Enhanced Photocatalytic Degradation of Caffeine Using Titanium Dioxide Photocatalyst Immobilized on Circular Glass Sheets under Ultraviolet C Irradiation

This work presents the development of titanium dioxide (TiO2) film immobilized on circular glass sheets for photocatalytic degradation of caffeine under ultraviolet C (UVC) irradiation. TiO2 was synthesized through the ultrasonic-assisted sol–gel method and immobilized on circular glass sheets by the doctor blade technique. Polyvinylpyrrolidone (PVP) was used to mix with the TiO2 precursor solution to enhance film adhesion on the glass surface. TiO2 film was mainly composed of anatase phase with a small amount of rutile phase. Caffeine removal was found to increase with increasing irradiation time. Caffeine (20 mg/L) in the synthetic wastewater could not be detected after 3 h of UVC irradiation. The reaction rate of caffeine degradation followed the pseudo-first-order model. The concentrated caffeine solutions required a longer irradiation time for degradation. The used TiO2-coated glass sheets could be easily separated from the treated wastewater and reusable. The caffeine removal efficiency of TiO2-coated glass sheets in each cycle maintained a high level (~100%) during fifteen consecutive cycles.


Introduction
Micro-organic pollutants are emerging substances mostly derived from human activities, such as pharmaceuticals and personal care products (PPCPs), perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA), pesticides, herbicides and many other compounds used in households, hospitals, industries, as well as agricultural areas. Although these substances have been detected in the natural water bodies with a trace concentration, it could be high enough to pose threats to the aquatic living organisms and affect the ecological balance [1,2].
improve the adhesion properties of TiO 2 film on supporting materials is the use of polymers as the binders [30,31]. Polyvinylpyrrolidone (PVP) is a thermoplastic polymer with nontoxic and hydrophilic properties [32]. PVP and its derivatives have been extensively applied in the industries of medicine, personal care products, beverages, cosmetics, paints, etc. [33]. There are several previously available reports about applications of PVP and its derivatives. For example, they were used as supports for metal catalytic systems [34,35], drug delivery [32], bone tissue generation [36], energy storage [37], and others. Furthermore, it has been reported that PVP can be used as a polymer binder to enhance the film-forming for TiO 2 coating. TiO 2 films with the assistance of PVP gave satisfactory results for photocatalytic oxidations of organic pollutants, such as Amido black-10B [30], methylene blue [31,38], and methyl orange [39]. However, the use of immobilized TiO 2 film as a photocatalyst in the degradation of caffeine remains scarce.
This work focuses on the development of TiO 2 film with good adherence to the surface of circular glass sheets for the degradation of caffeine. TiO 2 was synthesized via the ultrasonic-assisted sol-gel method and immobilized on circular glass sheets by the doctor blade technique. Polyvinylpyrrolidone was added to the TiO 2 sol in order to induce the polymeric film-forming and enhance the film adhesion on the glass surface. The prepared TiO 2 immobilized on circular glass sheets were used in the photocatalytic degradation of caffeine. The effects of irradiation time, initial caffeine concentration, and photocatalyst reusability were investigated.

Morphology of TiO 2 Film
Scanning electron microscopy (SEM) images of TiO 2 film immobilized on glass sheets are illustrated in Figure 1. There are no cracks in the TiO 2 film because polyvinylpyrrolidone in the TiO 2 gel solution can inhibit the violent evaporation of the solvent after coating on the solid substrates [31]. As can be seen in the cross-sectional image ( Figure 1C), the TiO 2 particles relatively flatten on the glass surface. The average TiO 2 film thickness is 1.61 ± 0.24 µm. TiO 2 films coated on substrates were composed of squeezed sphere-like TiO 2 particles because they were firmly pressed by a stirring rod during the doctor blade coating technique [31]. It was found that the glass surface is covered by TiO 2 particles with nearly orbicular shape from the top view ( Figure 1D).
Catalysts 2020, 10, x FOR PEER REVIEW 3 of 14 hydrophilic properties [32]. PVP and its derivatives have been extensively applied in the industries of medicine, personal care products, beverages, cosmetics, paints, etc. [33]. There are several previously available reports about applications of PVP and its derivatives. For example, they were used as supports for metal catalytic systems [34,35], drug delivery [32], bone tissue generation [36], energy storage [37], and others. Furthermore, it has been reported that PVP can be used as a polymer binder to enhance the film-forming for TiO2 coating. TiO2 films with the assistance of PVP gave satisfactory results for photocatalytic oxidations of organic pollutants, such as Amido black-10B [30], methylene blue [31,38], and methyl orange [39]. However, the use of immobilized TiO2 film as a photocatalyst in the degradation of caffeine remains scarce. This work focuses on the development of TiO2 film with good adherence to the surface of circular glass sheets for the degradation of caffeine. TiO2 was synthesized via the ultrasonic-assisted sol-gel method and immobilized on circular glass sheets by the doctor blade technique. Polyvinylpyrrolidone was added to the TiO2 sol in order to induce the polymeric film-forming and enhance the film adhesion on the glass surface. The prepared TiO2 immobilized on circular glass sheets were used in the photocatalytic degradation of caffeine. The effects of irradiation time, initial caffeine concentration, and photocatalyst reusability were investigated.

