In situ generation of hydroxyl radicals by B-doped TiO2 for efficient photocatalytic degradation of acetaminophen in wastewater

Acetaminophen (AP) is a widely used antipyretic analgesic belonging to the class of PPCPs, which is difficult to be effectively degraded by traditional water treatment processes. However, photocatalytic technology may be an effective approach. Herein, B-doped TiO2 photocatalytic materials were synthesized by sol–gel method, calcinated at 600℃ for 2 h, investigated by XRD, TEM, XPS, and other characterization methods. The photocatalytic efficiency and factors affecting the photocatalytic activity were assessed by degradation of AP under 365 nm UV light. Compared with undoped TiO2, 4%B-TiO2 nanopowder has smaller grain size, higher porosity, and lower bandgap energy of 3.11 eV. Scavenging experiments and ESR results show that •OH is the principal active species. Hence, the degradation efficiency of AP is as high as 98.8% in 30 min when adopting 10-mg/L AP initial concentration and 1-g/L 4%B-TiO2 loading, owing to efficient •OH generated by B-TiO2.


Introduction
PPCPs pose potential threats to public health and environmental ecosystems. Acetaminophen (AP) is a widely used nonsteroidal anti-inflammatory drug, which is difficult to be effectively degraded by traditional water treatment processes (Almeida andNunes 2019, Phong Vo et al. 2019). However, heterogeneous photocatalysis is an advanced oxidation process capable of degrading trace pollutants (Leyva et al. 2018;Loddo et al. 2018;Solis-Casados et al. 2018).
TiO 2 is the most promising and widely used material in the field of photocatalytic degradation, but it still has some limitations (Li 2015). TiO 2 doping and interfacial modification is one of the efficient ways to enhance the photodegradation efficiency (Varma et al. 2020;Vilé 2021). Considering the cost and stability of catalysts, we choose nonmetallic element-boron as dopant (Akpan and Hameed 2010). B-doped TiO 2 was prepared by many people in previous articles, but few examples have been reported on B-doped TiO 2 for the treatment of specific pollutants-AP. The reaction mechanism of B-doped TiO 2 is not clear specifically for AP contaminants. In order to further identify the active species produced and dominant during the reaction, we designed experiments to verify the mechanism of the reaction system.
In this paper, we prepared B-doped TiO 2 catalysts calcined at 600℃ for 2 h, designed experiments to explore the relatively optimal conditions, and characterized the best performing catalysts to investigate their physicochemical properties. The degradation mechanism of B-doped TiO 2 on AP which is not to be studied thoroughly is confirmed by quenching experiments and ESR spectrum. The theoretical basis for in situ generation of •OH provided by this study has important implications for the environment and human health. H 3 BO 3 and 1-mL ultrapure water were dissolved and stirred in 20-mL anhydrous ethanol, and the boric acid was used as a boron precursor. The latter was then dropped into the previous mixed solution under magnetic stirring through a separating funnel at a rate of 1-2 drops/s. The mixture was stirred for 2 h after dropping and transferred to a petri dish to left to age for 12 h to form a wet gel. Then it was dried in an oven at 60℃ for 5 h to acquire a dry gel and ground in a mortar to get fine powder. The pulver was placed in a muffle furnace, the temperature of which was raised up to 600℃ at a heating rate of 3℃/min, calcined at 600℃ for 2 h, and cooled to room temperature naturally, during which anatase was formed. To determine the optimal loading of boron nanoparticles for photocatalytic degradation of target pollutants, four atomic loadings were tested, namely, 2%, 4%, 6%, and 8%. For comparison, undoped TiO 2 was prepared without H 3 BO 3 according to the above operation processes.

Characterization
The materials were characterized by the following means. The morphology and microstructure of the samples were observed by FEI Talos F200X/Super-X high-resolution transmission electron microscope (HRTEM). The lattice spacing and particle diameter of the samples were measured by DigitalMicrograph and Nano Measurer software. The crystal phase of the powders was recorded on X'Pert PRO X-ray powder diffractometer (XRD, PANalytical, the Netherlands) equipped with Cu-Kα radiation. The average microcrystalline size of the samples was calculated with Scherrer equation. Elemental composition and chemical states on the surface of the samples were analyzed by X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, ThermoFisher, USA) with Al-Kα as the radiation source. High-resolution C1s, O1s, Ti2p, and B1s XPS spectra were recorded, and all the binding energies were referred to the contaminant carbon C1s peak (284.8 eV). The peaks of each spectrum were deconvoluted using CasaXPS software. Tougaard-type background and peak differentiating and imitating based on Lorentz-Gaussian function were adopted during the deconvolution process. The presence of functional groups and chemical bonds in molecules was identified through Fourier transform infrared spectra (FT-IR, Nicolet iS5, ThermoFisher, USA), which were recorded in the wavenumber range of 4000-400 cm −1 . Low-temperature nitrogen adsorption-desorption isotherms were recorded by ASAP 2460 specific surface area and porosity analyzer (Micromeritics, USA). Brunauer-Emmett-Teller (BET) specific surface area and Barrett-Joyner-Halenda (BJH) pore size distribution (pore diameter and pore volume of samples) were calculated based on the information of the isotherms. In the meantime, the isothermal adsorption curve type of the sample can be determined. The light absorption properties and the band gap of the photocatalysts were obtained by ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS, UV-3600i Plus, SHIMADZU, Japan), the spectra of which were recorded in the range of 200-800 nm. Photoluminescence (PL) spectra was tested using a FluoroMax-4 fluorescence spectrophotometer (HORIBA, USA). Electron spin resonance (ESR) signals of radicals were recorded using a BRUKER A300 spectrometer (Germany) with 5,5-dimethyl-1-pyrroline-N-oxide (DMPO) as the spin trap reagent. The absorbance of the solution containing pollutants after reaction was recorded on AOELAB A580 dual beam UV-Vis spectrophotometer (UV-Vis).

