − to form Br ·− 2 or Cl ·− 2 . HOBr/BrO − or HOCl/ClO − was formed by Br · /Br ·− 2 or Cl · /Cl ·− 2 recombination. HOBr/BrO − was reported to be a requisite intermediate in BrO − 3 formation via pure HO · mechanism (von Gunten and Oliveras, 1998 (Fang and Shang, 2012). Addition of organic matters could suppress BrO − 3 formation by scavenging Br · Wang Z. Y. et al., 2020). BrO − 3 was formed during oxidation of 2,4-bromophenol by UV/PDS while BrO − 3 was not formed in UV/H 2 O 2 . BrO − 3 was also formed in UV/PMS and a yield of BrO − 3 reached 100% at PMS concentration of 500 µM (Luo et al., 2020). A substantial conversion of Cl − into ClO − 3 was observed in UV/PDS at pH 3 and no ClO − 3 was observed at pH >5. It was proposed that Cl · formed from the reaction between SO ·− 4 and Cl − , initiated a cascade of subsequent pH-dependent reactions to form ClO − 3 (Lutze et al., 2015). In the process of oxidation, SO ·− 4 was the main reaction species, and all chloride chain reactions were initiated by SO ·− 4 (Qian et al., 2016). HOCl/ClO − was observed as an intermediate during the formation of ClO − 3 in UV/PDS (Hou et al., 2018), while there was no study referring to the formation of ClO − 3 in UV/PMS process. The comparison of BrO − 3 and ClO − 3 formation in the three UV/peroxide processes was also rarely reported.
The objective of this study was to compare (i) decontamination efficiencies of UV/peroxide processes (UV/PMS, UV/PDS, and UV/H 2 O 2 ) under various conditions and (ii) the formation of bromate and chlorate experimentally. Benzoic acid (BA), a recalcitrant organic compound, is mainly used in the food and pharmaceutical industries (Rayaroth et al., 2017). It has high reaction rate constants with both HO· and SO ·− 4 . Thus, BA was used as a probe to indicate the total oxidation capacity of available HO· and SO ·− 4 (Guan et al., 2011). Nitrobenzene (NB), a refractory pollutant, was used in chemical industry and released into the environment with the amount of about 19 million pounds each year through use, leakage, or industrial accidents (Wei et al., 2019). NB was selected as an indicator for HO · since it has high rate constant with HO· but quite low reaction rate constant with SO ·− 4 (Guan et al., 2011). Trichloromethane (TCM), a kind of disinfection by-product, has low rate constant with HO· and SO ·− 4 (Guan et al., 2018) and was used as representative of low reactivity organic pollutant. Firstly, total oxidation capacity of available HO · and SO ·− 4 in UV/peroxide processes and the inhibition of common radical scavengers [bicarbonate, carbonate, and natural organic matter (NOM)] on the UV/peroxide processes were investigated with BA as probe. Then, decontamination efficiencies were compared among UV/peroxides with BA, NB, and TCM as probes to present the performance of UV/peroxide processes on removal of different reactivity organic pollutants under various typical pH values. Finally, the formation of bromate and chlorate in pure water and tap water was monitored. The results would provide comprehensive comparison of UV/peroxide processes and guideline for selection of advanced oxidation process constructed based on UV disinfection unit.

Materials
Potassium peroxymonosulfate (PMS), potassium peroxodisulfate (PDS), BA, NB, sodium phosphate dibasic, sodium phosphate monobasic, sodium chloride, and potassium bromide were all ACS reagent grade and purchased from Sigma-Aldrich Company. Bromate standard for IC and chlorate standard for IC were also from Sigma-Aldrich Company. Hydrogen peroxide solution (35% w/w) was purchased from Alfa Aesar. HLPC grade phosphoric acid and methanol are available from DIMA-Tech and Thermo Fisher Science Inc. Gas chromatography (GC) grade chloroform (TCM) was purchased from Tianjin Chemical Reagent Co., Ltd. Suwannee River NOM (1R101N) was obtained from the International Humus Society, and the other reagents were of analytical reagent grade and purchased from China National Pharmaceutical Chemical Reagent Co., Ltd. All solutions were prepared in Milli-Q water (18.2 M cm) produced by Milli-Q Biocel water system.