Morphology of TiO2 Film
Scanning electron microscopy (SEM) images of TiO2 film immobilized on glass sheets are illustrated in Figure 1. There are no cracks in the TiO2 film because polyvinylpyrrolidone in the TiO2 gel solution can inhibit the violent evaporation of the solvent after coating on the solid substrates [31]. As can be seen in the cross-sectional image ( Figure 1C), the TiO2 particles relatively flatten on the glass surface. The average TiO2 film thickness is 1.61 ± 0.24 μm. TiO2 films coated on substrates were composed of squeezed sphere-like TiO2 particles because they were firmly pressed by a stirring rod during the doctor blade coating technique [31]. It was found that the glass surface is covered by TiO2 particles with nearly orbicular shape from the top view ( Figure 1D).

Crystalline Structure of TiO 2 Film
The result of the X-ray diffraction (XRD) analysis of TiO 2 film immobilized on a glass sheet is shown in Figure 2. It can be seen that the diffraction peaks are corresponding to anatase TiO 2 . The characteristic peak of anatase at the diffraction angle of 25.4 • was detected with the highest intensity. Apart from the anatase phase, a diffraction peak corresponding to the rutile phase was also detected with relatively less intensity. Therefore, the synthesized TiO 2 film in this study was mainly composed of the anatase phase with a small track of the rutile phase. Ohno et al. [40] proposed that the mixture of phase anatase and rutile phase is the most suitable form of TiO 2 photocatalysts. The synergistic effects of anatase and rutile phase could lead to higher photocatalytic performance in the degradation of organic pollutants in water, compared with isolated anatase or rutile phase. The crystallite size obtained using Scherrer's formula was 11.13 nm. Sirirerkratana et al. [31] reported the sol-gel synthesis of TiO 2 without applying ultrasonic wave. It was observed that the crystallite size of TiO 2 in this work (with the ultrasonic wave) was smaller than that reported by Sirirerkratana et al. [31]. Neppolian et al. [41] also reported that the TiO 2 prepared by ultrasonic-assisted sol-gel method could possess a smaller crystallite size than the sol-gel method. The reason was that ultrasonic waves could produce cavitation bubbles in the TiO 2 solution, resulting in the formation of small crystallite sizes of TiO 2 [42]. The result of the X-ray diffraction (XRD) analysis of TiO2 film immobilized on a glass sheet is shown in Figure 2. It can be seen that the diffraction peaks are corresponding to anatase TiO2. The characteristic peak of anatase at the diffraction angle of 25.4° was detected with the highest intensity. Apart from the anatase phase, a diffraction peak corresponding to the rutile phase was also detected with relatively less intensity. Therefore, the synthesized TiO2 film in this study was mainly composed of the anatase phase with a small track of the rutile phase. Ohno et al. [40] proposed that the mixture of phase anatase and rutile phase is the most suitable form of TiO2 photocatalysts. The synergistic effects of anatase and rutile phase could lead to higher photocatalytic performance in the degradation of organic pollutants in water, compared with isolated anatase or rutile phase. The crystallite size obtained using Scherrer's formula was 11.13 nm. Sirirerkratana et al. [31] reported the sol-gel synthesis of TiO2 without applying ultrasonic wave. It was observed that the crystallite size of TiO2 in this work (with the ultrasonic wave) was smaller than that reported by Sirirerkratana et al. [31]. Neppolian et al. [41] also reported that the TiO2 prepared by ultrasonic-assisted sol-gel method could possess a smaller crystallite size than the sol-gel method. The reason was that ultrasonic waves could produce cavitation bubbles in the TiO2 solution, resulting in the formation of small crystallite sizes of TiO2 [42].

