Sorptive removal of phenolic endocrine disruptors by functionalized biochar: competitive interaction mechanism, removal efficacy and application in wastewater

Abstract Sorptive removal of six phenolic endocrine disrupting chemicals (EDCs) estrone (E1), 17β-estradiol (E2), estriol (E3), 17α-ethynylestradiol (EE2), bisphenol A (BPA) and 4-tert-butylphenol (4tBP) by functionalized biochar (fBC) through competitive interactions was investigated. EDC sorption was pH dependent with the maximum sorption at pH 3.0–3.5 due to hydrogen bonds and π-π interactions as the principal sorptive mechanism. Sorption isotherm of the EDCs was fitted to the Langmuir model. Sorption capacities and distribution coefficient values followed the order E1 > E2 ≥ EE2 > BPA > 4tBP > E3. The findings suggested that EDC sorption occurred mainly through pseudo-second order and external mass transfer diffusion processes, by forming H-bonds along with π-π electron-donor–acceptor (EDA) interactions at different pH. The complete removal of ∼500 μg L−1 of each EDC from different water decreased in the order: deionised water > membrane bioreactor (MBR) sewage effluent > synthetic wastewater. The presence of sodium lauryl sulphonate and acacia gum in synthetic wastewater significantly suppressed sorption affinity of EDCs by 38–50%, hence requiring more fBC to maintain removal efficacy.


Introduction
Endocrine disrupting chemicals (EDCs) can cause adverse effects due to exogenous endocrine disruption in the reproductive, sexual differentiation, neurological and immune systems even at low concentrations (ng L -1 to µg L -1 ), and have attracted increasing attention [1][2]. Phenolic structure based compounds such as the natural estrogens 17β-estradiol (E2), estriol (E3) and estrone (E1), synthetic estrogen 17α-ethynylestradiol (EE2), and industrial compounds such as octylphenol, nonylphenol, 4-tert-butyl phenol (4tBP) and bisphenol A (BPA), are the most potent EDCs. EDCs are poorly removed in sewage treatment plants [2][3][4][5] and are a primary source of their discharge and occurrence in surface water, groundwater, seawater and sediment [4,6]. EDCs are relatively hydrophobic organic compounds according to their octanol-water partition coefficient (K ow ) values and have only one pK a value. The chemical structures and physicochemical properties of the EDCs are shown in Fig. 1 and Table 1, respectively.
Many separation processes such as coagulation, flocculation and precipitation have been used for the removal of EDCs from different water [3,5]. Conventional biological processes such as activated sludge, constructed wetlands, bio-filtration have shown limited removal of EDCs [3,5] while advanced treatment processes such as granular activated previously [22]. Briefly, biochar was activated by soaking in 50% orthophosphoric acid (oH 3 PO 4 ) with the impregnation ratio 1:1 (w/v, taking oH 3 PO 4 as 100%) for 3 h at 50 °C followed by heating at 600 °C for another 2 h. Prepared material was cooled in the reactor, washed several times with distilled water, and the pH was adjusted to ~7, followed by drying overnight at 100 °C. After activation, ~15-20% weight loss (activated burn off) was observed.
Average particle size of activated biochar was in the range 75-600 µm. As the activated biochar was enriched with different functional groups such as -COOH, -OH, C=O and C=C on its surface, (based on X-ray Photoelectron Spectroscopy (XPS) characterization) the prepared activated biochar was termed as fBC.

