Magnetic Template Anion Polyacrylamide–Polydopamine-Fe3O4 Combined with Ultraviolet/H2O2 for the Rapid Enrichment and Degradation of Diclofenac Sodium from Aqueous Environment

In this study, a novel system was set up by preparing a magnetic flocculant combining with ultraviolet/H2O2 to realize the rapid enrichment and degradation of diclofenac sodium (DCFS). For the magnetic flocculant, template anion polyacrylamide (TAPAM) with anion micro-block structure was prepared. Thereafter, polydopamine was used to modify TAPAM, Fe3O4 nanoparticles was grafted to the modified TAPAM by chelation, named template anion polyacrylamide-polydopamine-Fe3O4 (TAPAM-PDA-Fe3O4). Furthermore, the TAPAM-PDA-Fe3O4 preparation protocol was optimized by the response surface method (RSM). In the DCFS enrichment section, the rapid separation of flocs from water was realized by an external magnetic field and it indicated that the π–π stacking effect was dominant in neutral/alkaline condition, whereas charge neutralization was favored in acidic conditions. Meanwhile, a DCFS enrichment kinetic curve was much fitted by the pseudo-second-order kinetic model and DCFS enrichment isothermal curve was close to the Freundlich isothermal model, indicating the dependence of DCFS quantity enriched by TAPAM-PDA-Fe3O4 and a multilayer heterogeneous enrichment process. The degradation experiment confirmed that DCFS was effectively degraded by ultraviolet/H2O2/TAPAM-PDA-Fe3O4 and the maximum value of DCFS degradation efficiency reached 98.1%. Furthermore, the regeneration experiment showed that the enrichment and degradation efficiency of DCFS could maintain a relatively high level in the initial three recycles.


Introduction
Diclofenac sodium (DCFS) represents a rising concern due to its ecotoxicological potential and current, sustained release into the aquatic environment. Furthermore, many regions and countries, such as Europe, America, China, and so on, have added DCFS to the priority substances monitoring list of water policy [1][2][3]. Approximately a hundred tons of DCFS are sold annually worldwide, about 15% of which is excreted unchanged by human race and is frequently detected in domestic wastewater [4]. In addition, the ever-increasing global population is expected to cause a sharp increase in the emissions of DCFS in the future. Several studies have shown that DCFS exposure demonstrates bioaccumulation in invertebrates, which has an ecotoxicological risk to the aquatic food chain [5]. At present, due to the

TAPAM-PDA-Fe3O4 Synthesis Optimization
The preparation of TAPAM-PDA-Fe3O4 described in Text S1 was optimized by Box-Behnken Design (BBD) and response surface method (RSM). Six factors: pH (X1), mole ratio of PDAC/AMPS (X2), AMPS/AM (X3), AM/DA (X4), DA/Fe3O4 (X5) and irradiation time (X6) were selected as the independent variables. Intrinsic viscosity was chosen as the response variable. As shown in Table 1 and Supplementary Table S1, B54 (3 6 ) was used to improve the experiment and Supplementary Equation (S1) was used to predict the optimal condition.

TAPAM-PDA-Fe 3 O 4 Synthesis Optimization
The preparation of TAPAM-PDA-Fe 3 O 4 described in Text S1 was optimized by Box-Behnken Design (BBD) and response surface method (RSM). Six factors: pH (X 1 ), mole ratio of PDAC/AMPS (X 2 ), AMPS/AM (X 3 ), AM/DA (X 4 ), DA/Fe 3 O 4 (X 5 ) and irradiation time (X 6 ) were selected as the independent variables. Intrinsic viscosity was chosen as the response variable. As shown in Table 1 and Supplementary Table S1, B 54 (3 6 ) was used to improve the experiment and Supplementary Equation (S1) was used to predict the optimal condition.  (3 6 ).

Analytical Methods
FT-IR spectra were measured by a spotlight 200 FT-IR spectrometer (Nicolet IS10, Thermo Fisher Scientific, Waltham, MA, USA). X-ray photoelectron spectroscopy spectra was detected by the ESCALAB250Xi XPS spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). Chemical constituents and thermal stability of the products were analyzed by a DTG-60H synchronal thermal analyzer (SHIMADZU, Kyoto, Japan). The morphology and fractal dimension of magnetic flocculant were detected by the high-resolution field emission scanning electron microscope (SEM, Quattro S, Thermo Fisher Scientific, Waltham, MA, USA) coupled with Image-Pro software (plus 6.0, Media Cybernetics, Rockville, MA, USA). The magnetic properties and crystal phases of magnetic flocculant were analyzed by a vibrating sample magnetometer (PPMS DynaCool 9, Quantum Design, San Diego, CA, USA) and a X-ray diffractometer (7000, Shimadzu, Kyoto, Japan), respectively. A Nano ZS90Zetasizer (Malvern Instruments Co. Ltd, Malvern, UK) was used to analyze zeta potential of DCFS aqueous solution.