Effects of Irradiation Time and Photocatalyst Coated
The effects of irradiation time have been investigated with an initial caffeine concentration of 20 mg/L. Figure 3 shows the efficiencies of caffeine removal under ultraviolet C (UVC) irradiation over the uncoated and TiO2-coated glass sheets. For the uncoated glass sheet, the caffeine concentration was insignificantly reduced. The concentration maintained as high as the initial concentration after 3 h of UVC irradiation. This result is in agreement with the literature reported by Shu et al. [22], which studied the treatment of caffeine in water by direct photolysis at the wavelength range of 200-300 nm (UVC and UVB). It was found that direct photolysis is not having the ability to degrade caffeine in water because the reaction rate constant is very low. For TiO2-coated glass sheets, the adsorption of caffeine molecules on the surface of TiO2 film reached equilibrium within 1 h in the dark condition. After that, the UVC irradiation was performed. Caffeine was not completely detected after 3 h of irradiation. The efficiencies of caffeine removal were 69.52%, 90.11%, and 99.75% at 60, 120, and 180 min of UVC irradiation, respectively. This result indicated that the efficiency of caffeine removal could be enhanced by increasing irradiation time in the case of the photocatalyst-coated

Effects of Irradiation Time and Photocatalyst Coated
The effects of irradiation time have been investigated with an initial caffeine concentration of 20 mg/L. Figure 3 shows the efficiencies of caffeine removal under ultraviolet C (UVC) irradiation over the uncoated and TiO 2 -coated glass sheets. For the uncoated glass sheet, the caffeine concentration was insignificantly reduced. The concentration maintained as high as the initial concentration after 3 h of UVC irradiation. This result is in agreement with the literature reported by Shu et al. [22], which studied the treatment of caffeine in water by direct photolysis at the wavelength range of 200-300 nm (UVC and UVB). It was found that direct photolysis is not having the ability to degrade caffeine in water because the reaction rate constant is very low. For TiO 2 -coated glass sheets, the adsorption of caffeine molecules on the surface of TiO 2 film reached equilibrium within 1 h in the dark condition. After that, the UVC irradiation was performed. Caffeine was not completely detected after 3 h of irradiation. The efficiencies of caffeine removal were 69.52%, 90.11%, and 99.75% at 60, 120, and 180 min of UVC irradiation, respectively. This result indicated that the efficiency of caffeine removal could be enhanced by increasing irradiation time in the case of the photocatalyst-coated sample. Arfanis et al. [11] also reported that the efficiencies of caffeine removal at the natural pH increased by increasing UV irradiation time in the photocatalytic process.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 14 sample. Arfanis et al. [11] also reported that the efficiencies of caffeine removal at the natural pH increased by increasing UV irradiation time in the photocatalytic process. During irradiation of UV light over the TiO2 photocatalyst, the electron-hole pairs can be generated. The photo-excited electrons are transferred from the valence band to the conduction band, leaving the positive holes in the valence band. Subsequently, the photo-excited electrons and holes can migrate from bulk to the surface of TiO2, reacting with the surface adsorbed molecules to produce hydroxyl radicals. These radicals can degrade the surface of adsorbed caffeine molecules to obtain more readily degradable intermediate species and final harmless products according to the equations reported by Nosaka et al. [43], Li et al. [44], Abdellah et al. [45], and Ge et al. [46].
For the degradation mechanisms of caffeine by the photocatalytic process, the double bond of C4 and C8 in the caffeine molecule (C8H10N4O2; Mw = 194) was rapidly attacked by hydroxyl radicals. After that, dimethylparabanic acid (C5H6N2O3; Mw = 142) was formed as the first intermediate species and subsequently react with hydroxyl radicals to form di (N-hydroxymethyl) parabanic acid (C5H6N2O5, Mw = 174) as the second intermediate species. This second intermediate species could be gradually mineralized to CO2, NH3, and NH2Me [9].

