Experimental and Theoretical Study on the Transformation Behavior of Bisphenol S by Radicals Driven Persulfate Oxidation

An in-depth study on the degradation of bisphenol S (BPS) by both single-walled carbon nanotubes and heat activated persulfate (PS) was investigated in detail. The factors like materials dosage, initial substrate concentration, initial pH and water matrix on removal of BPS were evaluated and 10 µM BPS could be completely removed in 90 min under the optimal conditions of [BPS] 0 : [PS] 0 = 1: 100, T = 25 ℃ , pH 0 = 7.0, [N-SWCNTs] = 20 mg·L − 1 . Fast removal of BPS was also obtained when reaction temperature reached 65 ℃ without catalyst. There were 15 intermediates identied in total; and hydroxylation, sulfate addition, carboxylation, the cleavage of S − C bond and polymerization were considered as the main transformation pathways of BPS in both two systems based on LC-MS analysis. The proportion discrepancy of •OH and SO 4•− involved in two systems led to different distribution and abundance of observed products. The results of transition state calculation further conrmed the reaction potential of hydroxylation, hydrogen atom abstraction and sulfate addition, and the minimum reaction barriers were 22.20, 25.06 and 13.85 kJ/mol, respectively. The present work rstly reveals the overall transformation behavior of BPS in radicals-triggered PS system by combining experimental and theoretical study.


Text. S2. Preparation and characterization of nitrogen-doped nanocarbon based materials
Our previous work has proved that ammonium nitrate (NH 4 NO 3 ) is the best nitrogen source among NH 4 NO 3 , urea, polyaniline and indole when we explored the effects of N from different reagents doping on hydroxylated multi-walled carbon nanotubes to catalyze PS for UV-type filter elimination. 1 Therefore, in this study, NH 4 NO 3 was also used in order to further improve the removal efficiency of BPS. First, it was ground with carbonaceous materials at a mass ratio of 1:2 in an agate mortar. The above mixture was then placed into a tube furnace and kept at 350 ℃ for 1 h under continuous nitrogen gas flow. After calcined, the materials were reground and washed. Different NH 4 NO 3 -doped carbonaceous materials were abbreviated as N-MWCNTs, N-SWCNTs, N-CNT-OH, N-GO and N-GP.
Characterization of the prepared nanocarbon-based materials including scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectrometer (FT-IR), and BET surface area were performed and the used equipments in detail were recorded elsewhere. 1 Fig. S1 showed the SEM and TEM images of SWCNTs (a and c) and N-SWCNTs (b and d), which confirmed that the tubular structure of SWCNT still remained after nitrogen doping. In addition, N-SWCNTs were more easily dispersed than SWCNTs, revealing that nitrogen doping might improve the dispersion of CNTs, which further resulted in the increased specific surface area (560.44 and 619.59 m 2 ·g -1 of SWCNTs and N-SWCNTs respectively, listed in Table. S3). This phenomenon was also obtained in other materials except for GP (Table. S3), confirming nitrogen doping can enlarge the specific area of nanocarbon-based materials.
From XPS spectra (Fig. S2), the specific elemental composition of SWCNTs (a) and N-SWCNTs 4 (b) was summarized in Table. S2. Clearly, N did appear in N-SWCNTs samples, suggesting the successful doping of N into SWCNTs. Although the same synthesis method used in this and previous work, 1 more N content on the surface of N-SWCNTs was observed (1.16% versus 0.88%), which might be due to the existence of oxygen-containing functional groups on the surface of CNT-OH. 2 As seen from the FT-IR spectra (Fig. S4), there are three obvious absorption peaks for both SWCNTs and N-SWCNTs. The bands around 3415 and 1100 cm −1 were ascribed to the existence of hydroxyl groups (-OH), while the peak near 1560 cm −1 could be attributed to the C=O stretching vibration of the carboxylic group (-COOH). 3 Compared to SWCNTs, the intensities of these three peaks were significantly increased in N-SWCNTs, meaning that the contents of surface oxygenic functional groups on SWCNTs were improved in the process of material preparation.