Evaluation of photocatalytic performance
The photocatalytic degradation experiments were carried out in a photochemical reactor (YZ-GHX-A). The reactor is outfitted with attachments such as an electromagnetic stirrer, four flat-bottomed quartz test tubes (100 mL), a cold trap, and a cooling water circulation device, which can control temperature precisely and prevent the influence of water evaporation and temperature on the reaction system. A 500-W mercury lamp was used as the ultraviolet light source, and it was placed in the cold trap, which was located in the center of the reaction apparatus, and the working distance of the reaction system from the light source was 10 cm. The emission wavelength of the lamp is 365 nm, and the light intensity is 85 W/m 2 (Mostafa et al. 2021).
In a typical UV photocatalytic experiment, 0.05-g catalyst was dispersed in 50-mL solution by magnetic stirring. The initial concentration of each pollutant was 10 mg/L, and the initial pH was 7.0. The catalyst suspension under magnetic stirring was then exposed with a 500-W mercury lamp to trigger the photocatalytic reaction. Sufficient aliquots were aspirated every 5 min during the 30-min irradiation and offcenterred at 5000 rpm to remove residual catalyst particles for analysis.
The photodegradation performance was monitored by measuring the intensity change of characteristic absorbance of AP at the maximum absorption wavelength of 243 nm using an A580 UV-Vis spectrophotometer with a 10-mm path length quartz cell. The relative concentration was converted according to the AP standard curve method. The degradation data fit the model of pseudo-first-order kinetic. Calculate photocatalytic reaction rate constant and residual rate using Eqs. (1) and (2): where C 0 and C t are the initial and specific irradiation time t concentration of AP, respectively (mg/L); A 0 and A t are the initial and specific irradiation time t absorbance of AP, respectively; k is the rate constant of pseudo-first-order kinetic (min −1 ); and t is the reaction time (min). The degradation rate is calculated by (1 − In this experiment, the effects of calcination conditions, doping amount, catalyst dosage, and pollutant concentration on the photocatalytic reaction were investigated. The initial solution was prepared by diluting a concentrated solution of 1-g/L acetaminophen, using ultrapure water as solvent. Contaminant concentrations of 5-30 mg/L and catalyst concentrations of 0.25-2 g/L were tried in this experiment. The effects of solution pH (pH = 2-12) and H 2 O 2 concentration (0-20 mM) on photocatalytic degradation were investigated.
Similar blank experiments were performed in absence of catalyst or in the dark to demonstrate that the reactions were mainly actuated by the photocatalytic procedure. The above mercury lamp was replaced with a 500-W xenon lamp, and the paracetamol degradation test was carried out using simulated sunlight on the material that showed the best performance in the UV light experiment.
To test the cycle stability of the photocatalyst, the photocatalyst was recovered by off-center at 4000 rpm for 10 min at the end of the photocatalytic reaction and then dried in an oven at 60℃ to constant weight. The recovered catalyst was reused in five consecutive cycles to evaluate the degradation of photocatalytic performance due to catalyst deactivation. All experiments were conducted three times, and the results were averaged. The experimental error is within ± 5%. Finally, other pollutants were treated with 4%B-TiO 2 to explore the universality of this catalyst.
The active species shaped during the experiment by 4%B-TiO 2 were identified by indirect chemical probe technique, 0.01-M isopropanol (IPA), 0.001-M ethylenediamine tetraacetic acid disodium salt (EDTA-2Na), and 0.0005-M p-benzoquinone (BQ) were used as scavengers for hydroxyl radicals (•OH), photogenerated holes (h + ), and superoxide radicals (•O 2 − ), respectively. 5,5-Dimethyl-1-pyrroline-N-oxide (DMPO) was used as a spin trapping reagent, and electron spin resonance (ESR) spectra were recorded on a BRUKER A300 spectrometer under UV light exposure. The free radical trapping test was similar to the photoactivity test described above, except that a certain amount of scavengers were put into the reaction solution.