Effects of Initial Concentration
The effects of initial concentration were studied using five different caffeine concentrations, which were 1, 5, 10, 20, and 40 mg/L. The relationship between initial caffeine concentrations and photocatalytic removal efficiencies are revealed in Figure 4. At the same irradiation time, it was found that the removal efficiency of caffeine decreased with increasing initial caffeine concentration. The removal efficiencies achieved higher than 99% after 40, 100, 140, 180, and 180 min of UVC irradiation for the initial caffeine concentrations of 1, 5, 10, 20, and 40 mg/L, respectively. It can be concluded that the higher caffeine concentrations required more irradiation time than the lower caffeine concentrations for degradation. Luna et al. [13] also reported that the lower caffeine concentrations could be completely removed in shorter sunlight radiation time compared with the higher caffeine concentration. This phenomenon can be explained that some caffeine molecules in the concentrated solutions cannot contact the surface of TiO2 film to initiate the photocatalytic reaction [47]. During irradiation of UV light over the TiO 2 photocatalyst, the electron-hole pairs can be generated. The photo-excited electrons are transferred from the valence band to the conduction band, leaving the positive holes in the valence band. Subsequently, the photo-excited electrons and holes can migrate from bulk to the surface of TiO 2 , reacting with the surface adsorbed molecules to produce hydroxyl radicals. These radicals can degrade the surface of adsorbed caffeine molecules to obtain more readily degradable intermediate species and final harmless products according to the equations reported by Nosaka et al. [43], Li et al. [44], Abdellah et al. [45], and Ge et al. [46].
For the degradation mechanisms of caffeine by the photocatalytic process, the double bond of C4 and C8 in the caffeine molecule (C 8

Effects of Initial Concentration
The effects of initial concentration were studied using five different caffeine concentrations, which were 1, 5, 10, 20, and 40 mg/L. The relationship between initial caffeine concentrations and photocatalytic removal efficiencies are revealed in Figure 4. At the same irradiation time, it was found that the removal efficiency of caffeine decreased with increasing initial caffeine concentration. The removal efficiencies achieved higher than 99% after 40, 100, 140, 180, and 180 min of UVC irradiation for the initial caffeine concentrations of 1, 5, 10, 20, and 40 mg/L, respectively. It can be concluded that the higher caffeine concentrations required more irradiation time than the lower caffeine concentrations for degradation. Luna et al. [13] also reported that the lower caffeine concentrations could be completely removed in shorter sunlight radiation time compared with the higher caffeine concentration. This phenomenon can be explained that some caffeine molecules in the concentrated solutions cannot contact the surface of TiO 2 film to initiate the photocatalytic reaction [47]. The kinetic plots of ln (C0/C) versus irradiation time at various initial caffeine concentrations are revealed in Figure 5. It can be seen that photocatalytic degradation of caffeine at irradiation times of 0-80 min followed the pseudo-first-order kinetic model (Equation (1)). The apparent initial rate constant (k; min −1 ) can be determined by the integration form of pseudo-first-order kinetic model (Equation (2)), where C0 is the initial caffeine concentration (mg/L), t is irradiation time (min), and C is the remained concentration of caffeine at a different irradiation time (mg/L).  The kinetic plots of ln (C 0 /C) versus irradiation time at various initial caffeine concentrations are revealed in Figure 5. It can be seen that photocatalytic degradation of caffeine at irradiation times of 0-80 min followed the pseudo-first-order kinetic model (Equation (1)). The apparent initial rate constant (k; min −1 ) can be determined by the integration form of pseudo-first-order kinetic model (Equation (2)), where C 0 is the initial caffeine concentration (mg/L), t is irradiation time (min), and C is the remained concentration of caffeine at a different irradiation time (mg/L).  The kinetic plots of ln (C0/C) versus irradiation time at various initial caffeine concentrations are revealed in Figure 5. It can be seen that photocatalytic degradation of caffeine at irradiation times of 0-80 min followed the pseudo-first-order kinetic model (Equation (1)). The apparent initial rate constant (k; min −1 ) can be determined by the integration form of pseudo-first-order kinetic model (Equation (2)), where C0 is the initial caffeine concentration (mg/L), t is irradiation time (min), and C is the remained concentration of caffeine at a different irradiation time (mg/L).
The relationship between initial caffeine concentration and the apparent initial rate constant is shown in Figure 6. The apparent initial rate constant (k) at initial caffeine concentration of 5, 10, 20, and 40 mg/L were 0.0302, 0.0246, 0.0214, 0.0121 min −1 , respectively. It could be noted that the apparent initial rate constant (k) decreased with increasing initial caffeine concentration. These results are in agreement with other studies, which reported the photocatalytic degradation of caffeine using TiO 2 powder. The degradation rate of caffeine followed the pseudo-first-order kinetics model. The initial apparent rate constant (k) of caffeine degradation decreased with the increasing initial caffeine concentration [5,13].
Catalysts 2020, 10, x FOR PEER REVIEW 7 of 14 The relationship between initial caffeine concentration and the apparent initial rate constant is shown in Figure 6. The apparent initial rate constant (k) at initial caffeine concentration of 5, 10, 20, and 40 mg/L were 0.0302, 0.0246, 0.0214, 0.0121 min −1 , respectively. It could be noted that the apparent initial rate constant (k) decreased with increasing initial caffeine concentration. These results are in agreement with other studies, which reported the photocatalytic degradation of caffeine using TiO2 powder. The degradation rate of caffeine followed the pseudo-first-order kinetics model. The initial apparent rate constant (k) of caffeine degradation decreased with the increasing initial caffeine concentration [5,13].

