Magnetic gelatin-activated biochar synthesis from agricultural biomass for the removal of sodium diclofenac from aqueous solution: adsorption performance and external influence

ABSTRACT Pharmaceutical and personal care products have received increasing more attention worldwide. Diclofenac was found to be one of the most detected pharmaceuticals in environmental matrixes and has acute toxicity against micro-living organisms. The present study aimed to investigate the application of magnetic gelatin modified peanut shell biochar (PBG) as an environmentally friendly adsorbent for the removal of sodium diclofenac (DCF) in the aqueous phase. Through SEM, EDS, BET, XRD, XPS, and FTIR characterisation techniques, the surface morphology, element components, specific surface area as surface functional groups of PBG were different from the pristine biochar. Batch adsorption experiment showed that PBG performed very high DCF adsorption capacity (qmax = 348.9 mg/g at pH = 6.5, T = 298 K). The adsorption capacity was significantly affected by pH solution with a decrease trend when pH >6. The experimental results could be satisfactorily fitted by Pseudo-second order kinetic model, and Temkin isotherm model. Thermodynamic study indicated that the adsorption process was non-spontaneous and endothermic. Some external agents such as humic acid and common ions exhibited a negative effect on the removal of DCF in the co-solute system. The adsorption capacity of the PBG adsorbent remained at 71.5% even after five cycles of regeneration, which demonstrated the stability and reusability for further removal of DCF. Besides, the proposed adsorption mechanism might contain electrostatic interactions, hydrogen bonds interactions, and π-π interactions. Considering the simple fabrication approach and the excellent adsorption performance, the PBG adsorbent can be evaluated as a potential material for environmental remediation applications.


Introduction
Diclofenac (DCF) is a sterile topical nonsteroidal anti-inflammatory drug (NSAID) product for ophthalmic use throughout the world. The drug is normally used as the salt of sodium or potassium for improved solubility and adsorption [1]. Sodium diclofenac is designated chemically as 2-[2-(2,6-dichloroanilino)phenyl]acetate, monosodium salt, with an empirical formula of C 14 H 10 C 12 NO 2 Na, the molecular weight of 318.13 g/mol and osmolarity of 300 mOsmol/ml [2]. The global consumption of DCF was estimated about more than 1443 tons per year [3], and still increase annually due to its extensive use in the treatment of pain and inflammatory of human and animal. Europe Union listed DCF in the Watch List of the EU Water Framework Directive to assemble adequate monitoring information on water bodies [4]. Because of wide range of use, DCF was detected almost everywhere, such as hospital effluents, pharmaceutical production facility effluents [5], municipal wastewater treatment plants [6], groundwater [7], and even in drinking water [8,9]. Also, DCF also has been detected in soils and edible fruits and vegetables due to the usage of reclaimed wastewater in irrigation [10][11][12]. Thus, a great of concern on DCF potential risk has been raised because of its chronic toxicity towards aquatic living organism at trace concentrations such as significant changes on thyroid hormonal levels were observed on an Indian major carp (Cirrhinus mrigala) [13], causing changes on kidney histology of three-spined stickleback (Gasterosteus aculeatus) at low µm/L concentrations [14] or oxidative stress and tissue damage on mussel (Mytilus galloprovincialis) and common carp (Cyprinus carpio) [15,16]. Furthermore, research of Schwarz, et al [17] suggested that fish mortality was found to increase with rising diclofenac concentration, and the lowest observed effect concentration of 10 µg/L on the organismic level indicates the classification of DCF as a micropollutant that needs to be paid more attention. Indeed, it is necessary to remove this micropollutant from wastewaters before discharge into the environment.
At present, different methods including electrochemical-advanced oxidation processes (AOPs) [18,19], photocatalytic degradation [20], membranes separation [21], biodegradation [22,23], bio-filtration [24,25], and adsorption [26,27] have been applied for the purification and remediation of polluted water. Among these techniques, more attention has been paid for testing the application of adsorption technology for wastewater treatment. Adsorption has become one of the superior approaches to eliminate microorganic pollutants from water bodies due to its ease of operation, relatively low cost, and no sludge disposal [28]. Different types of adsorbents have been explored and utilised for polluted water treatment in the past few decades. Among them, modified biochar has attracted much attention because of its excellent adsorption capacities for heavy metal as well as organic pollutants in aqueous solution [29]. However, the subsequent separation of the powdered biochar from aqueous environment commonly requires multiple steps of centrifugation and filtration processes, which may constrain the application of biochar in wastewater treatment [30]. To solve this problem, transition metals (Fe, Ni, Co, etc.) were inserted into biochar surface to form magnetism. By this strategy, biochar becomes magnetic and easy to be isolated from aqueous solution using a magnetic separator. Hence, the risk of secondary pollution due to the delayed separation of biochar after adsorption can be avoided.
Gelatin is a traditional water-soluble biopolymer with distinctive advantages of nontoxicity, biodegradability, biocompatibility in physiological environments [31]. It is composed of animal's bone or meat and contains various functional groups (-OH, -NH 2 and -COOH) that act as binding sites via polar or ionic interaction. These characteristics have contributed to gelatin's proven records of applications in biomedical [32], the foods industry [33], and water purification [34]. Several studies used gelatin as an activating agent to modify the pristine structure of material [31,35,36]. The composite materials were then applied for the removal of heavy metal from water, such as Cr (VI), Cr (III), Pb (II), Hg (II), Cd (II), etc. However, the study about the application of gelatinactivated biochar for the removal of organic contaminants is limited. To fulfill this gap, this study aimed to prepare magnetic gelatin-activated peanut shell biochar (PBG) and utilised it as an adsorbent for the removal of microorganic pollutants from the water environment. Biochar was prepared from peanut shell through slow pyrolysis in N 2 atmospheric environment and its pristine structure was modified by a magnetic gelatin solution. The morphology, structure and chemical information of the modified biochar were lately examined by a variety of characterisation tools containing Scanning electron microscopy (SEM), Energy-dispersive X-ray spectroscopy (EDS), X-ray diffraction pattern (XRD), X-ray photoelectron spectroscopy (XPS), Fourier transforms infrared (FTIR). Batch adsorption experiments were conducted to explore the adsorption capacity of the asprepared nanocomposite towards sodium diclofenac. The purposes of our work were: (1) introducing a straightforward, cost-effective and eco-friendly approach to synthesis an agriculture biomass-derived material; (2) examining the adsorption performance of PBG on the removal of DCF from the aqueous mixtures; (3) investigating the effect of pH solution and co-solutes (common ions, humic acid) on adsorption capacity of magnetic gelatin-activated biochar towards DCF; and (4) discussing the potential adsorption mechanism between them.

