Kinetic and Thermodynamic Study of Methylene Blue Adsorption on TiO2 and ZnO Thin Films

In this work, we fabricated and characterized ZnO and TiO2 thin films, determining their structural, optical, and morphological properties. Furthermore, we studied the thermodynamics and kinetics of methylene blue (MB) adsorption onto both semiconductors. Characterization techniques were used to verify thin film deposition. The semiconductor oxides reached different removal values, 6.5 mg/g (ZnO) and 10.5 mg/g (TiO2), after 50 min of contact. The pseudo-second-order model was suitable for fitting the adsorption data. ZnO had a greater rate constant (45.4 × 10−3) than that of TiO2 (16.8 × 10−3). The removal of MB by adsorption onto both semiconductors was an endothermic and spontaneous process. Finally, the stability of the thin films showed that both semiconductors maintained their adsorption capacity after five consecutive removal tests.


Introduction
The world's population growth and the energy and water requirements by industries (e.g., petrochemical [1], pharmaceutical [2], textile [3], agrochemical [4], fuels [5], plastics [6]) have caused a severe threat to the environment. Water pollution makes water unsafe for fauna and humans, affecting different environmental systems [7]. One of the challenges for this century is to ensure that the population has access to safe water; the Organization for Economic Co-operation and Development (OECD) recommends that governments encourage the joint management of water quantity and quality [8]. Various techniques for water remediation have been implemented in the last decades (e.g., physical, chemical, and biological treatment technologies) [9]. Among these methods, the adsorption method (a physical method) has received attention due to its low cost and its effectiveness in removing contaminants from water [10]. During the adsorption process, the pollutant is retained on the substrate surface. Adsorption can be described as a chemical (covalent bond) or physical (weak electrostatic interactions) interaction between an adsorbate and adsorbent surface [11]. Different materials have been used to apply the adsorption process (e.g., zeolites, [12], alumina [13], clay [14], active carbon [15], biomass [16], semiconductors [17], MOF [18]). In the literature, there are various reviews on dye removal by adsorption using different materials [19][20][21]. Metal oxides have two synergic properties: (i) they can act as an adsorbent and (ii) as antimicrobial agents [22]. Furthermore, because semiconductors have variable oxidation states, large surface areas (e.g., as nanomaterials), and great versatility, they can be used for environmental control and contaminant removal [23]. Khoshhesab et al. reported that nanoparticles of ZnO had 92.3% of adsorption capacity in the removal of Congo red from a solution (75 ppm) after 120 min of contact [24]. Syarif et al. reported that nanoparticles of CuO had 61.0% of adsorption capacity in the removal of methylene blue from a solution (5 ppm) after 10 min of contact with CuO nanoparticles [25]. Noreen et al. utilized Fe 3 O 4 nanoparticles to remove a reactive blue dye from a solution, and reported 35 mg/g of adsorption capacity after 10 min of contact [26]. Abdullah et al. prepared MnO 2 2 of 12 nanoparticles to remove methylene blue from an aqueous solution, and reported 22.2 mg/g of adsorption capacity after 60 min of contact [27]. ZnO and TiO 2 are alternative adsorbents, as they are innocuous to the environment, they are chemically and physically stable, and have adequate surface properties (e.g., roughness, porosity, and surface area) [28,29].
Currently, in heterogeneous photocatalysis, as a previous step to the photocatalytic degradation process, the sorption/desorption equilibrium is required. However, the adsorption process studied is not commonly reported in photocatalytic studies [30]. Although there is a high potential of ZnO and TiO 2 as adsorbents, there are few reports on the thermodynamic study of dye adsorption onto the surface of these semiconductors. In this contribution, we synthesized and characterized ZnO and TiO 2 thin films and studied the kinetics and thermodynamics involved in the removal of MB by adsorption onto both thin films.