Reusability of Photocatalyst
Reusability of the photocatalyst was investigated by repeating the use of TiO2-coated glasses for photocatalytic degradation of caffeine in fifteen consecutive cycles. After the photocatalytic experiment completed in each cycle, the TiO2-coated glasses were easily separated from the treated wastewater. Afterward, the TiO2-coated glasses and new caffeine solution was replaced in the container for the next cycle test. The treated wastewater was taken to investigate the adhesion of TiO2 film on the glass surface by microcentrifuge at 12,000 rpm for 3 min. The water at the bottom of the microcentrifuge tubes was dropped on an SEM specimen stub. The water then evaporated. The stub was investigated by SEM. It was observed that no TiO2 powder settled at the bottom of the microcentrifuge tubes. This result indicated that the adhesion of TiO2 film on the glass surface was excellent. As can be seen in Figure 7, the removal efficiency of caffeine in each cycle maintained a high level (~100%) during fifteen consecutive cycles under UVC irradiation. This result indicated that the immobilization of TiO2 as a thin film on glass could solve the problem in the post-separation of TiO2 from treated wastewater, and they are reusable.

ln (C0/C) = kt
(2) Figure 6. Effects of initial caffeine concentration on the initial pseudo-first-order rate constant of caffeine degradation under UVC irradiation.

Reusability of Photocatalyst
Reusability of the photocatalyst was investigated by repeating the use of TiO 2 -coated glasses for photocatalytic degradation of caffeine in fifteen consecutive cycles. After the photocatalytic experiment completed in each cycle, the TiO 2 -coated glasses were easily separated from the treated wastewater. Afterward, the TiO 2 -coated glasses and new caffeine solution was replaced in the container for the next cycle test. The treated wastewater was taken to investigate the adhesion of TiO 2 film on the glass surface by microcentrifuge at 12,000 rpm for 3 min. The water at the bottom of the microcentrifuge tubes was dropped on an SEM specimen stub. The water then evaporated. The stub was investigated by SEM. It was observed that no TiO 2 powder settled at the bottom of the microcentrifuge tubes. This result indicated that the adhesion of TiO 2 film on the glass surface was excellent. As can be seen in Figure 7, the removal efficiency of caffeine in each cycle maintained a high level (~100%) during fifteen consecutive cycles under UVC irradiation. This result indicated that the immobilization of TiO 2 as a thin film on glass could solve the problem in the post-separation of TiO 2 from treated wastewater, and they are reusable.

