Removal of 17 β -Estradiol by Activated Charcoal Supported Titanate Nanotubes (TNTs@AC) through Initial Adsorption and Subsequent Photo-Degradation: Intermediates, DFT calculation, and Mechanisms

: A low-cost composite of activated charcoal supported titanate nanotubes (TNTs@AC) was developed via the facile hydrothermal method to remove the 17 β -estradiol (E2, a model of pharmaceutical and personal care products) in water matrix by initial adsorption and subsequent photo-degradation. Characterizations indicated that the modiﬁcation occurred, i.e., the titanate nanotubes would be grafted onto the activated charcoal (AC) surface, and the micro-carbon could modify the tubular structure of TNTs. E2 was rapidly adsorbed onto TNTs@AC, and the uptake reached 1.87 mg / g from the dual-mode model ﬁtting. Subsequently, the adsorbed E2 could be degraded 99.8% within 2 h under ultraviolet (UV) light irradiation. TNTs@AC was attributed with a unique hybrid structure, providing the hydrophobic e ﬀ ect, π − π interaction, and capillary condensation for E2 adsorption, and facilitating the electron transfer and then enhancing photocatalytic ability for E2-degradation. In addition, the removal mechanism of E2 was elucidated through the density functional theory calculation. Our study is expected to provide a promising material for environmental application. and W.L.; data curation, H.J.; writing—original draft preparation, H.J.; writing—review and editing, B.P., W.L., and H.J.; visualization, T.H.; supervision, H.J.; funding acquisition, W.L.


Synthesis of TNTs@AC
TNTs@AC was typically prepared via a facile one-step hydrothermal method from our previous study [26]. In brief, 1.2 g P25-type TiO 2 and 2.4 g AC were dispersed into 10 mol/L NaOH solution (66.7 mL DIW), followed by the overnight stirring i.e., 12 h at room temperature. Then, the mixed dispersion was put into a Teflon-made reactor and heated for 72 h at 130 • C. After cooling to room temperature, the separated black precipitates were fully washed using DIW to remove residual NaOH till close to neutral pH. After oven-drying at 80 • C for 4 h, TNTs@AC was ready. For comparison, pristine TNTs and treated AC were synthesized via a similar procedure without AC or TNTs, respectively.

Experimental Methods: Adsorption Kinetic and Isotherm, Photoregeneration
Adsorption kinetic tests were conducted in the dark using amber glass vials (43 mL, Teflon-lined caps) with E2 solutions (40 mL, 0.5 or 1 mg/L, pH 7.0 ± 0.2). Then, the kinetic tests were triggered by adding 0.02 g TNTs@AC, i.e., 0.5 g/L, into the respective E2 solutions and shaken at 200 rpm. Then, each of the 1 mL samples was obtained at pre-designed intervals and membrane-filtered via 0.22 µm polytetrafluoroethylene (PTFE) immediately. The filtrates were then analyzed for E2 concentration.
Adsorption isotherm tests were conducted in a similar fashion as for kinetics with various initial E2 concentrations (0.05-1.5 mg/L) and a fixed material dosage (0.5 g/L) at pH 7. The mixtures were equilibrated for 24 h under dark. For comparison, the adsorption isotherms by AC and pristine TNTs were also obtained.
The photocatalytic ability of TNTs@AC was evaluated under UV irradiation with 0.5 g/L material and at pH 7.0. Upon adsorption equilibrium, the suspension was gravity-settled for 1 h. After removing 90% supernatant, the residuals were put into a glass photoreactor with a quartz-made cover. Then, the reaction was started-up under UV irradiation (1.42 mW/cm 2 , 365 nm). At predetermined times, i.e., 0, 10, 20, 30, 60, 90, 120 min, the samples were collected and extracted by 20 mL methanol (80 • C for 4 h) to analyze the E2 concentration.
All experiments were conducted in duplicate. The experimental results were plotted by means of duplicates. In addition, error bars referred to relative deviation to the mean, indicating the reproducibility of data.

