Activation of persulfate by EDTA-2K-derived nitrogen-doped porous carbons for organic contaminant removal: Radical and non-radical pathways

Journal Pre-proofs Activation of persulfate by EDTA-2K-derived nitrogen-doped porous carbons for organic contaminant removal: radical and non-radical pathways Yawei Shi, Jiandong Zhu, Gang Yuan, Guozhu Liu, Qingfa Wang, Wenjuan Sun, Bin Zhao, Liang Wang, Hongwei Zhang PII: DOI: Reference:

S1385-8947(19)33424-2 https://doi.org/10.1016/j.cej.2019.124009 CEJ 124009

To appear in:

Chemical Engineering Journal

Received Date: Revised Date: Accepted Date:

18 October 2019 27 December 2019 31 December 2019

Please cite this article as: Y. Shi, J. Zhu, G. Yuan, G. Liu, Q. Wang, W. Sun, B. Zhao, L. Wang, H. Zhang, Activation of persulfate by EDTA-2K-derived nitrogen-doped porous carbons for organic contaminant removal: radical and non-radical pathways, Chemical Engineering Journal (2019), doi: https://doi.org/10.1016/j.cej.2019.124009

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier B.V.

Activation of persulfate by EDTA-2K-derived nitrogen-doped porous carbons for organic contaminant removal: radical and non-radical pathways Yawei Shia, Jiandong Zhua, Gang Yuanb, Guozhu Liub, Qingfa Wangb, Wenjuan Sunc, Bin Zhaoa, Liang Wang*a and Hongwei Zhangad

a State

Key Laboratory of Separation Membranes and Membrane Processes, School of

Environmental Science and Engineering, Tiangong University, Tianjin 300387, China b

School of Chemical Engineering and Technology, Tianjin University, Tianjin

300072, China c

School of Chemistry and Materials Science, Ludong University, Yantai 264025,

Shandong Province, China d

School of Environmental Science and Engineering, Tianjin University, Tianjin

300072, China

Abstract: Nitrogen-doped porous carbons prepared from ethylenediaminetetraacetic acid dipotassium salt (EDTA-2K) were employed in combination with sodium persulfate (PS) for the catalytic removal of sulfamethoxazole (SMX). Almost complete removal (99.5%) of SMX was obtained with the optimum carbon prepared at 800 °C (NC800), which was attributed to its high surface area, good electrical conductivity, rich defects as well as doped carbonyl and nitrogen groups. The reaction mechanism was proposed to be a combination of both radical and non-radical pathways. •OH, SO4•-, O2•- and 1O2 as reactive oxidative species were involved in the 1

reaction process based on quenching tests and electron spin resonance (ESR) measurements, while the non-radical pathway between PS and SMX with the assistance of NC800 also made a significant contribution as indicated from PS decomposition tests and linear sweep voltammetry (LSV) analysis. Adsorption played an important role in the reaction process, and this point was emphasized when interpreting the effects of quenchers and pH values. For potential practical applications of the NC800/PS system, experiments concerning catalyst stability, the catalytic removal of three other organic contaminants and SMX removal in real water matrices were also conducted. Keywords: Persulfate, porous carbon, sulfamethoxazole, radical pathway, non-radical pathway

*Corresponding author. Tel. /fax: +86 22 83955392. E-mail address: [email protected] (L. Wang).

2

1. Introduction Organic contaminants in water, such as antibiotics, dyes and phenolic compounds, are potential threats to environment and human health. Among various remediation methods, the advanced oxidation process (AOP) is considered one of the most attractive approaches due to its wide applicability and limited formation of additional toxic species in the effluent [1, 2]. In AOPs, hydrogen peroxide (H2O2), peroxymonosulfate (PMS) and persulfate (PS) as green oxidants are frequently employed. PS is traditionally employed as an inducer for polymerizations, and reactions involving PS are generally slow at ordinary temperatures [3]. Cobaltous-based decomposition of PMS was first reported in 1956 [4]. These sulfur-containing oxidants began to attract attention when they were found highly efficient for degradation of organic contaminants [5, 6]. Due to their solid nature, PMS and PS are easier to transport and store than liquid H2O2 for practical applications. Besides, sulfate radical (SO4•-) is usually involved in PMS/PS-based AOPs, which possesses a comparable redox potential to hydroxyl radial (•OH) but a much longer lifetime [7, 8]. Consequently, an increasing attention has been paid to PMS/PS-based AOP systems for the efficient removal of various organic contaminants in water environment [9-14]. One main concern is the release of sulfate anions when these sulfur-containing oxidants are used [15]. This can be considered acceptable by optimizing the oxidant dosage as well as further application of treated water [15]. Compared to PMS which is in a triple salt form, PS available in the form of 3

sodium or potassium persulfate (Na2S2O8 or K2S2O8) is cheaper and more stable [16, 17]. However, being more stable means it is more difficult to activated PS than PMS. Generally, energy inputs (heat [18, 19], ultraviolet and visible light[19, 20], ultrasound[21], etc.) or chemical agents (metal ions [19, 22], phenols [23], etc. ) are introduced and found to be efficient for PS activation. To reduce the additional costs in these methods, heterogeneous metal-based catalytic systems are also widely employed [24-27]. Moreover, carbon materials as metal-free catalysts have been extensively employed for persulfate activation recently, which avoided the leaching problem associated with metal based catalysts. A variety of carbon materials have been investigated, including carbon nanotube [28, 29], nanodiamond[30, 31], graphene [32, 33], biochar [34-36] and ordered mesoporous carbon[37, 38]. In general, a large surface area and suitable heteroatom doping of a carbon catalyst have been considered favorable for contaminant removal in a carbon/PS based system [34, 35]. To build large surface areas in carbon materials, the activation approach is highly efficient, but the use of corrosive activation agents (KOH, NaOH, H3PO4, etc.) is an inevitable disadvantage. Another approach for pore formation is the templating method, but the templates (silicon molecular sieves, surfactants, etc.) are always sacrificed after use, leading to a higher manufacturing cost. On the other hand, the post-treatment method is frequently used to introduce heteroatoms into carbon materials, but the process also suffers from drawbacks such as time-consuming and less uniform heteroatom doping. Recently, the in-situ carbon fabrication method has received increasing attention. In 4

this method, compounds containing both carbon and certain elements (Na, K, Zn, etc.,) are used as precursors and carbonized directly [39-43]. During the carbonization process, inorganic species capable of pore formation (Na2CO3 [39], K2CO3[39-41], ZnO [42, 43], etc.) are formed in-situ, acting as activation agents or templates and thus helping to build large surface areas in the resulting carbons. Besides, through the selection of suitable precursors, heteroatoms groups can be introduced in situ into the carbon products [44-46]. In other words, pore formation and heteroatom doping can be realized simultaneously in the in-situ approach, which is facile, efficient and cost-effective for the preparation of carbons with large surface areas and abundant heteroatom groups. In this work, a series of nitrogen-doped porous carbons were prepared by the in-situ method using ethylenediaminetetraacetic acid dipotassium (EDTA-2K) as the precursor. EDTA-2K was selected because it contained both potassium and nitrogen, which could help to introduce pores and nitrogen doping respectively into the carbon product. Besides, it was commercially available and much cheaper than other typical precursors such as metal organic frameworks. The carbons prepared from EDTA-2K were then employed in combination with PS for the catalytic removal of sulfamethoxazole (SMX), which was a typical sulfonamide antibiotic widely detected in the water environment [47, 48]. To the best of our knowledge, carbons prepared by carbonization of EDTA-2K or other EDTA-salts have not been used in previous works to activate persulfate. Based on radical quenching tests, ESR measurements, PS decomposition tests and LSV analysis, the reaction mechanism was proposed to be a 5