Discussion
TiO2 solution synthesized via the ultrasonic-assisted sol-gel method with polyvinylpyrrolidone showed a unique morphology and attached well to a circular glass sheet after coating. The thickness of the TiO2 film was properly controlled by tape strip in the doctor blade techniques. TiO2-coated glass exhibited high performance (~100%, within 3 h) and maintained a high performance after fifteen consecutive cycles in photocatalytic degradation of caffeine under UVC irradiation. Moreover, they were easy to handle and separate from the treated wastewater. In consideration of caffeine removal in the previous works (Table 1), it was found that the caffeine removal by the TiO2 powder system was almost complete within 3-8 h under solar light for the commercial TiO2 [5,13]. For sol-gel synthesized TiO2 powder, the removal was 80% within 6 h under ultraviolet A (UVA) irradiation [48]. Whereas, the maximum efficiencies of caffeine removal were in the range of 50-65% within 4 h for the works that used TiO2 film [11] and TiO2 nanotube [11] as photocatalysts under UVA irradiation. Although the previous works showed that the complete removal could be achieved by using TiO2 powder in the degradation of caffeine, it is still inconvenient to apply on the full scale. This is because the TiO2 powder needs to be separated from the effluent by the microfiltration unit before discharging into the natural water bodies. As described above, it can be attributed that the coating procedure and the experimental set up under UVC irradiation in this study are proven to be one of the effective techniques for the treatment of caffeine in water, which can overcome the limitation of post-separation in the TiO2 powder system. The photocatalytic reactor under UVC irradiation in this study can be scaled up for the treatment of caffeine contaminated wastewater (e.g., wastewater from coffee processing, coffee shops) in a closed system

Discussion
TiO 2 solution synthesized via the ultrasonic-assisted sol-gel method with polyvinylpyrrolidone showed a unique morphology and attached well to a circular glass sheet after coating. The thickness of the TiO 2 film was properly controlled by tape strip in the doctor blade techniques. TiO 2 -coated glass exhibited high performance (~100%, within 3 h) and maintained a high performance after fifteen consecutive cycles in photocatalytic degradation of caffeine under UVC irradiation. Moreover, they were easy to handle and separate from the treated wastewater. In consideration of caffeine removal in the previous works (Table 1), it was found that the caffeine removal by the TiO 2 powder system was almost complete within 3-8 h under solar light for the commercial TiO 2 [5,13]. For sol-gel synthesized TiO 2 powder, the removal was 80% within 6 h under ultraviolet A (UVA) irradiation [48]. Whereas, the maximum efficiencies of caffeine removal were in the range of 50-65% within 4 h for the works that used TiO 2 film [11] and TiO 2 nanotube [11] as photocatalysts under UVA irradiation. Although the previous works showed that the complete removal could be achieved by using TiO 2 powder in the degradation of caffeine, it is still inconvenient to apply on the full scale. This is because the TiO 2 powder needs to be separated from the effluent by the microfiltration unit before discharging into the natural water bodies. As described above, it can be attributed that the coating procedure and the experimental set up under UVC irradiation in this study are proven to be one of the effective techniques for the treatment of caffeine in water, which can overcome the limitation of post-separation in the TiO 2 powder system. The photocatalytic reactor under UVC irradiation in this study can be scaled up for the treatment of caffeine contaminated wastewater (e.g., wastewater from coffee processing, coffee shops) in a closed system until achieved complete removal. Afterward, the treated water can be discharged into the aquatic environment.

Preparation of TiO 2 Immobilized Glass Sheets
Circular transparent glass sheets were pretreated by immersion in 2-propanol for 30 min to remove impurities. They were then dried in an oven (Binder, Model FD115, Tuttlingen, Germany) at 105 • C for 1 h. The TiO 2 gel solution was prepared by the ultrasonic-assisted sol-gel method. PVP (0.8 g) was dissolved in methanol (20 mL) and stirred for 30 min. TBOT (4 g) and ACA (1.175 g) were subsequently added into the mixture. The mixture was stirred again for 30 min and checked in an ultrasonic clean bath (Crest Ultrasonics, Model 690HTAE, New York, NY, USA) at 50 kHz at room temperature for 30 min. After that, the TiO 2 gel solution (1 mL) was dropped on the surface of a glass sheet and immobilized by the doctor blade technique (using a glass rod as the blade with plastic films as the frames). The immobilized glasses were dried in an oven at 105 • C for 1 h and were then calcined in an electric muffle furnace (Carbolite, Model CWF 11/13, Sheffield, UK) at 500 • C for 2 h. The obtained TiO 2 film immobilized glass sheet is shown in Figure 8.
Catalysts 2020, 10, x FOR PEER REVIEW 10 of 14 calcined in an electric muffle furnace (Carbolite, Model CWF 11/13, Sheffield, UK) at 500 °C for 2 h. The obtained TiO2 film immobilized glass sheet is shown in Figure 8.