Enrichment Experiments
DCFS standard aqueous solution of 2.0 mg L −1 was prepared in the lab, and the corresponding predetermined concentration (0.1, 1.0, 1.5, 2.0 mg L −1 ) was accurately prepared by diluting standard aqueous solution with ultrapure water. Standard TAPAM-PDA-Fe 3 O 4 suspension of 10 mg mL −1 was prepared by adding 0.1 g TAPAM-PDA-Fe 3 O 4 into 10 mL deionized water and then 0.5, 1.0, 1.5, 2.0, 2.5 and 3.0 mL of that were separately added into 250 mL DCFS solution of predetermined concentration. A program-controlled ZR4-6 jar test apparatus (Shenzhen Zhongrun Water Industry Technology and Development Co. Ltd, Shenzhen, China) was used to carry out enrichment experiment at ambient temperature and the mixture was stirred vigorously (300 rpm) for 1 min, then was stirred moderately (160 rpm) for 4 min, and finally stirred slowly (40 rpm) for 5 min. The rapid separation of magnetic flocculant from aqueous environment was achieved by settling a circular magnet under the beaker and the sample was extracted from 1 cm under the aqueous solution surface to detect the remained concentration by HPLC-UV (WATERS, Milford, MA, USA) combined with the COSMOSIL 3C 18 -MS-II column (Nacalai Tesque, Inc., Kyoto, Japan) at 276-nm detection wavelength. A mixture of acetic acid solution (3.0%)/acetonitrile (80:20) was used as the mobile phase, the flow rate and the largest injection volume of which were 1.0 mL min −1 and 50 µL, respectively. All results of DCFS remained concentration were averaged with three averaged measurements and the scale bar was obtained by calculating the standard deviation of the three values. The removal efficiency is described by Equation (1): where C i and C f are for the initial and the final concentration of DCFS, respectively.

Degradation Experiments
The separated magnetic flocculant prepared in Section 2.4 was then added into the 25 mL quartz conical flask with 10 mL dilute sulfuric acid solution (0.1 M) to generate Fe 2+ by grafted Fe 3 O 4 nanoparticles, and the pH was adjusted to 4.5. The concentration of Fe 2+ in the sample after 60 min settlement was measured by the phenanthroline-colorimetric method. Then, a predetermined amount of H 2 O 2 (n (Fe 2+ ):n (H 2 O 2 ) = 1:4) was injected into the solution. Meanwhile, the DCFS solution treated directly with Fe 3 O 4 /H 2 O 2 /sulfuric acid was used as control. The concentration of Fe 2+ , H 2 O 2 and sulfuric acid in the control group was equal to that in the experimental group. The samples were both irradiated under UV environment at 365 nm wavelength by a GY-500 high-pressure mercury lamp for 60 min. The remaining amount of DCFS before and after irradiation was also detected by HPLC-UV (WATERS, Milford, MA, USA) and the degradation efficiency was calculated by Equation (1). The degradation products of DCFS were identified by ultraperformance liquid chromatography coupled with Vion ion mobility separation quadrupole time-of-flight tandem mass spectrometry (UPLC-Vion IMS Q-TOF-MS, ACQUITY I class, Waters, Milford, MA, USA).

Stability and Regeneration of TAPAM-PDA-Fe 3 O 4
Stability and regeneration experiments were carried out to investigate the potential of TAPAM-PDA-Fe 3 O 4 reutilization. After enrichment and degradation of sections of DCFS, the magnetic flocculant was extracted by an external magnetic field and revised by ethanol for three times, and it was used to repeat experiments in Sections 2.4 and 2.5, and processed for five cycles. The removal and degradation efficiency of DCFS was analyzed by the method mentioned above. Thereafter, the FT-IR spectra of used TAPAM-PDA-Fe 3 O 4 were obtained and virgin TAPAM-PDA-Fe 3 O 4 was used as control. Meanwhile, the Fe 2+ concentration of the solution settled for 60 min of every cycle containing sulfuric acid/TAPAM-PDA-Fe 3 O 4 was detected by the method cited in Section 2.5. Further, the original DCFS solution (equal amount of DCFS) of every cycle, which contained an equal mass of Fe 3 O 4 and the same concentration of sulfuric acid was used as control.