Chemicals and materials
Sodium diclofenac salt (purity 99%) was supplied by Shanghai Yien Chemical Technique Co., Ltd). The chemical identities of DCF are shown in Supplementary Information Table  S1. The other chemical reagents used were purchased from Shanghai Macklin Biochemical Co., Ltd (Shanghai, China). Peanut shell biomass -an agricultural residue -was chosen as the feedstock of biochar and collected from a market garden in Changsha (Hunan, China). Ultrapure water was obtained from Millipore Milli-Q water purification system. All the chemicals were used as received without any further purification.
To prepare DCF stock solution, the compound was dissolved in 10 mL pure methanol at a ratio of 5% w/v and continuously mixed into 90 mL ultrapure water to obtain 100 mL DCF stock solution (5 g/L). The working solution was obtained by diluting the stock with ultrapure water. The stock solution was stored in a refrigerator (<4°C) for later use.

Preparation of adsorbents
After collecting, the biomass (peanut shell) was washed with ultrapure water several times and dried naturally. It was ground with a masticator and passed through a 0.145 mm sieve, then transferred into a tube furnace and heated at 900°C for 2 hours with a heating rate of 5°C/min in atmospheric N 2 to obtain pristine biochar. The resulting biochar was cleaned with ultrapure water, dried in a vacuum at 60°C overnight and stored in a plasticsealed bag in a desiccator. The pristine biochar was labelled as PB.
To prepare magnetic gelatin-activated biochar (PBG), 0.4 g gelatin was added into a conical flask (volume 500 mL) containing 200 mL ultrapure water. The conical flask was put in a water bath at 358 K for 40 minutes. Simultaneously, 200 mL solution containing Fe 3 + and Fe 2+ was prepared by adding chlorine salts with concentrations of 0.2 M for Fe 3+ and 0.1 M for Fe 2+ , respectively. Gelatin solution and Fe ions solution were mixed in a 500 mL conical flask, an amount of peanut biochar was added into it at last. The mixture was put in an oscillator with a speed of 200 rpm for 150 minutes at 328 K. After the reaction, the modified biochar was washed several times with ultrapure water and separated from the solution by a centrifuge (speed 4000 rpm). The sample was dried in a vacuum at 60°C overnight and stored in a plastic-sealed bag in a desiccator. The modified biochar was labelled as PBG. The preparation process is graphically exhibited in Figure 1.

Adsorbent characterisations
The methodologies utilised to characterise the properties of adsorbents including: (1) Scanning electron microscopy (SEM) (S4800, Hitachi, Japan) to know about the morphologies of materials; (2) Particle size distribution analysis (PSD) (Mastersizer 2000, Malvern Instruments Ltd, UK) was employed to obtain information about the size and range of particles representative of PBG; (3) Energy-dispersive X-ray spectroscopy (EDS) (Oxford INCA-400, UK) with gold spray craft to obtain information about the elemental composition of materials; (4) X-ray diffraction (XRD) patterns were acquired on a diffractometer (Bruker AXS D8 Advance, Germany) with Cu Kα (λ = 0.15418 nm) radiation at an acceleration voltage of 40kV and 40 mA current, a 2θ interval of 0.02° and scan speed of 0.1s/step; (5) X-ray photoelectron spectroscopy (XPS) (ESCALAB 250Xi, Thermo Fisher, USA) under Al-Kα (hv = 1486.6 eV, power 150 W, 500 µm beam spot) X-ray radiation was used to analyse the surface chemistry and chemical states of elements; (6) Fourier Transformation InfraRed (FTIR) (IR Tracer-100, Shimadzu, Japan) in the wavenumber range of 400-4000 cm −1 to confirm the presences of functional groups on the surface of materials; (7) BET surface area and pore structure data of the adsorbents were investigated by the Brunauer-Emmett-Teller (BET) N 2 adsorptiondesorption approach at 77 K or liquid nitrogen temperature using Quantachrome Instruments (EVO, USA); (8) Zeta potential analysis was conducted to measure surface charge of the adsorbent by a zeta potential metre (Zetasizer Nano-ZS90, Malvern Instruments, UK). 25 mg of adsorbents was added to 25 mL ultrapure water and the pH value was adjusted in the range of 2.0-11.0 with negligible volumes of 0.01 M NaOH or 0.01 M HCl; (9) A vibrating sample magnetometer (VSM) (Quantum Design PPMS DynaCool, USA) was used to examine the magnetic properties of the sample; and (10) Thermogravimetry curves were measured using thermo-analytical equipment (STD Q600, TA Instruments, USA). The samples were heated from room temperature to 800°C under a nitrogen atmosphere with a heating rate of 10°/min.