combination of both radical and non-radical pathways. The effects of surface area, defects and carbonyl and nitrogen groups of the carbons on their performances for SMX removal were analyzed. For potential practical applications of the carbon/PS system, experiments concerning catalyst stability, the catalytic removal of three other organic contaminants and SMX removal in real water matrices were also conducted. 2. Experimental 2.1 Material and Characterization EDTA-2K was used as the precursor for preparation of nitrogen-doped porous carbons (NCs). The carbon prepared at the temperature of x °C was named as NCx. The carbons were characterized by X-ray diffraction (XRD), Raman, nitrogen sorption, CHNS elemental analysis (EA), X-ray photoelectron spectroscopy (XPS) and zeta potential analysis. Details for the preparation procedure, the regents and characterization methods employed were provided in section S1 in Supplementary Materials. 2.2 Catalytic oxidation experiments The oxidation experiments for SMX removal were conducted by mixing carbon catalyst, PS and SMX solution in glass flasks under magnetic stirring. The degradation of methyl orange, Rhodamine B or phenol was also conducted. The concentrations of the contaminants were analyzed by high performance liquid chromatography (HPLC) or UV-visible spectrometer. The apparent rate constant (kobs), adsorption amount (q) and equilibrium adsorption amount (qe) were calculated from Eq. (S1-3). The removal experiments were conducted at least in duplicates, and the 6

average results were reported with error bars. Detailed experimental procedures were provided in section S2 in Supplementary Materials. To investigate the solid concentration of SMX, spent NC800 was separated by filtration and then extracted with 30 mL methanol by sonication. The concentration of SMX in methanol was then measured by HPLC. For the stability tests of NC800, the spent catalyst was collected by filtration, washed with ethanol for three times and with water for one time and then dried at 120 °C for reuse. The catalyst after three runs was regenerated by calcination under nitrogen atmosphere at 400 °C for 2h. To identify the oxidation intermediates of SMX, a higher initial concentration (100 mg/L) was utilized, and both the aqueous sample at 120 min and the methanol sample obtained by extraction of the spent catalyst were analyzed by an HPLC system (Waters 2695, USA) coupled to a mass spectrometer (Waters micromass ZQ, USA) with an ESI source in the positive ion mode. To investigate the effect of water matrix, tap water or lake water was spiked with SMX. Tap water was obtained in Tiangong University and used directly. Lake water was obtained from the Pan Lake in Tiangong University and the supernate of the lake water after standing for 24h was utilized. To investigate the removal of natural organic matter (NOM) from lake water through adsorption or catalytic oxidation, the same experimental procedure was used except that no SMX was added, and the sample obtained at 120 min was analyzed by a total organic carbon (TOC) analyzer (Sievers M9, GE Analytical Instruments, USA) and a fluorescence spectrophotometer (F7000, Hitachi, Japan). ESR spectra were recorded to recognize the reactive oxidative species in the oxidation process. Decomposition 7

tests of PS and electrochemical measurements including LSV and electrochemical impedance spectroscopy (EIS) were performed for reaction mechanism analysis. The concentrations of PS were measured by a spectrophotometric method established previously [49, 50]. Details for fluorescence, ESR, LSV and EIS measurements were provided in section S3 in Supplementary Materials. 3. Results and discussion 3.1 Characterizations of the activated carbons The properties of resulting carbons using a series of EDTA alkali metal salts as precursors were compared and EDTA-2K was selected as the optimum one in the authors’ previous unpublished work, where characterizations of the EDTA-2K derived carbons (NC600-NC800) including XRD, XPS, CHNS elemental analysis and nitrogen sorption tests were reported (related data provided in Supplementary Materials for Review Only). In brief, the carbons were amorphous in nature, possessing larger surface areas and abundant nitrogen/oxygen functional groups. At a higher carbonization temperature of 900°C, little carbon was derived due to the more severe activation process. A commercial activated carbon (AC) was used as a reference in this work. The XRD pattern of AC also revealed its amorphous nature (Figure S1a), and the tiny peak around 26° may be attributed to a small graphitized local fraction in the carbon [51]. Nitrogen sorption (Figure S1b), XPS (Figure S1c, d) and CHNS elemental analysis of AC were also performed. For clarify, the specific surface areas, total pore volumes and bulk elemental compositions of the carbons were summarized in Table S1. The surface elemental compositions and amounts of 8

different types of oxygen and nitrogen functional groups were summarized in Table S2. The amorphous feature of the carbons were further confirmed from the Raman spectra (Figure S2), which possessed two broad peaks located around 1350 cm-1 (D band) and 1600 cm-1 (G band) respectively reflecting the disorder degree and graphitization level of the carbons. The relative intensity (ID/IG) increased with elevated carbonization temperature (Table S1), indicating the stronger in-situ activation process and thus the formation of more defects at a higher temperature. [34]. 3.2 Catalytic oxidation of SMX Since the EDTA-2K derived carbons were found to possess high surface areas and rich functionalities, their catalytic performances for SMX oxidation were further evaluated (Figure 1 and Figure S3). In this work, batch mode was used for the experiments because the powdered catalysts (<200 mesh) were not suitable for direct use in fixed-bed apparatus. Almost complete removal (99.5%) of SMX was achieved in an NC800/PS system in 120 min at a catalyst dosage of 0.050 g/L, while lower removals of 59.7% and 33.3% were observed for NC700 and NC600 (Figure 1a). Therefore, NC800 was selected as the representative catalyst for subsequent studies. For comparison, the removal of SMX using AC as the catalyst only reached 45.5%, and sole PS in the absence of any catalyst could hardly oxidize SMX (Figure 1a). A porous carbon prepared at 700 °C using EDTA-2Na as the precursor was also tested, and the obtained SMX removal was only 14% (Figure S4). This was probably 9

attributed to its limited surface area (680 m2/g), which was lower than all the other carbons investigated in this work. The effect of PS dosage on SMX removal was shown in Figure 1b. Adsorptive removal of SMX in the absence of PS reached 64.7%, and the addition of 1mM PS led to almost complete SMX removal. At a higher PS dosage of 2 mM, a similar SMX removal to the case of 1mM PS dosage was achieved but a slight decrease in kobs was observed, which was probably attributed to the self-quenching among excessive reactive oxidative species (ROSs) [10, 38]. When it came to the effect of catalyst dosage, a low value of 0.025 g/L was not enough to removal SMX efficiently (Figure 1c). Increasing the dosage to 0.075 g/L resulted in a 1.7 fold higher kobs compared to 0.050 g/L due to the availability of more adsorption/catalytic sites [38]. A higher catalyst dosage brought a larger surface area for the adsorption and catalytic reaction of SMX, leading to more efficient SMX removal. Considering the similar removal efficiencies, 0.050 g/L was chosen to decrease the cost of catalyst. As illustrated in Figure 1d, kobs decreased from 0.1069 to 0.0072 min-1 with an increasing SMX concentration from 30 to 100 mg/L. This was mainly attributed to the reduced availability of oxidative species and surface reactive sites relative to the SMX molecules [52]. Besides, a higher concentration could also lead to the formation of more intermediates [52] and thus compete with SMX. Despite of this, a high removal efficiency of 75.2% was achieved even at a high concentration of 100 mg/L SMX, and the removal efficiency could be further improved to 99.6% with a higher catalyst dosage. 10