TAPAM-PDA-Fe 3 O 4 Synthesis Optimization
The least square method shown in Supplementary Table S1 was used to measure the intrinsic viscosity regression model in consideration of the six factors. The regression model variance analysis and regression equation were presented in Supplementary Table S2 and Supplementary Equation (S2), respectively. As shown in Supplementary Table S2, the p-value was no more than 0.0001, suggesting statistical significance. Furthermore, X 1 , X 3 , X 6 , X 12 , X 32 , and X 62 were significant model terms. The interaction effects of X 1 , X 3 and X 1 , X 6 on the intrinsic viscosity of TAPAM-PDA-Fe 3 O 4 were depicted in Figure 2. The intrinsic viscosity TAPAM-PDA-Fe 3 O 4 increased slightly with the increase of X 3 (0.2-1.4) and then decreased as the X 3 further increased (1.4-5.0) as shown in Figure 2A. On the contrary, the intrinsic viscosity increased significantly with the increase of X 1 (5.0-8.5), then decreased slightly as the X 1 further increased (8.5-10.0). It was more significant that pH showed on intrinsic viscosity of TAPAM-PDA-Fe 3 O 4 than mole ratio of AMPS/AM as depicted in Figure 2A. The intrinsic viscosity of TAPAM-PDA-Fe 3 O 4 strengthened sharply with the increase of X 6 (30-75 min) and then decreased significantly as the X 6 further increased (75-120 min) as presented in Figure 2B. The other factors (X 2 , X 4 , X 5 ) had no significant effect on the intrinsic viscosity of TAPAM-PDA-Fe 3 O 4 indicated from Supplementary Table S2. Furthermore, the simulated maximum value of by RSM was 1721.5 mL g −1 (X 1 = 8.6, X 2 = 0.50, X 3 = 1.27 X 4 = 1.48, X 5 = 3.29, and X 6 = 86.0 min). Moreover, six parallel TAPAM-PDA-Fe 3 O 4 synthesis processes were carried out according to the theoretical optimal conditions. The actual intrinsic viscosity mean value was 1710.7 mL g −1 , which was close to the simulated value, indicating the appositeness of the theoretical optimal condition.  Figure 3A illustrates the FT-IR spectra of TAPAM-PDA-Fe3O4, TAPAM-Fe3O4, and APAM-Fe3O4, respectively. As for TAPAM-PDA-Fe3O4, the characteristic peaks at 3336 cm −1 and 3187 cm −1 belong to the phenolic O-H and N-H stretching vibration, the peak at 1650 cm −1 was attributed to the aromatic ring stretching vibration and N-H bending vibration, the peak at 1545 cm −1 inclined the N-H shearing vibration, the peak at 1411 cm −1 was caused by phenolic C-O-H bending vibration, and the peak at 1117 cm −1 was due to C-O vibration, suggesting that PDA was grafted onto TAPAM successfully [24,25]. Furthermore, the peak at 1604 cm −1 was due to the C=O stretching vibration from AM and AMPS. Meanwhile the -SO3H-stretching bands from AMPS was found at 1088 and 1039 cm −1 [26,27]. As for APAM-Fe3O4 and TAPAM-Fe3O4, it was found that the C=O peak area at 1643 cm −1 in TAPAM-Fe3O4 was larger than that at 1636 cm −1 in APAM-Fe3O4. Furthermore, the wide scan XPS spectra of the TAPAM-PDA-Fe3O4, TAPAM-Fe3O4 and APAM-Fe3O4 was depicted in Figure 3B for analyzing the Fe3O4 nanoparticles distribution on the flocculants surface. The photoelectron lines at binding energies of 709.2 eV corresponding to Fe2p were both found in curves of TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4, but they didn't exist in the curve of APAM-Fe3O4 (see dotted circles). As a result, PDAC used as the cationic template made the negative groups of AMPS more concentrated in the polymer chain and TAPAM grafted by PDA had abandon catechol groups, which may both contribute to make more Fe3O4 nanoparticles distribute on the flocculant surface and enhance the flocculant magnetic properties. Meanwhile, the binding energy of 166.3 eV related to S2P was identified in the XPS spectra of APAM-Fe3O4 and TAPAM-Fe3O4, while that wasn't recognized in TAPAM-PDA-Fe3O4 (see dotted circles). It was suggested that -SO3H-groups weren't exposed on the surface when PDA grafted onto the TAPAM, which may lead to a multilayer heterogeneous enrichment process. Furthermore, compared with the C1s spectra of TAPAM-Fe3O4 and APAM-Fe3O4, the C1s spectra of TAPAM-PDA-Fe3O4 was deconvoluted into five components corresponding to C=C (283.7 eV), C-C (284.1 eV), C-N (286.0 eV), C-O (287.2 eV) and C=O (288.2 eV), which was depicted in Figure S1 [28,29]. It was further confirmed that abundant aromastic rings with hydroxyl groups were grafted on TAPAM.  Figure 3A illustrates the FT-IR spectra of TAPAM-PDA-Fe 3 O 4 , TAPAM-Fe 3 O 4 , and APAM-Fe 3 O 4 , respectively. As for TAPAM-PDA-Fe 3 O 4 , the characteristic peaks at 3336 cm −1 and 3187 cm −1 belong to the phenolic O-H and N-H stretching vibration, the peak at 1650 cm −1 was attributed to the aromatic ring stretching vibration and N-H bending vibration, the peak at 1545 cm −1 inclined the N-H shearing vibration, the peak at 1411 cm −1 was caused by phenolic C-O-H bending vibration, and the peak at 1117 cm −1 was due to C-O vibration, suggesting that PDA was grafted onto TAPAM successfully [24,25]. Furthermore, the peak at 1604 cm −1 was due to the C=O stretching vibration from AM and AMPS. Meanwhile the -SO 3 H-stretching bands from AMPS was found at 1088 and 1039 cm −1 [26,27]. As for APAM-Fe 3 Figure S1 [28,29]. It was further confirmed that abundant aromastic rings with hydroxyl groups were grafted on TAPAM.