Experimental determinant
In the batch adsorption experiments, the initial concentration of DCF solution was 50 mg/L and 50 mL Erlenmeyer flasks were used. 5.0 mg adsorbents were added in 25 mL DCF solution (50 mg/L) then the flasks containing mixture solution were placed in an oscillator with a speed of 160 rpm at 298 K, pH ≈ 6.5 ± 0.1. After shaking at a predetermined time for equilibration adsorption, the mixture was filtered by chemical analytical filter paper (0.45 µm). Concentrations of DCF in the supernatants were determined using UV-Vis spectroscopy (UV-2550, Shimadzu) at λ = 276 nm. Detail of the method used to calculate adsorption capacity was described in Supplementary Information. All the experiments were run in duplicate.
Investigation of DCF (50 mg/L) adsorption kinetics was conducted as follows: 5.0 mg adsorbent was added into 25 mL in 50 mL Erlenmeyer flask for each sample. Sampling time was determined as 10, 20, 30, 60, 120, 240, 360, 480, 600 and 720 minutes. At each sampling time, the suspensions were filtered using chemical analytical filter paper and adsorption capacity was calculated as the difference between initial and final solution concentration of the adsorbate.
Adsorption isotherms were studied using the same procedure as depicted above excluded the initial concentration of DCF solution (in a range of 20-60 mg/L) and reaction temperature (288, 298, 308, and 318 K).
The effect of pH solution on the adsorption was studied. The pH value of the initial solution was adjusted in a range of 2.0-11.0 using negligible volumes of 0.01 M NaOH or 0.01 M HCl.
Experiments for exploring the effect of coexisting common ions on the adsorption process were conducted in the presence of different metal ions while the experimental conditions were maintained at temperature 298 K, pH ≈ 6.5 ± 0. The influence of humic acid on the adsorption of DCF was studied through adding different concentrations (0.005 to 0.03 mg/L) of humic acid to 25 mL DCF solution (50 mg/ L) containing 5.0 mg PBG at the temperature of 298 K and pH = 6.5.

Regeneration study
The regeneration of the adsorbent was carried out after the equilibrium condition was reached. The DCF-loaded PBG was added into 0.01 M NaOH solution (500 mL). The mixture was shaken at room temperature for 24 h. The same procedure was repeated for five time and the solid/liquid phases were separated using a magnet. In each cycle of adsorption, suspension liquid containing 5.0 mg PBG and 50 mg/L of DCF (25 mL) was shaken at 298 K for 12 h. The adsorption capacity of the regenerated PBG was calculated by Equation (2S) Supplementary Materials.