EDCs sorption on fBC in different water
Competitive sorption experiments of EDCs on fBC were conducted in 50 mL glass vials with Teflon-lined screw caps at 25 o C in duplicate. EDCs were dissolved in methanol to prepare the stock solution of 1.0 g L -1 . The final methanol content in the sorption system was <0.5% (v/v) to avoid the co-solvent effect. To study the effect of pH, the interaction of preequilibrated fBC (20 mL of solution at the same solution pH) with EDC mixture in solution (25 mL) was carried out at different pH values (1.86, 3.1, 4.0, 5. 1, 5.96, 7.85, 9.0 and 10.85) for 42 h at 25 o C. The initial concentration of each EDC in the mixture was adjusted to ~500 µg L -1 . Competitive sorption isotherm and kinetics experiments of EDCs in duplicate were also performed on an orbital shaker at 110 rpm, 25 o C for 48 h at pH ~3.0-3.25. Constant ionic strength was maintained using 0.01 M NaCl. The solid phase sorption (q s , µg g -1 ) and sorption distribution coefficient (K d , L kg -1 ) were calculated. The initial concentrations of each EDC were in the range ~250 µg L -1 to ~3000 µg L -1 in the mixture solution. The control experiments without sorbents were also performed. The sorbent dosage was selected for 15 to 95% sorption of each EDC at different concentrations. After equilibrium, the pH was measured and the solution was filtered through a 0.2 µm PTFE filter and analyzed by high-performance liquid chromatography (HPLC). Raw biochar showed low removal efficiency hence, the discussion on the results obtained with unmodified biochar is not presented here.
MBR effluent was filtered (1.2 µm) before being stored at 4 o C. MBR effluent and synthetic wastewaters were spiked with ~500 µg L -1 of each EDC in the mixture before interaction with fBC for 44 to 64 h at pH 3.0-3.25, (maximum sorption-based on the pH study) at 25 o C.
Different dosages of fBC were used to study the removal of EDCs in competitive mode. The concentrations of the target EDCs in the MBR effluent were lower than the limit of detection (LOD), hence EDCs were spiked to MBR effluent. Synthetic wastewater and MBR effluent were used to examine the effect of different constitutes (organics and inorganics) present in these waters, on the removal of EDCs. The physicochemical properties of MBR sewage effluent and synthetic wastewater are listed in Table S1.

Characterization of fBC
Microscopic analysis of fBC before and after EDC sorption was carried out using scanning electron microscope (SEM) (Zeiss Evo-SEM). Brunauer-Emmett-Teller (BET) nitrogen adsorption-desorption isotherms and Barrett-Joyner-Halenda (BJH) method were used to calculate specific surface area and the porosity of fBC using a Micromeritics 3 FlexTM surface characterization analyzer at 77 K. The bulk elemental analysis (C, H, O, N, P) of fBC was determined using Oxford energy dispersive X-ray spectroscopy (EDS) and XPS (Thermo Scientific, UK). A Renishaw inVia Raman spectrometer (Gloucestershire, UK) equipped with a Leica DMLB microscope (Wetzlar, Germany) and a 17 mW Renishaw helium neon laser source at 633 nm (with 50% of the laser intensity) with CCD array detector was used for Raman spectroscopy. The fBC was added to 0.01 M KCl solution at eight different pH values for 46 hours to measure its zeta potential (Nano-ZS, Malvern).

Analyses of EDCs
EDCs were analyzed by HPLC with an auto-sampler and a reverse-phase Zorbax Bonus RP C 18 column (5.0 µm, 2.1 1.50 mm, Agilent Technologies), with an injection volume of 100 µL. Mobile phase "A" was composed of acetonitrile and formic acid (99.9: 0.1) while mobile phase "B" consisted of Milli-Q water and formic acid (99.9: 0.1). The elution used 40% of "A" and 60% of "B" at a flow rate of 0.4 mL min -1 , which was changed to 0.3 mL min -1 at 0.1 min and maintained until 9.0 min. After 9 min, the flow rate was changed to 0.2 mL min -1 and maintained up to 24 min. EDCs were analyzed with a UV detector at 285 nm and a fluorescence detector at 280 nm (excitation wavelength) and 310 nm (excitation wavelength).
Fluorescence and UV wavelength were kept unchanged throughout the analysis. The method LOD for each EDC is given in Table S2.