TGA-DSC Analysis
Figure 4A-C show the Thermo Gravimetric and differential scanning calorimetry analysis (TGA-DSC) curves of APAM-Fe3O4, TAPAM-Fe3O4, and TAPAM-PDA-Fe3O4, respectively. It was observed in the TGA curves that four main stages for A and B and three main stages for C exist. The initial stages of the three TGA curves ranged 103.0-209.7 °C, 119.0-197.0 °C, and 164.5-229.6 °C with partial weight loss of about 5.77% for A, 3.29% for B, and 1.32% for C, owing to the confined water molecules bonded by hydrophilic groups in the molecular chains in the synthesis process, such as -SO3H- [30]. Thereafter, 10.3% weight loss (209.7-296.4 °C) for A and 14.6% weight loss (197.0-305.2 °C) for B were observed in the second stage. These results correspond to the thermal decomposition of -CO-NHgroups from A and B. The weight loss of 5.87% in the interval of 229.6-313.2 °C for C was corresponding to the thermal decomposition of aromatic rings and -CO-NH-groups [31,32]. In the third stage, 13.5% weight loss for A (296.4-357.1 °C), 14.0% weight loss for B (305.2-348.3 °C), and 7.54% weight loss for C (313.2-396.6 °C) were ascribed to the thermal decomposition of the -SO3Hgroups [33]. As for the fourth stage of A and B, weight loss of about 26.4% and 35.7% correspond to the thermal decomposition of the copolymer backbone [34]. It was decomposed completely above 473.9 °C for A, above 485.4 °C for B, and above 608.9 °C for C. The corresponding residual weight percentage of 44.9%, 31.9% and 78.6% was corresponding to the residual Fe3O4 and carbon black, respectively [35]. The DSC curves of A and B both showed apparent four peaks of heat absorption appearing at 195.    3.29% for B, and 1.32% for C, owing to the confined water molecules bonded by hydrophilic groups in the molecular chains in the synthesis process, such as -SO 3 H- [30]. Thereafter, 10.3% weight loss (209.7-296.4 • C) for A and 14.6% weight loss (197.0-305.2 • C) for B were observed in the second stage. These results correspond to the thermal decomposition of -CO-NH-groups from A and B. The weight loss of 5.87% in the interval of 229.6-313.2 • C for C was corresponding to the thermal decomposition of aromatic rings and -CO-NH-groups [31,32]. In the third stage, 13.5% weight loss for A (296.4-357.1 • C), 14.0% weight loss for B (305.2-348.3 • C), and 7.54% weight loss for C (313.2-396.6 • C) were ascribed to the thermal decomposition of the -SO 3 Hgroups [33]. As for the fourth stage of A and B, weight loss of about 26.4% and 35.7% correspond to the thermal decomposition of the copolymer backbone [34]. It was decomposed completely above 473.9 • C for A, above 485.4 • C for B, and above 608.9 • C for C. The corresponding residual weight percentage of 44.9%, 31.9% and 78.6% was corresponding to the residual Fe 3 O 4 and carbon black, respectively [35]. The DSC curves of A and B both showed apparent four peaks of heat absorption appearing at 195.

TGA-DSC Analysis
Figure 4A-C show the Thermo Gravimetric and differential scanning calorimetry analysis (TGA-DSC) curves of APAM-Fe3O4, TAPAM-Fe3O4, and TAPAM-PDA-Fe3O4, respectively. It was observed in the TGA curves that four main stages for A and B and three main stages for C exist. The initial stages of the three TGA curves ranged 103.0-209.7 °C, 119.0-197.0 °C, and 164.5-229.6 °C with partial weight loss of about 5.77% for A, 3.29% for B, and 1.32% for C, owing to the confined water molecules bonded by hydrophilic groups in the molecular chains in the synthesis process, such as -SO3H- [30]. Thereafter, 10.3% weight loss (209.7-296.4 °C) for A and 14.6% weight loss (197.0-305.2 °C) for B were observed in the second stage. These results correspond to the thermal decomposition of -CO-NHgroups from A and B. The weight loss of 5.87% in the interval of 229.6-313.2 °C for C was corresponding to the thermal decomposition of aromatic rings and -CO-NH-groups [31,32]. In the third stage, 13.5% weight loss for A (296.4-357.1 °C), 14.0% weight loss for B (305.2-348.3 °C), and 7.54% weight loss for C (313.2-396.6 °C) were ascribed to the thermal decomposition of the -SO3Hgroups [33]. As for the fourth stage of A and B, weight loss of about 26.4% and 35.7% correspond to the thermal decomposition of the copolymer backbone [34]. It was decomposed completely above 473.9 °C for A, above 485.4 °C for B, and above 608.9 °C for C. The corresponding residual weight percentage of 44.9%, 31.9% and 78.6% was corresponding to the residual Fe3O4 and carbon black, respectively [35].   10 µm/500 nm were shown in Figure 5. The corresponding fractal dimensions of 1.2786, 1.3154, and 1.3639 are also shown in Figure 5, respectively. Interestingly, the specific surface area of APAM-Fe 3 O 4 , TAPAM-Fe 3 O 4 and TAPAM-PDA-Fe 3 O 4 increased gradually while the average particle size was reduced in sequence. The surfaces of the three magnetic flocculants were not smooth, which was caused by the self-assembly of polymer chains to form particles, and the particles are intertwined to form a porous network structure. It is favorable that enhancing the flocculation and bridging effect in the flocculation process. Furthermore, the morphology of APAM-Fe 3 O 4 , TAPAM-Fe 3 O 4 , and TAPAM-PDA-Fe 3 O 4 changed gradually from fluffy to compact. This was caused by two reasons, on one hand, the template polymerization made the negative charge distribution more concentrated on the polymer chain so that the macromolecule backbone was more extended, which was favorable for the magnetic flocculant to form a lattice-like structure. On the other hand, the catechol groups from grafted PDA were strongly negative charged in the aqueous environment so that the internal charge repulsion of the molecular chain was increased, and the abundant aromatic rings from grafted PDA increased the inner-chain steric hindrance, which made the magnetic flocculant more spread in the aqueous environment. The lattice-like structure was further enhanced.

Morphological Analysis
A vacuum freeze dryer was used to treat the samples to preserve the surface morphology of samples. The micro-structures of APAM-Fe3O4, TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4 at a scale of 10 μm/500 nm were shown in Figure 5. The corresponding fractal dimensions of 1.2786, 1.3154, and 1.3639 are also shown in Figure 5, respectively. Interestingly, the specific surface area of APAM-Fe3O4, TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4 increased gradually while the average particle size was reduced in sequence. The surfaces of the three magnetic flocculants were not smooth, which was caused by the self-assembly of polymer chains to form particles, and the particles are intertwined to form a porous network structure. It is favorable that enhancing the flocculation and bridging effect in the flocculation process. Furthermore, the morphology of APAM-Fe3O4, TAPAM-Fe3O4, and TAPAM-PDA-Fe3O4 changed gradually from fluffy to compact. This was caused by two reasons, on one hand, the template polymerization made the negative charge distribution more concentrated on the polymer chain so that the macromolecule backbone was more extended, which was favorable for the magnetic flocculant to form a lattice-like structure. On the other hand, the catechol groups from grafted PDA were strongly negative charged in the aqueous environment so that the internal charge repulsion of the molecular chain was increased, and the abundant aromatic rings from grafted PDA increased the inner-chain steric hindrance, which made the magnetic flocculant more spread in the aqueous environment. The lattice-like structure was further enhanced.