Experimental Procedures
All the photochemical experiments were carried out in a cylindrical borosilicate glass container with a low-pressure mercury UV lamp (Heraeus, GPH 135t5l/4, 6 W output, 254 nm). The incident radiation intensities of the UV lamp (I 0 ) were 1.7 × 10 −6 Einstein s −1 (0.6 L solution) and 1.92 × 10 −6 Einstein s −1 (0.55 L solution). The optical path lengths (L) of the two reactor vessels were 2.70 cm (0.6 L solution) and 2.63 cm (0.55 L solution), respectively. Samples were extracted at predetermined intervals for each experiment and quenched with excess ascorbic acid or sodium nitrite. In the cases of tests in tap water, bromide was added to monitor the formation of bromate while no additional chloride was added for monitoring chlorate formation. Water quality parameters of tap water are shown in Table 5. All experiments were carried out at room temperature (20 ± 2 • C). The error bar represents the standard deviation of repeated experiments.

Analytical Methods
The concentrations of BA and NB were determined by highperformance liquid chromatography (HPLC) equipped with Waters 2,487 double λ detector and Waters symbol C18 column (4.6 mm × 150 mm, particle size 5 µm). TCM was quantified by gas chromatography (Agilent GC 6890). Details could be found in previous studies (Guan et al., 2011(Guan et al., , 2018. Concentrations of anions (chloride, bromide, chlorate, and bromate) were analyzed using a high-pressure ion chromatograph (Dionex Integrion) equipped with a Dionex AS19 column (4 × 250 mm). Isocratic eluent of 20 mM KOH, generated online by Dionex EGC 500 KOH, was used to separate the anions at the rate of 1.0 ml min −1 with a suppressor current of 50 mA. The injection volume was 200 µl and detection limits for bromate and chlorate were 0.01 and 0.01 µM. PMS concentration was standardized by iodometric titration (Ball et al., 1967). H 2 O 2 concentration was standardized based on its absorbance at 240 nm (ε = 40 M −1 ·cm −1 ) (Bader et al., 1988) and PDS concentration was quantified at 254 nm (ε = 20 M −1 ·cm −1 ) . Pseudo-first-order rate constant (k 0 ) of target compound degradation was obtained by fitting data of removal efficiency within 75%.
In some cases, such as UV/PMS and UV/H 2 O 2 processes at pH 11, the decontamination process could not be well-fitted by first-order reaction kinetics. Herein, we introduced the relative difference of removal efficiency (RDRE) of UV/peroxide processes to depict the inhibition effect of inorganic and organic carbon on UV/peroxide processes, which was calculated based on Equation (1).
where i is the index of sampling times ranging from 1 to n, n is the sample number. c i is the concentration of BA at sample time i in the absence of bicarbonate, carbonate, or NOM. c i ′ is the concentration of BA at the same sample time i in the presence of bicarbonate, carbonate, or NOM. c 0 is the initial concentration of BA in the absence of bicarbonate, carbonate, or NOM, and c ′ 0 is the initial concentration of BA in the presence of bicarbonate, carbonate, or NOM. Positive values of RDRE indicate the stimulation of additive, and negative values indicate the inhibition of additive. The larger the absolute value is, the stronger the stimulation or inhibition effect is.

Decontamination in the Presence of Common Radical Scavengers
Inorganic carbon (HCO − 3 and CO 2− 3 ) and NOM are widely present in surface water and groundwater, and regarded as free radical scavengers, leading to weakening the oxidation of target organic pollutants by advanced oxidation processes (Bennedsen et al., 2012). HO · and SO ·− 4 have different reactivity with inorganic and organic carbon, and the difference of reactivity would lead to a different effect on decontamination efficiency in HO· and SO ·− 4 -based oxidation process. BA was used as a probe    for both HO· and SO ·− 4 to compare the total oxidation capacity and further investigate the influence of inorganic and organic carbon on total oxidation capacity of the three processes.