Morphology observation
The HRTEM photographs of the undoped/doped samples are shown in Fig. 1. HRTEM observation show that there are little differences between undoped and B-doped samples in size and shape, and the prepared samples have high crystallinity, and most of the particles are spherical.
Representative HRTEM photographs and histograms of the particle size distribution of the samples are shown in Fig. 1(b, f). The average particle sizes of the two samples measured by HRTEM are 38.11 nm and 35.38 nm, respectively. The presence of boron reduces the catalyst particle size, consistent with the trend observed in XRD. However, it is quite different from the XRD results. The HRTEM measured values are all larger than the XRD measured and calculated values. It may be that the sampling number is too small or the overall particle size of the samples is too large.
The lattice gaps measured for different catalysts are shown in Fig. 1(d, h). Clear lattice fringes can be observed, which can be attributed to the most dominant and stable TiO 2 anatase (101) plane (Yadav et al. 2020). The incorporation of boron reduces the lattice gap of the sample slightly, which is consistent with the trend of XRD. Compared with the XRD results and the standard card 3.52 Å, the values of the lattice gap measured by the Fourier transform analysis of the samples are too large.
The EDS spectrum and element mapping of 4% B-TiO 2 are shown in Fig. 1(i) and (j), and the mass percentage information of each element in the sample is also provided. The weight percentages of B, Ti, and O elements measured by EDS analysis are 0%, 54.16%, and 45.84%, respectively, shown in Fig. 1(i). Because EDS is not an accurate quantitative method, the composition of the elements listed in the table deviates from the true state in the sample. Boron is a light element and is trace amount doped, so it is hard to detect. The element mapping in Fig. 1(j) shows that B, Ti, and O elements are uniformly distributed in 4%B-TiO 2 sample, which confirms the successful doping of B. The XRD patterns of 2%, 4%, 6%, and 8% B-TiO 2 show similar peaks, and peak positions of brookite or rutile TiO 2 can not be found in the patterns, indicating that it contains only anatase phase. The doping substitution of boron in the host lattice of anatase TiO 2 do not influence the crystal structure of the activator. The diffraction peak of B 2 O 3 could not be detected, probably because the boron content is less than 5 wt.%, which is lower than the detection limit. The (101) diffraction peak of anatase TiO 2 has the strongest intensity, which can be used to figure out the crystallite size of anatase TiO 2 crystals . The average crystallite size of the samples can be calculated using the following Scherrer Eq. (3): where D represents the grain size in the normal direction of the crystal plane and the measurement range is 3-200 nm; K is the shape factor and the value of K is 0.89; β is the width of half height of the diffraction peak; λ is X-ray irradiation wavelength; Cu-Kα wavelength is 1.54 Å; and θ is the X-ray diffraction angle.

Crystal structure analysis
As shown in Table 1, the calculated crystallite sizes of the 0-8% B-TiO 2 samples are 31. 28, 29.56, 25.62, 26.72, and 27.08 nm, respectively. The grain size of anatase TiO 2 decreases first and then increases with the addition of B doping content. The doped B element slightly hinders the growth of anatase TiO 2 crystals. The grain sizes of the (101) plane measured by XRD are 343, 321, 272, 286, and 290 Å, which are larger than the calculated values, but the trend of first decreasing and then increasing is consistent. The lattice spacing of (101) plane of the samples shows the same trend as crystallite sizes, but the values of the samples are relatively smaller compared to the standard card 3.52 Å. After boron doping, the TiO 2 unit cell shows the same trend in three dimensions. The unit cell volume of anatase TiO 2 crystals decreases from 0.1357 nm 3 in undoped TiO 2 sample to 0.1353 nm 3 in 6%B-TiO 2 and then increases to 0.1358 nm 3 in 8%B-TiO 2 . Table shows that the TiO 2 crystal shrinks first and then expands (Zhang et al. 2018).

Surface chemical composition and surface functional group analysis
XPS analysis of 4%B-TiO 2 is carried out to study the chemical states of elements and the bonding properties of the synthesized materials. Figure 3 shows the full spectra of the surface of the samples and the high-resolution XPS spectra of C1s, O1s, Ti2p, and B1s of 4%B-TiO 2 , respectively. Among them, the C1s peak with a lower binding energy of 284.12 eV is used for the calibration of the XPS results. The binding energies of C1s, O1s, Ti2p, and B1s are 284. 80, 530.24, 458.98, and 192.44 eV, respectively, shown in Fig. 3(a). Figure 3(b) shows the high-resolution spectrum of C1s. The C element can be attributed to residual carbon in the precursor solution (TBT) and foreign hydrocarbons in the XPS instrument itself. Moreover,284.80,286.52,and 289.01 eV correspond to carbonyl group (C-C), ether bond (C-O-C), and carboxyl group (O-C = O), respectively. Figure 3(c) shows the high-resolution spectrum of O1s. After peak deconvolution, O1s appears response peaks at 530.22, 532.01, and 533.35 eV, respectively. In these values, 530.22 eV is vested in Ti-O bond in the TiO 2 lattice (the standard binding energy is 529.9 eV), and 533.35 eV is assigned to the B-O bond of B 2 O 3 (the standard binding energy is 533.0 eV), indicating that the main structures of the catalyst are TiO 2 and B 2 O 3 . The 532.01 eV is between the standard binding energies of the Ti-O bond and the B-O bond. So we guess that B enters the TiO 2 lattice gap, replaces the oxygen in the TiO 2 lattice, and exists in the form of O-Ti-B (Bilgin Simsek 2017). Figure 3(d) represents the Ti2p high-resolution spectrum. After peak fitting, two peaks can be perceived at binding energies of 458.98 and 464.68 eV, which are attributed to Ti2p 3/2 and Ti2p 1/2 of TiO 2 , respectively (the standard binding energy of Ti2p 3/2 is 458.8 eV). Figure 3(e) shows the high-resolution spectrum of B1s. The XPS spectrum of B1s is symmetrical, and no peak separation is required. Only one peak is observed at 192.44 eV, indicating that B in this sample exists in a single form. The standard binding energies of B1s in B 4 C, TiB 2 , and B 2 O 3 are 186.5, 187.5, and 192.3 eV for C-B, Ti-B, and O-B bonds, respectively. The binding energy of B1s observed is located near the B-O bond, indicating that B exists on the surface of TiO 2 in the form of B 2 O 3 .
Figure 3(f) shows the FT-IR spectra of the samples, which are almost similar. A peak at 500 cm −1 can be observed in the samples, which is in the region of 625-310 cm −1 and is a broad and strong transmission band related to the stretching vibration of Ti-O bond. After the addition of boron, the increased intensity of transmission peak at 1396 cm −1 is caused by the stretching vibration of the B-O bond, and this type of vibrational mode occurs in the range of 1470-1065 cm −1 . The presence of boron-doped TiO 2 can be determined by the spectra, and B can be fixed on the TiO 2 surface through Ti-O-B bond.
The band appearing at 1648 cm −1 in both samples is due to the bending vibration of the OH group of H 2 O adsorbed on the TiO 2 powder. The broad transmission band at 3416 cm −1 is related to the stretching vibration of the hydroxyl groups of H 2 O molecules adsorbed on the samples (Koysuren and Koysuren 2019). The absorption band increases with the addition of boron. This is because boron behaves as a Lewis acid after being added to the TiO 2 lattice, which accelerates the adsorption of water on its surface. These materials were prepared by high temperature heat treatment and dried before IR measurements. All these IR results clearly show that the B-doped TiO 2 nanoparticles have high surface hydroxylation, which is reflected in