Figure 1 placed here As discussed above, due to the large surface area of NC800, adsorptive removal of SMX with it reached 64.7%, even higher than the catalytic systems with other three catalysts. Although the addition of PS greatly enhanced the removal efficiency to almost complete removal, it seemed that catalytic oxidation only made a small contribution of ~35%. However, this was not the truth. To address this issue, the spent NC800 after adsorption or catalytic oxidation was extracted with 30 mL of methanol. The optical photos of the two extracted samples were shown in Figure S5. Obviously, the sample obtained after catalytic oxidation appeared a yellow color, indicating the presence of some oxidation products on NC800 surface. Furthermore, the concentration of SMX in the methanol sample was measured by HPLC and found to be 12.46 mg/L, corresponding to 1.87 mg/L (=12.46 mg/L×30 mL/200 mL) in the treated SMX solution. Here most of adsorbed SMX had been desorbed (see Figure S6 and section S4 in Supplementary Materials). In other words, most of SMX (95.8%) was removed by catalytic oxidation and only a small part (3.7%) was removed by adsorption. For comparison, SMX concentration in the methanol sample obtained after adsorption was 206.95 mg/L, close to the mass balance value of 215.67 mg/L. The results clearly demonstrated the superiority of the NC800/PS catalytic system. As discussed latter, a high correlation coefficient was obtained when correlating the rate constants and the corresponding surface areas. That was to say, the final “apparent” contribution of adsorption was low, but adsorption played an important role in the catalytic process. 11

3.3 Reaction mechanism In general, PS can be activated to produce ROSs which are then involved in the oxidation of organic contaminants (Eq. (1)-(3)) [28, 38]. Based on previous works [53-55], PS accepted electrons from NC800, then the cleavage of its O-O bond led to the formation of SO4•- . S2O82-+2H2O→HO2-+2SO42-+3H+

(1)

S2O82-+ HO2-→SO42-+ SO4•-+ O2•-+H+

(2)

SO4•-+OH-→SO42-+•OH

(3)

To investigate the role of these species, a series of quenchers were added into the NC800/PS catalytic system. Methanol is known as an efficient scavenger for both •OH

and SO4•- (k•OH = 9.7 × 108 M-1 s-1, kSO4•- = 1.1 × 107 M-1 s-1) [38]. The addition

of 1M methanol (methanol/PS=1000) only reduced the reaction rate by 23% (Figure 2a and Figure S7a), indicating that the two free radicals were not dominantly responsible in the reaction system. Considering that the hydrophilic nature of methanol hindered its reaction with the surface adsorbed radicals [56], DMSO was further employed to quench surface-bounded •OH and SO4•- [34, 56, 57]. kobs decreased by 11% in the presence of DMSO, which implied that the surface-bounded radicals did not play the determining role neither. O2•- and singlet oxygen (1O2) as another two ROSs were reported to be involved in previous carbon/PS systems [28, 34].Thus, the effects of L-histidine and BQ as typical quenchers for 1O2 and O2•respectively[28, 31, 58] were further investigated. As illustrated in Figure 2a, the direct addition of L-histidine inhibited the reaction significantly, showing a SMX 12

removal of only 44.2%. It seemed that 1O2 was dominant in the reaction process. However, L-histidine as an amino acid has an amphoteric nature, and it has been used as a buffer previously [59]. As will be discussed later in Section 3.5, pH has an important effect on the adsorption/oxidation process. Actually, the suppression of reaction with L-histidine mainly originated from this pH divergence, and the suppression effect of L-histidine became limited after the impact of pH was mostly ruled out, indicating that 1O2 did not actually play the dominant role (see Section S5 in Supplementary Materials). Figure 2 placed here The removal efficiency of SMX also declined in the presence of BQ, but the drop in kobs could not be simply attributed to the quench of O2•-. The pH effect brought by L-histidine was not observed with BQ. The initial and final pH values were 4.66 and 3.16 with the addition of BQ, similar to the case in its absence (4.75 and 3.11). Among the four quenchers used, BQ possessed the lowest water solubility, thus it would compete strongly with SMX for hydrophobic adsorption onto the carbon surface. Remember that the adsorption of SMX has been found in close relationship with the reaction rate. As shown in Figure S7b, an apparent drop in adsorptive SMX removal was observed with the addition of BQ, which at least partially contributed to the declined kobs. In contrast, the effect of L-histidine on SMX adsorption was insignificant (Figure S7b). ESR measurements were also conducted to investigate the generation of ROSs. As shown in Figure S8a, the specific signals for DMPO-•OH and DMPO- SO4•- were 13

not found under the optimum experimental condition (NC800=0.050 g/L, PS= 1mM) probably due to their low concentrations, similar to the case in a previous report [60]. With a tenfold concentration increase of NC800 and PS (NC800=0.5 g/L, PS= 10 mM), the signals for DMPO-•OH and DMPO-SO4•- [34, 38] were observed although still relatively weak (Figure S8b). Similarly, the three-line signal of TEMP-1 O2 adduct was absent under the optimum experimental condition (Figure S8c) and was revealed at a tenfold concentration of NC800 and PS (Figure S8d). However, the signal of DMPO-O2•- was not found in neither case (Figure S8e-f), indicating its low concentration in the reaction system. The above quenching and ESR tests showed that ROSs only partly contributed to the oxidation of SMX in the reaction system. Thus, the non-radical pathway between PS and SMX was expected. PS alone could hardly oxidize SMX (Figure 1a), but decomposition of PS was observed when adding NC800 as the catalyst (Figure S9a), indicating that NC800 could activate PS. Interestingly, the decomposition rate of PS was obviously increased when SMX was added (Figure S9a). LSV analysis was further conducted to investigate the non-radical electron-transfer process. As illustrated in Figure 2b, an obvious current increase was recorded on the NC800 electrode with the addition of PS, suggesting the interaction between PS and NC800 for the possible formation of PS-carbon complexes[34]. Moreover, a broad oxidation hump indicated by the arrow in Figure 3b was observed only when both PS and SMX were added, which suggested a faster charge transfer and accelerated SMX oxidation ability in the in the ternary NC800/PS/SMX system. A similar situation was observed 14

in a previous report where a hump was found in the LSV pattern in the co-existence of PS and a carbon nanotube catalyst [49]. The persulfate decomposition and LSV results collectively suggested that the electron transfer in the ternary system could proceed through the formation of PS-carbon complexes and the subsequent oxidation of SMX by these reactive complexes [49]. In this non-radical pathway, the carbon network with rich free-flowing π electrons would act as a mediator, favoring the electron-transfer from SMX to the PS-carbon complexes [34, 38]. Based on the reaction mechanism proposed above, the contributions of different properties of the carbon catalysts were analyzed. Porous carbon materials were widely employed for the adsorptive removal of SMX, and it was found in our previous work [61] that the surface areas of porous carbons were closely positively related to their abilities for SMX adsorption. As illustrated in Figure S9b, a high correlation coefficient of 0.956 was obtained when correlating the rate constants and the corresponding surface areas. The involvement of non-radical pathway between PS and SMX helped to explain this result. A higher surface area would help to adsorb more SMX onto the carbon surface, and the non-radical oxidation mainly occurred there rather than in the bulk solution. Consequently, the apparent rate constants were closely positively related to the surface areas. A similar result between kobs and adsorption amount was observed previously when nitrogen-doped biochars were employed for PS activation [34]. After normalizing the rate constants to the surface areas, the obtained values of kobs/SBET still varied significantly from each other. This should be attributed to the different chemical properties among the carbons. 15