As shown in Figure 1, BA degradation in UV/H 2 O 2 and UV/PDS was faster than that in UV/PMS under neutral conditions (pH 7). The presence of 2.23 mgTOC·L −1 NOM showed inhibition on BA degradation in all the systems. Pseudofirst-order rate constant (k 0 ) of BA degradation was obtained by fitting data of BA removal within 75% and shown in Table 1. k 0 of BA degradation was 0.00263, 0.00326, and 0.00454 s −1 in UV/PMS, UV/PDS, and UV/H 2 O 2 at pH 7. By adding NOM, values of k 0 were reduced to 0.00160, 0.00203, and 0.00232 s −1 in UV/PMS, UV/PDS, and UV/H 2 O 2 , respectively. BA degradation appeared to be still the fastest in UV/H 2 O 2 after addition of NOM. The presence of NOM led to the relative decrease of k 0 of 39.16, 37.73, and 48.90% in UV/PMS, UV/PDS, and UV/H 2 O 2 , respectively, as shown in Table 2. The values of RDRE between UV/peroxide processes and UV/peroxide processes in the presence of NOM were −0.3468, −0.3069, and −0.3632 in UV/PMS, UV/PDS, and UV/H 2 O 2 ( Table 3). The two indexes, reflecting NOM effect on decontamination, both indicated that NOM showed the largest inhibition on UV/H 2 O 2 process and the smallest on UV/PDS process. Although HO· has a higher reactivity with BA (5.9 × 10 9 M −1 ·s −1 ) than SO ·− 4 (1.2 × 10 9 M −1 ·s −1 ), it also has a higher rate constant with NOM [1.4 × 10 4 (mgTOC·L −1 ) −1· s −1 ] than SO ·− 4 [2.2 × 10 3 (mgTOC·L −1 )M −1· s −1 ] (Reactions 2 and 3) (Guan et al., 2018). The value of k radical,BA /k radical,NOM for HO· (ratio of rate constant of HO· and BA to that of HO· and NOM, 4.21 × 10 5 mgTOC·L −1· M −1 ) was smaller than that for SO ·− 4 (5.45 × 10 5 mgTOC·L −1· M −1 ), which resulted in a slight higher inhibition of NOM on HO · than SO ·− 4 . This led to the larger inhibition of NOM on the UV/H 2 O 2 process and less on the UV/PDS process.
(3)  Table 1). UV/H 2 O 2 was still the most efficient process in the presence of bicarbonate for BA degradation. Figure 3 shows the effect of carbonate on BA degradation in UV/peroxides.  Table 1). Carbonate showed significant inhibition on decontamination in all the UV/peroxide processes. In the presence of carbonate, UV/PMS was the most efficient process for BA degradation. This was due to the larger quantity of HO· and SO ·− 4 produced by PMS photolysis since PMS has a quite large molar absorbance coefficient (146.4 M −1 cm −1 ) at pH 11 (Guan et al., 2011). Relative decrease of k 0 and RDRE caused by the addition of bicarbonate and carbonate (as shown in Tables 2, 3) indicated that bicarbonate and carbonate showed the largest inhibition on the UV/PDS process. Furthermore, carbonate showed larger inhibition on UV/peroxide processes than bicarbonate, while the difference of bicarbonate inhibition extent among UV/peroxide processes was more obvious than carbonate.
The rate constants of HO · and SO ·− 4 with bicarbonate were 8.5 × 10 6 M −1 ·s −1 and 3.6 × 10 6 M −1 ·s −1 (Reactions 4 and  6) . The values of k radical,BA /k radical,bicarboante (ratio of rate constant of radical and BA to that of radical and bicarbonate) for HO · and SO ·− 4 were calculated to be 6.94 × 10 2 and 3.33 × 10 2 , respectively. As a result, bicarbonate showed a slightly higher inhibition on SO ·− 4 than HO · and correspondingly a larger inhibition on the UV/PDS process. The rate constants of HO · and SO ·− 4 with carbonate were 3.9 × 10 8 M −1 ·s −1 and 6.5 × 10 6 M −1 ·s −1 (Reactions 5 and 7) . The values of k radical,BA /k radical,carboante (ratio of rate constant of radical and BA to that of radical and carbonate) for HO · and SO ·− 4 were calculated to be 1.51 × 10 1 and 1.84 × 10 2 , respectively. As a result, carbonate would show a higher inhibition on HO · than SO ·− 4 . Under alkaline conditions (pH 11), majority of SO ·− 4 was converted into HO · . Therefore, carbonate exhibited significant inhibition on the three processes and the difference of inhibition extent for UV/peroxides was not as much as bicarbonate.