Synthesis and Characterization of Thin Film Deposition
We used ammonium hydroxide and zinc acetate in the synthesis of ZnO powders, according to the procedure described in a previous report [31]. We used Degussa powder (P25) (Sigma-Aldrich, 99.5%, St. Louis, MO, USA) as a source of TiO 2 in the fabrication of TiO 2 thin films, according to the procedure described in a previous report [32]. We immobilized all catalysts on solid substrate to solve problems regarding catalyst removal after finishing the photocatalytic procedure [33]. We utilized the Doctor Blade technique for thin film deposition: First, we prepared a mixture of ZnO or TiO 2 powders, polyethylene glycol (PEG 5000) (Sigma-Aldrich, 99%, St. Louis, MO, USA), isopropyl alcohol (Sigma-Aldrich, 99%, St. Louis, MO, USA), and water. After suspension stabilization, the slurry was loaded into a soda lime substrate by the Doctor Blade method. Finally, the thin films were sintered at 500 • C for 1 h [31,34]. The thin films were characterized by diffuse reflectance spectroscopy measurements, providing information about the optical band gap energy of the semiconductors; by Raman spectroscopy assays, which allowed verifying the presence of ZnO and TiO 2 in the coatings; by X-ray diffraction measurements, which provided information about the crystalline structure of the thin films; and by scanning electron microscopy (SEM) assays, which allowed verifying their morphological properties.

Adsorption Kinetic and Thermodynamic Study
The semiconductors' films were immersed in a solution of methylene blue-MB (10 mL; 10 mg/L) (Sigma-Aldrich, ≥95%, St. Louis, MO, USA) contained in a glass batch reactor provided with an air bubbling system (0.5 L/min). The reactor was stored in the dark to study the MB adsorption process on the films. An aliquot was extracted at time zero and every 5 min thereafter for 50 min to determine the adsorption-desorption equilibrium time. We determined MB concentration by spectrophotometry at 665 nm using the Lambert-Beer law with a calibration curve (R 2 = 0.997). We determined the adsorption capacity of MB on the semiconductors according to [35]: where q t is the amount (mg) of MB adsorbed per gram of semiconductor (mg/g) at each time; C 0 is the initial MB concentration (mg/L); and m (g) is the amount of semiconductor. We applied the pseudo-first-order (PFO) and pseudo-second-order (PSO) models to fit experimental data according to these equations [35]: where q t is the amount of MB adsorbed per unit mass of the adsorbent (mg·g −1 ) at each time; q e is the maximum sorption capacity (mg ·g −1 ); and k 1 (min −1 ) and k 2 (g·mg −1 ·min −1 ) are the rate constants of the pseudo-first-and pseudo-second-order models, respectively. The fitting correlation coefficient (R 2 ) was used to determine the best-fitting kinetic models. Finally, we calculated standard enthalpy (∆H • ), standard entropy (∆S • ), and standard Gibbs free energy (∆G • ) for the adsorption process applying the Arrhenius equation [36]:

Raman Characterization
The Raman spectroscopy results are shown in Figure 1. Six Raman-active vibrational modes were observed for TiO 2 in Raman spectroscopy (e.g., A 1g + 2B 1g + 3E g ) [37]. Three Raman-active vibrational modes were observed for ZnO in Raman spectroscopy (e.g., A 1 + E 1 + E 2 ) [38]. Both catalysts shows the typical signals reported for such materials [39,40]. For the case of ZnO, the signals located at 274.5 cm −1 can be associated with oxygen vacancies into the semiconductor lattice [41,42].
Materials 2023, 16, x FOR PEER REVIEW the rate constants of the pseudo-first-and pseudo-second-order models, respectiv fitting correlation coefficient (R 2 ) was used to determine the best-fitting kinetic Finally, we calculated standard enthalpy (ΔH°), standard entropy (ΔS°), and s Gibbs free energy (ΔG°) for the adsorption process applying the Arrhenius equat

Raman Characterization
The Raman spectroscopy results are shown in Figure 1. Six Raman-active vib modes were observed for TiO2 in Raman spectroscopy (e.g., A1g + 2B1g + 3Eg) [37 Raman-active vibrational modes were observed for ZnO in Raman spectroscopy + E1 + E2) [38]. Both catalysts shows the typical signals reported for such materials For the case of ZnO, the signals located at 274.5 cm −1 can be associated with oxy cancies into the semiconductor lattice [41,42].  Figure 2 shows the (experimental and simulated) structural results for both T ZnO thin films. ZnO was polycrystalline, whose sample shows a plane of pre growth located at 2θ = 36.27. This signal is assigned to plane (101), where ZnO th show six other preferential growth planes, with all these reflections correspondin hexagonal wurtzite phase (JCPDS No. 36−1451) [43]. For the XRD-TiO2 pattern, was polycrystalline and was formed by two different crystalline structures: rutile #021-1276) and anatase (JCPDS #071-1166). During thin film deposition, we utilize source (Degussa-P25), this material being a mixture of those two crystalline phas