According to previous works [62, 63], defects and carbonyl groups (C=O) were found crucial for activation of persulfate. Besides, as discussed in section 3.1, four types of nitrogen-containing groups existed on the carbon surface, including pyridinic (N-6), pyrrolic (N-5), quaternary (N-Q), and pyridine-N-oxide (N-X). Among these species, N-6, N-5, and N-Q were thought favorable for improving kobs[25, 34, 64, 65]. As shown in Figure S9c, a positive trend between kobs/SBET and ID/IG was observed (R2=0.778). However, no clear trend was found between kobs/SBET and C=O amount (Figure S9d) or the total content of N-5, N-6, and N-Q (Figure S9e). The attempt to correlate kobs/SBET to the sum of N-5, N-6, N-Q and C=O also failed (data not shown). As verified by Cheng et al. [63], the defects were related to the generation of 1O2. Thus, the positive trend between kobs/SBET and ID/IG supported the involvement of 1O2 in the reaction mechanism. C=O was considered the main active site for the non-radical pathway between PS and the organic contaminant [63]. For the nitrogen groups, it was proposed previously [66] that nitrogen in the carbon framework could break the chemical inertness of the carbon lattice and produce charge transfer intermediates, promoting the non-radical pathway of direct charge transfer. However, although C=O and nitrogen groups were favorable for the direct charge transfer process, the conductivity of the carbon catalyst also played an important role [66]. Taking NC600 and NC800 for comparison, the former one possessed a larger total content of favorable functional groups (C=O, N-5, N-6 and N-Q) but a lower normalized rate constant (kobs/SBET). EIS analysis of the two carbons (Figure S9f) showed that a much smaller semicircle diameter in the Nyquist plot of NC800 was 16

revealed, which demonstrated that NC800 had a much better conductivity and a significantly enhanced ability for electron-transfer compared to NC600 [34, 38]. As a result, the correlation between kobs/SBET and amounts of the functional groups was insignificant. The oxidation of other three typical organic contaminants was also conducted, including Rhodamine B (a cationic dye), methyl orange (an anionic dye) and phenol. The removals for all three contaminants by the NC800/PS exceeded 97% (Figure S10a) and were higher than the corresponding removals without PS (Figure S10b), again demonstrating the superiority of the catalytic system. Moreover, the rate constants for these contaminants (Figure S10c) were plotted against the corresponding equilibrium adsorption capacitates calculated by the pseudo-second-order kinetic model (Figure S10d). Here the pseudo-second-order model was applied because the adsorption kinetic results were better fitted to it compared to the pseudo-first-order model (Figure S10e). As shown in Figure 2c, a positive correlation with R2=0.801 was obtained. Obviously, a higher adsorption capacity helped to bring more contaminants to the carbon surface, benefiting the non-radical oxidation process and thus leading to a faster removal rate. The total organic carbon (TOC) contents during the oxidation of SMX were measured furthermore. The TOC removal percentage fell behind the SMX removal percentage during the whole reaction process and reached 84.2% at 120 min (Figure 2d), indicating that some organic oxidation products were formed and remained in the solution. Here it should be noted that the removal by adsorption is not the real TOC 17

removal. The highly porous nature of the carbon favored the catalytic process, but it also led to the adsorption of some SMX and oxidation products. Thus, for the reported TOC removal, the removal by adsorption was involved in. Despite of this, these TOC measurements were still useful because they could tell about how many organic oxidation products still remained in water after the catalytic process, which was important from the viewpoint of water remediation. To recognize the oxidation products, LC-MS measurements for both the aqueous solution and the extract of spent NC800 after SMX oxidation were conducted. The [M+H]+ peak at m/z=254.3 (Figure S11a) was attributed to parent SMX. Besides, four organic oxidation products were detected, including P99, P210, P503 and P519. P99 (Figure S11b) was recognized as 3-amino-5-methyl-isoxazole (AMI), which was derived by the cleavage of S-N bond and was commonly observed previously in SMX oxidation [18, 67]. Although the coupling product from two AMI molecules was not detected, P210 (Figure S11c) was considered to be formed by further oxidation of the coupling product from AMI according to a previous report [68]. P503 (Figure S12a) was a dimer formed by the coupling of SMX [18], and the further oxidation of P503 led to the formation P519 (Figure S12b) [60, 68]. The plausible oxidation pathways of SMX based on LC-MS analysis have been illustrated in Figure 2e. The two organics framed by broken-line were not detected probably due to their low concentrations or transformation into other compounds. Based on the above discussion, the plausible oxidation mechanism of SMX with the NC800/PS system was illustrated in Figure 3. Both radical and non-radical 18

pathways were involved in oxidation of SMX. In the radical pathway, a serials of radicals including SO4•-, •OH and O2•- were produced and reactions took place between these species and SMX. In the non-radical pathway, SMX was adsorbed on the surface of NC800 and then electron transfer occurred from SMX to the PS-carbon complexes. The highly graphitic carbon framework as well as the carbonyl and nitrogen groups promoted this non-racial pathway. 1O2 (a ROS but not a radical) generated with the assistance of defects on the carbon and through recombination of O2•- and water also made some contribution. Figure 3 placed here 3.4 Stability tests The spent NC800 was washed with ethanol and water to investigate its reusability. The removal efficiency of SMX dropped to 67.6% in the second run (Figure 4a and Figure S13a), implying the partial deactivation of the catalyst. The surface area of NC800 after the catalytic run (NC800-used) dropped dramatically to 754 m2/g (Figure S13b and Table S3). No apparent change in the SEM morphology was observed (Figure S14) probably because the adsorbed oxidation products could not be revealed by SEM. XPS measurement (Figure 4b and Table S3) revealed the increased oxygen and nitrogen contents in NC800-used compared to NC800. Besides, sulfur was detected in NC800-used, which was not found in the fresh catalyst. The increased oxygen content may be partially attributed to oxidation of the carbon surface during the catalytic process considering that ammonium persulfate has been employed in previous reports [69, 70] for the oxidative modification of activated 19

carbons. Note that oxygen-containing functional groups were generally considered beneficial for PS activation and degradation of pollutants [28, 53, 54]. However, it was inferred that the increased oxygen content here was mainly originated from the adsorbed oxidation products but not oxygen functional groups (see section S6 in Supplementary Materials). The surface area of NC800 regenerated by washing (NC800-RW) was found to be 1358 m2/g (Figure S13b and Table S3), and the contents of oxygen, nitrogen and sulfur decreased to some extent (Figure 4b and Table S3). This was attributed to the partial removal of oxidation products. The removal efficiency of SMX further dropped to 46.3% in the third run (Figure 4a). After the third run, the spent catalyst was washed and then regenerated by calcination to obtain NC800-RC. The SMX removal efficiency was partially restored to 84.2% using NC800-RC as the catalyst (Figure 4a). The contents of oxygen, nitrogen and sulfur in NC800-RC decreased obviously although still slightly higher compared to the fresh one (Figure 4b and Table S3), indicating that the calcination process helped to remove most of the adsorbed organics. However, the surface area of NC800-RC was found to be only 1486 m2/g (Figure S13b and Table S3). It was not possible that the porous structure of NC800 was destroyed during calcination regeneration since the temperature (400 °C) was far below the preparation temperature (800 °C). Thus, it was inferred that the porous structure of NC800 was partially destroyed in the catalytic oxidation process. Actually, a decrease in the surface area was usually observed in previous reports [69, 70] when ammonium persulfate was used for carbon oxidation. As a result, although most of the adsorbed organics were removed during 20

the calcination process, the surface area of NC800-RC was still much lower than that of NC800. Similar to the results in previous works [30, 37, 71], the declined surface area and altered surface chemistry hindered the adsorption and catalytic oxidation of SMX, leading to weakened catalytic performance. Figure 4 placed here 3.5 Effect of solution environment The above experiments were conducted with DI water. For real water treatment, the pH, background ions and natural organic matters (NOM) may severely influence the removal efficiency of organic contaminants. The effect of initial pH on SMX removal was conducted first. The spontaneous pH of 50 mg/L SMX solution was 4.75, and the pH drop to 3.11 after the catalytic oxidation process was attributed to the generation of hydrogen ions (Eq. (4)-(5)). As illustrated in Figure 5a and Figure S15a, a similar rate constant was obtained when the initial solution was adjusted to more acidic (pH=3.05). However, when the pH was adjusted to more alkaline values, a continuous decrease in the rate constant was observed. At an initial pH of 10.94, the rate constant dropped so dramatically that the removal efficiency of SMX at 120 min decreased to only 26.1%. The removal of SMX with the PS/NC800 system was further conducted in the presence of 1mM phosphate buffer at pH 5.8-8.0 (Figure S15b). Similar to the case in the absence of buffer, kobs also declined with increasing pH values. It was reported previously that the activation of PS could be facilitated in alkaline conditions because it was easier for negatively charged OH- to donate an electron compared to H2O [38]. In this work, the decreased rate constant at higher pH 21