Destruction of Different Organic Target Compounds Figure 4 shows the comparison of BA degradation by UV/peroxides. k 0 of BA degradation was obtained by fitting data of BA removal within 75% and shown in  (He et al., 2013;Yang et al., 2017;Luo et al., 2019;Sun et al., 2019;Xiao et al., 2020a). Production rates of radicals from UV/peroxide processes mainly depend on molar extinction coefficients and photolysis quantum yields of peroxides. Molar extinction coefficients of H 2 O 2 and its dissociated form HO − 2 were 19.6 and 229 M −1 ·cm −1 (Baxendale and Wilson, 1957) while quantum yield of H 2 O 2 photolysis at 254 nm was = 0.5 (Crittenden et al., 1999). Molar extinction coefficients of HSO − 5 (monovalent form of PMS) and SO 2− 5 (divalent form of PMS) were 13.8 and 149.5 M −1 ·cm −1 while quantum yield of PMS photolysis at 254 nm was = 0.52 (Guan et al., 2011). Molar extinction coefficient and photolysis quantum yield of PDS at 254 nm were reported to be varied in different studies. The values were ε = 20.07 M −1 ·cm −1 and = 0.7 , ε = 22.07 M −1 ·cm −1 and = 0.5 (Qian et al., 2016), and ε = 21.2 M −1 ·cm −1 and = 0.567 (Heidt et al., 1948). By considering quantum yield of PDS photolysis  as = 0.5, the photo-production rates of total radicals at pH 7 from UV/H 2 O 2 and UV/PDS were not in big difference, both of which were larger than radical photo-production rate from UV/PMS. This was almost in accordance with the initial degradation of BA at pH 7. With time extension, BA degradation in UV/H 2 O 2 appeared strengthened as compared with that in UV/PDS. This might be due to the additional production of HO· via reaction between H 2 O 2 and quinones (Koppenol and Butler, 1985), the intermediate product of BA oxidation. As pH decreased from 7 to 3, degradation rate of BA was almost not affected in UV/H 2 O 2 and UV/PMS, while degradation rate of BA was slightly slowed down in UV/PDS. As pH increased from 7 to 11, BA degradation was significantly enhanced in UV/PMS. It was mainly due to the increased photo-production of HO· and SO ·− 4 , which originated from the increased molar absorption coefficient from 14.3 to 146.4 M −1 ·cm −1 (Guan et al., 2011). Figure 5 and Table 4 show the comparison of UV/peroxide processes on NB degradation. At the oxidant concentration of 100 µM, k 0 values of NB degradation in UV/PMS were 0.00120, 0.00106, and 0.00483 s −1 at pH 3, pH 7, and pH 11, respectively. The values of k 0 in UV/PDS were 0.00107, 0.00079, and 0.00204 s −1 while k 0 values in UV/H 2 O 2 were 0.00249, 0.00204, and 0.00168 s −1 at pH 3, pH 7, and pH 11, respectively.  The performance of UV/peroxides declined in the sequence of UV/H 2 O 2 > UV/PMS > UV/PDS for NB degradation under acidic and neutral conditions. Meanwhile, UV/PMS also showed excellent decontamination for NB degradation under alkaline conditions. The declined decontamination of UV/H 2 O 2 from pH 7 to 11 might be due to the increased capture of HO· by hydrogen peroxide and fast depletion of hydrogen peroxide (Crittenden et al., 1999), although the increase of pH would lead to the increased molar absorption coefficient of hydrogen peroxide (Baxendale and Wilson, 1957), intending to increase photo-production of HO·. The rate constants of PDS with SO ·− 4 and HO · were 6.3 × 10 5 M −1 ·s −1 and 1.4 × 10 7 M −1 ·s −1 (Guan et al., 2018). The rate constants of NB with SO ·− 4 and HO · were ≤10 6 M −1 s −1 and 3.9 × 10 9 M −1 ·s −1 (Guan et al., 2011). The ratios of k radical,NB (rate constant of radical and NB) to k radical,PDS (rate constant of radical and PDS) were ≤1.6 and 2.8 × 10 2 for SO ·− 4 and HO · . Correspondingly, the ratios of k radical,NB ·c NB to k radical,PDS ·c PDS were ≤0.1 and 18.4 for SO ·− 4 and HO · by considering the initial concentrations of NB and PDS. More SO ·− 4 was captured by the parent oxidant PDS than HO · . The enhanced decontamination of UV/PDS with pH increasing was mainly due to the conversion of SO ·− 4 into HO·, since HO· has a higher radical usage efficiency than SO ·− 4 . Figure 6 and Table 4 show the comparison of TCM degradation by UV/peroxides. UV/H 2 O 2 showed limited degradation of TCM