XRD Patterns
It was shown in Figure 6A that the XRD pattern curves of Fe3O4, TAPAM-PDA-Fe3O4, TAPAM-Fe3O4 and APAM-Fe3O4, respectively. The characteristic diffraction peaks of Fe3O4 in TAPAM-PDA-Fe3O4 and TAPAM-Fe3O4 were confirmed, indicating that the Fe3O4 crystal phase was unchanged after the polymerization section. However, there were not any characteristic peaks of Fe3O4 observed in APAM-Fe3O4, suggesting that the Fe3O4 nanoparticles were enmeshed into the polymer chain,

XRD Patterns
It was shown in Figure 6A  which corresponds to the XPS analysis results. In addition, a broad peak at 22.7 °C , was observed, indicating the wrinkled surface structure of organic fraction.

Magnetic Properties
As shown in Figure 6B, the saturation magnetization value was measured to be 25.3 emu g −1 for TAPAM-PDA-Fe3O4, 5.7 emu g −1 for TAPAM-Fe3O4, 3.8 emu g −1 for APAM-Fe3O4, and 85.1 emu g -1 for Fe3O4. The saturation magnetization value of TAPAM-PDA-Fe3O4 was relatively much higher than that of TAPAM-Fe3O4 or APAM-Fe3O4. It suggested that Fe3O4 nanoparticles chelated with grafted PDA was contributed to enhance the magnetization of flocculant. Although the saturation magnetization of TAPAM-PDA-Fe3O4 decreased compared with Fe3O4 particles, it could significantly facilitate the separation of flocs by an external magnetic field.

DCFS Initial Concentration Effect
The enrichment capability of TAPAM-PDA-Fe3O4 for removing various concentration DCFS was investigated and APAM-Fe3O4 and TAPAM-Fe3O4 were used as control. For all three kinds of magnetic flocculants, the removal efficiency was higher as the DCFS initial concentration was 0.1 mg L −1 instead of 1.0, 1.5 and 2.0 mg L −1 , as shown in Figure 7A-C. It was because that steric hindrance between DCFS and polymer chain was enhanced when DCFS initial concentration exceeded a certain value, which negatively affected the removal efficiency. It was also found that TAPAM-PDA-Fe3O4 had better DCFS removal efficiency than APAM-Fe3O4 or TAPAM-Fe3O4. The maximum removal efficiency of APAM-Fe3O4, TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4 was 49.1%, 66.2%, and 99.1% as the DCFS concentration was 0.1 mg L −1 , respectively. This phenomenon could be explained as follows. On one hand, the anion block structure made the negative charge more concentrated and strengthened the charge neutralization effect. On the other hand, the TAPAM modified by PDA had abandoned aromatic rings and absorbed DCFS through π-π stacking effect, which further improved the DCFS removal efficiency. As a result, the 0.1 mg L −1 DCFS was used in the following sections.

Magnetic Properties
As shown in Figure 6B, the saturation magnetization value was measured to be 25.

DCFS Initial Concentration Effect
The enrichment capability of TAPAM-PDA-Fe 3 O 4 for removing various concentration DCFS was investigated and APAM-Fe 3 O 4 and TAPAM-Fe 3 O 4 were used as control. For all three kinds of magnetic flocculants, the removal efficiency was higher as the DCFS initial concentration was 0.1 mg L −1 instead of 1.0, 1.5 and 2.0 mg L −1 , as shown in Figure 7A-C. It was because that steric hindrance between DCFS and polymer chain was enhanced when DCFS initial concentration exceeded a certain value, which negatively affected the removal efficiency. It was also found that TAPAM-PDA-Fe 3 O 4 had better DCFS removal efficiency than APAM-Fe 3 O 4 or TAPAM-Fe 3 O 4 . The maximum removal efficiency of APAM-Fe 3 O 4 , TAPAM-Fe 3 O 4 and TAPAM-PDA-Fe 3 O 4 was 49.1%, 66.2%, and 99.1% as the DCFS concentration was 0.1 mg L −1 , respectively. This phenomenon could be explained as follows. On one hand, the anion block structure made the negative charge more concentrated and strengthened the charge neutralization effect. On the other hand, the TAPAM modified by PDA had abandoned aromatic rings and absorbed DCFS through π-π stacking effect, which further improved the DCFS removal efficiency. As a result, the 0.1 mg L −1 DCFS was used in the following sections.

DCFS Initial pH Effect
The enrichment performance of APAM-Fe3O4, TAPAM-Fe3O4, and TAPAM-PDA-Fe3O4 was investigated under the conditions of various DCFS initial pH values. As seen in Figure 8A, in the range of pH 1.0-3.0, the DCFS removal efficiency was increased when the sample was treated by APAM-Fe3O4. In the interval of pH 3.0-8.0, the DCFS removal efficiency was maintained between 68.2%-72.4%. However, it decreased dramatically to 3.0% when the initial pH further increased to 12.0. As for TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4, the removal efficiency curves of DCFS had the similar trend. In the range of pH 4.0-9.0, the enrichment performance of TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4 was both better than that of APAM-Fe3O4. Furthermore, the enrichment performance of TAPAM-PDA-Fe3O4 was much better in the interval of pH 5.0-9.0, and DCFS removal efficiency was above 90.0% and the maximum value was above 95.0%. As a result, pH 8.0 was used in DCFS enrichment kinetics and isotherms analysis.

