Experimental and Theoretical Study of Sorption Capacity of Hexagonal Boron Nitride Nanoparticles: Implication for Wastewater Purification from Antibiotics

The constant accumulation of antibiotics and their degradation products in wastewater as a result of human activity poses a serious threat to humanity and other living beings. To contribute to solving this important problem, hollow hexagonal boron nitride nanoparticles (BNNPs) with a spherical shape and smooth surface were synthesized, which were characterized as an efficient adsorbent for wastewater treatment from three types of antibiotics: ciprofloxacin (CIP), tetracycline (TC), and benzylpenicillin (BP). As follows from DFT calculations, the interaction of antibiotic molecules (AM) with the BN surface is neither purely physical nor purely chemical, and negative binding energy (BE) indicates that the adsorption process is spontaneous and endothermic. The calculated electron density redistributions at the AM/BN interfaces show that antibiotics interact with BN mainly through oxygen-containing groups. In addition, this interaction causes the BN surface to bend, which increases both the BE and the contact area. The removal efficiency of antibiotics (Re, %) depends on their initial concentration. At an initial concentration of 10 µg/mL, Re50 and Re100 were observed after 24 h and 14 days, respectively. With an increase in the initial concentration to 40 μg/mL, Re50 and Re100 were achieved after 5 and 28 days (with the exception of ciprofloxacin (~80% Re)). The maximum sorption capacity of BNNPs (qe) was determined to be 297.3 mg/g (TC), 254.8 mg/g (BP), and 238.2 mg/g (CIP), which is significantly superior to many other systems. Tetracycline is adsorbed much faster than the other two antibiotics, which is confirmed by both theoretical and experimental data. Based on the results of the DFT analysis, a simple and efficient sorbent regeneration strategy was proposed, which ensures complete removal of antibiotics after 14 (BP), 21 (TC), and 10 (CIP) days. Thus, the obtained results clearly show that BNNPs are promising sorbents for various classes of antibiotics, including aminoglycosides, tetracyclines, and β-lactams.


Introduction
The uncontrolled, ever-growing accumulation of antibiotics and their residues in the environment is an acute modern problem. Their presence in water and soil is a potential hazard to the environment, humans, and other living beings. Many therapeutic agents are not completely metabolized, which leads to the penetration of active drug molecules into the biological environment, the emergence of new contamination sources, and the wide spread of bacteria and microorganisms with multidrug resistance [1][2][3]. Modern wastewater facilities do not allow efficient removal of antibiotic residues from the environment [4,5], which leads to their accumulation in ecological systems [6,7]. Global studies of river pollution with antibiotics have shown that 65% of surveyed rivers in 72 countries on 6 continents are contaminated with antibiotics [8]. In some rivers, the concentrations were so high that they posed a real danger to both the ecosystem and human health. Therefore, the development of effective approaches to the removal of antibiotics from the aquatic environment is of great importance.

Structural Characterization of BNNPs
The BNNP morphology was analyzed by scanning electron microscopy (SEM) on a JSM-7600F (JEOL, Tokyo, Japan) instrument equipped with an energy dispersive Xray (EDX) detector (Oxford Instruments, High Wycombe, UK). The microstructure and phase composition of BNNPs were studied using an FEI Tecnai G2 Spirit Twin transmission electron microscope (Thermo Fisher Scientific, New York, NY, USA) operated at an accelerated voltage of 120 kV and an Ultima IV X-ray diffractometer (Rigaku, Wilmington, MA, USA). Chemical bonds after the antibiotic absorption on the BN surface were determined by Fourier-transformed infrared (FTIR) spectroscopy in the total reflection mode in the range of 400-4000 cm −1 with a resolution of 4 cm −1 . The specific surface area of the samples was measured by low-temperature nitrogen adsorption at 77 K on a Nova 1200e instrument (Quantachrome, Boynton Beach, FL, USA). Before measurements, the samples were degassed at 150 • C. The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) theory.