that about 30% of TCM was obtained by 15 min under all the investigated pH values at the oxidant concentration of 500 µM. UV/PDS showed excellent performance on TCM removal that more than 95% of TCM was degraded by 15 min under all pH conditions. k 0 values of TCM degradation at pH 3, pH 7, and pH 11 were 0.00223, 0.00265, and 0.00117 s −1 in the UV/PMS process, 0.01524, 0.01018, and 0.00626 s −1 in the UV/PDS process, and 0.00046, 0.00039, and 0.00033 s −1 in the UV/H 2 O 2 process, respectively. UV/peroxide performance was in the increased sequence of UV/H 2 O 2 < UV/PMS < UV/PDS under all pH conditions. During TCM degradation, Cl − was formed as the final product. Cl · would be generated when Cl − coexisted with SO ·− 4 and/or HO · (Lutze et al., 2015;Guan et al., 2018). The limited removal of TCM by UV/H 2 O 2 might be due to the scavenging of radicals (SO ·− 4 , HO · , Cl · , and phosphate radical) by hydrogen peroxide, leading to the low efficiency of radicals for TCM degradation in UV/H 2 O 2 . In UV/PDS, TCM removal decreased as pH increased from 7 to 11. The rate constants of PDS with SO ·− 4 , HO · , and Cl · were 6.3 × 10 5 M −1 ·s −1 , 1.4 × 10 7 M −1 ·s −1 , and 8.8 × 10 6 M −1 ·s −1 (Guan et al., 2018). The rate constants of TCM with SO ·− 4 , HO · , and Cl · were 2 × 10 6 M −1 ·s −1 , 6.3 × 10 7 M −1 ·s −1 , and 6.6 × 10 7 M −1 ·s −1 (Guan et al., 2018). The ratios of k radical,TCM (rate constant of radical and TCM) to k radical,PDS (rate constant of radical and PDS) were 3.2, 4.5, and 7.5 for SO ·− 4 , HO · , and Cl · , respectively. The ratios reflected the radical efficiency toward TCM against PDS, similar to the radical participation ratio (RPR) reported in UV/PMS (Guan et al., 2018). In the presence of chloride, SO ·− 4 was fast converted into Cl · and Cl ·− 2 under acidic conditions and increasing pH would lead to conversion of Cl · to HO · (Lutze et al., 2015;Guan et al., 2018). This would lead to the declined TCM degradation as pH increased from 7 to 11. The variation of TCM destruction with pH was similar to the trend of 2-methylisoborneol and geosmin removal vs. pH, which was ascribed to the distribution of phosphate ions (Xie et al., 2015). As pH increased, H 2 PO − 4 dissociated to HPO 2− 4 and HPO 2− 4 has higher rate constants with HO · and SO ·− 4 than H 2 PO − 4 . The enhanced scavenging effect of phosphate buffer might also contribute to decrease of TCM removal as pH increased.
The higher removal efficiency for TCM by UV/PDS, as compared with the other two UV/peroxide processes, might be due to the lower scavenging of radicals (SO ·− 4 , HO · , and Cl · ) by the parent oxidant (PDS) in the reaction system. Based on the analysis of removal of BA, NB, and TCM by the UV/peroxide processes, it could be obtained that the decontamination rate would mainly depend on the molar absorption coefficient and radical quantum yield for the target compounds with high rate constants toward both HO · and SO ·− 4 , while the scavenging effect of the parent oxidant for radicals should also be considered besides molar absorption coefficient and radical quantum yield when choosing the superior process of decontamination rate for the target compounds with low rate constants toward SO ·− 4 or toward both HO · and SO ·− 4 . It would be suggested that UV/PDS might be a good choice for removing chlorosubstituted organic compounds with low rate constant with radicals, while UV/PMS was recommended to be used under alkaline conditions and UV/H 2 O 2 would be used under acidic and neutral conditions for destructing organic pollutants with high rate constant with radicals.

Bromate and Chlorate Formation
Bromate and chlorate were reported to be formed in UV/PDS in the presence of bromide and chloride (Fang and Shang, 2012;Lutze et al., 2015). However, whether chlorate was formed in the presence of chloride in UV/PMS system was rarely reported. The comparison of bromate and chlorate formation in UV/peroxide processes was also little reported. Hence, bromate and chlorate formation was comparatively investigated in the UV/peroxide systems in pure water and tap water. Figure 7 shows that BrO −