Enrichment Mechanism Analysis
The enrichment mechanism of TAPAM-PDA-Fe3O4 was further investigated. It was illustrated in Figure 8B that DCFS zeta potential obtained both before and after TAPAM-PDA-Fe3O4 treatment at different pH value. The DCFS isoelectric point was located at pH 4.2 before TAPAM-PDA-Fe3O4 treatment. However, the DCFS isoelectric point was around pH 7.8 after TAPAM-PDA-Fe3O4 addition, which was ascribed to the charge neutralization between -NH + groups in DCFS and -SO3Hgroups in TAPAM-PDA-Fe3O4. Furthermore, high DCFS maximum removal efficiency at pH 9.0 known from Figure 8A indicated the involvement of π-π stacking between DCFS and TAPAM-PDA-Fe3O4. At pH 10.0, the maximum removal efficiency reached 44.3%, which indicates that under relatively strong alkaline conditions, the electrostatic repulsion between the negatively charged groups also leads to a decrease in the π-π stacking efficiency, and the enmeshment/sweeping effect

DCFS Initial pH Effect
The enrichment performance of APAM-Fe3O4, TAPAM-Fe3O4, and TAPAM-PDA-Fe3O4 was investigated under the conditions of various DCFS initial pH values. As seen in Figure 8A, in the range of pH 1.0-3.0, the DCFS removal efficiency was increased when the sample was treated by APAM-Fe3O4. In the interval of pH 3.0-8.0, the DCFS removal efficiency was maintained between 68.2%-72.4%. However, it decreased dramatically to 3.0% when the initial pH further increased to 12.0. As for TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4, the removal efficiency curves of DCFS had the similar trend. In the range of pH 4.0-9.0, the enrichment performance of TAPAM-Fe3O4 and TAPAM-PDA-Fe3O4 was both better than that of APAM-Fe3O4. Furthermore, the enrichment performance of TAPAM-PDA-Fe3O4 was much better in the interval of pH 5.0-9.0, and DCFS removal efficiency was above 90.0% and the maximum value was above 95.0%. As a result, pH 8.0 was used in DCFS enrichment kinetics and isotherms analysis.

Enrichment Mechanism Analysis
The enrichment mechanism of TAPAM-PDA-Fe3O4 was further investigated. It was illustrated in Figure 8B that DCFS zeta potential obtained both before and after TAPAM-PDA-Fe3O4 treatment at different pH value. The DCFS isoelectric point was located at pH 4.2 before TAPAM-PDA-Fe3O4 treatment. However, the DCFS isoelectric point was around pH 7.8 after TAPAM-PDA-Fe3O4 addition, which was ascribed to the charge neutralization between -NH + groups in DCFS and -SO3Hgroups in TAPAM-PDA-Fe3O4. Furthermore, high DCFS maximum removal efficiency at pH 9.0 known from Figure 8A indicated the involvement of π-π stacking between DCFS and TAPAM-PDA-Fe3O4. At pH 10.0, the maximum removal efficiency reached 44.3%, which indicates that under relatively strong alkaline conditions, the electrostatic repulsion between the negatively charged groups also leads to a decrease in the π-π stacking efficiency, and the enmeshment/sweeping effect played a considerable role. In a nutshell, the whole process is depicted in Figure 9. First, DCFS was absorbed by abundant aromatic rings in TAPAM-PDA-Fe3O4 through π-π stacking in the interval of

Enrichment Mechanism Analysis
The enrichment mechanism of TAPAM-PDA-Fe 3 O 4 was further investigated. It was illustrated in Figure 8B that DCFS zeta potential obtained both before and after TAPAM-PDA-Fe 3 O 4 treatment at different pH value. The DCFS isoelectric point was located at pH 4.2 before TAPAM-PDA-Fe 3 O 4 treatment. However, the DCFS isoelectric point was around pH 7.8 after TAPAM-PDA-Fe 3 O 4 addition, which was ascribed to the charge neutralization between -NH + groups in DCFS and -SO 3 H-groups in TAPAM-PDA-Fe 3 O 4 . Furthermore, high DCFS maximum removal efficiency at pH 9.0 known from Figure 8A indicated the involvement of π-π stacking between DCFS and TAPAM-PDA-Fe 3 O 4 . At pH 10.0, the maximum removal efficiency reached 44.3%, which indicates that under relatively strong alkaline conditions, the electrostatic repulsion between the negatively charged groups also leads to a decrease in the π-π stacking efficiency, and the enmeshment/sweeping effect played a considerable role. In a nutshell, the whole process is depicted in Figure 9. First, DCFS was absorbed by abundant aromatic rings in TAPAM-PDA-Fe 3 O 4 through π-π stacking in the interval of 7.0 < pH < 9.0. Second, it was charge neutralization between DCFS and -SO 3 H-groups in TAPAM-PDA-Fe 3 O 4 that worked in the range of 1.0 < pH < 4.2. At last, large flocs were formed by the aggregation of small flocs through the external magnetic field.