Optical properties and band structures
Figure 4(a) shows the UV-Vis diffuse reflectance spectra of TiO 2 and 4%B-TiO 2 . The light absorption edge of TiO 2 is about 396 nm, indicating that it has almost no visible light catalytic activity. Due to the formation of 4%B-TiO 2 by B doping, the light absorption edge is obviously red-shifted to the visible light region around 431 nm, indicating that B-doped TiO 2 has visible light absorption properties. As mentioned earlier, anatase TiO 2 has a tendency to refine the grain size after loading, which leads to a blue-shift of the absorption edge due to the quantum size effect. However, due to the surface effect, the large surface tension of the nanoparticles makes the lattice distorted, the lattice constant becomes smaller, etc. The redshift of the light absorption band will be observed relative to the coarse crystal material. The two factors work together, and the characterization show that the latter has a stronger influence. The increase of the internal stress inside the particle will lead to the change of the band structure and the narrowing of the energy level spacing, thereby causing the red shift (Ye et al. 2020). Figure 4(b) shows the hν-(Ahν) 2 plots of the two composites. The bandgap energy of semiconductors was calculated using the Tauc-plot method. Herein, A, h, ν, and hν represent the absorbance, Planck's constant, light frequency, and photon energy, respectively. The hv-(Ahν) 2 diagram is applicable to the direct bandgap semiconductor anatase TiO 2 . The onset of absorption is at the intersection of the line extending from the linear region of the graph and the abscissa axis. It can be seen that TiO 2 has a wide bandgap of 3.21 eV, and the bandgap is narrowed to 3.11 eV after B doping. Based on UV-Vis spectroscopy and Tauc curve analysis, borondoped TiO 2 exhibits higher UV light absorption (Koysuren and Koysuren 2019). At the same time, the reduction of the band gap of 4%B-TiO 2 broadens the light absorption edge of TiO 2 to the visible light range, which is beneficial for the catalyst to capture long-wavelength sunlight and improves its photoconversion efficiency (Li et al. 2020).
Figure 4(c) shows the photoluminescence spectra (PL) of the samples, which reflects the separation and recombination capabilities of photogenerated charges and holes. All samples exhibit regular PL curves in the range of 270-800 nm. TiO 2 shows a higher PL signal, and the PL signal of 4%B-TiO 2 is relatively lower. The rapid recombination of electron-hole pairs in TiO 2 limits its photocatalytic activity. The lower PL curve of 4%B-TiO 2 indicates that the intensity of photoluminescence is weak, and the sample generates less photons after the recombination of photogenerated electrons and holes. The recombination of photogenerated carriers is suppressed, indicating that 4%B-TiO 2 has better photocatalysis performance (Liu et al. 2022c;Vile et al. 2021).