Characterizations
To investigate the morphology of TNTs@AC, the high-resolution transmission electron microscopy (HRTEM) was performed on Tecnai30 microscopy (FEI, USA) at 300 kV. To verify the crystalline phase and structure, X-ray diffraction (XRD) was conducted on a D/max-2500 Powder X-ray diffractometer (Rigaku, Japan) with a scanning step of 0.02 at 40 mA and 100 kV. For zeta potentials, Nano-ZS90 Zetasizer (Malvern Instruments, UK) was applied.

Analytical Methods
E2 was analyzed on an Agilent 1100 liquid chromatography system (HPLC, USA) equipped with a 3.5 µm SB-C18 column (4.6 × 150 mm). The detector wavelength was placed at 210 nm. The mobile phase consisted of 60% acetonitrile and 40% phosphoric acid solution (0.085%, v/v) with a 1 mL/min flow rate, giving the detection limit of~1 µg/L.
Photocatalytic degradation intermediates/byproducts of E2 were detected on an Agilent HP 1100 LC-MS n Trap system (USA) with a Zorbax RX-C18 column.

DFT Calculation Method
All the computational calculations based on density functional theory (DFT) were conducted on the Gaussian 16 software (Version C.01, Wallingford, CT, USA) [27]. The optimization of geometry and electrostatic potential energy (ESP) of E2 were calculated at the level of B3LYP/6-31+G(d,p) [28][29][30]. Details on the optimized geometry structure and energy of E2 are listed in Figure S1 and Table S1 Supplementary materials of (SM). Visual molecular dynamics (VMD 1.9.1) was used to render and generate the colored iso-surface map of ESP [31,32]. Figure 1 displays the HRTEM images of TNTs@AC. As shown in our previous report by Liu et al. [26], neat activated charcoal was in bulk with the flat surface. TNTs exhibited hollow multilayered nanotubes with open-top structures, which have the inner diameter as 4.5 nm (Figure 1).

Material Characterizations
After the hydrothermal treatment with P25-TiO 2 , the formed titanate nanotubes would be grafted onto the AC surface, as shown in Figure 1a, which is consistent with the other report [26]. In addition, the tiny portion in nanoscale of AC, i.e., micro-carbon, could modify the tubular structure of TNTs (Figure 1b), which transfers the original open-end structure of the pristine TNTs as the narrow throat.   /g). However, the total pore volume for TNTs@AC (0.52 cm 3 /g) is much lower than that of TNTs (1.26 cm 3 /g), indicating that the micro-carbon may coat on the narrow throat of nanotubes [33]. In addition, the pHPZC values are derived as 6.8, 2.6, and 3.1 for AC, TNTs, and TNTs@AC [26], respectively. Figure 2 presents the adsorption kinetics of E2 by TNTs@AC. The removal efficiency of E2 reaches 99.8% and 91.5% for the initial E2 concentration of 0.5 and 1.0 mg/L, respectively. For both cases, the adsorption equilibrium reaches within 360 min.

Adsorption Kinetics of E2 by TNTs@AC
The equilibrium E2 uptake (qe, mg/g) and removal efficiency (E, %) were calculated as follows: where C0, Ce, V, and m are the initial or equilibrium E2 concentrations (mg/L) in the aqueous phase, the solution volume (L), and the materials mass, respectively. Pseudo-first-order and pseudo-second-order kinetic models were able to interpret the experimental results [12,[34][35][36][37], as described below: Pseudo-first-order model: Pseudo-second-order model:  /g). However, the total pore volume for TNTs@AC (0.52 cm 3 /g) is much lower than that of TNTs (1.26 cm 3 /g), indicating that the micro-carbon may coat on the narrow throat of nanotubes [33]. In addition, the pH PZC values are derived as 6.8, 2.6, and 3.1 for AC, TNTs, and TNTs@AC [26], respectively. Figure 2 presents the adsorption kinetics of E2 by TNTs@AC. The removal efficiency of E2 reaches 99.8% and 91.5% for the initial E2 concentration of 0.5 and 1.0 mg/L, respectively. For both cases, the adsorption equilibrium reaches within 360 min.