values was mainly attributed to the decreased adsorption ability. As discussed in section 3.3, the adsorption of SMX favored its oxidation because the non-radical pathway mainly occurred on the carbon surface. Due to the amphoteric nature of SMX (pKa (base)=1.6, pKa (acid)=5.7) [42], negatively charged species (SMX-) is dominant at pH>5.7. On the other hand, the isoelectric point (IEP) of NC800 was determined to be 3.8 based on zeta-potential measurements (Figure S15c). With an increase in the initial pH, the surface of NC800 turned to be more negatively charged, leading to increased electrostatic repulsion. Consequently, the adsorption amounts of SMX declined at higher pH values (Figure S15d). The result here again indicated that the non-radical pathway played an important role in the oxidation process. Figure 5 placed here The effects of Cl- and HCO3- as typical background ions and humic acid (HA) as a major NOM component were displayed in Figure 5b. Cl- could be oxidized by SO4•to produce Cl• and then rapidly reacted with another Cl- to form Cl2•-, which possessed a lower oxidation potential compared to SO4•- (Eq. (4)-(5)). The decline in kobs upon the addition of 20 mM Cl- was insignificant because free radicals did not play the determining role as discussed above. Cl- +SO4•-→Cl• + SO42Cl•+Cl-→Cl2••OH+HCO -→H O+ 3 2

(4) (5)

HCO3•

SO4•-+HCO3-→SO42-+ HCO3•

(6) (7)

When 20 mM HCO3- was added, kobs for SMX removal dropped dramatically 22

(Figure 5b and Figure S15e). HCO3- could react with SO4•- and •OH to form HCO3• (Eq. (6)-(7), which was a much weaker oxidative species [72]. However, since the free radicals were not dominant in the oxidation process, the significant suppression with HCO3- could not be simply attributed to the quenching effect. Actually, the basic nature and the buffering effect of HCO3- also contributed to the decreased kobs [73] (see Section S7 in Supplementary Materials). The presence of HA also suppressed the removal of SMX to some extent (Figure 5b). HA may get adsorbed to the carbon surface through hydrophobic, - stacking or other interactions [74], covering the active sites and hindering the adsorption of SMX. On the other hand, HA may also act as a competitor in the oxidation process [75], consuming PS and those ROSs. To further investigate the effects of water matrices, tap water or lake water was spiked with SMX and then treated with the NC800/PS system. The removal efficiency of SMX dropped to 82.3% and 71.4% for tap and lake water respectively, lower than the 99.5% removal obtained in DI water (Figure S16). The properties of the water matrices were compared in Table S4. The real waters possessed higher pH values and conductivities than DI water, both of which contributed to the declined removal percentages. Moreover, lake water and tap water both contained some NOM components as indicated from their TOC values. The negative effect of HA as a typical NOM on SMX removal has been investigated above. Furthermore, lake water without SMX spiking was treated with NC800 or NC800/PS. To determine the NOM components before and after treatment, 3D EEM fluorescence measurements were conducted (Figure 6a-c). The spectra could be divided into five regions (Region I-V) 23

according to previous works [76, 77]. Region I and II were related to aromatic proteins such as tyrosine- and tryptophan-like compounds while Region III was associated with fulvic acid-like substances [76]. Besides, Region IV and V were attributed to soluble microbial byproduct-like and humic acid-like organics, respectively [77]. As displayed in Figure 6a, all those types of organics could be observed in fresh lake water, and fulvic acid/humic acid-like organics were the main components. The intensity of the fluorescence signals decreased after adsorption treatment with NC800 (Figure 6b). After treating with the NC800/PS system, more evident decrease in fluorescence intensity was observed (Figure 6c). The EEM fluorescence results clearly demonstrated that the NOM components in lake water could be adsorbed and oxidized by the NC800/PS system, which was also indicated from the change in TOC concentration before and after treatment (Figure 6d). Thus, it was safe to conclude that the declined SMX removal percentages in real waters resulted from the convoluted effects of pH, inorganic ions and NOM components. Figure 6 placed here Although these results provided some information for potential practical use, it should be admitted that the catalyst in this work is still far from actual use. Anyway, PS or PMS based AOP is still a young technology compared to others such as Fenton oxidation, and the limited report concerning actual water treatment on an industrial level is still one of the main shortcoming of this process. However, more and more research in this field may help to provide more options for future selection of suitable water treatment technologies based on the properties of real water sources. 24

4. Conclusions Using EDTA-2K as the precursor, a series of nitrogen-doped porous carbons were prepared and employed for the catalytic removal of SMX in a PS-based AOP system. The carbon prepared at 800 °C (NC800) possessed the optimum performance for SMX removal, which was attributed to its high surface area, good electrical conductivity and rich defects as well as doped carbonyl and nitrogen groups. ROSs including •OH, SO4•-, 1O2 and 1O2 were involved in the reaction process based on quenching tests and ESR measurements. Moreover, the non-radical electron-transfer pathway between PS and SMX with the assistance of carbon was also proposed to make a great contribution as indicated from PS decomposition tests and LSV analysis. Adsorption played an important role in the reaction process, and the suppression effect of BQ, L-histidine and high pH values were all related to decreased adsorption of SMX. The TOC removal percentage was lower than SMX removal, indicating that some organic oxidation products were formed and were found to be AMI, dimer of SMX and others from LC-MS measurements. Stabilities test showed that deactivation of the catalyst was mostly attributed to decreased surface area and increased surface oxygen content, which could be partially restored by calcination regeneration. The NC800/PS system was efficient for catalytic removal of other three organic contaminants including Rhodamine B, methyl orange and phenol as well. For practical applications, real water matrices were spiked with SMX and then treated with the NC800/PS system. The removal efficiency of SMX dropped to 82.3% and 71.4% for tap and lake water respectively, which was attributed to their higher pH 25

values as well as co-existing ions and NOM components in them. Conflicts of interest There are no conflicts of interest to declare. Acknowledgements Financial support from the National Natural Science Foundation of China [Grant No. 51908409, 51978465, 51638011, 21806070], the Science and Technology Plans of Tianjin [Grant No. 19JCZDJC39800], Natural Science Foundation of Shandong Province [Grant No. ZR2018PB107] and China Postdoctoral Science Foundation [Grant No. 2018M641655] are gratefully acknowledged. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version. References [1] R. Dewil, D. Mantzavinos, I. Poulios, M.A. Rodrigo, New perspectives for advanced oxidation processes, J. Environ. Manage., 195 (2017) 93-99. [2] D.B. Miklos, C. Remy, M. Jekel, K.G. Linden, J.E. Drewes, U. Hübner, Evaluation of advanced oxidation processes for water and wastewater treatment – A critical review, Water Res., 139 (2018) 118-131. [3] D.A. House, Kinetics and mechanism of oxidations by peroxydisulfate, Chem. Rev., 62 (1962) 185-203. [4] D.L. Ball, J.O. Edwards, The Kinetics and mechanism of the decomposition of Caro's acid. I, J. Am. Chem. Soc., 78 (1956) 1125-1129. [5] G.P. Anipsitakis, D.D. Dionysiou, Degradation of organic contaminants in water 26