Modeling of sorption kinetics, thermodynamics and isotherms
The sorption data were fitted to four kinetic models, namely pseudo first-order (PFO), pseudo second-order (PSO), the Weber−Morris intra-particle diffusion model (IDM) and the external mass transfer models as follow: where K i is the apparent diffusion rate constant (g µg -1 min -1/2 ), q t (µg g -1 ) is the sorbed mass at time t, q s (µg g -1 ) is the equilibrium sorbent mass, K 1 (min -1 ) is the PFO kinetic rate constant, K 2 (µg g -1 min -1 ) is the PSO kinetic rate constant, and C is a constant (µg g -1 ) that provides the thickness of the boundary layer [22].C 0 and C t (µg L -1 ) represent the concentrations of EDCs in solution at the beginning and at time t, respectively, β (cm min -1 ) is the external mass transfer coefficient, and S (cm -1 ) is the specific surface of fBC for external mass transfer. βS value was calculated from the slope of the C t /C 0 versus t plot [5].
Gibbs free energy (∆G°, kJ mol -1 ) of EDCs sorption onto fBC at 25 0 C was estimated with different concentrations using equation 5 [5]: where K d (L kg -1 ) is the apparent individual sorption distribution coefficient and can be defined by the ratio of sorbed EDC concentration (q s , µg g -1 ) to aqueous EDC concentration (C w , µg L -1 ), using equation 6: where V (L) is the solution volume, and M (g) is the sorbent mass.
The sorption data were also fitted to the Langmuir isotherm model which is represented below [23]: where q max is the maximum adsorption capacity (µg g -1 ) and K L is the Langmuir fitting parameter (L µg -1 ). Parameters were estimated by nonlinear regression weighted by the dependent variable.
The sorption data obtained were used to calculate the total competitive sorption capacities using summarized Langmuir sorption (equations 8 and 9). Since all the competitive solutes interact and compete for the same sorption sites in fBC, the overall maximum sorption capacity of fBC can be estimated by the additive contribution of each EDC's maximum sorption capacity [22].
where q max1 , q max2 , q max3 , q max4 , q max5 and q max6 are the maximum Langmuir sorption capacity for E1, E2, E3, EE2, BPA and 4tBP, respectively. q max.(total) is the total maximum Langmuir sorption capacity of the fBC for EDC mixtures, estimated by the summation of individual EDC's maximum sorption capacities.

Characterization of fBC and EDCs
The structure of the carbon network in fBC was analysed by SEM, BET, XPS, FTIR and Raman spectra. SEM images of fBC after sorption experiments showed the development of flakes like structure on the fBC surface (Fig. S1). Fig. 2a represents the nitrogen adsorptiondesorption isotherm plot of fBC. This clearly indicates that isotherm plot was found to be type II isotherm. This isotherm features N 2 -uptake increment up to a relative pressure of 0.5 then slightly reduced followed by an increment of N 2 adsorption (relative pressure up to 1.0) to reach a plateau. This result suggested the existence of mesopore (2-50 nm) and macrospore (> 50 nm) structure of fBC and the isotherm indicate unrestricted monolayer-multilayer adsorption [24,25]. More clearly, the arrow (in Fig.2a) point, the beginning of the most linear middle section of the isotherm indicating the point at which monolayer coverage is complete and multilayer adsorption is about to begin. On the other hand, Fig. 2b shows the presence of mesopores (~2-50 nm) in fBC core structure which has been calculated from the adsorption branch of the isotherm by the BJH method [26]. Thus, the fBC mostly contained meso and macroporous structures and tis indicated that the activation method did not lead to the development of microporous (< 2.0 nm) structures. Also, BET and Langmuir surface area were found to be 1.18 and 8.22 m 2 g -1 , respectively which was much lower than the reported values [27]. The BJH adsorption cumulative surface area of pores was also found to be 1.37 m 2 g -1 and Dubinin-Astakov micropore surface area was found to be 0.52 m 2 g -1 ( Table S3).
The Raman spectroscopy on fBC showed two characteristic peaks at 1341 and 1588 cm -1 (Fig. 3), which correspond to the D-band (disordered structure) and G-band (graphitic structures) of sp2-type carbon present in fBC [22,28]. The relative band intensity ratio ( is 1.04 (> 1), demonstrated the functionalization in fBC. It is important to note that Raman spectra of fBC featured a strong D-band, which illustrated a slightly more amorphous character (disordered) of the carbon in fBC owing to more oxygenated functional groups on its structure as functionalization of biochar was carried out using acid. D-band surface defect was possible by the introduction of other elements onto carbon structure during fBC preparation.
XPS results indicated that fBC was rich in different functional groups especially -C=C-, -C-O and -O-C=O [29]. Carbon (C1s) spectra of fBC showed that fBC surface was composed of aromatic carbon mostly -C=C-(284.8 eV) due to long chain arene unit, -C-O (286.3 eV), -C=O (287.8 eV), -COOH (289 eV) and π-π* (292.35 eV) due to functional groups ( Fig. 4a). Also, fBC surface contained oxygenated and phosphorous-based functional groups/complexes. O1s spectra showed that oxygen content was mostly in the form of organic carbon (at 533.3 and 531.62 eV) (Fig. 4b) [30]. The P2p XPS spectrum of fBC showed a peak at 133.79 eV in the form pentavalent tetra coordinated phosphorus (PO 4 i.e. C-O-PO 3 ), as in polyphosphates and/or phosphates (Fig. 4c) [31]. The survey peaks showed the same results ( Fig. 4d). The elemental composition of fBC was found to be 81.76% C, 13.32% O, 0.8% N and 2.3% P from XPS.