Text. S4 The efficiency of different nanocarbon-based materials
Under the conditions of 1.0 mM PS, 25 ℃, 30 mg·L -1 catalyst and initial pH 7.0, the adsorption and degradation kinetics of 10 µM BPS were carried out in nanocarbon-based catalyst/PS system. As shown in Fig. S5, all the materials could reach adsorption-desorption equilibrium after 30 min shaking. 24.0%, 63.3%, 14.6%, 1.0% and 9.8% of 10 µM BPS were adsorbed by MWCNTs, SWCNTs, CNT-OH, GO and GP, respectively, while a bit increase was achieved after N doping except for CNT-OH (their saturated adsorption capacities were similar). The adsorption content of each catalyst was highly surface area-dependent and thus large surface area (560.44 m 2 ·g -1 , listed in Table. S3) led to the highest adsorption for SWCNTs case. After adding PS, BPS was gradually degraded. Compared to PS alone, in which only 4.8% BPS removal was obtained, the removal efficiency of BPS in nanocarbon-based materials/PS systems ranged from 16.7% to 100.0% not including GO. The above results demonstrated that the presence of carbonaceous materials significantly improved the removal efficiency of BPS; and both the adsorption and the oxidative degradation contributed to the removal of BPS, which were further enhanced after N doping (Figs.1 (d) and S5), in good agreement with other findings. [4][5][6] This promoted effect could be attributed to the facts that N doping interrupts graphitic carbon configuration and then generates more active sites, changes electron states of the carbon network 7 and enhances the π-π bond of the carbon matrix. 4 The highest removal of BPS was observed in MWCNTs and SWCNTs situations (84.7%~100.0%) whether N doping or not, in comparison with CNT-OH, GP, and GO (2.0%~21.7%). This might be explained by the fact that the bent graphene sheets of CNTs included partially delocalized π electrons; these π electrons could provide better access to PS for target compounds. 8

Text. S5 Catalytic mechanism
Accroding to recent reaserches, radical and non-radical mechanisms, including sulfate radicals 10 hydroxyl radicals (OH), 11 singlet oxygen generation ( 1 O 2 ), 12, 13 PS-catalyst complex and electron transfer, 14,15 are proved to exist in PS activation driven by nanocarbon-based materials. To explore the possible reactive species in PS/N-SWCNTs system, EPR and radical quenching experiments were conducted and the results were shown in Fig. S9. From EPR spectra ( Fig. S9 (a)), it was seen that obvious peak of DMPO-OH adduct (hyperfine splitting constants of α N = α H = 14.9 G) was observed in PS+DMPO system, in comparison to single DMPO and N-SWCNTs. After adding N-SWCNTs, this signal intensity was largely improved. It was unexpected that no peak of DMPO-SO 4 adduct was recorded in all studied systems, which was contradictory with the identified sulfate addition intermediates (P5). A small amount of SO 4 − , fast transformation of DMPO-SO 4 adduct to DMPO-OH adduct and very strong intensity of DMPO-OH adduct might together lead to the disappearance of sulfate radicals signal. 16 The radical quenching experiments were conducted to further validate the above observations of EPR spectra. Common SO 4 − and OH scavengers namely methanol and tert-tutanol were used.
Herein, tert-tutanol can quench OH at a rate constant of (3.8-7.6) ×10 8 M -1 s -1 , while methanol can quench both SO 4 − and OH and the corresponding rate constants were 2.5×10 7 and 9.7×10 8 M -1 s -1 . As shown in Fig. S9 (b), compared to control, the residual of BPS in reaction solution after adding 100 mM tert-butanol increased from 6.1% to 26.2% in 60 min, and more BPS (72.2%) remained in solution at higher tert-butanol dosage. In contrary to tert-butanol, the addition of methanol did not show any inhibitory effect on BPS degradation even though its molar concentration was 50000 times as high as BPS. Similar findings were also obtained in previous studies. 17 This quenching inability 9 of methanol might be related with its more hydrophilic property than tert-butanol, which resulted in hard approach to the catalyst surface, and further failed quenching radicals on the surface of catalyst.
From combined analysis of EPR results and quenching tests, we concluded that surface •OH radical (•OH ads ) was the primary ROS contributing to BPS removal.Other ROS like SO 4 − and 1 O 2 might be also involved in the oxidative reaction, which was reflected by the incomplete quenching phenomenon of tert-butanol. Duan       Note: Take C (2) as an example. Note: Take C (2) as an example.