Characterizations
The TiO2 film immobilized glass sheets were cut into 1 × 1 cm 2 and coated by gold for determination of the surface and the cross-section morphology by a JEOL JSM-6610LV scanning electron microscope (SEM, Peabody, Massachusetts, MA, USA) operated at 15 kV with the magnifications of 5000× and 10,000×. The sheets were cut into 2 × 2 cm 2 to characterize the crystalline structure of TiO2 by a AXS Bruker D8 DISCOVER X-ray diffractometer (XRD, Coventry, UK) with Cu-Kα radiation at the wavelength of 0.15406 nm. The Scherrer's formula (Equation (3)) was used to determine the crystallite size of TiO2 (τ) from the broadening of the (101) reflection of anatase [49].
where K is the form of factor (0.9), β is the width of the peak at half the maximum intensity after subtraction of the instrumental noise, and θ is the diffraction angle.

Photocatalytic Reactor
The photocatalytic reactor in this experiment is shown in Figure 9. The setup is in a rectangular plywood chamber with 50 × 50 × 70 cm 3 in dimensions. The irradiation was performed using four ultraviolet C (UVC) 16.7 W lamps (Philips, TUV F17T8, wavelength = 253.7 nm). The UVC lamps were installed on the top of the chamber. Note that UVC radiation was chosen as a light source because of its high energy, which can induce high performance in the photocatalytic process within a short irradiation time. The photocatalytic reactor under UVC irradiation in this study can be scaled up for the treatment of actual caffeine wastewater in a closed system until achieved complete removal. Afterward, the treated water will be discharged into the aquatic environment. Human and other life forms in the aquatic environment will not expose to the UVC ray from this process. The containers for caffeine solutions (inside diameter = 9 cm, height = 1.5 cm) were placed in a mechanical shaker (60 rpm) at a distance of 25 cm from the irradiation source. The immobilized glass sheets were placed at the bottoms of the glass containers.

Characterizations
The TiO 2 film immobilized glass sheets were cut into 1 × 1 cm 2 and coated by gold for determination of the surface and the cross-section morphology by a JEOL JSM-6610LV scanning electron microscope (SEM, Peabody, Massachusetts, MA, USA) operated at 15 kV with the magnifications of 5000× and 10,000×. The sheets were cut into 2 × 2 cm 2 to characterize the crystalline structure of TiO 2 by a AXS Bruker D8 DISCOVER X-ray diffractometer (XRD, Coventry, UK) with Cu-K α radiation at the wavelength of 0.15406 nm. The Scherrer's formula (Equation (3)) was used to determine the crystallite size of TiO 2 (τ) from the broadening of the (101) reflection of anatase [49].
where K is the form of factor (0.9), β is the width of the peak at half the maximum intensity after subtraction of the instrumental noise, and θ is the diffraction angle.

Photocatalytic Reactor
The photocatalytic reactor in this experiment is shown in Figure 9. The setup is in a rectangular plywood chamber with 50 × 50 × 70 cm 3 in dimensions. The irradiation was performed using four ultraviolet C (UVC) 16.7 W lamps (Philips, TUV F17T8, wavelength = 253.7 nm). The UVC lamps were installed on the top of the chamber. Note that UVC radiation was chosen as a light source because of its high energy, which can induce high performance in the photocatalytic process within a short irradiation time. The photocatalytic reactor under UVC irradiation in this study can be scaled up for the treatment of actual caffeine wastewater in a closed system until achieved complete removal. Afterward, the treated water will be discharged into the aquatic environment. Human and other life forms in the aquatic environment will not expose to the UVC ray from this process. The containers for caffeine solutions (inside diameter = 9 cm, height = 1.5 cm) were placed in a mechanical shaker (60 rpm) at a distance of 25 cm from the irradiation source. The immobilized glass sheets were placed at the bottoms of the glass containers.

Photocatalytic Experiment
The caffeine solution was prepared in deionized water. The solution (25 mL) was added into the prepared container. An immobilized glass sheet was soaked in the solution for 1 h in the dark to attain equilibrium adsorption. Subsequently, the UVC lamps were turned on and samples were collected every time intervals of 20 min until 180 min. For the initial concentration experiment, caffeine solutions were prepared at concentrations of 1, 5, 10, 20, and 40 mg/L. In the case of a reusability experiment, the immobilized glass sheet was tested for fifteen consecutive cycles. The treated caffeine solution was changed before the next cycle.