Effect of pH on competitive sorption of EDCs
The effect of pH for competitive EDCs' sorption against the solid-phase concentration of each EDC is shown in Fig. 5a. The sorption of individual EDC by fBC was highly pH dependent and was found to be moderate at very low pH ~1.85 (where fBC became positive). At this pH, q s values of all EDCs were found to be low, which might be due to the repulsion between the positively charged fBC (zeta potential value was positive) and protonated EDCs (Fig. 5b).
However, the EDC sorption at this pH might be due to the electron-donor-acceptor (EDA) interactions between oppositely charged arene units. Increase in pH from 1.85 to 3.5 increased the sorption capacity significantly. The q max values of individual EDCs were ~4110, ~3356, ~3333, ~3350, ~2765 and ~2725 µg g -1 for E1, E2, EE2, BPA, 4tBP and E3, respectively. The maximum sorption of EDCs could be due to EDA interactions along with strong hydrogen bonds formation [23,32]. Further increase in pH from 3.5 to pH 5.0 led to a significant reduction of the sorption of each EDC. However, when pH was increased above 5.0, another high q s value for EDC was observed at pH near 8.0. This was due to pK a values (pH = pK a + log[salt/acid]) of each EDC and surface hydroxyl groups on fBC, which were responsible for the formation of strong hydrogen bonds together with EDA interactions with the fBC surface functional groups. Further pH increases up to 10.85 caused a decrease of q s value of each EDC because of the highly repulsion between the negatively charged fBC and the EDCs.
Hydrogen bond formations, as well as EDA interactions, were not strong as solution pH was above the pK a values of the EDCs. The E1 sorption was the highest among the EDCs studied.
This was mainly due to the presence of C=O functional group on its structure which might help in EDA interactions. Thus, q max values for all EDCs in mixture mode were obtained at pH 3.0-3.5 where significant EDC interactions with fBC occurred. Detailed sorption mechanism at different pH is discussed in section 3.5.

EDC sorption kinetics
The kinetics of the competitive sorption of EDCs onto fBC is shown in Fig. 6 Table 2).
To further evaluate the competitive sorption, the kinetic data were fitted with external mass transfer and intraparticle diffusion models as they play a major role in sorption process ( Fig. 6, Table 2). The third step was very slow, thus, could not be treated as a rate-accelerating step. In addition, linear regression coefficient values were significant (r 2 > 0.90). The linear regression fittings for individual EDC did not pass through the origin, i.e. deviating from the origin or near saturation. This might be due to the difference in the mass transfer rate of the EDCs in the initial and final stages of sorption. From intercept (C) data it was found that the intra-particle diffusion was not the sole rate-limiting step [5]. The external mass transfer also played an important role in controlling the sorption rate. The regression coefficients (r 2 > 0.950) of all EDCs showed that the competitive sorption of EDCs could be represented by the external mass transfer model (Fig. 6b).
The Boyd plot (equation 10) was used to explore whether intra-particle and external mass transfer processes exerted any significant influence on the sorption rate of EDCs [5]: From the Boyd plot, it can be observed that none of the sorption data lines pass thorough the origin (Fig. 6c), indicating that the external mass transfer governed the sorption of EDCs on fBC. This finding is consistent with the previous study for BPA sorption using MWCNT where the external mass transfer was solely responsible for the sorption of BPA [5]. Thus, the sorption kinetics of EDCs could be described by PSO and external mass transfer models.