Structural Characterization
We utilized a PowderCell package to simulate the experimental XRD data simulation, we employed the rutile and anatase forms of TiO2, and hexagonal (ZnO) crystalline structures. We applied the Rietveld method (Bragg-Brentano g with the March-Dollase as model to preferred orientation; with the plane the pla as plane's orientation. The X-ray source was Cu K radiation ( = 0.1544426 n   Figure 2 shows the (experimental and simulated) structural results for both TiO 2 and ZnO thin films. ZnO was polycrystalline, whose sample shows a plane of preferential growth located at 2θ = 36.27. This signal is assigned to plane (101), where ZnO thin films show six other preferential growth planes, with all these reflections corresponding to the hexagonal wurtzite phase (JCPDS No. 36−1451) [43]. For the XRD-TiO 2 pattern, the TiO 2 was polycrystalline and was formed by two different crystalline structures: rutile (JCPDS #021-1276) and anatase (JCPDS #071-1166). During thin film deposition, we utilized a TiO 2 source (Degussa-P25), this material being a mixture of those two crystalline phases [44]. determine grain size of the semiconductors [45]. The domain grain size of ZnO was 34.4 nm, and 24.1 nm and 38.8 nm for anatase and rutile structures, respectively. These results correspond to those of previous reports by other authors [44,46].

Morphological Characterization
Morphological properties are determined by the experimental conditions and deposition method [47]. We synthesized ZnO using the sol-gel method, and we utilized Degussa P25 as the TiO2 source. Figure 3 shows the morphological results for TiO2 and ZnO. These results show that the thin films' surfaces are heterogeneous and porous, that TiO2 and ZnO are composed of microaggregates of different sizes, and that the agglomerated particles have two different spherical sized (50-80 nm to TiO2 and ~220nm to ZnO). Figure  3a shows typical morphological properties for Degussa P25 TiO2 [48]. The quasi-spherical ZnO nanoparticles are a commonly reported result when the sol-gel method is employed as a synthesis method [49]. Various authors have reported that the surface properties of the semiconductors are affected by synthesis method employed for their fabrication [23]. 20   We utilized a PowderCell package to simulate the experimental XRD data. In the simulation, we employed the rutile and anatase forms of TiO 2 , and hexagonal wurtzite (ZnO) crystalline structures. We applied the Rietveld method (Bragg-Brentano geometry with the March-Dollase as model to preferred orientation; with the plane the plane (101) as plane's orientation. The X-ray source was Cu K α radiation (λ = 0.1544426 nm); the pseudo-Voigt 1 function iterations 300; and the ϕ factor was 1.9. This methodology was suitable to identify the crystalline phases in each thin film. Table 1 lists the crystalline parameters obtained from the simulations. We employed the Debye-Scherrer equation to determine grain size of the semiconductors [45]. The domain grain size of ZnO was 34.4 nm, and 24.1 nm and 38.8 nm for anatase and rutile structures, respectively. These results correspond to those of previous reports by other authors [44,46].

Morphological Characterization
Morphological properties are determined by the experimental conditions and deposition method [47]. We synthesized ZnO using the sol-gel method, and we utilized Degussa P25 as the TiO 2 source. Figure 3 shows the morphological results for TiO 2 and ZnO. These results show that the thin films' surfaces are heterogeneous and porous, that TiO 2 and ZnO are composed of microaggregates of different sizes, and that the agglomerated particles have two different spherical sized (50-80 nm to TiO 2 and~220nm to ZnO). Figure 3a shows typical morphological properties for Degussa P25 TiO 2 [48]. The quasi-spherical ZnO nanoparticles are a commonly reported result when the sol-gel method is employed as a synthesis method [49]. Various authors have reported that the surface properties of the semiconductors are affected by synthesis method employed for their fabrication [23].  Figure 4 shows optical results for the ZnO and TiO2 semiconductors. Both of them show a high reflectance of approximately (or greater than) 60% after 360 nm. ZnO and TiO2 are not active under visible irradiation due to their high band gap (Eg). We determined the Eg value using the Kubelka-Munk remission function [50]. Figure 4b shows the Eg estimation for each thin film. The estimated bad gaps for the thin films are shown in Figure 4b [51]. These results correspond to those of previous reports for ZnO and TiO2 Degussa P25 [52,53]. The spectroscopic and structural characterization verified the presence of ZnO and TiO2 in the coatings synthesized.