Materials characterisations
The surface shape and morphology of biochar pre-and post-treatment were analysed through SEM images ( Figure 2) and EDS spectra ( Figure S1a, b). It can be observed that at low magnitude images (Figure 2(a, c)), the pristine biochar (PB) sample exhibited a relatively smooth surface while the surface of modified biochar (PBG) is rough and uneven with different sizes of deposited fragments. These insoluble-deposited fragments were probable iron particles. These changes on the sample surface were confirmed by high magnitude images ( Figure 2(b and d)). Moreover, the EDS spectra of PB and PBG depicted in Figure 3(a, b) also endorsed the well succeed magnetic gelatin biochar synthesis methodology. While PB is composed of C, O, K elements (Figure S1 b), 7.11% of Fe appeared in PBG characteristic elements (Figure S1 a). This further indicated that Fe elements were successfully coated on the PBG surface through the reaction between biochar with magnetic gelatin. Besides, the average particles size distribution of PBG is displayed in Figure 3(c). It could be seen that the PBG has the size at a wide range between 0.6 μm and 1000.2 μm.
Nitrogen adsorption-desorption isotherms were carried out to characterise the specific surface areas and pore size of biochar before and after modification ( Figure 3) and the result of BET analysis is listed in Table S2. As shown in Figure 3(c), the N 2 sorption isotherm of PBG is classified as type I according to the IUPAC classification [37], which is a typical characteristic of microporous materials. Furthermore, there is a wide knee (hysteresis loop not well defined) in the PBG adsorption/desorption isotherms, indicating the presence of mesopores on the material surface [38]. From the pore size distribution curve ( Figure 3(d)), it could be observed the presence of micropores pore and mesopores with the diameter in large range (1.2-127 nm), concentrated at the range of 2-8 nm. After a process of pyrolysis at 900°C, the BET surface area of PB was 222.74 m 2 /g, however, it sharply reduced to 70.15 m 2 /g after modifying by magnetic gelatin solution. The obtained result may be due to the microporous structure of the sample because it was known that microporous solids have relatively small external surfaces [37]. Additionally, the magnetic solution included Fe 2+ and Fe 3+ was used in the modification process, thus, the iron particles might block the small pores on the material surface, led to its BET surface area significantly less than the pristine biochar.
XRD technique was used to identify the crystal structures of samples. The XRD patterns of PB and PBG are shown in Figure 4a. A broad diffraction peak at 25° and a prominent peak at 26.9° can be observed in the XRD pattern of PB and PBG, respectively, which could be ascribed to amorphous carbon [39]. There is a minor peak at 39.3° corresponding to (103) plane (JCPDS 37-1492) on the X-ray diffraction pattern of PBG, which can be assigned to the crystalline structure [40]. Besides, three characteristic peaks were showed for iron oxide (2θ = 35.3°, 56.1°, and 61.5°), and the peak positions could be indexed to (311), (511), and (440) (JCPDS No 19-629), respectively [41]. The results indicated that the iron ions were successfully loaded on the surface of the biochar, which provided the conditions for biochar magnetic separation. The magnetisation curve of the resulting modified sample is displayed in Figure 4(b) at room temperature. Specific pertinent data of PBG showed ferromagnetic property with a saturation magnetisation value of 32.0 emu/g, which indicated that PBG was enough to realise solid-liquid separation by using a permanent magnet [29]. This magnetisation characteristic confirmed that a considerable amount of iron oxide (Fe 3 O 4 ) was successfully loaded on the PBG surface, which was demonstrated above by XRD pattern analyses.
The thermo-gravimetric analysis (TGA) curves of PB and PBG are presented in Figure 4 (c). As can be seen, the major weight loss of the samples occurred in two stages, however, their weight loss temperature peaks are different. While the PB sample presented slight weight loss (8.1%) between 15 and 60°C at the first state, a mass loss of 11.8% being observed in the TGA curve of PBG sample. This first loss is linked to the evaporation of adsorbed water. In the second stage, the weight loss of PB insignificantly changed at a wide range of temperatures (60-800°C) because of the lignin -main component of peanut shell biochar-is commonly known as a difficult component for thermal decomposition (often ranging from 160 to 800°C) [42]. However, a distinct weight loss (12.5%) appeared at 100-800°C on the TGA curve of PBG, which could be ascribed to the destruction of magnetic iron oxides and oxygen-containing functional groups (such as carboxyl, hydroxyl groups on PBG).
FTIR analysis is employed to find out the functional groups contained in the samples. The FTIR spectra of PB and PBG are illustrated in Figure 4 XPS characterisation was used to further investigate the chemical nature of the flexible substrate before and after modification. XPS spectra obtained for pristine biochar and modified biochar (elemental survey scan and individual scans for primary elements observed) are presented in Figure S2a. As could be seen in Figure S2a, the PB sample contains carbon, oxygen while more elements such as nitrogen, chlorine, and iron were observed on the PBG surface, indicating that N and iron particles of the magnetic gelatin solution were loaded on the biochar surface. The XPS peak deconvolution of C 1s and O1 s are presented in Figure 5(a, b) for PBG, and Figure S2 b, c for PB sample, respectively. For N 1s and Fe 2p, the XPS spectra of these atoms are displayed as Figure 5(c, d). The XPS of four main components (C 1s, O 1s, N 1s and Fe 2p) of the PBG sample after DCF adsorption is presented in Figure 6(a, b, c, d) and will be further discussed in adsorption mechanism part. The C 1s spectra of PB were observed at the peak of 284.8 eV, and 286.0 eV which can be assigned the carbon atom in C-C/C = C and C-O moieties, respectively. Whereas the C 1s spectra of PBG could be resolved into four component peaks centred at 284.5 eV, 286,3 eV 288,4 eV and 285.2 eV, which were attributable to C-C/C = C, C-O, C = O species and C sp 3 , respectively [43,44]. This result indicated that C = O was added into pristine biochar after modification process. The O 1s spectrum of PBG was well fitted into four peaks at 530.    [29]. The peak at 718.1 eV could be assigned to Fe o and the peak at 725.0 eV was attributable to Fe 2p 1/2 , confirming the presence of Fe 3+ species on the surface of PBG [47]. The proportion of Fe 0 was only 9.3%, while the proportion of Fe 3+ and Fe 2+ was 66.1% and 24.6%, respectively.
The material characterisations described above confirmed that PBG was successfully synthesised. Its removal sodium diclofenac capacity was then examined by batch experiments along with some external factors such as temperature, contact time, pH value, coexist ion. The results will be discussed in the next section.

The influence of adsorbent dosage
The effect of adsorbent amount was determined by set of experiment with different adsorbent amount (ranging from 2.5 to 20 mg). For this adsorption experiment,  Figure 7(a). It could be seen that the removal efficiency increased with the increase of adsorbent dosage. With a dosage of 2.5 mg, PBG performed a removal efficiency of 19.2%, then it increased from 47.9% to 83.5% with an adsorbent dose varying from 5 to 20 mg. In contrast, the amount of adsorbed DCF per unit weight of the PBG adsorbent (adsorption capacity) decreased when increasing the mass of the adsorbent. This result might attributable to the partial aggregation of biochar at higher concentration, which might decrease the active sites on the surface of biochar, led to the reduction of adsorption capacity [48].
The highest adsorption capacity result for DCF was attained at 5.0 mg with 47.3% of DCF removal, corresponding an adsorbed mount of 239.5 mg/g, while 20 mg reached 83.5% but with q e value of 104.4 mg/g (Figure 7(a)). Therefore, 5.0 mg of adsorbent dose was chosen for further batch adsorption experiment.