N 2 adsorption-desorption experiments
Figure 5(a) shows the low-temperature nitrogen adsorption-desorption isotherm of the samples. 4%B-TiO 2 shows a type IV isotherm with an H1-type hysteresis loop, which is a characteristic of mesoporous materials (Yadav et al. 2020). The reason is as follows: the rear section of the isotherm is convex again, and an adsorption hysteresis loop appears in the middle section (corresponding to the system in which the porous adsorbent has capillary condensation), and the hysteresis loop is more common in type IV isotherms. The hysteresis loops are divided into six categories in the latest IUPAC classification. There is a saturated adsorption plateau on the adsorption isotherm, which indicates that the pore size distribution is relatively uniform. H1-type hysteresis loop can be observed in mesoporous materials with relatively narrow pore size distribution and in spherical particle aggregates with relatively uniform size. Undoped TiO 2 belongs to the type IV isotherm. The explanation is as follows: the undoped sample shows a type II isotherm, which reflects the typical physical adsorption process on nonporous or macroporous adsorbents. However, according to the results calculated by the BJH model, the pore size of TiO 2 does not belong to the macropore range. Considering the case where there is a large amount of adsorption but no hysteresis loop in the intermediate pressure part of the type IV isotherm. When the relative pressure is about 0.2-0.3, according to the Kelvin equation, it can be seen that the pore radius is very small, and the effective pore radius is only as the molecular size of a few adsorbates. The capillary condensation does not occur, and the adsorption and desorption isotherms coincide. Figure 5(b) shows the BJH pore size distribution maps of the two samples, and the synthesized nanocomposites exhibit narrow pore size distribution ranging from 2 to 15 nm. Both of them can be regarded as mesoporous materials. The mesopore distribution of the samples is uniform, mainly concentrates at about 6 nm. The average pore size of the doped catalyst is relatively larger than that of the undoped catalyst (Rabhi et al. 2021). According to HRTEM images, it can be demonstrated that the formation of mesoporous structure is assigned to the aggregation of TiO 2 crystallites. This morphological and porosity property of the 4%B-TiO 2 composite is beneficial for the practical application of photocatalysis, as it provides enhanced interfacial contact between the photocatalyst and the reactants, which is beneficial for the adsorption of contaminant molecules, increasing the number of active centers and promoting charge transfer. All these advantages are hopeful to motivate the photocatalytic efficiency of B-doped TiO 2 in the degradation of trace pollutants (Liu et al. 2020d).
High specific surface area and dense porosity are important properties for efficient photocatalytic materials. According to the results of N 2 physical adsorption and desorption at 77 K, the specific surface area is calculated using the BET formula, and the average pore diameter and total pore volume are calculated using the BJH model, as shown in Table 2. The outcomes show that the specific surface area and total pore volume aggrandize significantly with the addition of boron doping content. The specific surface area of 4%B-TiO 2 is 17.2272 m 2 /g, which is much larger than that of TiO 2 (4.0049 m 2 /g). The pore volume of the samples shows the similar increasing tendency, with the pore volumes of TiO 2 and 4%B-TiO 2 being 0.005095 and 0.046075 cm 3 /g, respectively. Both results suggest that the doping of B into TiO 2 can broaden the space between the aggregated TiO 2 nanoparticles, leading to an augment in the specific surface area of the photocatalyst (Liu et al. 2021b).  increases from 69.9 to 98.8% with the increase of B concentration and then decreases to 95.0%. Therefore, 4% B doping is the optimal B doping amount to promote the photocatalytic efficiency of TiO 2 . Considering the catalyst degradation efficiency and synthesis cost, 4%B-TiO 2 was used in the photocatalytic experiments. Compared with TiO 2 , the enhanced performance of 4%B-TiO 2 may be due to its smaller particle size, larger specific surface area, and lower band gap. This has been supported by XRD, HRTEM, BET, and UV-Vis characterization studies.

Effect of calcination conditions
Figure 6(b) shows that with the increase of calcination temperature, the degradation ratio of AP on 4%B-TiO 2 photocatalyst reaches 98.8% at 600℃ and the degradation effect is the best. When the temperature rises above 600℃, the rutile type appears, resulting in a decrease in catalytic performance. The experimental results and XRD characterization show that high activity anatase is formed at a temperature of 600℃. In the preliminary experiment, the degradation rate drops sharply at 300℃ and 800℃, brookite is formed at a low temperature of 300℃, and a rutile phase is formed at a high temperature of 800℃. Too high or too low temperature will affect the photocatalytic activity of 4%B-TiO 2 . Particle sintering occurs when the calcination temperature is too high (Liu et al. 2020b;Tao et al. 2022). Figure 6(c) displays that the longer the calcination time, the better the photocatalytic activity of the catalyst. However, the fitting curves are glued together, and the calcination time has little effect on the AP degradation rate. As long as the calcination is performed at 600℃, the effect of the calcination time on the degradation effect is extremely small, so we adopt the empirical value of 2 h. Therefore, 4%B-TiO 2 catalyst formed by calcination at 600℃ for 2 h is adopted in the succeeding experiments.

Effect of photocatalyst dosage and initial concentration of AP
Figure 6(d) presents the effect of catalyst loading on the removal rate of 50 ml of 10-mg/L AP solution. When the amount of 4%B-TiO 2 aggrandizes from 0.25 to 1 g/L, the AP degradation rate increases from 76.3 to 98.8%. However, when the dosage of 4%B-TiO 2 further augments to 2 g/L, the AP removal rate is 98.4%, which is not much improved, but the degradation rate is slightly faster in the early stage. This may be the opacity of the reaction solution caused by the excess catalyst, which brings down the light transmittance, thereby lowering the light-receiving surface area of the catalyst, resulting in the masking and scattering effects on photon absorption. Increasing the dosage of catalyst cannot provide a more significant improvement, while reducing the dosage of catalyst requires prolonging the reaction time. Considering the practical application cost of the catalyst, the catalyst concentration of 1 g/L is taken in the following tests (Liu et al. 2021c).
With 1-g/L catalyst loading and 5-30-mg/L initial AP concentration, the fitting curves of the residual rate of photocatalytic degradation are shown in Fig. 6(e). When the initial AP concentration is 10 mg/L, AP is almost completely removed (98.8%) within 30 min. However, when 30-mg/L AP is added, the removal rate of pollutants within 30 min decreases significantly. It may be that the high concentration of pollutants shields the UV light, or it may be that the generated intermediates compete with APs for limited active substances (Abdelaal and Mohamed 2013). The absorbance of 10-mg/L AP solution is about 0.72, and the absorbance measured at this concentration is more accurate. Therefore, in a typical AP solution degradation experiment, the substrate concentration of 10 mg/L is employed.