Analysis of Caffeine Concentration
The collected sample was filtered by a nylon syringe filter (diameter of 0.22 μm). After that, the filtered solution was analyzed by high performance liquid chromatography (HPLC) using an Alliance Waters e2695 instrument (Milford, Massachusetts, MA, USA) equipped with a photodiode array detector (Waters 2998) and a Vertisep AQS C18 HPLC column (4.6 mm × 150 mm, 5 μm). A solvent was pumped at a fixed flow rate of 1 mL/min. The mobile phase consisted of ultra-pure water (50%) and methanol (50%). Each sample (10 μL) was automatically injected for analysis of caffeine concentration at the wavelength of 254 nm. The peak area of caffeine was displayed on the chromatogram after 3 min of the sample injection.

Conclusions
TiO2 synthesized through the ultrasonic-assisted sol-gel method was successfully immobilized as a film on the circular glass sheets. Polyvinylpyrrolidone was used to mix with the TiO2 precursor solution to enhance film adhesion on the glass surface. The direct photolysis under UVC irradiation was proven to be ineffective for caffeine removal. On the other hand, the efficiency of caffeine removal by the photocatalytic oxidation process strongly depended on the time of UVC irradiation and the initial concentration of caffeine. The completed degradation of caffeine in aqueous solutions could be achieved by increasing irradiation time up to 180 min. The degradation rate of caffeine followed the pseudo-first-order kinetic model and tended to increase with decreasing initial caffeine concentration. Furthermore, TiO2-coated glass sheets could be reused as a photocatalyst for water treatment without a significant decrease in its activity after fifteen consecutive cycles.

Photocatalytic Experiment
The caffeine solution was prepared in deionized water. The solution (25 mL) was added into the prepared container. An immobilized glass sheet was soaked in the solution for 1 h in the dark to attain equilibrium adsorption. Subsequently, the UVC lamps were turned on and samples were collected every time intervals of 20 min until 180 min. For the initial concentration experiment, caffeine solutions were prepared at concentrations of 1, 5, 10, 20, and 40 mg/L. In the case of a reusability experiment, the immobilized glass sheet was tested for fifteen consecutive cycles. The treated caffeine solution was changed before the next cycle.

Analysis of Caffeine Concentration
The collected sample was filtered by a nylon syringe filter (diameter of 0.22 µm). After that, the filtered solution was analyzed by high performance liquid chromatography (HPLC) using an Alliance Waters e2695 instrument (Milford, Massachusetts, MA, USA) equipped with a photodiode array detector (Waters 2998) and a Vertisep AQS C18 HPLC column (4.6 mm × 150 mm, 5 µm). A solvent was pumped at a fixed flow rate of 1 mL/min. The mobile phase consisted of ultra-pure water (50%) and methanol (50%). Each sample (10 µL) was automatically injected for analysis of caffeine concentration at the wavelength of 254 nm. The peak area of caffeine was displayed on the chromatogram after 3 min of the sample injection.

Conclusions
TiO 2 synthesized through the ultrasonic-assisted sol-gel method was successfully immobilized as a film on the circular glass sheets. Polyvinylpyrrolidone was used to mix with the TiO 2 precursor solution to enhance film adhesion on the glass surface. The direct photolysis under UVC irradiation was proven to be ineffective for caffeine removal. On the other hand, the efficiency of caffeine removal by the photocatalytic oxidation process strongly depended on the time of UVC irradiation and the initial concentration of caffeine. The completed degradation of caffeine in aqueous solutions could be achieved by increasing irradiation time up to 180 min. The degradation rate of caffeine followed the pseudo-first-order kinetic model and tended to increase with decreasing initial caffeine concentration. Furthermore, TiO 2 -coated glass sheets could be reused as a photocatalyst for water treatment without a significant decrease in its activity after fifteen consecutive cycles.
Author Contributions: Conceptualization-P.K. and S.C.; Analysis and discussion of the characterization results-R.M., P.K., and S.C.; Supervision-P.K.; Experiments and writing of the original manuscript-R.M.; Manuscript corrections-P.K., P.P., and S.C. All authors have read and agreed to the published version of the manuscript.