The influence of pH
The adsorption of DCF on PB and PBG was examined from pH 2.0 to 11.0. The results of two adsorbents DCF adsorption capacity and their zeta potential are displayed in Figure 7(c, d). It could be observed that the adsorption capacity of PBG was stabilised at pH 2.0-6.0, slightly decreased at pH 7.0, and dramatically decreased at pH > 7.0. While PB adsorption capacity reached the highest value at pH 2.0, it slowly decreased when pH at the value of 3.0-4.0 and continue to decrease at pH 5.0-6.0 before remaining at low sorption capacity in pH range 7.0-11.0. The results in Figure 7(c, d) demonstrated that q e values decreased with increasing pH values for both adsorbents, suggesting for both adsorbents removal efficiency was always favoured by acidic pH. Following the aims of this study, further discussing the effect of pH as well as other ambient conditions will be conducted for the modified biochar as a target adsorbent.
Zeta potential is a characterisation property of biochar and has a considerable role when exploring the relationship between pH solution and adsorption ability of the adsorbent. As could be seen in Figure 7  and a negative charge at higher of this value. The highest DCF adsorption capacity was observed at pH = 3.0 (245.6 mg/g) and this result matched with the zeta potential measured of the adsorbent. In general, the solution pH has much influence on the adsorption performance of biochar-derived adsorbent. Thus, further discussion about the effect of pH on the adsorption process will be described at the mechanism part.

Adsorption kinetics
To understand the efficiency of an adsorbent, it's necessary to illuminate the effect of contact time on adsorption. The adsorbed quantity of DCF over PB and PBG was evaluated in the time range between 10 and 720 minutes to reach the equilibrium state. The rate of DCF adsorption by PBG was fast, with 79% of the ultimate adsorption occurred in the first 30 min. After 120 min, the adsorption capacity raised slowly and reached the equilibrium state after 10 hours reaction. It was found by kinetic experiment that q e reach a value of 238.2 mg/g when the experimental conditions were: C o = 50 mg/L, pH≈6.5, m/V = 0.2, and T = 298 K.
Adsorption kinetic models can be applied to study the chemical reactions, reaction rate, and particle diffusion behaviour of adsorbate onto adsorbent microparticle. In this study, pseudo-first-order (PFO), pseudo-second-order (PSO), Elovich, and the intra-particle diffusion models were used to simulate the experimental data. The PFO kinetic model assumes that the adsorption only occurs at specific binding sites (located on the surface of the adsorbent) and adsorption itself is the rate-limiting step during the entire transportadsorption process. The PSO kinetic model is often applied to explore chemical adsorption, which may occur the sharing or transferring of electron on the surface of adsorbent and adsorbate [49]. Elovich kinetic model is well known as a useful tool for studies that involve chemisorption of gases on a solid surface without desorption of the products, the rate may decrease with time due to an increase in surface coverage [50,51]. The theoretical aspects and mathematical expressions of each model are given in Supplementary Materials. The fit quality and the accuracy of the kinetic parameters were measured through determination coefficient (R 2 ), adjusted determination coefficient (R 2 adj ), sum of squared errors (SSE), average relative error (ARE) and Chi-square test (χ 2 ). The expressions of the error functions are given in Supplementary Materials. Figure 8(a, b) and Table 1 shows the kinetic experimental results of different models for DCF adsorption onto adsorbents and the simulated parameters, respectively. According to Table 1, the higher values of determination coefficient (R 2 > 0.97), adjusted determination coefficient (R 2 adj > 0.97) and the lower values of sum of squared errors (SEE < 232), average relative error (ARE < 2.2%), Chi-square (χ 2 < 1.22). Based on this result, it can be affirmed that the non-linear model of PSO was the most appropriate in adequately describing the kinetic studies of the as-prepared adsorbents. Meanwhile, the calculated q e value that was generated by the PSO model (235.8 mg/g) was more in line with the experimental data (238.2 mg/g), further indicating the feasibility of the PSO model to fit the kinetic DCF adsorption results. Thus, the chemisorption probably involves in the DCF adsorption process [52].
Weber and Morris's intraparticle diffusion model was applied to investigate the mass transfer steps in the DCF adsorption process. Figure 8(b) and Table 1 present the fitting of experiment data by intra-particle diffusion model, and simulated parameters, respectively. It was known that a solid/liquid sorption process may contain three steps: external mass transfer (film diffusion), intra-particle diffusion and adsorption on the internal surface of the adsorbent [29]. As could be seen (Figure 8(b)), the plot of q t versus t 1/2 of PBG was multi-linear, which suggested that three steps took place in the adsorption process. The first sharper portion may be due to the film diffusion of solute molecules or the diffusion of DCF through the solution to the external surface of PBG. The second part was the intraparticle diffusion on the pores of PBG and the third part meant the equilibrium stage, which might due to the diffusion was stable when a smaller number of available sorption sites of the adsorbent or low DCF concentration in the solution. Thus, the adsorption of DCF by PBG might include both surface sorption and intraparticle diffusion mechanism.