Adsorption Kinetics of E2 by TNTs@AC
The equilibrium E2 uptake (q e , mg/g) and removal efficiency (E, %) were calculated as follows: where C 0 , C e , V, and m are the initial or equilibrium E2 concentrations (mg/L) in the aqueous phase, the solution volume (L), and the materials mass, respectively. Pseudo-first-order and pseudo-second-order kinetic models were able to interpret the experimental results [12,[34][35][36][37], as described below: Pseudo-first-order model: Pseudo-second-order model: where q t and q e are the E2 uptakes (mg/g) at time t (min) and equilibrium, respectively; k 1 (min −1 ) and k 2 (g/(mg·min)) are the respective rate constants. Table S3 (Supplementary Material) lists the fitted parameters of pseudo-first-order and pseudo-second-order kinetic models for E2 adsorption onto TNTs@AC. The pseudo-second-order model provides better-fitted results (R 2 > 0.998). The corresponding calculated equilibrium adsorption capacity (q e,cal ) for initial E2 concentration of 0.5 and 1.0 mg/L is 1.01 and 2.01 mg/g, respectively, which is close to the experimental results (q e,exp = 0.999 and 1.908 mg/g, respectively). where qt and qe are the E2 uptakes (mg/g) at time t (min) and equilibrium, respectively; k1 (min −1 ) and k2 (g/(mg•min)) are the respective rate constants. Table S3 (Supplementary Material) lists the fitted parameters of pseudo-first-order and pseudosecond-order kinetic models for E2 adsorption onto TNTs@AC. The pseudo-second-order model provides better-fitted results (R 2 > 0.998). The corresponding calculated equilibrium adsorption capacity (qe,cal) for initial E2 concentration of 0.5 and 1.0 mg/L is 1.01 and 2.01 mg/g, respectively, which is close to the experimental results (qe,exp = 0.999 and 1.908 mg/g, respectively).  Figure 3a presents the adsorption isotherms of E2 by AC, TNTs, and TNTs@AC in the E2 concentration range from 0.1 to 1.5 mg/L. It is clearly shown that TNTs@AC offer the highest adsorption E2 uptake. In addition, TNTs exhibit the negligent adsorption capacity for E2. The linear, Langmuir, Freundlich, and Dual-mode isotherm models were applied to simulate the experimental results as follows [26,[38][39][40]:

Adsorption Isotherms of E2 by Various Materials
Langmuir model: Freundlich model: Dual-mode model: where qe and Ce: the solid and aqueous phase E2 concentrations (mg/g), respectively, at equilibrium, Kd: distribution coefficient (L/g), KF: Freundlich capacity constant (mg/g• (L/mg) 1/n ), n: Freundlich intensity constant, b: Langmuir affinity coefficient (L/mg); QL or Qmax: maximum Langmuir adsorption capacity (mg/g).  Figure 3a presents the adsorption isotherms of E2 by AC, TNTs, and TNTs@AC in the E2 concentration range from 0.1 to 1.5 mg/L. It is clearly shown that TNTs@AC offer the highest adsorption E2 uptake. In addition, TNTs exhibit the negligent adsorption capacity for E2. The linear, Langmuir, Freundlich, and Dual-mode isotherm models were applied to simulate the experimental results as follows [26,[38][39][40]:

