E ect of Natural Organic Matter on the Ozonation Mechanism of Trimethoprim in Water

: Trimethoprim (TMP) is often used for the treatment of various bacterial infections. It can be detected in water, and it is di ﬃ cult to be biodegraded. In this study, the degradation mechanism of TMP through ozonation and the e ﬀ ect of humic acids (HA) were investigated. Excessive ozone (pH 6, 0 ◦ C) could reduce the content of TMP to less than 1% in 30 s. However, when ozone (O 3 ) was not excessive (pH 6, 20 ◦ C), the removal e ﬃ ciency of TMP increased with the increase of O 3 concentration. Four possible degradation pathways of TMP in the process of ozonation were speculated: hydroxylation, demethylation, carbonylation, and cleavage. The presence of HA in water inhibit the generation of ozonation products of TMP. The excitation-emission matrices (EEM) analysis showed that with the extension of ozonation time, the ﬂuorescence value in the solution decreased and the ﬂuorescence peak blue shifted. These results indicated that the structure of HA changed in the reaction and was competitively degraded with TMP. According to the free radical quenching test, the products of pyrolysis, direct hydroxylation and demethylation were mainly produced by indirect oxidation.


Introduction
In recent years, the emerging persistent organic matter attracted great attention in the environmental protection industry [1][2][3]. Owing to their high consumption and refractory chemical properties [4], antibiotics are considered as a new persistent organic substances [5][6][7]. Antibiotics enter the aquatic environment through industrial and medical wastewater from the pharmaceutical industry [8,9], as well as human and animal feces [10]. Trimethoprim (TMP) is a commonly used antibacterial synergetic agent which can be combined with sulfonamides to greatly enhance their antibacterial activity [11,12]. It is widely used in the treatment of human diseases and animal bacterial infections [13,14]. Only 25% of TMP can be metabolized by human body after ingestion [15], and the remaining TMP and its metabolites are discharged out of the body through the digestive system and subsequently enter domestic wastewater [16][17][18]. TMP has been detected in wastewater from sewage treatment plants, surface water environment, drinking water, and other water bodies [19][20][21][22][23]. Even in large-scale wastewater treatment plants with advanced technologies such as chlorination and ultraviolet disinfection radiation, the concentration of TMP in the effluent ranges from the ng/L to the µg/L level [24,25]. TMP in water cannot be effectively removed through the existing traditional wastewater treatment processes [26][27][28]. Therefore, exploring an efficient and environmentally friendly technology to degrade TMP in water is of great significance. At present, some advanced technologies and materials have been studied to treat refractory wastewater. For instance, it was reported that Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). All of these chemicals were used without any pre-treatment.
Ultrapure water used for sample preparation was obtained from the Milli-Q water purification system (Millipore Synergy 185, US). The pharmaceutical sample solutions were stored in the dark at 4 • C. All the experiments were performed at pH 6.0 ± 0.1.

Ozonation Experiments Setup
The ozonation experimental system for the study was consisted of an oxygen cylinder, an O 3 gas generator (Model 3S-X, KIRIS, China), and a reaction reactor device. The O 3 gas was generated by passing pure O 2 feed gas (99.9%) through the O 3 generator equipped with a gaseous flow meter. Then continuously sparged O 3 gas into the bottom of the reactor using a fine bubble diffuser at a constant flow rate of 1 L/min. Excess O 3 was passed into two gas absorption bottles containing 2% KI solution.

Removal of TMP in O 3 Solution
Four experimental conditions were prepared, including TMP (10 mg/L) in ultrapure water, in methanol solution (0.25 mol/L), in HA solution (15 mg/L), and in mixed solution of HA (15 mg/L) and methanol (0.25 mol/L). The solutions were designed to study the effect of the additional NOM and free radical quencher (methanol for ·OH) on the ozonation of TMP.
First, O 3 gas was introduced into ultrapure water at 0 • C (in ice bath). The concentration of O 3 in the water increased gradually, and the corresponding concentration of O 3 solution was obtained by controlling the aeration time. Afterward, 50 mL of the water matrix (the concentration was twice that of the above four conditions) was added into a conical flask, and 50 mL of O 3 solution from the reactor was added to the conical flask. In this way, the concentration of the corresponding substance in the mixed solution meet the set experimental conditions. The solution was mixed before determining its concentration to ensure that the reaction was complete (in 1 min), and sodium thiosulfate solution was added to the sample to stop the reaction. At the same time, 5 mL of O 3 solution was obtained to determine the concentration of O 3. The initial concentration of dissolved O 3 was calculated by dividing the concentration in O 3 solution by 2. To ensure accuracy, all experiments were conducted in triplicate.
The effects of HA and OH on TMP degradation were studied by comparing the degradation of TMP in four experimental conditions.

