Physical Origin of Diminishing Photocatalytic E ﬃ ciency for Recycled TiO 2 Nanotubes and Ag-Loaded TiO 2 Nanotubes in Organic Aqueous Solution

: Arrays of titania nanotubes (TiO 2 NTs) were developed by electrochemical anodization and doped with silver on their surface by photodeposition to achieve TiO 2 NTs / Ag. It is found that only anatase TiO 2 NTs were formed, with the preferential growth direction perpendicular to the titanium substrate, and with the length and diameter of ~2 µ m and 90–120 nm, respectively. The presence of Ag on the surface of TiO 2 NTs was also conﬁrmed. The TiO 2 NTs and TiO 2 NTs / Ag were used as photocatalysts to decolorize the methylene blue (MB) aqueous solution. The photodegradation e ﬃ ciency (PDE) is as high as 83% for TiO 2 NTs and 98% for TiO 2 NTs / Ag photocatalysts. This work focused on the investigation of the stability and recyclability of these photocatalysts in terms of e ﬃ ciency and its physical origin by surface analysis using X-ray photoelectron spectroscopy (XPS). It is found that PDE diminishes from 83% to 76% in TiO 2 NTs upon eight recycling runs and from 98% to 80% in TiO 2 NTs / Ag upon six recycling runs. The XPS analysis revealed that the physical origin of diminishing e ﬃ ciency is the carbon contamination on the surface of recycled TiO 2 NTs and a combination of carbon contamination and Ag leaching in recycled TiO 2 NTs / Ag.


Introduction
Titania nanotubes (TiO 2 NTs) continue to attract attention from researchers interested in photocatalytic degradation of organic pollutants in water. The parent material TiO 2 is stable and inexpensive and the TiO 2 NTs have several physical properties ideal for photodegradation: their ordered porous architecture, with a unidirectional electrical channel, large surface area, and long charge carrier lifetime [1][2][3][4][5][6][7]. Two well-known drawbacks of using TiO 2 in solar photocatalysis applications are: (i) its wide band gap, which requires UV frequencies to initiate the charge transfer mechanism and (ii) the rapid electron hole recombination rate of photogenerated carriers [8][9][10][11][12]. The main strategy Figure 1 shows the field emission scanning electron microscopy (FESEM) micrographs of TiO 2 NTs and TiO 2 NTs/Ag-0.1 M, -0.5 M and 1.0 M samples. The TiO 2 NT arrays are apparently well-ordered. The inner diameter of the average nanotube is 90-120 nm and its average length is~2 µm. Inspecting Figure 1, there is a qualitative difference between the structural characteristics of TiO 2 NTs and TiO 2 NTs/Ag-0.1 M, on the one hand, and those of TiO 2 NTs/Ag-0.5 M and 1.0 M on the other. That is, the smallest Ag concentration of 0.1 M does not have any apparent effect on the morphology of the titania nanotubes. At higher concentrations, there is a clear indication of silver deposition on the surface of TiO 2 NTs. This is relevant to the putative LSPR effect discussed above. If Ag atoms are incorporated as isolated impurities (isolated atoms or very small clusters of atoms), they will modify optical properties mainly via localized states in the band gap. However, if the silver forms clusters large enough then it can support surface plasmons excitable by visible light. While we have not established the actual morphology of the Ag in the TiO 2 NTs, the SEM images should be considered alongside the visible absorbance shown below.
Catalysts 2020, 10, x FOR PEER REVIEW 3 of 13 to the observed nanostructure. Finally, we used X-ray photoelectron spectroscopy (XPS) on the optimally loaded sample to track the change in the local structure surrounding each atomic species in the sample that occurs with repeated catalytic cycles. In this way, we identify the most likely mechanism for diminishing photocatalytic performance.