Competitive sorption of EDCs and Gibbs free energy
The variations of solid phase concentration of EDCs with aqueous equilibrium concentration are represented in Fig. 7 attributed to the functionalization of the biochar, which resulted in the formation of additional sorption sites with increased functional groups and an increase in specific surface area and micro pore volume [22]. The observed interactions were comparable with reports for sorption of various solutes by different carbon nanomaterials such as SWCNT, MWCNT and fullerene [15]. For example, carbon nanomaterials were applied for the sorption of BPA and EE2 (seperately) and it was observed that the solid phase concentrations (1×10 4 -1×10 5 µg g -1 ) were within the range observed in this study for competitive sorption, if the same initial EDC concentration was maintained [15]. This implies that mixture of EDCs did not have adverse effect on the sorption of individual EDC using fBC. In comparison, limestone sediment was found less effective for the sorption interactions with BPA, E2, EE2, 4-tert-octylphenol and 4n-nonylphenol, with the Freundlich constant values being significantly lower with prolonged interaction time than in this study [2]. Ying et al. [2] found the concentrations in sediment were 45, 70, 80 and 1750 µg kg -1 for BPA, E2, EE2, 4-tert-octylphenol, respectively.
Similarly, lower Langmuir and Freundlich isotherm parameters were reported for BPA and EE2 interactions with sewage sludge [33]. Lorphensri et al. [21] also found EE2 removal using alumina, silica and a hydrophobic medium (porapak P) was significantly lower than in this study. Furthermore, there was a significant linear relationship between log K d and log C w (r 2 > 0.98) (Fig. S3). Increase in EDCs concentration reduced K d value of each EDC which resulted in a decrease in removal efficiency. This was due to the gradual saturation of fBC surface at higher solute concentrations, leading to lower removal efficiencies.
The sorption spontaneity of EDCs at different concentrations (~250 to 3000 µg L -1 ) onto fBC was examined and Gibbs free energy was calculated ( values showed the spontaneous nature of the sorption of EDCs onto fBC.

Competitive sorption mechanism
Interaction mechanism of each EDC with fBC can be explained based on experimental findings. fBC core structure consisted of meso-and macro-pores. The BET surface area and Dubinin-Astakov micropore surface area were 1.183 and 0.516 m² g -1 , respectively. BJH adsorption average pore diameter of fBC was ~8.0 nm which might enough for diffusion of EDCs onto fBC pores as the apparent molecular size of EDCs molecules was well below the average pore diameter of fBC (Table 1). However, from the kinetics study, it was found that both external mass transfer (sorption by active sites) and PSO (multilayer sorption) model parameters satisfactory described sorption kinetic behaviour of EDCs on fBC, suggesting the role of surface functional groups present in fBC for EDCs sorption. The mechanism of EDCs sorption by surface functional groups of fBC is described below at different pH conditions. Varying pH can affect the protonation-deprotonation transition of functional groups on any carbonaceous materials and change the chemical speciation for ionisable organic compounds [5]. The zeta potential values of fBC suspension in aqueous solution at various pH were determined, and point of zero charge was found to be 2.2 (Fig. 5b). Lower sorption of any compound at this pH was highly expected. At pH below 2.2, the zeta potential value of fBC was found to be the positive indicating surface of fBC was protonated under highly acidic conditions. EDCs molecules might also become protonated at pH 1.85 (e.g. due to hydroxyl or ketonic groups in EDCs). Hence, electrostatic repulsion of the same charge of quadruples might lead to lower sorption of EDCs at pH < 2.2. However, highly acidic condition (at pH 1.85) was still favourable to sorb EDCs by fBC to some extent. The EDA interactions can explain the sorption of EDCs at this pH. Chemical structures of all EDCs contain at least one hydroxyl group in arene unit, i.e. phenolic group, and due to resonanceeffect of arene unit, nearby carbon atoms in hydroxyl group (in arene unit) could act as π-electron donor site (Figs 1 and S4). Meanwhile, fBC consisted of ketonic and carboxylic -