with sulfate radicals generated by the conjunction of peroxymonosulfate with cobalt, Environ. Sci. Technol., 37 (2003) 4790-4797. [6] G.P. Anipsitakis, D.D. Dionysiou, Radical generation by the interaction of transition metals with common oxidants, Environ. Sci. Technol., 38 (2004) 3705-3712. [7] T. Olmez-Hanci, I. Arslan-Alaton, S. Gurmen, I. Gafarli, S. Khoei, S. Safaltin, D. Yesiltepe Ozcelik, Oxidative degradation of Bisphenol A by carbocatalytic activation of persulfate and peroxymonosulfate with reduced graphene oxide, J. Hazard. Mater., 360 (2018) 141-149. [8] X. Duan, W. Li, Z. Ao, J. Kang, W. Tian, H. Zhang, S.-H. Ho, H. Sun, S. Wang, Origins of boron catalysis in peroxymonosulfate activation and advanced oxidation, J. Mater. Chem. A, 7 (2019) 23904-23913. [9] L. Hu, G. Zhang, M. Liu, Q. Wang, S. Dong, P. Wang, Application of nickel foam-supported Co3O4-Bi2O3 as a heterogeneous catalyst for BPA removal by peroxymonosulfate activation, Sci. Total. Environ., 647 (2019) 352-361. [10] R. Luo, M. Li, C. Wang, M. Zhang, M.A. Nasir Khan, X. Sun, J. Shen, W. Han, L. Wang, J. Li, Singlet oxygen-dominated non-radical oxidation process for efficient degradation of bisphenol A under high salinity condition, Water Res., 148 (2019) 416-424. [11] J. Liang, X. Xu, W. Qamar Zaman, X. Hu, L. Zhao, H. Qiu, X. Cao, Different mechanisms between biochar and activated carbon for the persulfate catalytic degradation of sulfamethoxazole: Roles of radicals in solution or solid phase, Chem. 27

Eng. J., 375 (2019) 121908. [12] J. Yu, L. Tang, Y. Pang, G. Zeng, J. Wang, Y. Deng, Y. Liu, H. Feng, S. Chen, X. Ren, Magnetic nitrogen-doped sludge-derived biochar catalysts for persulfate activation: Internal electron transfer mechanism, Chem. Eng. J., 364 (2019) 146-159. [13] G. Zhang, Z. Wu, H. Liu, Q. Ji, J. Qu, J. Li, Photoactuation healing of α-FeOOH@g-C3N4 catalyst for efficient and stable activation of persulfate, Small, 13 (2017) 1702225. [14] J. Wang, M. Yang, R. Liu, C. Hu, H. Liu, J. Qu, Anaerobically-digested sludge conditioning by activated peroxymonosulfate: Significance of EDTA chelated-Fe2+, Water Res., 160 (2019) 454-465. [15] R.R. Solis, I.F. Mena, M.N. Nadagouda, D.D. Dionysiou, Adsorptive interaction of peroxymonosulfate with graphene and catalytic assessment via non-radical pathway for the removal of aqueous pharmaceuticals, J. Hazard. Mater., 384 (2020) 121340. [16] T. Zhang, Y. Chen, Y. Wang, J. Le Roux, Y. Yang, J.P. Croue, Efficient peroxydisulfate activation process not relying on sulfate radical generation for water pollutant degradation, Environ. Sci. Technol., 48 (2014) 5868-5875. [17] X. Pan, L. Yan, C. Li, R. Qu, Z. Wang, Degradation of UV-filter benzophenone-3 in aqueous solution using persulfate catalyzed by cobalt ferrite, Chem. Eng. J., 326 (2017) 1197-1209. [18] Y. Ji, Y. Fan, K. Liu, D. Kong, J. Lu, Thermo activated persulfate oxidation of antibiotic sulfamethoxazole and structurally related compounds, Water Res., 87 28

(2015) 1-9. [19] X. Pan, L. Yan, R. Qu, Z. Wang, Degradation of the UV-filter benzophenone-3 in aqueous solution using persulfate activated by heat, metal ions and light, Chemosphere, 196 (2018) 95-104. [20] T.K. Lau, W. Chu, N.J. Graham, The aqueous degradation of butylated hydroxyanisole by UV/S2O82-: study of reaction mechanisms via dimerization and mineralization, Environ. Sci. Technol., 41 (2007) 613-619. [21] J.M. Monteagudo, H. El-taliawy, A. Durán, G. Caro, K. Bester, Sono-activated persulfate oxidation of diclofenac: Degradation, kinetics, pathway and contribution of the different radicals involved, J. Hazard. Mater., 357 (2018) 457-465. [22] M. Nie, C. Yan, M. Li, X. Wang, W. Bi, W. Dong, Degradation of chloramphenicol by persulfate activated by Fe2+ and zerovalent iron, Chem. Eng. J., 279 (2015) 507-515. [23] M. Ahmad, A.L. Teel, R.J. Watts, Mechanism of persulfate activation by phenols, Environ. Sci. Technol., 47 (2013) 5864-5871. [24] C. Alexopoulou, A. Petala, Z. Frontistis, C. Drivas, S. Kennou, D.I. Kondarides, D. Mantzavinos, Copper phosphide and persulfate salt: A novel catalytic system for the degradation of aqueous phase micro-contaminants, Appl. Catal. B-Environ., 244 (2019) 178-187. [25] Y. Shang, C. Chen, P. Zhang, Q. Yue, Y. Li, B. Gao, X. Xu, Removal of sulfamethoxazole from water via activation of persulfate by Fe3C@NCNTs including mechanism of radical and nonradical Process, Chem. Eng. J., (2019) 122004. 29

[26] C. Wang, J. Kim, V. Malgras, J. Na, J. Lin, J. You, M. Zhang, J. Li, Y. Yamauchi, Metal-organic frameworks and their derived materials: Emerging catalysts for a sulfate radicals-based advanced oxidation process in water purification, Small, 15 (2019) e1900744. [27] W. Li, Y. Zhang, Y. Liu, X. Cheng, W. Tang, C. Zhao, H. Guo, Kinetic performance of peroxymonosulfate activated by Co/Bi25FeO40: radical and non-radical mechanism, J. Taiwan. Inst. Chem. E., 100 (2019) 56-64. [28] X. Cheng, H. Guo, Y. Zhang, X. Wu, Y. Liu, Non-photochemical production of singlet oxygen via activation of persulfate by carbon nanotubes, Water Res., 113 (2017) 80-88. [29] H. Sun, C. Kwan, A. Suvorova, H.M. Ang, M.O. Tadé, S. Wang, Catalytic oxidation of organic pollutants on pristine and surface nitrogen-modified carbon nanotubes with sulfate radicals, Appl. Catal. B-Environ., 154 (2014) 134-141. [30] X. Duan, C. Su, L. Zhou, H. Sun, A. Suvorova, T. Odedairo, Z. Zhu, Z. Shao, S. Wang, Surface controlled generation of reactive radicals from persulfate by carbocatalysis on nanodiamonds, Appl. Catal. B-Environ., 194 (2016) 7-15. [31] H. Lee, H.I. Kim, S. Weon, W. Choi, Y.S. Hwang, J. Seo, C. Lee, J.H. Kim, Activation of persulfates by graphitized nanodiamonds for removal of organic compounds, Environ. Sci. Technol., 50 (2016) 10134-10142. [32] H. Chen, K.C. Carroll, Metal-free catalysis of persulfate activation and organic-pollutant degradation by nitrogen-doped graphene and aminated graphene, Environ. Pollut., 215 (2016) 96-102. 30