Adsorption Isotherms of E2 by Various Materials
Linear model: Langmuir model: Freundlich model: Dual-mode model: where q e and C e : the solid and aqueous phase E2 concentrations (mg/g), respectively, at equilibrium, K d : distribution coefficient (L/g), K F : Freundlich capacity constant (mg/g·(L/mg) 1/n ), n: Freundlich intensity constant, b: Langmuir affinity coefficient (L/mg); Q L or Q max : maximum Langmuir adsorption capacity (mg/g).  Table 1 lists the fitted parameters of various isotherm models for E2 adsorption onto different materials. The dual-model isotherm model offers the best-fitted results (R 2 ≥ 0.995) than the other three classic isotherm models. The maximum adsorption capacity (QL) of three materials follows the order: TNTs@AC (1.44 mg/g) > AC (0.89 mg/g) > TNTs (0.16 mg/g), which agree with the BET surface area (Table S2, Supplementary Material). For AC, the hydrophobic effect and π−π interaction are the dominant mechanisms for E2 adsorption [12]. The strong hydrophobic E2 molecule [5] is difficult to be adsorbed onto TNTs, and thus the low E2 uptake is attributed to the electrostatic attraction between neutral E2 molecules and negatively charged TNTs (Table S2, Supplementary Material). However, for TNTs@AC composites, not only AC surface was modified by TNTs, but TNTs were attached by micro-carbon. Thus, the higher uptake of E2 by TNTs@AC than neat AC and TNTs was reached via the hydrophobic and π−π interactions contributed by modified AC and capillary condensation from the micro-carbon-modified TNTs [15,26].   Table 1 lists the fitted parameters of various isotherm models for E2 adsorption onto different materials. The dual-model isotherm model offers the best-fitted results (R 2 ≥ 0.995) than the other three classic isotherm models. The maximum adsorption capacity (Q L ) of three materials follows the order: TNTs@AC (1.44 mg/g) > AC (0.89 mg/g) > TNTs (0.16 mg/g), which agree with the BET surface area (Table S2, Supplementary Material). For AC, the hydrophobic effect and π−π interaction are the dominant mechanisms for E2 adsorption [12]. The strong hydrophobic E2 molecule [5] is difficult to be adsorbed onto TNTs, and thus the low E2 uptake is attributed to the electrostatic attraction between neutral E2 molecules and negatively charged TNTs (Table S2, Supplementary Material). However, for TNTs@AC composites, not only AC surface was modified by TNTs, but TNTs were attached by micro-carbon. Thus, the higher uptake of E2 by TNTs@AC than neat AC and TNTs was reached via the hydrophobic and π−π interactions contributed by modified AC and capillary condensation from the micro-carbon-modified TNTs [15,26]. To further explore the adsorption mechanism of E2 onto TNTs@AC via the capillary condensation, the initial E2 concentration is increased from 0.25 to 3.0 mg/L, as shown in Figure 3b. The fitted parameters by various isotherm models are listed in Table 2. It is evidently showed that there is a two-stage isotherm profile, i.e., Langmuir adsorption at lower C e (<0.06 mg/L) following the linear uptake at higher C e , which is similar with the observation by Liu et al. [26], considering the dual-mode adsorption, i.e., Langmuir-type adsorption and capillary condensation. In addition, the maximum adsorption capacity (Q L ) is increased to 1.87 mg/g at high E2 concentration ( Table 2) than 1.44 mg/g at low E2 concentration (Figure 3a), indicating TNTs@AC could adsorb E2 not only by surface hydrophobic and π−π interactions but also by the capillary condensation. The detailed E2 adsorption mechanism onto TNTs@AC is discussed in Section 3.5.  Figure 4 shows that the photo-degradation of adsorbed E2 by AC could be negligible (<10%), and the mass loss could be attributed to the extraction method. However, the adsorbed E2 on TNTs@AC is significantly degraded, i.e., 99.8% and 98.7%, under UV irradiation within 120 min for initial E2 concentration of 0.5 and 1.0 mg/L, respectively, indicating the TNTs@AC composites exhibited the excellent photocatalytic activity. The details on the E2 degradation mechanism and enhanced photocatalysis for TNTs@AC are elucidated in Section 3.5.