Adsorption Kinetic Study
The adsorption of dye onto a semiconductor surface is a principle that relies on two steps: (i) diffusion of reactants onto the semiconductor surface and (ii) adsorption of reactants onto the semiconductor surface. The first step (the diffusion process) (i) follows the classic laws of diffusion (e.g., Fick's law) [54]. The second step (the adsorption process) (ii) can be a physical or a chemical process. During chemisorption, the dye molecule or ion attaches itself to a specific surface by a chemical bond and, in the physical adsorption, the dye molecules attach onto the adsorbent surface under the influence of van der Waals forces and hydrogen bonding [55]. The adsorption kinetic process can be studied through various theoretical methods (e.g., pseudo-first, pseudo-second, the intraparticle diffusion, Elovich) [35].  Figure 4 shows optical results for the ZnO and TiO 2 semiconductors. Both of them show a high reflectance of approximately (or greater than) 60% after 360 nm. ZnO and TiO 2 are not active under visible irradiation due to their high band gap (E g ). We determined the E g value using the Kubelka-Munk remission function [50]. Figure 4b shows the E g estimation for each thin film. The estimated bad gaps for the thin films are shown in Figure 4b [51]. These results correspond to those of previous reports for ZnO and TiO 2 Degussa P25 [52,53]. The spectroscopic and structural characterization verified the presence of ZnO and TiO 2 in the coatings synthesized.  Figure 4 shows optical results for the ZnO and TiO2 semiconductors. Both of them show a high reflectance of approximately (or greater than) 60% after 360 nm. ZnO and TiO2 are not active under visible irradiation due to their high band gap (Eg). We determined the Eg value using the Kubelka-Munk remission function [50]. Figure 4b shows the Eg estimation for each thin film. The estimated bad gaps for the thin films are shown in Figure 4b [51]. These results correspond to those of previous reports for ZnO and TiO2 Degussa P25 [52,53]. The spectroscopic and structural characterization verified the presence of ZnO and TiO2 in the coatings synthesized.

Adsorption Kinetic Study
The adsorption of dye onto a semiconductor surface is a principle that relies on two steps: (i) diffusion of reactants onto the semiconductor surface and (ii) adsorption of reactants onto the semiconductor surface. The first step (the diffusion process) (i) follows the classic laws of diffusion (e.g., Fick's law) [54]. The second step (the adsorption process) (ii) can be a physical or a chemical process. During chemisorption, the dye molecule or ion attaches itself to a specific surface by a chemical bond and, in the physical adsorption, the dye molecules attach onto the adsorbent surface under the influence of van der Waals forces and hydrogen bonding [55]. The adsorption kinetic process can be studied through various theoretical methods (e.g., pseudo-first, pseudo-second, the intraparticle diffusion, Elovich) [35].