UV-Vis-Spectrophotometry
Antibiotic concentrations were measured on a Flame UV-Vis spectrophotometer (Ocean Optics, Orlando, FL, USA) in the wavelength range from 190 to 500 nm. A calibration curve was plotted at different antibiotic concentrations (from 0.5 to 4000 µg/mL). The antibiotic concentration in the BNNP solution at each time point was compared with the calibration curve. The optical density (λ max, nm) was also taken into account at an antibiotic concentration of 1 mg/mL. Drug release experiments were carried out in triplicate for each type of antibiotic [39][40][41][42][43].
Each type of antibiotic was completely dissolved in deionized water to prepare a stock solution, which was further diluted to obtain working solutions at antibiotic concentrations of 10, 20, and 40 µg/mL. Then, 50 mg of BNNPs were added to 10 mL of each antibiotic solution and immediately subjected to sonication to obtain a homogeneous suspension. Adsorption tests were carried out at room temperature. Blank experiments were also carried out in the absence of adsorbents to ensure that there were no impurities from the walls of the glass flask.

BNNP Purification from Adsorbed Antibiotics
A mixture of acetonitrile and isopropanol has been shown to be effective in purifying NPs from organic pollutants [44]. In this work, we have developed and studied the following three different methods for purifying BNNPs from antibiotics: Purification was carried out for 28 days. The concentrations of antibiotics in solution on days 1, 3,5,7,9,11,14,21, and 28 were determined using a UV spectrophotometer. The obtained spectra were integrated and compared with the concentration curve. The experiments were carried out in triplicate, and the obtained data were averaged.

Sorption Kinetic Analysis
To plot adsorption kinetic curves, 2 mL of the supernatant was taken at certain time intervals (8 h, 1, 3, 5, 7, 11, 14, 21, and 28 days) and then the antibiotic concentrations were determined by measuring absorbance values of the solutions using UV-visible spectrophotometry.
The antibiotic removal efficiency (Re, %) was calculated using the following equation: where C 0 and C t are antibiotic concentrations at the initial and current (t) time points, respectively (mg/L). The adsorption capacity (q e , mg/g) of the BNNPs was determined after they were kept in an antibiotic solution at a concentration of 1 mg/mL for 56 days.
Value q e was calculated according to the following equation [45]: where C 0 and C e are initial and equilibrium antibiotic concentrations, respectively (mg/L); V-volume of antibiotic solution (L); W-amount of adsorbent (g). Although the antibiotic concentration in wastewater is quite low, in this work we considered concentrations of 10, 20, and 40 µg/mL. All experiments were performed in triplicate.

DFT Calculations of Antibiotic Adsorption on BN Surface
Theoretical analysis of atomic structure and stability of the antibiotic-BN nanosystems was performed using density functional theory (DFT) [46,47] within the generalized gradient approximation (GGA) using normalized Trullier-Martins [48] pseudopotentials in the SIESTA software package [49]. The systems were modeled as supercells with a sufficiently large vacuum gap (at least 20 Å) to neglect intermolecular interactions in the non-periodic direction. The plane wave energy cutoff was set to 200 Ry. To calculate the equilibrium atomic structures, the Brillouin zone was chosen according to the Monkhorst-Pack scheme [50] and selected according to the size of the unit cell from 8 × 8 × 1 to 2 × 2 × 1. To calculate the electronic properties, the cutoff of k-grid was 16 × 16 × 1.
Although the DFT method is widely used to calculate electronic structure, it poorly describes the strength of dispersion and the van der Waals interactions, which can regulate the physical absorption process. Therefore, for modeling antibiotic-BN systems, the Grimme correction method (DFT-D method) [51] was used.