Effect of initial pH and H 2 O 2
Figure 6(f) analyzes the influence of initial pH on AP removal efficiency. In the photocatalytic degradation experiments, the pH value of the AP solution is adjusted in the range of 2 to 12 using HCl and NaOH solutions. The initial pH of the stock solution is about 7. When the initial solution pH is varied in the range of 4-10, 10-mg/L AP solution could be almost degraded entirely after 30-min illumination, and the photocatalytic removal efficiency of AP only fluctuates slightly. In consequence, the 4-10 pH range is the best condition for the degradation of AP by 4%B-TiO 2 under UV light. In order to simplify the experimental process, reduce the introduction of other ions, and lower the expense of water treatment; the original pH = 7 is employed as the optimum pH. The degradation efficiency is as high as 98.8% within 30 min at pH = 7. When the initial pH is 2, the removal rate of AP after 30 min of UV irradiation is 66.7%. When the pH adjusts to 12, the degradation rate of AP is 45.3%. This indicates that neither extremely acidic nor alkaline conditions are favorable for the degradation of AP. Especially under strong alkaline conditions, the degradation efficiency drops sharply.
TiO 2 particles are easy to agglomerate under strong acid conditions (pH = 2-3), so the surface area and the photon absorption rate decrease, thereby reducing the degradation rate of AP. The decrease of removal rate in strong alkaline solution may be related to the electrostatic repulsion between 4%B-TiO 2 and AP. When the pH value increases to 12, the number of AP accumulates deprotonated from AP, the number of AP molecules decreases, and the absorbance also decreases, which can be observed in the figure. There is electrostatic repulsion between the negatively charged TiO 2 surface (pH pzc = 6.3) and the anionic form of AP molecules (pK a = 9.7), which results in poor adsorption of AP and reduced AP degradation rate (Rabhi et al. 2021). Figure 7(a) shows the effect of the addition of 0-20 mM H 2 O 2 on the reaction system under UV light exposure. In a typical AP solution degradation experiment within 30 min, as the H 2 O 2 concentration increases from 0 to 20 mM, the absorbance ratio of AP after reaction increases from 0.01 to 0.78. The photocatalytic removal rate decreases with the rise of H 2 O 2 . This may be that both AP pollutants and added H 2 O 2 have absorbance at around 240 nm. The more H 2 O 2 added, the greater the absorbance measured at 240 nm, the phenomenon of which can be observed in the figure ).

Effect of adsorption and direct photolysis
In a typical AP solution degradation experiment, the mercury lamp was replaced by a xenon lamp and we prolonged the reaction time. Figure 7(b) showed 10.7% degradation of AP by 4%B-TiO 2 after the irradiation of simulated visible light for 90 min. Due to insufficient excitation energy, the catalyst could only degrade a small amount of AP, and the photocatalytic activity is extremely low. The photocatalytic efficiency is highly dependent on the light source. Take into account the fact that low photon energy in visible light, the UV light source is adopted in all experiments (Cai et al. 2022).
Since adsorption and direct photolysis may conduce to AP removal, it is necessary to evaluate their effects on AP degradation. Figure 7(b) shows that the absorbance of AP solution (10 mg/L) is almost constant after stirring with 4%B-TiO 2 for 30 min in the dark. This indicates that the adsorption of AP on surface of catalyst is negligible. In the absence of photocatalyst and in the presence of only light, the removal rate of AP solution is 5.7% after UV light illumination for 30 min, so photolysis is negligible relative to photocatalysis (Bilgin Simsek 2017).
In the presence of 4%B-TiO 2 , the absorbance at 243 nm decreases significantly after UV light irradiation for 30 min. The photodegradation rate of AP is greatly enhanced by the presence of 4%B-TiO 2 and UV light. Figure 7(c) shows the UV-Vis absorption spectra of the reaction solution at different reaction times. The absorption of AP at 243 nm almost completely disappears after UV irradiation for 30 min. The absorbance at maximum (243 nm) reflects the concentration change of AP over time (Liu et al. 2020c).