Photoregeneration of TNTs@AC
Water 2020, 12, x 7 of 12 To further explore the adsorption mechanism of E2 onto TNTs@AC via the capillary condensation, the initial E2 concentration is increased from 0.25 to 3.0 mg/L, as shown in Figure 3b. The fitted parameters by various isotherm models are listed in Table 2. It is evidently showed that there is a two-stage isotherm profile, i.e., Langmuir adsorption at lower Ce (<0.06 mg/L) following the linear uptake at higher Ce, which is similar with the observation by Liu et al. [26], considering the dual-mode adsorption, i.e., Langmuir-type adsorption and capillary condensation. In addition, the maximum adsorption capacity (QL) is increased to 1.87 mg/g at high E2 concentration ( Table 2) than 1.44 mg/g at low E2 concentration (Figure 3a), indicating TNTs@AC could adsorb E2 not only by surface hydrophobic and π−π interactions but also by the capillary condensation. The detailed E2 adsorption mechanism onto TNTs@AC is discussed in Section 3.5.  Figure 4 shows that the photo-degradation of adsorbed E2 by AC could be negligible (<10%), and the mass loss could be attributed to the extraction method. However, the adsorbed E2 on TNTs@AC is significantly degraded, i.e., 99.8% and 98.7%, under UV irradiation within 120 min for initial E2 concentration of 0.5 and 1.0 mg/L, respectively, indicating the TNTs@AC composites exhibited the excellent photocatalytic activity. The details on the E2 degradation mechanism and enhanced photocatalysis for TNTs@AC are elucidated in Section 3.5.   Figure 5 elucidates the removal, i.e., adsorption and photo-degradation, mechanisms of E2 by TNTs@AC. For the E2 uptake onto TNTs@AC, both TNTs and AC play significant roles in the adsorption. First, E2 molecules have high hydrophobicity, i.e., K owv = 4.01 [5], resulting in E2 that is favorable to attach on the AC surface via the hydrophobic and π−π interactions [12,41]. Second, based on the theoretical calculation of electrostatic potential (ESP) energy, the red sides have the highest ESP value to 61 kcal/mol (Figure 6a), which can be attracted to the negatively charged TNTs via the weak electrostatic attraction. Third, from the Figure 6b, the three dimensions of E2 molecules are 7.87, 14.43, and 7.64 Å, which is much smaller than the inner diameter of TNTs, i.e., 4.5 nm [42] (Figure 1), indicating the E2 molecules can enter the tubular structure, leading to the capillary condensation by micro-carbon-modified TNTs (Figure 3). Thus, the adsorption capacity of E2 was significantly enhanced for TNTs@AC (Figure 3).

Removal Mechanism of E2 by TNTs@AC
Water 2020, 12, x 8 of 12 Figure 5 elucidates the removal, i.e., adsorption and photo-degradation, mechanisms of E2 by TNTs@AC. For the E2 uptake onto TNTs@AC, both TNTs and AC play significant roles in the adsorption. First, E2 molecules have high hydrophobicity, i.e., Kow=4.01 [5], resulting in E2 that is favorable to a ach on the AC surface via the hydrophobic and π−π interactions [12,41]. Second, based on the theoretical calculation of electrostatic potential (ESP) energy, the red sides have the highest ESP value to 61 kcal/mol (Figure 6a), which can be attracted to the negatively charged TNTs via the weak electrostatic attraction. Third, from the Figure 6b, the three dimensions of E2 molecules are 7.87, 14.43, and 7.64 Å, which is much smaller than the inner diameter of TNTs, i.e., 4.5 nm [42] (Figure 1), indicating the E2 molecules can enter the tubular structure, leading to the capillary condensation by micro-carbon-modified TNTs (Figure 3). Thus, the adsorption capacity of E2 was significantly enhanced for TNTs@AC (Figure 3).  For the photo-degradation of adsorbed E2 by TNTs@AC, under the UV irradiation, TNTs can generate the valence band (VB) and conduction band (CB) ( Figure 5). The photo-induced electrons (eTi − ) and holes (hTi + ) are produced in CB and VB, respectively (Equation (9)). Then, the generated holes can directly oxidize the water or hydroxide ion to produce hydroxyl radical (•OH) (Equations (10) and 11). Meanwhile, upon the surface modification, the coated micro-carbon and grafted AC can act as the new electron acceptor to transfer photo-generated electrons to AC/micro-carbon phase, then inhibit the electron-hole recombination (Equation (12)). The oxygen molecule will obtain the electron from AC or micro-carbon phase to produce the superoxide radical (•O2 − ) (Equation (13)). In addition, •OH also can be derived from •O2 − , based on Equation 14. The generated radicals and h + are charge  Figure 5 elucidates the removal, i.e., adsorption and photo-degradation, mechanisms of E2 by TNTs@AC. For the E2 uptake onto TNTs@AC, both TNTs and AC play significant roles in the adsorption. First, E2 molecules have high hydrophobicity, i.e., Kow=4.01 [5], resulting in E2 that is favorable to a ach on the AC surface via the hydrophobic and π−π interactions [12,41]. Second, based on the theoretical calculation of electrostatic potential (ESP) energy, the red sides have the highest ESP value to 61 kcal/mol (Figure 6a), which can be attracted to the negatively charged TNTs via the weak electrostatic attraction. Third, from the Figure 6b, the three dimensions of E2 molecules are 7.87, 14.43, and 7.64 Å, which is much smaller than the inner diameter of TNTs, i.e., 4.5 nm [42] (Figure 1), indicating the E2 molecules can enter the tubular structure, leading to the capillary condensation by micro-carbon-modified TNTs (Figure 3). Thus, the adsorption capacity of E2 was significantly enhanced for TNTs@AC (Figure 3).  For the photo-degradation of adsorbed E2 by TNTs@AC, under the UV irradiation, TNTs can generate the valence band (VB) and conduction band (CB) ( Figure 5). The photo-induced electrons (eTi − ) and holes (hTi + ) are produced in CB and VB, respectively (Equation (9)). Then, the generated holes can directly oxidize the water or hydroxide ion to produce hydroxyl radical (•OH) (Equations (10) and 11). Meanwhile, upon the surface modification, the coated micro-carbon and grafted AC can act as the new electron acceptor to transfer photo-generated electrons to AC/micro-carbon phase, then inhibit the electron-hole recombination (Equation (12)). The oxygen molecule will obtain the electron from AC or micro-carbon phase to produce the superoxide radical (•O2 − ) (Equation (13)). In addition, •OH also can be derived from •O2 − , based on Equation 14. The generated radicals and h + are charge For the photo-degradation of adsorbed E2 by TNTs@AC, under the UV irradiation, TNTs can generate the valence band (VB) and conduction band (CB) ( Figure 5). The photo-induced electrons (e Ti − ) and holes (h Ti + ) are produced in CB and VB, respectively (Equation (9)). Then, the generated holes can directly oxidize the water or hydroxide ion to produce hydroxyl radical (•OH) (Equations (10) and (11)). Meanwhile, upon the surface modification, the coated micro-carbon and grafted AC can act as the new electron acceptor to transfer photo-generated electrons to AC/micro-carbon phase, then inhibit the electron-hole recombination (Equation (12)). The oxygen molecule will obtain the electron from AC or micro-carbon phase to produce the superoxide radical (•O 2 − ) (Equation (13)).