[33] M. Pedrosa, G. Drazic, P.B. Tavares, J.L. Figueiredo, A.M.T. Silva, Metal-free graphene-based catalytic membrane for degradation of organic contaminants by persulfate activation, Chem. Eng. J., 369 (2019) 223-232. [34] S. Zhu, X. Huang, F. Ma, L. Wang, X. Duan, S. Wang, Catalytic removal of aqueous contaminants on N-doped graphitic biochars: Inherent roles of adsorption and nonradical mechanisms, Environ. Sci. Technol., 52 (2018) 8649-8658. [35] J. Yu, L. Tang, Y. Pang, G. Zeng, H. Feng, J. Zou, J. Wang, C. Feng, X. Zhu, X. Ouyang, J. Tan, Hierarchical porous biochar from shrimp shell for persulfate activation: A two-electron transfer path and key impact factors, Appl. Catal. B-Environ., 260 (2020) 118160. [36] J. He, Y. Xiao, J. Tang, H. Chen, H. Sun, Persulfate activation with sawdust biochar in aqueous solution by enhanced electron donor-transfer effect, Sci. Total. Environ., 690 (2019) 768-777. [37] X. Duan, H. Sun, M. Tade, S. Wang, Metal-free activation of persulfate by cubic mesoporous carbons for catalytic oxidation via radical and nonradical processes, Catal. Today, 307 (2018) 140-146. [38] L. Tang, Y. Liu, J. Wang, G. Zeng, Y. Deng, H. Dong, H. Feng, J. Wang, B. Peng, Enhanced activation process of persulfate by mesoporous carbon for degradation of aqueous organic pollutants: Electron transfer mechanism, Appl. Catal. B-Environ., 231 (2018) 1-10. [39] M. Sevilla, A.B. Fuertes, A general and facile synthesis strategy towards highly porous carbons: carbonization of organic salts, J. Materi. Chem. A, 1 (2013) 13738. 31

[40] J. Dai, S. Tian, Y. Jiang, Z. Chang, A. Xie, R. Zhang, Y. Yan, Facile synthesis of porous carbon sheets from potassium acetate via in-situ template and self-activation for highly efficient chloramphenicol removal, J. Alloy. Compd., 732 (2018) 222-232. [41] J. He, J. Dai, A. Xie, S. Tian, Z. Chang, Y. Yan, P. Huo, Preparation of macroscopic spherical porous carbons@carboxymethylcellulose sodium gel beads and application for removal of tetracycline, RSC Adv., 6 (2016) 84536-84546. [42] I. Ahmed, B.N. Bhadra, H.J. Lee, S.H. Jhung, Metal-organic framework-derived carbons: Preparation from ZIF-8 and application in the adsorptive removal of sulfamethoxazole from water, Catal. Today, 301 (2018) 90-97. [43] X. Li, H. Yuan, X. Quan, S. Chen, S. You, Effective adsorption of sulfamethoxazole, bisphenol A and methyl orange on nanoporous carbon derived from metal-organic frameworks, J. Environ. Sci., 63 (2017) 250-259. [44] X. Wu, B. Ding, C. Zhang, B. Li, Z. Fan, Self-activation of potassium gluconate derived nitrogen and sulfur dual-doping hierarchical porous carbons for asymmetric supercapacitors with high energy densities, Carbon, 153 (2019) 225-233. [45] M. Sevilla, G.A. Ferrero, N. Diez, A.B. Fuertes, One-step synthesis of ultra-high surface area nanoporous carbons and their application for electrochemical energy storage, Carbon, 131 (2018) 193-200. [46] S. Yu, N. Sun, L. Hu, L. Wang, Q. Zhu, Y. Guan, B. Xu, Self-template and self-activation synthesis of nitrogen-doped hierarchical porous carbon for supercapacitors, J. Power Sources, 405 (2018) 132-141. [47] W. Ben, B. Zhu, X. Yuan, Y. Zhang, M. Yang, Z. Qiang, Occurrence, removal 32

and risk of organic micropollutants in wastewater treatment plants across China: Comparison of wastewater treatment processes, Water Res., 130 (2018) 38-46. [48] W. Gwenzi, N. Chaukura, Organic contaminants in African aquatic systems: Current knowledge, health risks, and future research directions, Sci. Total. Environ., 619-620 (2018) 1493-1514. [49] H. Lee, H.-J. Lee, J. Jeong, J. Lee, N.-B. Park, C. Lee, Activation of persulfates by carbon nanotubes: Oxidation of organic compounds by nonradical mechanism, Chem. Eng. J., 266 (2015) 28-33. [50]

C.

Liang,

C.-F.

Huang,

N.

Mohanty,

R.M.

Kurakalva,

A

rapid

spectrophotometric determination of persulfate anion in ISCO, Chemosphere, 73 (2008) 1540-1543. [51] Y. Cao, K. Wang, X. Wang, Z. Gu, Q. Fan, W. Gibbons, J.D. Hoefelmeyer, P.R. Kharel, M. Shrestha, Hierarchical porous activated carbon for supercapacitor derived from corn stalk core by potassium hydroxide activation, Electrochim. Acta, 212 (2016) 839-847. [52] X. Pan, C. Li, J. Chen, J. Liu, J. Ge, J. Yao, S. Wang, Z. Wang, R. Qu, A. Li, The photodegradation of 1,3,6,8-tetrabromocarbazole in n-hexane and in solid-mediated aqueous system: Kinetics and transformation mechanisms, Chem. Eng. J., 375 (2019) 121986. [53] Y. Wu, J. Guo, Y. Han, J. Zhu, L. Zhou, Y. Lan, Insights into the mechanism of persulfate activated by rice straw biochar for the degradation of aniline, Chemosphere, 200 (2018) 373-379. 33

[54] K. Zhu, X. Wang, D. Chen, W. Ren, H. Lin, H. Zhang, Wood-based biochar as an excellent activator of peroxydisulfate for Acid Orange 7 decolorization, Chemosphere, 231 (2019) 32-40. [55] K. Zhu, X. Wang, M. Geng, D. Chen, H. Lin, H. Zhang, Catalytic oxidation of clofibric acid by peroxydisulfate activated with wood-based biochar: Effect of biochar pyrolysis temperature, performance and mechanism, Chem. Eng. J., 374 (2019) 1253-1263. [56] Y. Yao, C. Lian, G. Wu, Y. Hu, F. Wei, M. Yu, S. Wang, Synthesis of “sea urchin”-like carbon nanotubes/porous carbon superstructures derived from waste biomass for treatment of various contaminants, Appl. Catal. B-Environ., 219 (2017) 563-571. [57] J.C. Espinosa, S. Navalón, M. Álvaro, H. García, Reduced graphene oxide as a metal‐free catalyst for the light‐assisted fenton‐like reaction, ChemCatChem, 8 (2016) 2642-2648. [58] W. Hu, Y. Xie, S. Lu, P. Li, T. Xie, Y. Zhang, Y. Wang, One-step synthesis of nitrogen-doped sludge carbon as a bifunctional material for the adsorption and catalytic oxidation of organic pollutants, Sci. Total. Environ., 680 (2019) 51-60. [59] P. Stärtzel, H. Gieseler, M. Gieseler, A.M. Abdul-Fattah, M. Adler, H.-C. Mahler, P. Goldbach, Freeze drying of L-arginine/sucrose-based protein formulations, Part I: Influence of formulation and arginine counter ion on the critical formulation temperature, product performance and protein stability, J. Pharm. Sci., 104 (2015) 2345-2358. 34