Stability and wide applicability of photocatalyst
Catalyst recycling is the key step in evaluating the practical application of photocatalysts. Figure 7(d) shows the Fig. 7 Effects of different factors on photocatalytic degradation activity. a H 2 O 2 dosage; b light source, adsorption, photolysis; c UV-Vis absorption spectra at different reaction times; d stability experiments; e-f different pollutants photocatalytic activity of the recovered 4%B-TiO 2 catalyst. The experimental was carried out with 10-mg/L AP, 1-g/L catalyst addition, and initial pH = 7 in 30-min reaction time. At the end of each cycle, the used photocatalyst was separated from the contaminant solution by centrifugation, and the separated catalyst was washed 3 times with ultrapure water and dried to constant weight. The dried catalyst was used for the next cycle. The photocatalytic degradation efficiencies are 98.8%, 95.0%, 89.2%, 88.7%, and 80.8%, respectively in 5 cycles. The AP degradation efficiency decreases by 18% in the fifth cycle compared with the first cycle. The decrease in photocatalytic degradation efficiency may be attributed to the poisoning of the catalyst by pollutants and secondary products.
In the process of recycling catalyst, weigh the dried powder before the experiment, and calculate the recycling rate relative to the previous one. The recycling rates are 85.46%, 82.40%, 78.60%, 73.37%, and 67.18%, respectively. The loss of the catalyst was caused during the centrifugal washing process and adhering to the plastic centrifuge tube wall after drying. Assuming that the loss value is constant, with the increase of the number of cycles, the total mass of the catalyst decreases gradually, and the loss rate increases gradually. Therefore, the recovery rate showed a downward trend. The results show that 4% B-TiO 2 nanophotocatalyst has good recyclability and strong stability, which can prolong the service life of the catalyst, can reduce the cost of catalyst, and is expected to be used for environmental remediation.
In order to further investigate the remarkable photocatalytic activity of 4%B-TiO 2 , its photocatalytic degradation of various trace pollutants (acetaminophen (AP), ethylparaben (EP), salicylic acid (SA), tetracycline hydrochloride (TC), oxytetracycline hydrochloride (OTC), sulfamethazine (SM2), sulfamethoxazole (SMZ), phenol (PN), bisphenol A (BPA), and dyes (methyl orange (MO), methylene blue (MB), rhodamine B (RhB), crystal violet (CV)) are investigated and summarized in Fig. 7(e-f). The experimental conditions are the same as the previous tests, with 10-mg/L initial pollutant concentration and 1-g/L catalyst loading. By comparing the curves in the figure, it is found that 4%B-TiO 2 could successfully degrade trace pollutants and dyes, but the degradation rates are inconsistent. This difference is attributed to the types and number of pollutants, the spatial structure, and the conditions of the bonds of the molecules. Overall, 4%B-TiO 2 nanophotocatalyst has good performance in photocatalytic degradation of refractory pollutants and dyes ).

Photocatalytic activity interpretation
The active substances produced during the photocatalytic reaction have strong oxidizing power and can decompose organic pollutants. So reactive species (RSs) quenching experiments are performed to monitor the existence of •OH, h + and •O 2 − . Other conditions are the same as described in "Evaluation of Photocatalytic Performance" (Sin et al. 2021). Figure 8(a-b) shows the absorbance ratios of the quenching experiments. In the quenching experiment using AP as model pollutant, because the characteristic wavelength of AP is too low at 243 nm, EDTA-2Na and BQ have strong absorbance at low wavelengths. Even if no photocatalytic reaction is performed, the addition of quencher EDTA-2Na and BQ will also greatly affect the absorbance measurement of AP itself (Liu et al. 2022b). Therefore, the pollutant was replaced in the quenching experiment, and AP was replaced with MB with a characteristic wavelength of 664 nm, and a quantitative experiment was performed, as shown in Fig. 8(a). EDTA-2Na and BQ do not show significant absorbance in the visible light range above 400 nm (Liu et al. 2022a). In this experiment, it can be seen that the same catalyst has different degradation efficiencies for different pollutants. It is assumed that the effect of the quenching agent on the reaction system is the same, that is, the quenching experiment is carried out with assumptions, and it is not even accurate. Meanwhile, non-radical degradation processes are not taken into account in this research. In the quenching experiment with MB as the model pollutant, the optimal dosage of the quencher was tested, and the contribution rate was calculated under the optimal dosage of the quencher. The pollutant was replaced with the model pollutant AP, and the quenching experiment was carried out again. The result is shown in Fig. 8(b).
In the photocatalytic degradation of MB using 4%B-TiO 2 , the rate constants of the blank control and the solutions containing the quenchers IPA, EDTA-2Na, and BQ are 0.1668, 0.0096, 0.0653, and 0.0494 min −1 , respectively, as shown in Fig. 8(a). The photocatalytic effect is the best when no quencher is added. It can be calculated that the contributions of •OH, h + , and •O 2 − to MB degradation are 94.24%, 60.87%, and 70.40%, respectively. Therefore, •OH is the main player in MB degradation, and •OH plays a more important role than h + and •O 2 − in the reaction system. The results also indicate that the added quencher may affect other active species while quenching specified free radical (Fan et al. 2021;Meng et al. 2017).
To further investigate RSs produced by the 4%B-TiO 2 system, ESR technique was adopted to monitor the ESR signal using DMPO as capture reagent. Figure 8 (c, d, e) displays that, no ESR signal is perceived in the dark, which demonstrates the significance of light for the generation of RSs. The three characteristic peaks of the adduct DMPO-•OH (quartet 1:2:2:1 signal ratio), DMPO-h + (triplet 1:1:1 signal ratio), and DMPO-•O 2 − (quartet 1:1:1:1 signal ratio or sextet) are tested under UV light illumination. The signal intensity increases with prolonging of exposure time. The higher the peak height, the higher the concentration of RSs. In the ESR characterization of 4%B-TiO 2 , stable •OH, h + , and •O 2 − can be generated continuously in the 4%B-TiO 2 system, proving that these three RSs plays an important part in degrading AP during reaction process, which is consistent with the results of quenching experiments (Jin et al. 2022;Liu et al. 2020d).