Ozonation of TMP in O 3 Contactor
The experimental condition in this study was similar to that in Section 2.3.1 except that the reactor was stirred continuously using a magnetic stirrer at 20 • C, and O 3 gas was directly introduced into four water substrates to react with the pollutants. Samples were obtained at corresponding times, and sodium thiosulfate solution was added to the sample to stop the reaction. To ensure that the amount of the products was above the detection limit of the HPLC-MS/MS, the initial TMP concentration was set at 10 mg/L. The experiments were carried out in a continuously aerated reactor containing TMP solution with an initial pH of 6.0 ± 0.1, and the amounts of the residual parent compounds and the formation of intermediate products were detected by HPLC-MS/MS at regular time intervals. Samples for EEM detection were also taken before the beginning of the experiment and at the 5th, 10th, and 30th min of the experiment.

Analytical Methods
The concentration of O 3 was determined according to the "Standard Methods for the Examination of Water and Wastewater (21st Ed)-APHA (2005)-Method 4500-O 3 -Indigo Colorimetric Method". More detailed information and procedures about indigo method can be found in the literature [55].
Fluorescence spectrometer (FS5, Edinburgh Instruments, Livingston, UK) was used to measure the changes of the fluorescent substances in the reaction process. The parameters were set as follows: scanning speed of 12,000 nm/min, data interval of 5.0 nm, excitation light bandwidth of 3.0 nm, and emission light bandwidth of 3.0 nm. The excitation wavelength (E x ) was set to 230-450 nm and the emission wavelength (E m ) was 260-550 nm.
The toxicity of TMP and its degradation products were predicted using the Toxicity Estimation Software Tool (V4.2.1, USEPA, 2016).

Removal Efficiency of TMP in O 3 Saturated Water
The mass transfer rate of O 3 increases with the increase of O 3 concentration [56]. Therefore, the concentration of O 3 played an important effect on the degradation of TMP. Liu et al. [57] carried out ozonation experiments on eight kinds of antibiotics, including ampicillin, azithromycin, erythromycin, clarithromycin, ofloxacin, sulfamethoxazole, trimethoprim, and tetracycline, and the results showed that ozonation rate was highly depended on O 3 concentration and contact time, and the removal efficiency of antibiotics increased with the increase of O 3 concentration. As shown in Figure 1a, with the increase of O 3 concentration, the removal efficiency of TMP increased. O 3 dosage was positively correlated with TMP removal efficiency.
Fluorescence spectrometer (FS5, Edinburgh Instruments, UK) was used to measure the changes of the fluorescent substances in the reaction process. The parameters were set as follows: scanning speed of 12,000 nm/min, data interval of 5.0 nm, excitation light bandwidth of 3.0 nm, and emission light bandwidth of 3.0 nm. The excitation wavelength (Ex) was set to 230-450 nm and the emission wavelength (Em) was 260-550 nm.
The toxicity of TMP and its degradation products were predicted using the Toxicity Estimation Software Tool (V4.2.1, USEPA, 2016).

Removal Efficiency of TMP in O3 Saturated Water
The mass transfer rate of O3 increases with the increase of O3 concentration [56]. Therefore, the concentration of O3 played an important effect on the degradation of TMP. Liu et al. [57] carried out ozonation experiments on eight kinds of antibiotics, including ampicillin, azithromycin, erythromycin, clarithromycin, ofloxacin, sulfamethoxazole, trimethoprim, and tetracycline, and the results showed that ozonation rate was highly depended on O3 concentration and contact time, and the removal efficiency of antibiotics increased with the increase of O3 concentration. As shown in Figure 1a, with the increase of O3 concentration, the removal efficiency of TMP increased. O3 dosage was positively correlated with TMP removal efficiency. After adding HA, the removal efficiency of TMP was lower than that of the TMP solution without adding any substance. Moreover, HA may cause competitive degradation, thereby inhibiting the degradation of TMP [58]. When the concentration of O3 was not excessive, methanol can inhibit After adding HA, the removal efficiency of TMP was lower than that of the TMP solution without adding any substance. Moreover, HA may cause competitive degradation, thereby inhibiting the degradation of TMP [58]. When the concentration of O 3 was not excessive, methanol can inhibit the removal efficiency of TMP solution in the presence and absence of HA. It can be seen in Figure 1a, when the O 3 concentration was approximately 15 mg/L, the removal efficiency of TMP in the O 3 solution was 32%. After adding HA, the removal efficiency reduced to 25%. Under the condition of low concentration of O 3 , the inhibition of methanol on TMP degradation was evident. The removal efficiency of TMP was only 8% in the solution with HA and 15% without HA. Given that methanol has quenching effect on ·OH, when O 3 was not excessive, the indirect oxidation of O 3 occupied the dominant position in reaction.