Characterization of BN Nanoparticles
SEM and TEM micrographs of the as-synthesized BNNPs are shown in Figure 1a,b. Hollow BNNPs has a spherical shape with a thin shell (shell thickness of 30-60 nm) and a smooth surface. The size of BNNPs ranges from 100 to 400 nm, while most NPs (>70%) are from 200 to 300 nm in size. Figure 1c shows the FTIR spectrum of BNNPs. Two high-intensity peaks can be attributed to out-of-plane B-N-B bending (780 cm −1 ) and in-plane B-N stretching (1370 cm −1 ) vibrations [52]. A small peak at 520 cm −1 and a shoulder at 1500 cm −1 correspond to the B 2 O 3 and O-B-O bonds [53]. The XRD pattern of BNNPs is presented in Figure 1d are from 200 to 300 nm in size. Figure 1c shows the FTIR spectrum of BNNPs. Two high-intensity peaks can be attributed to out-of-plane B-N-B bending (780 cm -1 ) and in-plane B-N stretching (1370 cm -1 ) vibrations [52]. A small peak at 520 cm -1 and a shoulder at 1500 cm -1 correspond to the B2O3 and O-B-O bonds [53]. The XRD pattern of BNNPs is presented in Figure 1d. In addition to the main peaks from the (002), (100), (101), and (004) crystallographic h-BN planes (ICDD card No. 00-034-0421), there are additional maxima corresponding to BNO (ICDD card No. 00-37-1234) and B2O3 (ICDD card No. 00-06-0297) phases. Thus, the oxidized state of BNNPs observed in the XRD pattern ( Figure 1d) is in good agreement with the FTIR spectroscopy data (Figure 1c). The specific surface area was calculated using the BET model. The specific surface area of nanoparticles with several BN layers, determined by the N2 adsorption method, was 33.6 m 2 /g.

Interaction Mechanisms of Antibiotic Molecules with BN Surface: Theoretical Insight
The optimal positions of antibiotics relative to the BN surface were theoretically studied. Due to the large size of BNNPs, BN was considered an infinite sheet. First, the preferred positions of antibiotics on the BN surface were determined. For this, six positions were considered for each antibiotic type, corresponding to the main functional groups of antibiotic molecules ( Figure 2, Table 1). The specific surface area was calculated using the BET model. The specific surface area of nanoparticles with several BN layers, determined by the N 2 adsorption method, was 33.6 m 2 /g.

Interaction Mechanisms of Antibiotic Molecules with BN Surface: Theoretical Insight
The optimal positions of antibiotics relative to the BN surface were theoretically studied. Due to the large size of BNNPs, BN was considered an infinite sheet. First, the preferred positions of antibiotics on the BN surface were determined. For this, six positions were considered for each antibiotic type, corresponding to the main functional groups of antibiotic molecules ( Figure 2, Table 1).
As shown in Figure 2a, ciprofloxacin (CIP) contains the following configuration: carboxylic group (P1), cyclopropyl group (P2), pipyrazine ring (P3), fluorine group (P4), oxo group (P5), and a parallel arrangement of the dihydroquinoline ring-vertical stacking (P6). In tetracycline, dimethylamino group (P1), a hydroxyl group, including the simultaneous interaction of both amino groups (P2), amide group (P3), the peripheral region of anthracycline (P4), as well as parallel arrangement with OH groups oriented from (P5) or to (P6) BN plane were considered (Figure 2b). In the case of benzylpenicillin, the following positions were identified: phenylacetyl group (P1), amino group (P2), thioether group (P3),  group (P5), and a parallel arrangement of the dihydroquinoline ring-vertical stacking (P6). In tetracycline, dimethylamino group (P1), a hydroxyl group, including the simultaneous interaction of both amino groups (P2), amide group (P3), the peripheral region of anthracycline (P4), as well as parallel arrangement with OH groups oriented from (P5) or to (P6) BN plane were considered ( Figure 2b). In the case of benzylpenicillin, the following positions were identified: phenylacetyl group (P1), amino group (P2), thioether group (P3), carboxylic group (P4), and two parallel arrangements with the S group oriented from (P5) or to (P6) BN plane ( Figure 2c).  For each position, the binding energy (BE) was calculated as follows: where Etot is the total energy of the system, EBNNPs and EAB are the energies of a freestanding substrate and an antibiotic molecule, respectively. The BE values for antibiotics with different charges on the h-BN surface are present in Table 2. Negatively charged forms of antibiotics (absence of a proton on the carboxylic group) are marked as "−1", positively charged forms (an additional proton on the amino groups) are denoted as "+1" and uncharged forms of antibiotics are marked as 0. The most stable position for all antibiotic types is a vertical arrangement, regardless of the molecule charge. In the case of tetracycline, the vertical stacking positions P5 and P6 show rather similar BEs (difference up to 1 eV). For benzopenicillin, the most favorable structure is P5 (the thio group is oriented from the BN plane) for the negatively charged form and P6 (the thio group is oriented to the BN plane) for the neutral form of benzopenicillin. Structures P2 and P3 also have rather low BEs. It should be noted that the structure of the positively charged benzylpenicillin was not considered in the simulation since it is unstable and is destroyed during optimization. All negatively charged forms of the considered antibiotics (absence of a proton in the -COO-groups) bind more strongly to the surface of BN than neutral or positively charged forms.