Adsorption isotherms
Adsorption isotherm models are widely used as a quantitative method to characterise adsorbate equilibrium between aqueous and solid phase at a constant ambient temperature [53]. In this study, several typical isotherm models were used to fit the experimental data including Langmuir, Freundlich and Temkin. The introduction and equations regarding those models are given in Supplementary Materials. The non-linear forms of these isotherm models were used to fit the adsorption isotherm experimental data of DCF uptake onto PBG and PB. The statistical evaluation of the fitted models was conducted using five mathematically error functions which are described Supplement Materials. The results and models' parameters are displayed in Figure 8(c, d) and Table 2, respectively. As shown in Figure 8(c and d), the uptake of DCF by PBG and PB increased with the increase of DCF concentration at all tested temperatures. As could be seen in Table 2  The Temkin model presented a relatively high values of R 2 and R 2 adj (> 0.98), besides, the results of sum square error (SEE = 560), Chi-square (χ 2 = 58) and average relative error (ARE = 6.56%) were lower than those of Langmuir isotherm. Thus, it could be confirmed that the adsorption behaviour of DCF onto PBG better fitted the Temkin model.

Influence of temperature and thermodynamic study
To explore the effect of temperature on the adsorption of DCF, the adsorption experiments were performed in the temperature range of 288-318 K at pH 6.5 and 50 mg/L initial DCF concentration and the result is shown in Figure S3. The adsorption of DCF appears to be significantly affected by the temperature which influences the quantity adsorbed. The experimental results indicated that as the temperature rose from 288 K to 318 K, the adsorption capacity decreased from 227.3 mg/g to 188.0 mg/g for PBG.
The thermodynamic parameters include the standard change in Gibb's energy (ΔG 0 ), the change of enthalpy/heat (ΔH 0 ) and the change of entropy (ΔS 0 ) are calculated from the temperature-dependent adsorption isotherms (Figure 11(a)) via corresponding equations (given in the Supplementary Materials). The results are presented in Table 3. It could be seen that ΔG 0 value is positive at all tested temperatures, indicating that the adsorption was non-spontaneous. Moreover, the value of q e decreased with the increase of temperature (e.g. an adsorption capacity of 256.3 mg/g was observed at 15°C (288 K) and that reduced along with an increase in temperature to 209.9 mg/g at 45°C (318 K) with the DCF initial concentration of 60 mg/L), meanwhile the values of ΔG 0 became lower with decreasing temperature, suggested that lower temperature was beneficial to the adsorption of DCF onto PBG. Several previous studies have reported similar results about the thermodynamic characteristic of DCF adsorption on derived from biochar based on biomass feedstock such as Isabel grape bagasse [54], pinewood biochar [55]. These authors ascribed the decline in sorption capacity to two main reasons: the solubility of the drug in water and the energy exchange that occurred during the process. Accordingly, raising temperature may cause an increase in the solubility of DCF, then the adsorbate particles tend to be more affinity against the solvent than the adsorbent. The decline in the interaction between DCF molecules and the PBG surface was also affected by the increasing temperature since the higher temperature, the higher vapour pressure, thus the agitation of dissolved chemical species was then increased. This phenomenon gave the result of reducing the force of the attraction between adsorbate and adsorbent.
Furthermore, the positive value of ΔH 0 and ΔG 0 values in the range of 0-20 kJ/mol, suggesting that the adsorption process was endothermic and physio-sorption [56]. A positive value of ΔS 0 revealed the randomness in the solid-solution interface increased when DCF molecules adsorbed onto the PBG surface.

The influence of humic acid
Humic acid (HA) is the natural organic substances that most frequently existing organic compound in water [57]. It is composed of abundant organic functional groups such as phenolic, hydroxyl, carboxylic, amine, and quinone, which may affect the adsorption process of DCF in the aqueous system. The variation of q e in the presence of HA is displayed in Figure 7(b). Obviously, DCF adsorption capacity decreased when humic acid concentration (C HA ) increased. At low concentration (0.005 mg/L), HA has an insignificant effect on the adsorption of DCF by PBG. However, at higher concentrations, it strongly reduced the adsorbed amount of DCF onto PBG, particularly q e decreased by 78.8% (50.5 mg/L) when C HA was 0.03 mg/L. It may because in the aqueous environment, the organic molecular groups of HA dissociated than the negative charges bound could be built, which can interact with other organic components in the system. Thus, the decline of DCF adsorption might result in the formation of solute DCF-HA complexes in aqueous solutions. In addition, DCF is known as a weak acid with a pK a about 4.0 (Table  S1), so its molecules carried a negative charge at the adjusted pH value of the experimental condition (pH≈6.5) while HA is also negatively charged in water. Thus, they can compete during adsorption onto the PBG surface through π-π interactions and thereby leads to the decrease of DCF adsorption capacity of the as-prepared modified biochar.