Removal Efficiency of TMP in O 3 Contactor
As shown in Figure 1b, O 3 can rapidly degrade TMP into other small molecular compounds within 1 min. The addition of HA and methanol had little effect on the experimental results, indicating that TMP can be directly oxidized by O 3 molecules.

Degradation Pathway of TMP
The intermediate products of TMP and the abundance of each product during ozonation were determined by using HPLC-MS/MS. The intermediate products produced during TMP degradation and their potential pathways are shown in Figure 2. The main transformation pathway of TMP in the degradation process was hydroxylation, demethylation, carbonylation, and cleavage [59,60]. In these reactions, hydroxylation was the main way of TMP degradation, since ·OH can be connected to any ring or methylene bridge of a TMP molecule [34]. The direct hydroxylation products of TMP can be expressed by the formula C 14 H 19 N 4 O 3+x [34], including T12 (m/z = 307) to T16 (m/z = 371) [61]. Carbonylation was also one of the main reaction pathways, which mainly occurred on a methylene bridge and a pyrimidine ring. The reactive bridging methylene group of TMP was attacked to form a carbonylation product T1 (α-ketotrimethoprim m/z = 305). The carbonylation of T2 (m/z = 325) occurred to a pyrimidine ring, and then it was attacked by ·OH to produce hydroxylation products including T3 (m/z = 341) to T5 (m/z = 373). Products T6 (m/z = 127), T7 (m/z = 141), and T8 (m/z = 143) confirmed the methylene bridge cracking of TMP. The demethylation products of T9 (m/z = 277), T10 (m/z = 263), and T11 (m/z = 249) followed the formula of C 14-y H 19-2y N 4 O 3+x (y varying from 1 to 3) [62]. Demethylation was always accompanied by hydroxylation products. T17 (m/z = 293) to T30 (m/z = 313), which were produced under the joint action of demethylation and hydroxylation, were the hydroxylation products of the above three substances. The toxicity of TMP and its degradation products was evaluated using "Toxicity Estimation Software Tool" (V4.2.1, USEPA, 2016) (Table S2), indicating that no products with significant toxicity were formed during the reaction. TMP was degraded rapidly within 5 min after the start of the reaction, and intermediate products were formed at the same time. The variation of the abundance of the intermediate products during the experiment is shown in Figure 3 to 6. T1 and T2 were both carbonylation products of TMP, which accumulated rapidly at the beginning of the reaction, and then rapidly degraded. T1 remained stable in the later stage, whereas T2 might continue to react with the oxidant to produce T3, T4, and  TMP was degraded rapidly within 5 min after the start of the reaction, and intermediate products were formed at the same time. The variation of the abundance of the intermediate products during the experiment is shown in Figures 3-6. T1 and T2 were both carbonylation products of TMP, which accumulated rapidly at the beginning of the reaction, and then rapidly degraded. T1 remained stable in the later stage, whereas T2 might continue to react with the oxidant to produce T3, T4, and T5. The reaction trend of T2 was the same with that of T3, which was completely degraded to form T5 within 2 min. The amount of T5 in solution was low and degraded slowly until the region was stabilized. Owing to the quenching effect of methanol on ·OH, methanol can inhibit the production of most intermediates. As shown in Figure 3, after adding methanol, the removal efficiency of TMP decreased slightly, but the change was not significant, which indicated that direct oxidation was the dominant role in carbonylation. The maximum abundance of cleavage reaction products was only 0.01%, which degraded rapidly after accumulation to the peak. Figure 4a-c showed that the addition of methanol reduced the amount of T7 by half, and almost no other cleavage products were produced, which indicating that indirect oxidation was the dominant role in cleavage reaction. As shown in Figure 4d-f, the peak accumulation of demethylation product T9 to T11 occurred at approximately 2 min after the reaction, and then decreased slowly. As demethylation products, T9 and T11 were also affected by methanol. The addition of methanol not only affected the demethylation of TMP, but also inhibited the subsequent replacement of the second methoxyl group by the hydroxyl group. The quenching of ·OH by methanol restricted hydroxylation reaction and significantly reduced the hydroxylation products T12 to T16 in Figure 5. T12 was the direct hydroxylation product of TMP, which accumulated rapidly in the early stage of the reaction, and then degraded rapidly to form the subsequent polyhydroxylation products (T16 and other intermediate products), which remained stable after 2 min. The amount of T16 had little change during the reaction process. When methanol was added to the solution, the direct hydroxylation product of T12 was significantly reduced, which indicated that indirect oxidation was dominant in the reaction process. The common products of hydroxylation and demethylation are shown in Figure 6. The generation of T17 to T30 was close to that of hydroxylation product T15, but significantly higher than those of demethylation products including T9 to T11. The reason may be that TMP was attacked by ·OH at the same time of demethylation, or the hydroxylated product of TMP undergoes a demethylation reaction, which resulting in a relatively high abundance of T21. However, due to the complexity of the reaction, determining the dominant reaction was difficult.