C=O functional groups in its arene units (as confirmed by XPS) which might act as π-electron
acceptor site for the interactions (due to resonance) and the graphene unit in fBC can act as a π-electron donor site. Thus, stronger EDA interactions (from C=O and COOH as π-electron acceptor while EDCs as a π-electron donor) would be the main reason for the sorption of all EDCs at this pH (Fig. S4). The π-π electron-donor-donor (EDD) interactions between phenolic -OH of EDCs and surface -OH group and graphene unit of fBC are not significant, and considered weaker than EDA interactions. Thus, EDD interactions at very low pH can be excluded. The interactions of fBC and EDCs can also be predicted from pH shift tests before and after adsorption experiments. The variation of solution pH is shown in Fig. 5b. The equilibrium pH slightly increased after sorption (from 1.86 to 2.09) indicating either release of hydroxyl ions or consumption of proton either by fBC or EDCs. Moreover, hydroxyl ions might exchange proton in the solution for neutralization leading to increase the solution pH.
However, proton exchange for fBC at pH ~1.85 is not favourable for -C=O, -COOH and -OH groups as their pK a values are higher. Hence, surface -OH groups or quaternary nitrogen groups (tiny fraction) of fBC as well as EDCs functional groups such as -OH and -CH might form hydrogen bond along with EDA interactions, i.e. excess hydrogen ions took part in hydrogen bond formations hence equilibrium pH was increased. Therefore, sorption at pH ~1.85 was observed due to EDA interactions and hydrogen bond formations.
The maximum solid phase concentration and K d values of all EDCs were observed at pH 3.0-3.5 (Fig. 5a), and the main reason might be due to the formation of strong hydrogen bonds and strong EDA interactions. Based on the pH shift study, equilibrium pH shifted slightly to higher pH (pH 3.10-4.0 to 3. 25-4.18). The increase in equilibrium pH indicated hydroxyl groups released into solution, which was more significant in the control experiments than sorption experiment (Fig. 5b). Although the pH shift seemed less significant in the sorption experiments, the release of protons from fBC functional groups (e.g. -COOH) and then forming strong hydrogen bonds (-COOH in fBC and phenolic -OH in EDCs) might be the main reason for resisting equilibrium pH to decrease (Fig. 5b). mechanism. The surface -COOH has pK a value of ~3.0-5.0 [22,34,35]. Functional groups -COOH and C=O could act as strong π-electron acceptor while the phenolic group in EDCs and graphene unit of fBC could act as π-electron donor site (Fig. S4). Thus, stronger EDA interactions could be the main sorption mechanism. EDD interactions might not be significant as graphene surface could act as π-electron donor site as well as EDCs could act as π-electron donor. Hence, EDA interactions were the main mechanism for higher sorption of EDCs with additional contribution from the CAHB formations. Zhang et al. [5] studied the separation of BPA using MWCNT and proposed the π-π stacking interactions between the bulk π-system of MWCNT surface and BPA molecule in a wide pH range of 4.0 to 10.0 [5]. They also mentioned that MWCNT could function as hydrogen bond donor to form hydrogen bonds with -OH functional groups on BPA. However, they did not mention any specific group in MWCNT responsible for π-π stacking interactions, while our observations indicated that π-π stacking interactions (EDA) only came from the surface -COOH/C=O and -C=C-groups.
Also, Jung et al. [11] studied the adsorption of emerging contaminants such as EE2, BPA, sulfamethoxazole, atrazine, carbamazepine, diclofenac and ibuprofen using activated biochar at different pH (3.5, 7.0, 10.5) and suggested that maximum adsorption was mainly due to π-π interactions. Therefore, maximum interactions might be due to the formation of negative CAHB as well as strong hydrogen bonds, with the main contribution from EDA interactions leading to the higher interaction of EDCs with fBC.
With the increase in pH from 3.0 to 5.0, the solid phase concentration of all EDCs decreased. When pH was increased from 5.0 to 8.0, another peak interaction value for all EDCs was found although not as high as at pH ~3.0-3.5. All the EDCs have one pK a value.
So, near pH 8-9 proton release from any EDC is highly possible as pH= pK a + log [salt/acid].
On the other hand, any carbonaceous surface -OH group has pK a value of ~8.0-10.0 [5,34].
Proton exchange by EDCs molecules was calculated and found their ∆G o values were favourable for proton exchange in solution at this condition (  (Fig. 5b) [35][36][37]. In comparison with control experimental equilibrium pH, the change of equilibrium pH was also significant indicating the release of protons from surface -OH groups of fBC (Fig. 5b). The π-π EDA interactions could also play an important role under this condition. However, EDA interaction might only come from fBC surface -C=O groups (π-electron acceptor), graphene unit of fBC (as a π-electron donor) and EDCs (as a π-electron donor) whereas, EDA interactions might not be effective for the surface carboxylic group (pK a value is near pH 3.0-5.0) (Fig. S 4) [35][36][37]. EDD interactions might be less effective. Furthermore, at pH > 9.0, each EDC may exist as an anion, and the sorption was significantly impeded due to the electrostatic repulsive force between negatively charged fBC surface and EDCs anions. The ionized forms of EDCs were the predominant fraction at pH > pK a , and the hydrogen bonds and hydrophobic interactions between fBC and each ionized EDC were much weaker than those between fBC and nonionized EDCs. Besides, both EDC and fBC were negatively charged, and the electrostatic repulsion between them can also weaken their adsorption to some extent at pH > 10. Hence, lower sorption at pH above 9.0 was observed. In addition, the pH shift result was not significant at pH 10.85 indicating the insignificant role of the hydrogen bond formation.