Physicochemical Properties of TiO2NTs and TiO2NTs/Ag
Figure1 shows the field emission scanning electron microscopy (FESEM) micrographs of TiO2NTs and TiO2NTs/Ag-0.1 M, -0.5 M and 1.0 M samples. The TiO2NT arrays are apparently wellordered. The inner diameter of the average nanotube is 90-120 nm and its average length is ~2 µm. Inspecting Figure 1, there is a qualitative difference between the structural characteristics of TiO2NTs and TiO2NTs/Ag-0.1 M, on the one hand, and those of TiO2NTs/Ag-0.5 M and 1.0 M on the other. That is, the smallest Ag concentration of 0.1 M does not have any apparent effect on the morphology of the titania nanotubes. At higher concentrations, there is a clear indication of silver deposition on the surface of TiO2NTs. This is relevant to the putative LSPR effect discussed above. If Ag atoms are incorporated as isolated impurities (isolated atoms or very small clusters of atoms), they will modify optical properties mainly via localized states in the band gap. However, if the silver forms clusters large enough then it can support surface plasmons excitable by visible light. While we have not established the actual morphology of the Ag in the TiO2NTs, the SEM images should be considered alongside the visible absorbance shown below. The crystalline phase of TiO2NTs and TiO2NTs/Ag annealed at 500 °C was examined using Xray diffraction (XRD). In the XRD profiles (Figure 2), the diffraction peaks at 25.2°, 37.8°, 47.9°, 53.8°, 54.9° and 62.9° could be indexed to (101), (004), (200), (105), (211) and (204) crystalline planes of the anatase TiO2. For nanoparticles, anatase is usually the most stable of the three possible TiO2 crystal structures-and is slightly advantageous in charge separation applications because of its surface chemistry. There was no significant change in the crystal structure that occurred with Ag concentration. While the FESEM indicated that Ag was incorporated in the form of metallic Ag, no Ag crystal peaks were detected in the XRD patterns of TiO2NT/Ag-0.1 M, -0.5 M and -1.0 M samples. The deposited Ag is likely too small and few in number to be detected. The crystalline phase of TiO 2 NTs and TiO 2 NTs/Ag annealed at 500 • C was examined using X-ray diffraction (XRD). In the XRD profiles (Figure 2), the diffraction peaks at 25.2 • , 37.8 • , 47.9 • , 53.8 • , 54.9 • and 62.9 • could be indexed to (101), (004), (200), (105), (211) and (204) crystalline planes of the anatase TiO 2 . For nanoparticles, anatase is usually the most stable of the three possible TiO 2 crystal structures-and is slightly advantageous in charge separation applications because of its surface chemistry. There was no significant change in the crystal structure that occurred with Ag concentration. While the FESEM indicated that Ag was incorporated in the form of metallic Ag, no Ag crystal peaks were detected in the XRD patterns of TiO 2 NT/Ag-0.1 M, -0.5 M and -1.0 M samples. The deposited Ag is likely too small and few in number to be detected. To verify the presence of AgNPs on TiO2NTs, the UV-vis diffuse reflectance spectra for TiO2NTs, TiO2NTs/Ag-0.1 M, -0.5 M and -1.0 M were acquired and shown in Figure 3. We see similar optical properties for the 0.1 M sample and pure TiO2. The lowest concentration of Ag had no effect on visible light absorption. However, for the TiO2NTs/Ag-0.5 M and -1.0 M samples, a broad absorption peak centered between 450 and 500 nm emerges. This peak is attributed to the local surface plasmon resonance (LSPR) of AgNPs. The data again suggests that there is a qualitative difference between samples with an Ag concentration as low as 0.1 M compared to those with a concentration above 0.5 M. Our results strongly imply that the AgNPs only form above some minimum concentration of Ag. Below this concentration, we expect that the silver is deposited in a different form, perhaps as isolated impurities. If the associated impurity levels were mainly near band edges, they would have no effect on visible absorbance, which appears to be the case for 0.1 M Ag. Previous studies [39,40] have addressed the size-dependence of the visible absorbance spectra for AgNPs. A broad visible band is present for AgNPs as small as 3 nm (though