Water 2020, 12, x FOR PEER REVIEW 7 of 17 T5. The reaction trend of T2 was the same with that of T3, which was completely degraded to form T5 within 2 min. The amount of T5 in solution was low and degraded slowly until the region was stabilized. Owing to the quenching effect of methanol on ·OH, methanol can inhibit the production of most intermediates. As shown in Figure 3, after adding methanol, the removal efficiency of TMP decreased slightly, but the change was not significant, which indicated that direct oxidation was the dominant role in carbonylation. The maximum abundance of cleavage reaction products was only 0.01%, which degraded rapidly after accumulation to the peak. Figure 4a-c showed that the addition of methanol reduced the amount of T7 by half, and almost no other cleavage products were produced, which indicating that indirect oxidation was the dominant role in cleavage reaction. As shown in Figure 4d-f, the peak accumulation of demethylation product T9 to T11 occurred at approximately 2 min after the reaction, and then decreased slowly. As demethylation products, T9 and T11 were also affected by methanol. The addition of methanol not only affected the demethylation of TMP, but also inhibited the subsequent replacement of the second methoxyl group by the hydroxyl group. The quenching of ·OH by methanol restricted hydroxylation reaction and significantly reduced the hydroxylation products T12 to T16 in Figure 5. T12 was the direct hydroxylation product of TMP, which accumulated rapidly in the early stage of the reaction, and then degraded rapidly to form the subsequent polyhydroxylation products (T16 and other intermediate products), which remained stable after 2 min. The amount of T16 had little change during the reaction process. When methanol was added to the solution, the direct hydroxylation product of T12 was significantly reduced, which indicated that indirect oxidation was dominant in the reaction process. The common products of hydroxylation and demethylation are shown in Figure 6. The generation of T17 to T30 was close to that of hydroxylation product T15, but significantly higher than those of demethylation products including T9 to T11. The reason may be that TMP was attacked by ·OH at the same time of demethylation, or the hydroxylated product of TMP undergoes a demethylation reaction, which resulting in a relatively high abundance of T21. However, due to the complexity of the reaction, determining the dominant reaction was difficult.      The addition of HA had a great influence on the degradation of TMP and the formation of intermediate products, which completely inhibited the formation of the carbonylation products T1 and T2. As for the cleavage products, HA inhibited the production of the products in both of the solutions in the presence and absence of methanol. The inhibition effect of HA on indirect oxidation in the reaction was clear, but when HA and methanol existed together, the product abundance in the solution was almost zero except for T7. HA had little effect on demethylation products T9 and T10, but it inhibited the subsequent demethylation product T11. When methanol was added to the solution, only T9 can be produced. This result meant that indirect oxidation reaction played an important role in the demethylation reaction. HA significantly inhibited the hydroxylation products except T12 to T16.    The addition of HA had a great influence on the degradation of TMP and the formation of intermediate products, which completely inhibited the formation of the carbonylation products T1 and T2. As for the cleavage products, HA inhibited the production of the products in both of the solutions in the presence and absence of methanol. The inhibition effect of HA on indirect oxidation in the reaction was clear, but when HA and methanol existed together, the product abundance in the solution was almost zero except for T7. HA had little effect on demethylation products T9 and T10, but it inhibited the subsequent demethylation product T11. When methanol was added to the solution, only T9 can be produced. This result meant that indirect oxidation reaction played an Normalized abundance(%)