EDC removal from sewage effluent vs. synthetic wastewater
When fBC was applied to remove a mixture of EDCs from MBR sewage effluent (spiked at 508.4, 525.9, 532.9, 534.5, 465.8, 460.5 µg L -1 of E3, BPA, 4tBP, E2, E1 and EE2, respectively) at pH 3.0-3.25 and 25 o C, pronounced differences were observed for different EDCs in terms of their complete sorption onto fBC (Fig. 8). Synthetic wastewater also contains a mixture of different organic acids, organic compounds and inorganic salts (

Conclusions
In this work fBC with enhanced functional groups, specific surface area, and meso-and macro-poreswas successfully prepared for the removal of EDC mixture from water and wastewater. The sorptive removal of EDCs by fBC through competitive interactions reached equilibrium within 42 h, with the external mass transfer diffusion process as the rate-limiting step. PSO well modeled the sorption kinetics. EDC sorption was highly pH dependent, with the maximum sorption occurring at pH ~3.0-3.5. The sorption equilibrium followed the Langmuir isotherm model, suggesting monolayer coverage. In term of sorption mechanism, EDC sorption mainly occurred through π-π EDA interactions and by forming different hydrogen bonds. Overall, the sorption capacity and distribution coefficient values decreased as E1 > E2 ≥ EE2 > BPA > 4tBP > E3 due to the difference in the EDCs' hydrophobicity.
Water composition had a pronounced effect on EDC removal, as shown by the highest removal in deionised water, followed by MBR sewage effluent, and finally synthetic wastewater. The presence of sodium lauryl sulphonate and acacia gum in synthetic wastewater caused a significant reduction in the competitive sorption of EDCs on fBC. Thus, fBC can be successfully applied for the removal of EDC mixtures from water and wastewater, although appropriate pre-treatment may be required to remove the interfering substances such as certain surfactants.         Highlights fBC removed ~100% of EDC mixture from water and wastewater.
Sorption in wastewater was reduced by 38-50% due to sodium laryl sulphonate.
π-π EDA interactions with H-bond formation were main sorption mechanism.