For all considered antibiotics, their BE with BN is negative and adsorption clusters are more stable than freestanding structures; therefore, adsorption on BN is an exothermic and spontaneous process. The distance between the antibiotic (AB) molecule and the BN surface is more than 2.4 Å. For each position, the binding energy (BE) was calculated as follows: where E tot is the total energy of the system, E BNNPs and E AB are the energies of a freestanding substrate and an antibiotic molecule, respectively. The BE values for antibiotics with different charges on the h-BN surface are present in Table 2. Negatively charged forms of antibiotics (absence of a proton on the carboxylic group) are marked as "−1", positively charged forms (an additional proton on the amino groups) are denoted as "+1" and uncharged forms of antibiotics are marked as 0. The most stable position for all antibiotic types is a vertical arrangement, regardless of the molecule charge. In the case of tetracycline, the vertical stacking positions P5 and P6 show rather similar BEs (difference up to 1 eV). For benzopenicillin, the most favorable structure is P5 (the thio group is oriented from the BN plane) for the negatively charged form and P6 (the thio group is oriented to the BN plane) for the neutral form of benzopenicillin. Structures P2 and P3 also have rather low BEs. It should be noted that the structure of the positively charged benzylpenicillin was not considered in the simulation since it is unstable and is destroyed during optimization. All negatively charged forms of the considered antibiotics (absence of a proton in the -COO-groups) bind more strongly to the surface of BN than neutral or positively charged forms.
For all considered antibiotics, their BE with BN is negative and adsorption clusters are more stable than freestanding structures; therefore, adsorption on BN is an exothermic and spontaneous process. The distance between the antibiotic (AB) molecule and the BN surface is more than 2.4 Å.
However, the BE is too high for normal physical interaction. To determine the type of interaction in the BN@AB hybrid for each antibiotic type, the electron density redistri- bution between the entire system (ρ total BNNPs@AB ) and each individual part (ρ BN and ρ AB for freestanding h-BN and antibiotic molecule, respectively) was calculated as follows: The structures of BN@AB after geometry optimization and redistribution of electron density near the BN surface for the neutral form of antibiotics are presented in Figure 3.
The obtained results show that the interaction of AB molecules with the BN surface is neither purely physical nor purely chemical. This interaction changes the geometric parameters of the BN surface ( Figure 3). As a result of the interaction of AB molecules with functional groups on the BN surface, the electron density is redistributed, which causes the BN surface to bend. The formation of a concave on the BN surface leads to an increase in both the contact area and the BE.
The electron density redistribution shows that antibiotics interact with BN mainly through oxygen-containing groups. In the area of -COOH, hydroxyl, and oxo groups, the charge density is transferred to the BN surface, while for other groups (fluorine for CIP, sulfur for BP, and nitrogen-containing groups), the charge is redistributed only on the AB molecule. This explains the higher BE of tetracycline compared to benzylpenicillin and ciprofloxacin, which contain a large number of oxygen-containing groups. However, the BE is too high for normal physical interaction. To determine the type of interaction in the BN@AB hybrid for each antibiotic type, the electron density redistribution between the entire system ( BNNPs@AB ) and each individual part ( BN and AB for freestanding h-BN and antibiotic molecule, respectively) was calculated as follows: The structures of BN@AB after geometry optimization and redistribution of electron density near the BN surface for the neutral form of antibiotics are presented in Figure 3.
The obtained results show that the interaction of AB molecules with the BN surface is neither purely physical nor purely chemical. This interaction changes the geometric parameters of the BN surface ( Figure 3). As a result of the interaction of AB molecules with functional groups on the BN surface, the electron density is redistributed, which causes the BN surface to bend. The formation of a concave on the BN surface leads to an increase in both the contact area and the BE.