Adsorption Kinetic Study
The adsorption of dye onto a semiconductor surface is a principle that relies on two steps: (i) diffusion of reactants onto the semiconductor surface and (ii) adsorption of reactants onto the semiconductor surface. The first step (the diffusion process) (i) follows the classic laws of diffusion (e.g., Fick's law) [54]. The second step (the adsorption process) (ii) can be a physical or a chemical process. During chemisorption, the dye molecule or ion attaches itself to a specific surface by a chemical bond and, in the physical adsorption, the dye molecules attach onto the adsorbent surface under the influence of van der Waals forces and hydrogen bonding [55]. The adsorption kinetic process can be studied through various theoretical methods (e.g., pseudo-first, pseudo-second, the intraparticle diffusion, Elovich) [35]. Figure 5a,b shows the adsorption kinetics on TiO 2 and ZnO. Figure 5 indicates that the TiO 2 thin films reach 10.5 mg/g and the ZnO thin films reach 6.5 mg/g after 50 min of contact. These differences can be assigned to morphological properties and grain size. Table 2 lists the fitting results of the two models implemented. Table 2 indicates that the PSO model showed was suitable (greatest R value) to describe the adsorption process for both semiconductors. ZnO has a greater k 2 value than that of TiO 2 and a smaller q e value than that of TiO 2 , thus indicating that the ZnO surface saturates faster than the TiO 2 surface, a behavior that can be associated to reduced grain size of TiO 2 thin films. In the PSO model, the electrostatic interaction onto the surface affects the interaction with MB molecules. The MB dye is a cationic dye; the isoelectric point of TiO 2 in water (7.0 [56]) is smaller than the isoelectric point of ZnO (9.5 [57]); and under experimental conditions, the ZnO surface is positively charged, then TiO 2 would have more effective interaction with MB than ZnO thin films would. Furthermore, the grain size of TiO 2 (anatase 84.2%) is smaller than that of ZnO (see Table 1), and the specific surface area of TiO 2 should be greater than that of ZnO, increasing the MB adsorption capacity of TiO 2 in comparison with that of ZnO thin films [58].
Materials 2023, 16, x FOR PEER REVIEW 6 of 12 Figure 5a,b shows the adsorption kinetics on TiO2 and ZnO. Figure 5 indicates that the TiO2 thin films reach 10.5 mg/g and the ZnO thin films reach 6.5 mg/g after 50 min of contact. These differences can be assigned to morphological properties and grain size. Table 2 lists the fitting results of the two models implemented. Table 2 indicates that the PSO model showed was suitable (greatest R value) to describe the adsorption process for both semiconductors. ZnO has a greater k2 value than that of TiO2 and a smaller qe value than that of TiO2, thus indicating that the ZnO surface saturates faster than the TiO2 surface, a behavior that can be associated to reduced grain size of TiO2 thin films. In the PSO model, the electrostatic interaction onto the surface affects the interaction with MB molecules. The MB dye is a cationic dye; the isoelectric point of TiO2 in water (7.0 [56]) is smaller than the isoelectric point of ZnO (9.5 [57]); and under experimental conditions, the ZnO surface is positively charged, then TiO2 would have more effective interaction with MB than ZnO thin films would. Furthermore, the grain size of TiO2 (anatase 84.2%) is smaller than that of ZnO (see Table 1), and the specific surface area of TiO2 should be greater than that of ZnO, increasing the MB adsorption capacity of TiO2 in comparison with that of ZnO thin films [58].   [62]. Konicki et al. reported AC of BY28 and BR46 dyes onto Graphene Oxide was 68.5 and 76.9 mg/g, respectively [63]. Finally, the pseudo-second model has been reported by various authors as a suitable fitting model for dye adsorption on different adsorbent types. Table 3 lists reports fitting kinetic data with pseudo-second model.   [63]. Finally, the pseudo-second model has been reported by various authors as a suitable fitting model for dye adsorption on different adsorbent types. Table 3 lists reports fitting kinetic data with pseudo-second model. Figure 6 shows the thermodynamic calculation applying the Arrhenius equation to MB adsorption onto the thin films of both semiconductors (Equation (6)). The ∆H • and ∆S • values were calculated from Figure 6. Table 3 lists the thermodynamic results. The removal of MB by using semiconductor oxides was a spontaneous process (∆G < 0, for both materials). This result is due to the morphological properties of the semiconductors' surface. Furthermore, the adsorption process was endothermic and more stable for TiO 2 than for ZnO. The positive ∆S values of both semiconductor oxides could be associated with a degree of hydration of cationic MB molecules in the solution [64]. The MB remotion was more favored on TiO 2 than on ZnO. Table 3 lists the thermodynamic results reported by other authors. Results show a variation range depending on both adsorbent and dye type. The ∆G • values for all studies listed in Table 3 are negative. It indicates that the dye adsorption onto adsorbents was spontaneous. This spontaneity of the process increases when the temperature increases. Bennabi et. al. reported that this behavior is associated with decreasing thickness of the boundary layer surrounding the adsorbent surface with temperature increasing. This effect improves the mass transfer of the dye to the adsorbent surface [65].