Removal Mechanism of E2 by TNTs@AC
In addition, •OH also can be derived from •O 2 − , based on Equation 14. The generated radicals and h + are charge of E2 degradation (Equations (15) and (16)), which perform the photo-regeneration for TNTs@AC under UV irradiation.

Conclusions
In this study, low-cost activated charcoal supported titanate nanotubes material via the one-step facial hydrothermal method was prepared to remove E2 through initial adsorption and subsequent photo-degradation. TNTs@AC exhibited good adsorption capacity for E2 (1.44 mg/g at low E2 Photo-degradation pathway of adsorbed E2 by AC and TNTs@AC. (Initial E2 concentration = 0.5 or 1.0 mg/L, material dosage = 0.5 g/L, temperature = 25 ± 0.2 • C, volume of solution = 40 mL, solution pH = 7.0 ± 0.2).

Conclusions
In this study, low-cost activated charcoal supported titanate nanotubes material via the one-step facial hydrothermal method was prepared to remove E2 through initial adsorption and subsequent photo-degradation. TNTs@AC exhibited good adsorption capacity for E2 (1.44 mg/g at low E2 concentration and 1.87 mg/g at high E2 concentration) based on the dual-mode model fitting due to the modification for both AC and TNTs and simultaneously performed excellent photocatalytic degradation under UV irradiation (99.8% degradation of adsorbed E2 within 2 h). HRTEM images indicated that the formed titanate nanotubes would be grafted onto the AC surface, and the micro-carbon could modify the tubular structure of TNTs, which build the well-designed structure to facilitate the adsorption and electron transfer. High E2 uptake was achieved through hydrophobic effect, π−π interaction, and capillary condensation. In addition, based on the DFT theoretical calculation, the benzene ring side of the E2 molecule had the highest ESP value as 61 kcal/mol, which favored attaching onto materials. The three dimensions of E2 molecules, i.e., 7.87, 14.43, and 7.64 Å, less than the inner diameter of TNTs, i.e., 4.5 nm, guaranteeing the occurrence of the capillary condensation. Moreover, the coated micro-carbon and grafted AC can act as the new electron acceptor to transfer photo-generated electrons to AC/micro-carbon phase, then inhibit the electron-hole recombination. Therefore, TNTs@AC is a promising material for E2 removal and could provide the potential application via easy regeneration.