The electron density redistribution shows that antibiotics interact with BN mainly through oxygen-containing groups. In the area of -COOH, hydroxyl, and oxo groups, the charge density is transferred to the BN surface, while for other groups (fluorine for CIP, sulfur for BP, and nitrogen-containing groups), the charge is redistributed only on the AB molecule. This explains the higher BE of tetracycline compared to benzylpenicillin and ciprofloxacin, which contain a large number of oxygen-containing groups.  Electron density redistribution at the BN/AB interface. The loss and gain of charge are denoted by yellowish and bluish colors, respectively. The boron, nitrogen, carbon, oxygen, and hydrogen atoms are marked by green, gray, brawn, red, and beige colors, respectively. The isosurface constant value is 0.01 eV/Å .

Kinetics of Antibiotic Adsorption on BNNPs
First, the absorption spectra of antibiotics were obtained at a maximum concentration of 200 µg/mL in the wavelength range from 190 to 500 nm using UV-visible spectrophotometry ( Figure 4). Absorption peaks were observed at wavelengths of 288 and First, the absorption spectra of antibiotics were obtained at a maximum concentration of 200 μg/mL in the wavelength range from 190 to 500 nm using UV-visible spectrophotometry ( Figure 4). Absorption peaks were observed at wavelengths of 288 and 300 nm (benzylpenicillin), 276 and 353 nm (tetracycline), and 288, 270, 323, and 333 nm (ciprofloxacin).  The kinetics of change in the antibiotic concentration in the BN suspension was studied for three different initial concentrations (10, 20, and 40 µ g/mL) at a temperature of 24°C and pH = 6 for 28 days ( Figure 5). The amount of antibiotic adsorbed on the surface of BNNPs per unit time depended on the solution's initial concentration as follows: the higher the initial antibiotic concentration, the lower the adsorption rate. When testing benzylpenicillin and tetracycline at initial concentrations of 10 and 20 µ g/mL, 50% of the antibiotic was adsorbed in the first 2 days. At a concentration of 40 µ g/mL, the adsorption time increased to 4 days. In the case of ciprofloxacin, 50% of the antibiotic was adsorbed on days 2, 4, and 6 at concentrations of 10, 20, and 40 µ g/mL, respectively. At an initial concentration of 10 μg/mL, 100% adsorption was observed on the 9th (tetracycline) and 11th (benzylpenicillin and ciprofloxacin) days. With an increase in the initial concentration to 20 μg/mL, the time for complete adsorption increased to 14 (tetracycline) and 21 (benzylpenicillin and ciprofloxacin) days. At the maximum studied concentration of 40 µ g/mL, only tetracycline and benzylpenicillin showed 100% adsorption on the 28th day, while with the use of ciprofloxacin, the sorption process continued.  The kinetics of change in the antibiotic concentration in the BN suspension was studied for three different initial concentrations (10, 20, and 40 µg/mL) at a temperature of 24 • C and pH = 6 for 28 days ( Figure 5). The amount of antibiotic adsorbed on the surface of BNNPs per unit time depended on the solution's initial concentration as follows: the higher the initial antibiotic concentration, the lower the adsorption rate. When testing benzylpenicillin and tetracycline at initial concentrations of 10 and 20 µg/mL, 50% of the antibiotic was adsorbed in the first 2 days. At a concentration of 40 µg/mL, the adsorption time increased to 4 days. In the case of ciprofloxacin, 50% of the antibiotic was adsorbed on days 2, 4, and 6 at concentrations of 10, 20, and 40 µg/mL, respectively. At an initial concentration of 10 µg/mL, 100% adsorption was observed on the 9th (tetracycline) and 11th (benzylpenicillin and ciprofloxacin) days. With an increase in the initial concentration to 20 µg/mL, the time for complete adsorption increased to 14 (tetracycline) and 21 (benzylpenicillin and ciprofloxacin) days. At the maximum studied concentration of 40 µg/mL, only tetracycline and benzylpenicillin showed 100% adsorption on the 28th day, while with the use of ciprofloxacin, the sorption process continued.