Possible photodegradation mechanism
The energy band theory of 4%B-TiO 2 is explained as follows. By calculation, the E CB and E VB of TiO 2 are about − 0.29 and 2.91 eV, respectively. In general, nonmetallic doping reduces the E g of semiconductors by raising the VB site through atomic orbital hybridization. Therefore, the E CB and E VB of 4%B-TiO 2 can be determined as − 0.29 and 2.82 eV vs. NHE, respectively (Sin et al. 2021 − , because the CB potential of 4%B-TiO 2 (-0.29 eV vs. NHE) is not more negative than that of O 2 /•O 2 − (− 0.33 eV vs. NHE) redox potential. This is inconsistent with the outcomes of the quenching experiment and ESR spectrum. Considering that the potential difference between them is not large, it may be that there is an error between the theoretical calculation and the true value of the catalyst or there may be other secondary factors that affect the generation of •O 2 − (Jin et al. 2022;Wang et al. 2021). Finally, the holes themselves have strong oxidizing properties, which can oxidize pollutants directly. In summary, the mechanism diagram Fig. 9 is roughly consistent with the identification results of RSs and ESR characterization.
Combining the photocatalytic efficiency of 4%B-TiO 2 with the band potential, active species recognition, and ESR signal capture results, a possible mechanism for the photocatalytic degradation of AP can be proposed. The specific reaction steps are shown in the following Eqs. (4-11) (Kumar et al. 2014 O, and other stable inorganic compounds. The generation of free radicals effectively suppresses the recombination of electron-hole pairs, prolongs the lifetime of carriers, and enables them to efficiently migrate to the surface of catalyst. The improvement of the separation ability of photogenerated charge carriers results in higher degradation efficiency (Ganesh 2018).
The secondary influencing factors mentioned above can be explained by ligand-metal charge transfer (LMCT) in surface adsorption sensitization and oxygen vacancy. Ligand-metal Radicals + Pollutant → Intermediate products → CO 2 + H 2 O charge transfer enhances the photoreactive activity of TiO 2 by sensitizing TiO 2 to form ligand TiO 2 complexes. Several studies have shown that small organic molecules (SOM) can be used for semiconductor surface modification (Chen et al. 2020;Hasan et al. 2018;Qin et al. 2022;Zhang et al. 2023).
Although SOM or TiO 2 cannot respond to visible light by themselves, surface complexes of SOM and TiO 2 can produce visible-driven LMCT. The surface ligands of modified TiO 2 generally contain carboxyl (− COOH) and hydroxyl (− OH) groups, such as catechol, chlorophenol, glucose, and organic acids. The functional groups on these ligands can be used as coordination centers to complex with semiconductors to form metal-organic complexes. In this paper, AP is used not only as substrate for degradation but also LMCT sensitizer. REDOX substance TiO 2 is easy to form oxygen vacancy under preparation condition of ionic doping, high-temperature calcination, and UV light reaction conditions. The existence of oxygen vacancy leads to the lack of electrons in the Ti atom near the oxygen vacancy, forming an electron-poor center on the Ti atom. The center is coupled with the organic pollutant AP that can give electron groups to form a LMCT complex under visible light response. Under the excitation of visible light, electrons transfer from the HOMO molecular orbital of the ligand to the CB of TiO 2 (Liu et al. 2021a;Vile et al. 2021). Thus, photocatalytic reaction process mediated by ROS was enhanced to promote the conversion of O 2 to •O 2 − . The action of LMCT and the existence of oxygen vacancy change the CB of TiO 2 and reduce the band gap. More electrons could be stimulated to the CB under UV light to produce •O 2 − , which participate in the degradation of AP.

Conclusion
The obtained consequences indicates that the best removal rate of 98.8% can be obtained after 30-min UV light illumination on 10-mg/L AP solution with 1 g/L 4%B-TiO 2 . Its photocatalytic performance depends on the B doping amount and calcination temperature, which are the determinants for the structure, porosity, interface, and optical Fig. 9 Schematic diagram of the mechanism for photocatalytic degradation of AP with 4%B-TiO 2 properties. Compared with undoped TiO 2 , 4%B-TiO 2 has characteristics of smaller grain size (25.62 nm), higher specific surface area (17.23m 2 /g), and lower band gap energy (3.11 eV). It is because of these structures and properties, the optimal catalyst 4%B-TiO 2 can generate efficient •OH in situ when degrading AP. And the quenching experiments and ESR characterization show that the attack of •OH on AP molecules is dominant. This research is meaningful because the degradation of AP by 4%B-TiO 2 is a good example for the treatment of pollutants like PPCPs. And the exploration of its mechanism also contributes to human health and ecological environment.
Author contribution Caofan Xiao conceived, designed the experiments, and wrote the manuscript. Xueqi Chen and Xiumei Tao provided suggestions for the study. Xian Liu supervised the progress of experiments, provided suggestions, and reviewed papers. Xun Wang provided suggestions and financial support. Lei Zhu reviewed the first draft. All authors discussed the results and contributed to the final manuscript.
Funding This work is supported by the National Natural Science Foundation of China (51672196). This article is also supported by the Graduate Innovation and Entrepreneurship Fund Project of Wuhan University of Science and Technology (JCX2021016).
Data availability All relevant data generated or analyzed during this study are included in this article.

Declarations
Ethics approval This manuscript does not contain any studies with human participants or animals performed by any of the authors.

Competing interests
The authors declare no competing interests.