The influence of co-existing common ions
Like other pharmaceutical products, the nonsteroidal anti-inflammatory drugs frequently exist with the presence of metal ions in water matrixes due to industrial wastewater and other sources of inputs. The effect of metal ions on the removal of DCF by PBG was explored in the presence of Ca 2+ , Mg 2+ , Mn 2+ , Zn 2+ , and Fe 2+ . Experimental results and simulated data fitted by Pseudo-second-order kinetic model are presented in Figure 9(a), and Table S4, respectively. In general, the presence of metal ions hindered the adsorption capacity of PBG towards DCF (for all tested common ions), however, the degree of influence varied with each type of ion. It was reported that divalent metal cations have a strong tendency to be adsorbed by biochar through sorption mechanisms such as electrostatic interaction, cation exchange, surface complexation, and π electron-rich domain on the aromatic structure of biochar and metal precipitation [58]. In a determined concentration of co-existing cations (0.05 M), the reaction of DCF and adsorbent was happening slower, suggesting that competition in the adsorption sites between DCF molecules and coexist ions has occurred. As could be seen, Zn 2+ strongly inhibited the removal of DCF by PBG. The q e at equilibrium was only 106.4 mg/g, corresponding to 45.3% of the amount DCF adsorbed without co-existing metal ion. Two covalent ions Ca 2+ and Fe 2+ had the same effect on the adsorption process, which significantly reduced the value of q e (169.4 mg/g and 173.8 mg/g, respectively). Besides, the adsorption capacity of DCF adsorbed to PBG in the presence of Mn 2+ was highest, compared to Mg 2+ and other metal ions, suggesting it has the lowest competitiveness during adsorption. The removal of DCF from aqueous solution may be influenced by the presence of different anions that are the very common factors in the aquatic system. Here, we explored the effects of Cl − , NO 3 − , CO 3

2-
, SO 4 2and PO 4 3anions as co-solutes on DCF removal by magnetic gelatin modified biochar. As shown in Figure 9(b), DCF adsorption capacity wasn't significantly affected by the presence of investigated anions excluding SO 4 2-. In terms of adsorption rate, kinetic model parameters are presented in Table S4. Among the coexist anions, PO 4 3had a slightly positive influence on adsorption rate (q e = 172.3 mg/g after 10 min reaction), while other anions decelerated slowly this rate. The existence of anions did not occupy the activity surface sites on the PBG surface. This may be mainly attributed to the electrostatic repulsion between the negatively charge PBG surface and the anions. Hence, they had a small effect on the removal of DCF.

Regeneration evaluation
The regeneration and reuse of any adsorbent suggest its application as an efficient and environmentally friendly material. The reusability tests were conducted up to 5 cycles and the acquired results are shown in Figure 9(c). The FTIR spectra of adsorbent PBG (fresh, DCF adsorbed, and recycled ones) are presented in Figure 9(d). Stretching bands at 633, 710, 1089 and 1582 cm −1 were observed for both DCF and DCF adsorbed PBG, verifying the adsorption of DCF onto PBG. After five regeneration cycles, the DCF uptake by PBG on the recycling adsorbent still remained at 170.3 mg/g. The adsorption rate remained steady at approximate 71.5% in the fifth cycle compared with that of the first cycle, indicating that the PBG could be reused for DCF adsorption due to the excellent regeneration performance. Thus, PBG is proposed as a recyclable and easy adsorbent to prepare for the removal of DCF from aqueous phase. Further experiment would be carried with other anti-inflammatory drugs to confirm the advantage characteristics of PBG adsorbent.

Comparison with other adsorbents
As listed in Table 4, adsorbents derived from pine sawdust, waste red bricks, tea waste, and others are good potential adsorbents for the removal of DCF from water and their adsorption capacities are 263.7, 292.3, 315.0 and 136.0 mg/g, respectively. Compared with those materials, the one-step synthesis material prepared in this study showed a very high removal efficiency (95.7%) and significantly higher adsorption capacity (q max = 348.9 mg/ g) at 298 K with relatively low dosage. Thus, the as-prepared material can be considered as an effective adsorbent for the removal of DCF at various environmental applications.
Further adsorption experiments can be employed to answer the question whether PBG presents good adsorption capacity towards other non-steroidal anti-inflammatory drugs as well as pharmaceutical compounds.