of 200 μg/mL in the wavelength range from 190 to 500 nm using UV-visible spectropho tometry (Figure 4). Absorption peaks were observed at wavelengths of 288 and 300 nm (benzylpenicillin), 276 and 353 nm (tetracycline), and 288, 270, 323, and 333 nm (ciproflox acin).  The kinetics of change in the antibiotic concentration in the BN suspension was stud ied for three different initial concentrations (10, 20, and 40 µ g/mL) at a temperature o 24°C and pH = 6 for 28 days ( Figure 5). The amount of antibiotic adsorbed on the surface of BNNPs per unit time depended on the solution's initial concentration as follows: the higher the initial antibiotic concentration, the lower the adsorption rate. When testing ben zylpenicillin and tetracycline at initial concentrations of 10 and 20 µ g/mL, 50% of the an tibiotic was adsorbed in the first 2 days. At a concentration of 40 µ g/mL, the adsorption time increased to 4 days. In the case of ciprofloxacin, 50% of the antibiotic was adsorbed on days 2, 4, and 6 at concentrations of 10, 20, and 40 µ g/mL, respectively. At an initia concentration of 10 μg/mL, 100% adsorption was observed on the 9th (tetracycline) and 11th (benzylpenicillin and ciprofloxacin) days. With an increase in the initial concentra tion to 20 μg/mL, the time for complete adsorption increased to 14 (tetracycline) and 21 (benzylpenicillin and ciprofloxacin) days. At the maximum studied concentration of 40 µ g/mL, only tetracycline and benzylpenicillin showed 100% adsorption on the 28th day while with the use of ciprofloxacin, the sorption process continued. The antibiotic removal efficiency (Re, %, Equation (1)) was calculated depending o the initial antibiotic concentration (Figure 5a-c, insets). The removal rate in the first day was high in all cases, which can be explained by the abundance of free adsorption sites o the surface of BNNPs that were not occupied by antibiotic molecules. At an initial concen tration of 10 µ g/mL, a Re50 was observed in the first 24 h. With an increase in concentratio to 20 and 40 μg/mL, the time increased to 3 and 5 days, respectively. It can be seen that, a low initial concentrations (10 and 20 µ g/mL), the BNNPs are effective for all three type of antibiotics, and their complete removal from the solution is achieved on days 14 an 21, respectively. At the maximum initial concentration of 40 µ g/mL, Re100 is observed o the 28th day (benzylpenicillin and tetracycline), while ~20% of ciprofloxacin still remain The Re can be arranged in the following order: TC > BP > CIP.
Experimental sorption data were compared with the results of DFT calculations (Fig  ure 5d). For theoretical simulations, the BN cell size with one antibiotic molecule per on h-BN sheet varied from 7 × 13 Å (~43 wt%) to 26 × 27 Å (~9.2 wt%). After reaching a con centration of 13-14 wt% (one antibiotic molecule per 20 × 21 Å BN cell), the BE values o all antibiotics decrease. This indicates that the adsorption process is thermodynamicall favorable up to sufficiently high antibiotic concentrations. The BE values for the BN@T nanohybrid are lower than those of the BN@CIP and BN@BP counterparts, which is i good agreement with the experimental data that tetracycline is adsorbed much faster tha the other two antibiotics. At ~35 wt% and high concentrations (10 × 13 Å and less h-BN cell), the BE sharply increases. This is due to the influence of steric factors; antibiotic mo ecules begin to interact with each other, and the force of their repulsion makes a larg contribution to the BE.