[60] J. Yan, J. Li, J. Peng, H. Zhang, Y. Zhang, B. Lai, Efficient degradation of sulfamethoxazole by the CuO@Al2O3 (EPC) coupled PMS system: Optimization, degradation pathways and toxicity evaluation, Chem. Eng. J., 359 (2019) 1097-1110. [61] Y. Shi, G. Liu, L. Wang, H. Zhang, Activated carbons derived from hydrothermal impregnation of sucrose with phosphoric acid: remarkable adsorbents for sulfamethoxazole removal, RSC Adv., 9 (2019) 17841-17851. [62] X. Duan, H. Sun, J. Kang, Y. Wang, S. Indrawirawan, S. Wang, Insights into heterogeneous

catalysis

of

persulfate

activation

on

dimensional-structured

nanocarbons, ACS Catal., 5 (2015) 4629-4636. [63] X. Cheng, H. Guo, Y. Zhang, G.V. Korshin, B. Yang, Insights into the mechanism of nonradical reactions of persulfate activated by carbon nanotubes: Activation performance and structure-function relationship, Water Res., 157 (2019) 406-414. [64] C. Liu, Y. Wang, Y. Zhang, R. Li, W. Meng, Z. Song, F. Qi, B. Xu, W. Chu, D. Yuan, B. Yu, Enhancement of Fe@porous carbon to be an efficient mediator for peroxymonosulfate activation for oxidation of organic contaminants: Incorporation NH2-group into structure of its MOF precursor, Chem. Eng. J., 354 (2018) 835-848. [65] S. Yang, L. Li, T. Xiao, Y. Zhang, D. Zheng, Promoting effect of ammonia modification on activated carbon catalyzed peroxymonosulfate oxidation, Sep. Purif. Technol., 160 (2016) 81-88. [66] N. Wang, W. Ma, Z. Ren, Y. Du, P. Xu, X. Han, Prussian blue analogues derived porous nitrogen-doped carbon microspheres as high-performance metal-free 35

peroxymonosulfate activators for non-radical-dominated degradation of organic pollutants, J. Materi. Chem. A, 6 (2018) 884-895. [67] S. Wang, J. Wang, Radiation-induced degradation of sulfamethoxazole in the presence of various inorganic anions, Chem. Eng. J., 351 (2018) 688-696. [68] M. Xu, J. Li, Y. Yan, X. Zhao, J. Yan, Y. Zhang, B. Lai, X. Chen, L. Song, Catalytic degradation of sulfamethoxazole through peroxymonosulfate activated with expanded graphite loaded CoFe2O4 particles, Chem. Eng. J., 369 (2019) 403-413. [69] N. Li, X. Ma, Q. Zha, K. Kim, Y. Chen, C. Song, Maximizing the number of oxygen-containing functional groups on activated carbon by using ammonium persulfate and improving the temperature-programmed desorption characterization of carbon surface chemistry, Carbon, 49 (2011) 5002-5013. [70] K.M. Mena Aguilar, Y. Amano, M. Machida, Ammonium persulfate oxidized activated carbon fiber as a high capacity adsorbent for aqueous Pb(II), Journal of Environmental Chemical Engineering, 4 (2016) 4644-4652. [71] S. Wang, L. Xu, J. Wang, Nitrogen-doped graphene as peroxymonosulfate activator and electron transfer mediator for the enhanced degradation of sulfamethoxazole, Chem. Eng. J., (2019) 122041. [72] H. Fu, P. Zhao, S. Xu, G. Cheng, Z. Li, Y. Li, K. Li, S. Ma, Fabrication of Fe3O4 and graphitized porous biochar composites for activating peroxymonosulfate to degrade p-hydroxybenzoic acid: Insights on the mechanism, Chem. Eng. J., (2019) 121980. [73] P. Zhang, X. Tan, S. Liu, Y. Liu, G. Zeng, S. Ye, Z. Yin, X. Hu, N. Liu, 36

Catalytic degradation of estrogen by persulfate activated with iron-doped graphitic biochar: Process variables effects and matrix effects, Chem. Eng. J., 378 (2019) 122141. [74] W. Libbrecht, A. Verberckmoes, J.W. Thybaut, P. Van Der Voort, J. De Clercq, Tunable large pore mesoporous carbons for the enhanced adsorption of humic acid, Langmuir., 33 (2017) 6769-6777. [75] X. Cheng, P. Li, W. Zhou, D. Wu, C. Luo, W. Liu, Z. Ren, H. Liang, Effect of peroxymonosulfate oxidation activated by powdered activated carbon for mitigating ultrafiltration membrane fouling caused by different natural organic matter fractions, Chemosphere, 221 (2019) 812-823. [76] X.S. He, B.D. Xi, Z.M. Wei, Y.H. Jiang, Y. Yang, D. An, J.L. Cao, H.L. Liu, Fluorescence excitation-emission matrix spectroscopy with regional integration analysis for characterizing composition and transformation of dissolved organic matter in landfill leachates, J. Hazard. Mater., 190 (2011) 293-299. [77] W. Chen, P. Westerhoff, J.A. Leenheer, K. Booksh, Fluorescence excitation− emission matrix regional integration to quantify spectra for dissolved organic matter, Environ. Sci. Technol., 37 (2003) 5701-5710.

37

Figure Captions Figure 1 Effect of catalyst type (a), PS dosage (b), catalyst dosage (c) and initial concentration (d) on SMX removal. Experimental conditions unless otherwise stated: catalyst type=NC800, PS dosage=1 mM, catalyst dosage=0.050 g/L, initial concentration=50 mg/L, temperature=30 °C. Insets are the apparent rate constants. An additional catalyst dosage value of 0.1 g/L was used at an initial concentration of 100 mg/L. Figure 2 Effect of quenchers on SMX oxidation (a), LSV obtained by the NC800 electrode (b), correlation between kobs and adsorption amounts of different contaminants (c), TOC removal (d) and plausible oxidation pathways of SMX (e). Figure 3 Plausible reaction mechanism for SMX oxidation with the NC800/PS system. Figure 4 The stability tests of NC800 for SMX removal (a) and XPS spectra of fresh used and regenerated NC800 (b) Experimental conditions: PS dosage=1 mM, catalyst dosage=0.050 g/L, initial concentration=50 mg/L, temperature=30 °C. Figure 5 Effect of initial pH (a) and co-existing ions/humic acid (b) on SMX removal. Figure 6 3D EEM fluorescence spectra of fresh lake water (a), lake water treated with NC800 (b) and lake water treated with NC800/PS as well as the TOC values before and after treatment (d). Experimental conditions: catalyst type=NC800, PS dosage=0 or 1 mM, catalyst dosage=0.050 g/L, temperature=30 °C.

38

There are no conflicts of interest to declare.

Figure 1 Effect of catalyst type (a), PS dosage (b), catalyst dosage (c) and initial concentration (d) on SMX removal. Experimental conditions unless otherwise stated: catalyst type=NC800, PS dosage=1 mM, catalyst dosage=0.050 g/L, initial concentration=50 mg/L, temperature=30 °C. An additional catalyst dosage value of 0.1 g/L was used at an initial concentration of 100 mg/L.

39

Figure 2 Effect of quenchers on SMX oxidation (a), LSV obtained by the NC800 electrode (b), correlation between kobs and adsorption amounts of different contaminants (c), TOC removal (d) and plausible oxidation pathways of SMX (e).

40

Figure 3 Plausible reaction mechanism for SMX oxidation with the NC800/PS system.

Figure 4 The stability tests of NC800 for SMX removal (a) and XPS spectra of fresh used and regenerated NC800 (b) Experimental conditions: PS dosage=1 mM, catalyst dosage=0.050 g/L, initial concentration=50 mg/L, temperature=30 °C.

41

Figure 5 Effect of initial pH (a) and co-existing ions/humic acid (b) on SMX removal.

Figure 6 3D EEM fluorescence spectra of fresh lake water (a), lake water treated with NC800 (b) and lake water treated with NC800/PS as well as the TOC values before and after treatment (d). Experimental conditions: catalyst type=NC800, PS dosage=0 or 1 mM, catalyst dosage=0.050 g/L, temperature=30 °C. 42

Highlights EDTA-2K-dervied carbons employed to activate persulfate for sulfamethoxazole removal. Up to 99.5% removal of 50 mg/L sulfamethoxazole was obtained within 120min. Surface area, conductivity, defects and heteroatoms contributed in the process.

43