Adsorption mechanisms
Based on the combined data of characterisation of PBG as well as the kinetic, isotherm, and thermodynamic, solution pH effect studies, it's possible to suggest a mechanism for adsorption of DCF onto PBG adsorbent. In an adsorption process, the adsorbed quantity of adsorbate onto the adsorbent often depends on the surface area and pores volume if the special interaction sites are not involved [71]. Considering the results of BET characterisation (Table S2) along with the similar pore sizes distribution curves of two adsorbents ( Figure 3) and the big difference in DCF adsorption capacity (Table 1) between the pristine and modified samples, suggesting the presence of some special interactions rather than the simple contribution of surface area. As mentioned above, the pH of solution plays an important role in exploring possible adsorption mechanisms at aqueous phase because the surface charges of adsorbent and adsorbate vary with the change of pH value, thus determine their surface mutual interactions. Figure 9(a) shows the effect of solution pH on the DCF adsorption capacity and zeta potential of the PBG sample. Electrostatic attraction is usually employed to understand the adsorption process of various pharmaceutical compounds from water [48,71]. Considering the point zero charge of PBG (pK pzc = 4.75) and pK a of DCF (4.0) [72], the q e value should be low at pH>4.0 since the possible electrostatic repulsion would occur between positively charged biochar surface and anion contaminants. However, the observed q e is relatively high at a wide range of pH value (2.0 to 6.0), indicating the electrostatic interaction is not the only one mechanism can be used to explain the observation. In addition, the degradation of adsorption capacity with increasing solution pH value may be explained by the increasing negative surface charge of PBG and the number of deprotonated moieties (-COOH and -OH) of DCF. Accordingly, electrostatic repulsion occurred between the negatively charged of PBG surface and negative DCF, resulted in reducing DCF removal efficiency at high pH.
The hydrogen bonding (H-bonding) interactions have been widely applied to interpret the adsorption mechanism of numerous organic contaminants in aqueous phase by various porous adsorbents [73][74][75]. The DCF molecule has three H-bond acceptors and two donors. Besides, the surface of PBG contains the carboxylic acid group (-C(=O)OH), which may interact with DCF molecules through H-bonding interaction. Suppose that DCF is used as H-donor for the H-bonding, the obtain q e could be very low since at solution pH>4 (pK a value of DCF), the DCF molecules are presented in the unionised form and deprotonated. However, experimental results show that DCF adsorption capacity reached a value higher 244.8 mg/g at pH = 5.0 then slightly decreased to 238.6 mg/g at pH = 6.0. Therefore, the contribution of DCF as an H-donor can be eliminated. Accordingly, the PBG can be considered as H-donor due to the phenolic group or carboxylic acid group on its surface. Despite of the q e value was significantly lower at pH>7, the DCF molecules were still absorbed onto the adsorbent (q e = 29.7 mg/g at pH = 10.0) due to the H atom of the phenolic group is stable up to pH≈10. The adsorption mechanism was mainly attributed to H-bonding interaction with H-acceptance and H-donation corresponding to DCF molecules and PBG, respectively.
It has been extensively investigated that adsorption of organic compounds involving phenolic groups induces the formation of π-π* bonds. Moreover, the adsorbent characterisation analysis reveals the presence of -C = O and C = C functional groups, which might act as π-electron donors for the interactions whereas the benzene rings in DCF structure present an electron-acceptor character [76]. Results from Figure 7(c) show that at relatively high pH condition (pH = 11.0), DCF molecules was still absorbed onto PBG surface (q e = 26.9 mg/g), indicating the presence of π-π interaction. Even though, the contribution of this interaction to the adsorption process was not significant since the value of q e was proven to be pH-dependent but the number of π-electrons of DCF and PBG doesn't change with pH value. XPS spectra of PBG before and after adsorption were studied to further confirm the interaction mechanisms discussed above. Figure 6(a, b, c, d) present high-resolution XPS spectra of C 1s, O 1s, N 1s and Fe 2p of PBG after adsorption, respectively. As could be seen, after DCF adsorption, the C-C/C = C (284.5 eV) showed a shift to 284.9 eV, further confirming that π-π might be responsible for DCF removal. In addition, for the N 1s peaks of PBG after adsorption (Figure 6(c)), the peak at 400.1 and 400.7 eV related to C-N/N-O and -NH 2 groups after DCF loaded on PBG also transferred to 399.8 and 400.4 eV, respectively [77]. These changes indicated that the new bond (−CO−NH−) was formed and it might pose a significant effect on the adsorption of DCF. Figures 5(b) and 6(b) show O 1s spectra of PBG before and after adsorption. As could be observed, the positions of all decomposed O 1s peaks shifted and the change in ratio of different O 1s spices after adsorbed DCF. It was probably because oxygen-containing functional groups (such as −COOH and −OH) of PBG reacted with thy hydroxyl and carboxyl groups of DCF through H-bonding [78]. Meanwhile, the peaks of Fe 2p spectra were insignificantly changed after reaction, suggesting the ironloaded on the adsorbent was not reduced during the reaction [79].
Based on the discussion above, the possible adsorption mechanism can be depicted in Figure 10.

Conclusion
In this study, a novel adsorbent was successfully synthesised by activating peanut shell biochar using magnetic gelatin solution. The as-prepared material was applied for the removal of DCF from aqueous solution. The maximum adsorption capacity (q max ) of PBG for DCF was 348.9 mg/g at 298 K, which was the highest value of DCF adsorption compared to that of other given adsorbents from Table 4. Results of adsorbent characterisation indicated the addition of magnetic gelatin supported the physiochemical properties and remarkably enhanced the oxygen-containing groups (as confirmed by XPS and FTIR). The DCF removal by PBG was relatively stable at pH in a range of 2.0 to 6.0 and increased inversely with pH>7.0, indicating that adsorption capacity dependents on solution pH value. The simulated data of kinetic and isotherm studies were evaluated using different error functions including R 2 , R 2 adj , SSE, χ 2 and ARE. The results indicated that the experimental data can be well described by the pseudo-second-order model and Temkin isotherm model, respectively. The parameters obtained from the thermodynamic study suggested that the adsorption process was non-spontaneous and endothermic. The coexist metal ions including Ca 2+ , Mg 2+ , Mn 2+ , Zn 2+ , and Fe 2+ negatively affected the removal of DCF by competing for adsorption sites on the PBG surface and the order of influence can be suggested as Zn 2+ >Fe 2+ ≈Ca 2+ >Mg 2+ ≈Mn 2+ . Whereas the addition of co-solute anions including Cl − , NO 3 − , CO 3 2-, and PO 4 3had little influence on the adsorption process. Besides, q e significantly decreased with an increase in humic acid concentration. After five times recycling, the adsorption capacity of PBG for DCF removal was reduced about 28.5%, which indicated a high stability and reusability of PBG. Furthermore, the large adsorption affinity of PBG for DCF might be mainly attributed to hydrogen bonds, following by electrostatic interactions and π-π interactions. Considering the simple fabrication approach and the excellent adsorption performance towards sodium diclofenac-one of the most widely used nonsteroidal anti-inflammatory drug over the world, the PBG adsorbent can be evaluated as a potential material for environmental remediation applications.