DCFS Enrichment Kinetics Analysis
The DCFS enrichment kinetics treated by TAPAM-PDA-Fe3O4 were investigated by pseudo-first order (PFO), pseudo-second order (PSO), and intraparticle diffusion models, which were expressed as Supplementary Equation (S3)-(S5). All the obtained data were simulated by the three kinetic models. As a result, the simulated parameters were depicted in Table 2. Compared with other two models, the PSO model correlation coefficients were much closer to 1.0, meanwhile the q value calculated by PSO model is much closer to the actual value. This confirmed that DCFS removal rate was dependent on the quantity of DCFS flocculated onto the surface of TAPAM-PDA-Fe3O4 [36,37]. The PSO model simulated curves of DCFS enrichment kinetics by TAPAM-PDA-Fe3O4 are presented in Figure 10A. The DCFS removal rate increased sharply in the first 30 min, but slowed down thereafter. The foremost DCFS rapid enrichment was due to the accessibility of aromatic rings and -SO3H-groups in TAPAM-PDA-Fe3O4. The succedent DCFS slow enrichment was due to the saturation of the flocculation sites.

DCFS Enrichment Kinetics Analysis
The DCFS enrichment kinetics treated by TAPAM-PDA-Fe 3 O 4 were investigated by pseudo-first order (PFO), pseudo-second order (PSO), and intraparticle diffusion models, which were expressed as Supplementary Equation (S3)-(S5). All the obtained data were simulated by the three kinetic models. As a result, the simulated parameters were depicted in Table 2. Compared with other two models, the PSO model correlation coefficients were much closer to 1.0, meanwhile the q value calculated by PSO model is much closer to the actual value. This confirmed that DCFS removal rate was dependent on the quantity of DCFS flocculated onto the surface of TAPAM-PDA-Fe 3 O 4 [36,37]. The PSO model simulated curves of DCFS enrichment kinetics by TAPAM-PDA-Fe 3 O 4 are presented in Figure 10A. The DCFS removal rate increased sharply in the first 30 min, but slowed down thereafter. The foremost DCFS rapid enrichment was due to the accessibility of aromatic rings and -SO 3 H-groups in TAPAM-PDA-Fe 3 O 4 . The succedent DCFS slow enrichment was due to the saturation of the flocculation sites. q e (mg g −1 ) k 1 × 10 2 (g mg −1 min −1 ) R 2 q e (mg g −1 ) Where q e and q max,exp are the enrichment capacity of the magnetic flocculant at equilibrium and the actual maximum enrichment capacity of the magnetic flocculant, respectively. k 1 and k 2 are the rate constant of first-order and second-order flocculation, respectively. k p is the intraparticle diffusion rate constant and R 2 is the correlation coefficient.

DCFS Enrichment Isotherms Analysis
The obtained experimental data were simulated by Dubinin-Radushkevich, Freundlich, and Langmuir isotherm models, which were expressed as Supplementary Equation (S6)-(S8). The final fitted parameters are all depicted in Table 3. It was conformed that Freundlich model correlation coefficients were much close to 1.0, suggesting that a multilayer heterogeneous flocculation of DCFS occurred onto the TAPAM-PDA-Fe3O4 [38,39]. In addition, the Freundlich isotherm simulated curves of DCFS at different temperatures were depicted in Figure 10B. The DCFS enrichment efficiency at 318.15 K was better than that at 308.15 K or 328.15 K. This indicated that the Brownian motion principle between DCFS and TAPAM-PDA-Fe3O4 had a great influence on the enrichment process. To some extent, it increased the collision possibility between DCFS and TAPAM-PDA-Fe3O4, which was positively contributed to DCFS enrichment. While beyond a certain limit, confined DCFS molecules by TAPAM-PDA-Fe3O4 tended to escape, which weakened DCFS enrichment process.

DCFS Enrichment Isotherms Analysis
The obtained experimental data were simulated by Dubinin-Radushkevich, Freundlich, and Langmuir isotherm models, which were expressed as Supplementary Equation (S6)-(S8). The final fitted parameters are all depicted in Table 3. It was conformed that Freundlich model correlation coefficients were much close to 1.0, suggesting that a multilayer heterogeneous flocculation of DCFS occurred onto the TAPAM-PDA-Fe 3 O 4 [38,39]. In addition, the Freundlich isotherm simulated curves of DCFS at different temperatures were depicted in Figure 10B. The DCFS enrichment efficiency at 318.15 K was better than that at 308.15 K or 328.15 K. This indicated that the Brownian motion principle between DCFS and TAPAM-PDA-Fe 3 O 4 had a great influence on the enrichment process. To some extent, it increased the collision possibility between DCFS and TAPAM-PDA-Fe 3 O 4 , which was positively contributed to DCFS enrichment. While beyond a certain limit, confined DCFS molecules by TAPAM-PDA-Fe 3 O 4 tended to escape, which weakened DCFS enrichment process.

Degradation Performance
As shown in Figure 11A, the total amount of DCFS in the two samples separately was equal (25 µg) and when the DCFS simulated wastewater (0.1 mg L −1 ) was directly treated with UV/H 2 O 2 /Fe 3 O 4 , the degradation efficiency was low because of the low DCFS concentration. Meanwhile, the maximum degradation efficiency was 31.1% and the residual amount of DCFS was 17.2 µg after 30 min UV irradiation. However, the maximum degradation efficiency reached 98.1% and the DCFS residual amount was 0.48 µg when it was under UV/H 2 O 2 /TAPAM-PDA-Fe 3 O 4 enrichment-degradation (E-D) treatment. This suggested that DCFS degradation could be realized efficiently when the enriched DCFS was re-dissolved into the dilute sulfuric acid solution (pH = 4.5) which was 0.04 times the original wastewater volume. Meanwhile, the degradation products of DCFS were identified by UPLC-Vion IMS QTOF-MS. Among the products, bis (6-methylheptyl) phthalate and 2, 6-dichloeoaniline were identified, except diclofenac sodium, which was depicted in Supplementary Figure S2. However, the degradation pathways of diclofenac sodium need further identification, and will be investigated in our next study. Moreover, various degradation processes of diclofenac sodium were summarized and compared in Table 4. It can be seen that the UV/H 2 O 2 /TAPAM-PDA-Fe 3 O 4 process is a promising method to treat low concentration diclofenac sodium solution.