Adsorption Thermodynamic Study
Materials 2023, 16, x FOR PEER REVIEW Figure 6 shows the thermodynamic calculation applying the Arrhenius equatio adsorption onto the thin films of both semiconductors (Equation (6)). The ΔH° and ues were calculated from Figure 6. Table 3 lists the thermodynamic results. The rem MB by using semiconductor oxides was a spontaneous process (ΔG < 0, for both m This result is due to the morphological properties of the semiconductors' surface. more, the adsorption process was endothermic and more stable for TiO2 than for Z positive ΔS values of both semiconductor oxides could be associated with a degre dration of cationic MB molecules in the solution [64]. The MB remotion was more on TiO2 than on ZnO. Table 3 lists the thermodynamic results reported by other Results show a variation range depending on both adsorbent and dye type. The ΔG for all studies listed in Table 3 are negative. It indicates that the dye adsorption ont bents was spontaneous. This spontaneity of the process increases when the temper creases. Bennabi et. al. reported that this behavior is associated with decreasing thic the boundary layer surrounding the adsorbent surface with temperature increasi effect improves the mass transfer of the dye to the adsorbent surface [65].  Results verified that the adsorption process is an important step and indicated that such a process should be studied during photocatalytic tests.

Recyclability Study
To verify the potential application of semiconductors in continuous remediation water systems, we determined the recyclability of both semiconductor oxides in the MB adsorption during various cycles. Figure 7 shows the stability results of the studied semiconductors. The adsorption process was repeated five consecutive times. Figure 7 shows that after the fifth cycle, the removal performance reduced by 5% for TiO 2 and 2% for ZnO. Such stable results are associated with the stability of the semiconductor oxides, and with the chemistry of the substrate (soda lime glass) and method of thin film deposition.
These results indicate that the thin films were suitable and reusable for MB adsorption after five cycles. These results are relevant to improve continuous flow remediation systems where adsorbents are incorporated in suspension form. Thin films can avoid additional separation steps, reducing the economic implementation of these systems.
are associated with the stability of the semiconductor oxides, and with the chemistry of the substrate (soda lime glass) and method of thin film deposition. These results indicate that the thin films were suitable and reusable for MB adsorption after five cycles. These results are relevant to improve continuous flow remediation systems where adsorbents are incorporated in suspension form. Thin films can avoid additional separation steps, reducing the economic implementation of these systems.

Conclusions
We fabricated ZnO and TiO2 thin films. The morphological, optical, and spectroscopic characterizations verified the presence of ZnO and TiO2 in the coatings. Furthermore, the XRD simulation identified the crystalline structures of both semiconductors: TiO2 (anatase 84.2%-rutile 15.8%) and ZnO (wurtzite). The pseudo-second-order model was suitable to fit the kinetic results. Furthermore, TiO2 (qe 10.5 mg/g) was more effective in MB removal than ZnO (qe 6.5 mg/g). The MB adsorption onto both semiconductors was a spontaneous and endothermic process: TiO2 (ΔG = −2.9 kJ/mol; ΔH = 50.6 kJ/mol) and ZnO (ΔG = −7.1 kJ/mol; ΔH = 40.0 kJ/mol). Finally, the recycling test showed that the semiconductors were suitable after five consecutive adsorption tests. All the above results verified the significance of the adsorption process. The present authors consider that adsorption studies should be included during photocatalytic tests.  Acknowledgments: The authors would like to thank Universidad del Atlántico.

Conflicts of Interest:
The authors declare no conflict of interest.

Conclusions
We fabricated ZnO and TiO 2 thin films. The morphological, optical, and spectroscopic characterizations verified the presence of ZnO and TiO 2 in the coatings. Furthermore, the XRD simulation identified the crystalline structures of both semiconductors: TiO 2 (anatase 84.2%-rutile 15.8%) and ZnO (wurtzite). The pseudo-second-order model was suitable to fit the kinetic results. Furthermore, TiO 2 (q e 10.5 mg/g) was more effective in MB removal than ZnO (q e 6.5 mg/g). The MB adsorption onto both semiconductors was a spontaneous and endothermic process: TiO 2 (∆G = −2.9 kJ/mol; ∆H = 50.6 kJ/mol) and ZnO (∆G = −7.1 kJ/mol; ∆H = 40.0 kJ/mol). Finally, the recycling test showed that the semiconductors were suitable after five consecutive adsorption tests. All the above results verified the significance of the adsorption process. The present authors consider that adsorption studies should be included during photocatalytic tests.