EEM Analysis
O 3 could oxidize not only the TMP, but also the HA in water. HA have good fluorescence characteristics, which can be quantitatively and qualitatively analyzed by using fluorescence spectroscopy. The changes of fluorescence substances during the experiment are shown in Figure 7. Fluorescence peak (b) was at E x /E m = 270 nm/490 nm, and the peak value was close to 9000. Peak (a) at E x /E m = 425 nm/475 nm was not evident. After the O 3 was introduced into the solution, Peak (a) and Peak (b) were shifted towards shorter wavelengths (blue-shifted) from 480 nm to 450 nm along the excitation axis. The fluorescence value of peak (a) was higher than that of peak (b). Moreover, the fluorescence value gradually decreased with the progress of the reaction. The emission wavelength of peak (a) was short, which indicated that the substance represented by peak (a) has simple structure and low aromatization level. The decrease of peak (a) value after the introduction of O 3 indicated that the substances with simple structure were preferentially degraded. The fluorescence intensity of peaks (a) and (b) decreased rapidly to 74.84% and 43.28% after 5 min of ozonation, and gradually decreased to 28.18% and 13.66% after 30 min of ozonation. The changes of fluorescence substances of solution with TMP and HA are also shown in Figure 7. Compared with those in the absence of TMP, the peak patterns of the fluorescence substances changed significantly after TMP was introduced. The peak value increased obvious, and peak (a) was shifted towards shorter wavelengths (blue-shifted), which indicated that more aromatics were formed.

Conclusion
Excessive O3 can degrade TMP rapidly, and the addition of HA and methanol had little effect on the reaction, indicating that TMP could be directly oxidized by O3 molecules. When O3 was not excessive, the removal efficiency of TMP increased with the increase of O3 dosage. The addition of The structure of HA was changed by the reaction of O 3 in the solution. The decrease of fluorescence value in solution indicates that O 3 can degrade HA and its products. The change of peak position in Figure 7 shows that HA not only formed competitive degradation with TMP, but also affected the ozonation of TMP.

Conclusions
Excessive O 3 can degrade TMP rapidly, and the addition of HA and methanol had little effect on the reaction, indicating that TMP could be directly oxidized by O 3 molecules. When O 3 was not excessive, the removal efficiency of TMP increased with the increase of O 3 dosage. The addition of HA and methanol inhibited the degradation of TMP. Therefore, indirect oxidation also played an important role in the removal of TMP.
The reaction pathways proposed were hydroxylation, demethylation, cleavage, and carbonylation. In particular, hydroxylation was the dominant reaction. The existence of HA in water may affect the generation of cleavage and carbonylation products of TMP significantly, and the formation of hydroxylated products and demethylation products also decreased slightly. According to the free radical quenching test, the cleavage and demethylation products were mainly produced by indirect oxidation. With the extension of ozonation time, the fluorescence value in the solution decreased and the fluorescence peak blue shifted, indicating that the structure of HA changed in the reaction and was competitively degraded with TMP.
Ozone oxidation technology not only has a fast degradation rate, but also can remove odor without producing secondary pollution. Although ozone oxidation only leads to partial oxidation of drugs, due to the high removal rate and the small quantities of products, partial oxidation is enough to reduce pharmacological activity and toxicity. Ozonation has a high potential for the oxidation of pharmaceuticals in drinking water and wastewater. This study could provide theoretical basis for the control of TMP and its ozonation products during water treatment.
Author Contributions: N.Z. and R.Y. make equal contributions to conception and design, and/or acquisition of data, and/or analysis, interpretation of data and writing the manuscript; B.Z., F.W., and H.C. participate in revising the manuscript critically for important intellectual content; and give final approval of the version to be submitted and any revised version. All authors have read and agreed to the published version of the manuscript.