After completion of the adsorption experiments using the maximum concentratio of antibiotics (40 µ g/mL), BN@AB samples were analyzed by FTIR spectroscopy ( Figur  6). In all cases, high-intensity peaks were present, corresponding to vibrations of bonds i BNNPs (B-N-B at 780 cm −1 , B-N at 1370 cm −1 ) [52]. Additionally, characteristic peaks o various functional groups present in antibiotics were observed, which confirms their suc cessful adsorption on the BNNP surface. The benzylpenicillin adsorption is confirmed b The antibiotic removal efficiency (Re, %, Equation (1)) was calculated depending on the initial antibiotic concentration (Figure 5a-c, insets). The removal rate in the first days was high in all cases, which can be explained by the abundance of free adsorption sites on the surface of BNNPs that were not occupied by antibiotic molecules. At an initial concentration of 10 µg/mL, a Re 50 was observed in the first 24 h. With an increase in concentration to 20 and 40 µg/mL, the time increased to 3 and 5 days, respectively. It can be seen that, at low initial concentrations (10 and 20 µg/mL), the BNNPs are effective for all three types of antibiotics, and their complete removal from the solution is achieved on days 14 and 21, respectively. At the maximum initial concentration of 40 µg/mL, Re 100 is observed on the 28th day (benzylpenicillin and tetracycline), while~20% of ciprofloxacin still remains. The Re can be arranged in the following order: TC > BP > CIP.
Experimental sorption data were compared with the results of DFT calculations (Figure 5d). For theoretical simulations, the BN cell size with one antibiotic molecule per one h-BN sheet varied from 7 × 13 Å (~43 wt%) to 26 × 27 Å (~9.2 wt%). After reaching a concentration of 13-14 wt% (one antibiotic molecule per 20 × 21 Å BN cell), the BE values of all antibiotics decrease. This indicates that the adsorption process is thermodynamically favorable up to sufficiently high antibiotic concentrations. The BE values for the BN@TC nanohybrid are lower than those of the BN@CIP and BN@BP counterparts, which is in good agreement with the experimental data that tetracycline is adsorbed much faster than the other two antibiotics. At~35 wt% and high concentrations (10 × 13 Å and less h-BN cell), the BE sharply increases. This is due to the influence of steric factors; antibiotic molecules begin to interact with each other, and the force of their repulsion makes a large contribution to the BE.
After completion of the adsorption experiments using the maximum concentration of antibiotics (40 µg/mL), BN@AB samples were analyzed by FTIR spectroscopy (Figure 6). In all cases, high-intensity peaks were present, corresponding to vibrations of bonds in BNNPs  [53]. The BNNPs@TC spectrum contains vibrational peaks characteristic of tetracycline at 3342-3325 cm −1 , which can be attributed to the stretching vibrations of N-H and O-H bonds. Adsorption bands in the range of 3064-3003 cm −1 and 2955-2835 cm −1 are ascribed to C-H and CH 3 stretching vibrations, respectively. The presence of the C=C bond appears as adsorption bands at 1622-1569 cm −1 . The characteristic peaks at 1454 and 1357 cm −1 can be due to the bending vibrations of C-H and CH 3 , respectively. The peaks at 1247-1000 cm −1 and 995 cm −1 were assigned to the C-H in-plane deformation and the C-N stretching vibration, respectively [54]. The peak at 1270 cm −1 observed in the FTIR spectrum of the BNNPs@CIP sample is due to the C-N bond. The remaining adsorption peaks were identified as C=C bond (1680-1620 cm −1 ), C=O bond (1725-1705 cm −1 ), and C-H bond (3150-2750 cm −1 ). A broad maximum in the range 3700-3000 cm −1 is associated with the vibrations of N-H and O-H bonds [55].  The maximum sorption capacity (qe) was determined using Equation (2). After keeping the BNNPs in an antibiotic solution with a concentration of 1 mg/mL for 56 days, the following qe values were obtained: 297.29 mg/g (TC), 254.76 mg/g (BP), and 238.17 mg/g (CIP). Table 3 compares the values of the maximum adsorption capacity of benzylpenicillin, tetracycline, and ciprofloxacin using different sorbents. It can be seen that BNNPs show a much higher efficiency in removing antibiotic molecules compared to many other materials.  The maximum sorption capacity (q e ) was determined using Equation (2). After keeping the BNNPs in an antibiotic solution with a concentration of 1 mg/mL for 56 days, the following q e values were obtained: 297.29 mg/g (TC), 254.76 mg/g (BP), and 238.17 mg/g (CIP). Table 3 compares the values of the maximum adsorption capacity of benzylpenicillin, tetracycline, and ciprofloxacin using different sorbents. It can be seen that BNNPs show a much higher efficiency in removing antibiotic molecules compared to many other materials.