Degradation Performance
As shown in Figure 11A, the total amount of DCFS in the two samples separately was equal (25 μg) and when the DCFS simulated wastewater (0.1 mg L −1 ) was directly treated with UV/H2O2/Fe3O4, the degradation efficiency was low because of the low DCFS concentration. Meanwhile, the maximum degradation efficiency was 31.1% and the residual amount of DCFS was 17.2 μg after 30 min UV irradiation. However, the maximum degradation efficiency reached 98.1% and the DCFS residual amount was 0.48 μg when it was under UV/H2O2/TAPAM-PDA-Fe3O4 enrichmentdegradation (E-D) treatment. This suggested that DCFS degradation could be realized efficiently when the enriched DCFS was re-dissolved into the dilute sulfuric acid solution (pH = 4.5) which was 0.04 times the original wastewater volume. Meanwhile, the degradation products of DCFS were identified by UPLC-Vion IMS QTOF-MS. Among the products, bis (6-methylheptyl) phthalate and 2, 6-dichloeoaniline were identified, except diclofenac sodium, which was depicted in Supplementary  Figure S2. However, the degradation pathways of diclofenac sodium need further identification, and will be investigated in our next study. Moreover, various degradation processes of diclofenac sodium were summarized and compared in Table 4. It can be seen that the UV/H2O2/TAPAM-PDA-Fe3O4 process is a promising method to treat low concentration diclofenac sodium solution.   In consideration of the stability of TAPAM-PDA-Fe 3 O 4 , it was reused for five cycles to enrich and degrade DCFS, and the FT-IR spectra of the revised magnetic flocculant are shown in Figure 11C. Compared with the initial TAPAM-PDA-Fe 3 O 4 , the characteristic peaks could be all found at 3182, 1652, and 1604 cm −1 in the revised magnetic flocculant. While the intensity of the characteristic peak at 1411 cm −1 , which is ascribed to the phenolic C-O-H bending vibration, was weaker than the virgin TAPAM-PDA-Fe 3 O 4 curve. This was due to the irreversible self-assembly of TAPAM-PDA-Fe 3 O 4 and parts of the catechol groups were enmeshed inside the magnetic flocculant. However, the functional groups of TAPAM-PDA-Fe 3 O 4 remained relatively stable. Meanwhile, the DCFS removal efficiency and DCFS degradation efficiency both remained in the relatively high value in the initial three circles, which were separately above 90.0% and 87.0%, as shown in Figure 11B. However, the removal and degradation efficiency in the fourth and fifth cycles degraded sharply, and the corresponding minimum values were 65.1% and 59.5%, respectively. The phenomena could be explained as follows. The decrease of DCFS removal efficiency was owing to the irreversible enmeshment of TAPAM-PDA-Fe 3 O 4 , which led to the decreasing amount of the active function groups on the surface of the magnetic flocculant. Furthermore, the decrease of DCFS degradation efficiency was caused by two aspects. On one hand, the decrease of DCFS removal efficiency would negatively affect the concentration of the concentrated DCFS solution. On the other hand, the amount of Fe 3 O 4 nanoparticles grafted on the surface of TAPAM-PDA-Fe 3 O 4 decreased through reacting with sulfuric acid, which led to the sequence decrease of Fe 2+ concentration of DCFS solution in the fourth and fifth degradation cycles, as shown in Figure 11D. However, there was little effect on the initial three removal and degradation efficiency cycles.

Conclusions
In this study, TAPAM-PDA-Fe 3 O 4 was successfully prepared to achieve a high-level enrichment of DCFS, up to 99.1%. Furthermore, the rapid separation of flocs from aqueous environment was realized by external magnetic field and higher maximum removal efficiency could be achieved with TAPAM-PDA-Fe 3 O 4 (99.1%) instead of APAM-Fe 3 O 4 (49.0%) or TAPAM-Fe 3 O 4 (66.2%). In neutral/alkaline condition, it was π-π stacking that had a significant impact on flocculation performance. However, in acidic conditions, charge neutralization played a leading role. In the degradation section, it was found that the degradation efficiency of DCFS by UV/H 2 O 2 /TAPAM-PDA-Fe 3 O 4 E-D treatment could reach 98.2%, compared with which was 31.1% of UV/H 2 O 2 /Fe 3 O 4 direct treatment. Meanwhile, the TAPAM-PDA-Fe 3 O 4 could be reused effectively for three circles to realize effective enrichment and degradation of DCFS. In summary, we have designed an integrated DCFS removal system, which can reach above 90.0% removal efficiency and above 87.5% degradation efficiency in the first three cycles. Looking forward, it is possible for the UV/H 2 O 2 /TAPAM-PDA-Fe 3 O 4 system to treat wastewater containing DCFS effectively in relatively loose external conditions, thus holding great promise for further use in actual sewage treatment.

Conflicts of Interest:
The authors declare no conflict of interest.