BNNP Purification Strategy from Adsorbed Antibiotics
To be cost-effective, a sorbent used in wastewater treatment must easily regain its adsorption properties after being cleaned of contaminants so that it can be reused. There are chemical, mechanical, dispersive, and mixed methods of cleaning sorbents. Chemical cleaning is based on the use of various chemicals to remove antibiotics. Mixtures of acetonitrile and isopropyl alcohol are highly effective for cleaning contaminated particles [44]. According to the results of our DFT analysis, the BE of positively charged forms of antibiotics corresponding to an acidic environment is lower than that of neutral or negatively charged forms ( Figure 5d). Thus, three different methods for cleaning BNNPs from antibiotics were developed and tested, as described below. Figure 7 shows the efficiency of BNNP purification from adsorbed antibiotics. Method 1 ensured complete removal of antibiotics after 14 (benzylpenicillin), 21 (tetracycline), and 10 (ciprofloxacin) days. Already on the first day, 40, 35, and 45% of adsorbed benzylpenicillin, tetracycline, and ciprofloxacin, respectively, are released. When using Method 2, the maximum degree of purification from benzylpenicillin, tetracycline, and ciprofloxacin is 72.5, 70, and 80% after 29, 28, and 32 days, respectively, and on the first day, only 20, 27.5, and 32.5% of each antibiotic are desorbed. The efficiency of Method 3 was in the middle between Methods 1 and 2. The degree of BNNP purification from benzylpenicillin, tetracycline, and ciprofloxacin is 25, 30, and 35% (after one day) and 77.5, 77.5, and 85% (after 28 days), respectively.
After cleaning the adsorbents from antibiotics, antibiotic-contaminated solutions can be properly disposed of or recycled using ozonation, chlorination, or Fenton oxidation methods [9,12].

Conclusions
Here we demonstrate that spherical hollow hexagonal BN nanoparticles (BNNPs) with an average size of 200-300 nm are promising sorbents for wastewater treatment from various types of antibiotics. The maximum sorption capacity of tetracycline, bencylpeniciline, and ciprofloxacin is 297.3, 254.8, and 238.2 mg/g, respectively. The removal efficiency of antibiotics (Re, %) depends on their initial concentration; Re50 and Re100 are achieved after 24 h and 14 days (10 µ g/mL), 3 and 21 days (20 µ g/mL) and 5 and 28 days (40 µ g/mL), respectively. The mechanisms of antibiotic adsorption and subsequent sorbent purification were analyzed using DFT calculations. The obtained results show that the interaction of antibiotic molecules (AM) with the BN surface is neither purely physical nor purely chemical. Antibiotics interact with BN mainly through oxygen-containing groups, and the adsorption process is spontaneous and endothermic. In addition, this interaction causes the BN surface to bend, which increases both the binding energy

Conclusions
Here we demonstrate that spherical hollow hexagonal BN nanoparticles (BNNPs) with an average size of 200-300 nm are promising sorbents for wastewater treatment from various types of antibiotics. The maximum sorption capacity of tetracycline, bencylpeniciline, and ciprofloxacin is 297.3, 254.8, and 238.2 mg/g, respectively. The removal efficiency of antibiotics (Re, %) depends on their initial concentration; Re 50 and Re 100 are achieved after 24 h and 14 days (10 µg/mL), 3 and 21 days (20 µg/mL) and 5 and 28 days (40 µg/mL), respectively. The mechanisms of antibiotic adsorption and subsequent sorbent purification were analyzed using DFT calculations. The obtained results show that the interaction of antibiotic molecules (AM) with the BN surface is neither purely physical nor purely chemical. Antibiotics interact with BN mainly through oxygen-containing groups, and the adsorption process is spontaneous and endothermic. In addition, this interaction causes the BN surface to bend, which increases both the binding energy and the contact area. Based on the results of the DFT analysis, a simple and efficient BNNP purification strategy was proposed. Combined treatment in a mixture of acetonitrile and ethanol solutions at pH = 4.4 (acetate buffer) ensures complete removal of antibiotics after 14 (BP), 21 (TC), and 10 (CIP) days. Our encouraging results are of great importance for the further development of efficient and cost-effective h-BN-based adsorbents for water purification from therapeutic contaminants.