its shape depends on both the size and shape of the nanoparticle). The presence of nanoparticles of this size would cause a dramatic increase in visible absorption, as is clearly seen for concentrations above 0.5 M. The surface composition of TiO2NTs and the oxidation states of Ag on the surface of TiO2NTs were investigated by XPS. Figure 4a shows the wide-scan XPS spectra in which titanium, oxygen, carbon, nitrogen and silver (in TiO2NTs/Ag) are detected. That is, we investigate the local structure of all native ions, the Ag dopants, and additional species present due to adventitious contamination on the surface. A tiny amount of nitrogen may come from electrolyte (ammonium fluoride solution). The high-resolution XPS spectra of Ag 3d, Ti 2p and O 1s are depicted in Figure 4. The XPS spectra To verify the presence of AgNPs on TiO 2 NTs, the UV-vis diffuse reflectance spectra for TiO 2 NTs, TiO 2 NTs/Ag-0.1 M, -0.5 M and -1.0 M were acquired and shown in Figure 3. We see similar optical properties for the 0.1 M sample and pure TiO 2 . The lowest concentration of Ag had no effect on visible light absorption. However, for the TiO 2 NTs/Ag-0.5 M and -1.0 M samples, a broad absorption peak centered between 450 and 500 nm emerges. This peak is attributed to the local surface plasmon resonance (LSPR) of AgNPs. The data again suggests that there is a qualitative difference between samples with an Ag concentration as low as 0.1 M compared to those with a concentration above 0.5 M. Our results strongly imply that the AgNPs only form above some minimum concentration of Ag. Below this concentration, we expect that the silver is deposited in a different form, perhaps as isolated impurities. If the associated impurity levels were mainly near band edges, they would have no effect on visible absorbance, which appears to be the case for 0.1 M Ag. Previous studies [39,40] have addressed the size-dependence of the visible absorbance spectra for AgNPs. A broad visible band is present for AgNPs as small as 3 nm (though its shape depends on both the size and shape of the nanoparticle). The presence of nanoparticles of this size would cause a dramatic increase in visible absorption, as is clearly seen for concentrations above 0.5 M. To verify the presence of AgNPs on TiO2NTs, the UV-vis diffuse reflectance spectra for TiO2NTs, TiO2NTs/Ag-0.1 M, -0.5 M and -1.0 M were acquired and shown in Figure 3. We see similar optical properties for the 0.1 M sample and pure TiO2. The lowest concentration of Ag had no effect on visible light absorption. However, for the TiO2NTs/Ag-0.5 M and -1.0 M samples, a broad absorption peak centered between 450 and 500 nm emerges. This peak is attributed to the local surface plasmon resonance (LSPR) of AgNPs. The data again suggests that there is a qualitative difference between samples with an Ag concentration as low as 0.1 M compared to those with a concentration above 0.5 M. Our results strongly imply that the AgNPs only form above some minimum concentration of Ag. Below this concentration, we expect that the silver is deposited in a different form, perhaps as isolated impurities. If the associated impurity levels were mainly near band edges, they would have no effect on visible absorbance, which appears to be the case for 0.1 M Ag. Previous studies [39,40] have addressed the size-dependence of the visible absorbance spectra for AgNPs. A broad visible band is present for AgNPs as small as 3 nm (though its shape depends on both the size and shape of the nanoparticle). The presence of nanoparticles of this size would cause a dramatic increase in visible absorption, as is clearly seen for concentrations above 0.5 M. The surface composition of TiO2NTs and the oxidation states of Ag on the surface of TiO2NTs were investigated by XPS. Figure 4a shows the wide-scan XPS spectra in which titanium, oxygen, carbon, nitrogen and silver (in TiO2NTs/Ag) are detected. That is, we investigate the local structure of all native ions, the Ag dopants, and additional species present due to adventitious contamination on the surface. A tiny amount of nitrogen may come from electrolyte (ammonium fluoride solution). The high-resolution XPS spectra of Ag 3d, Ti 2p and O 1s are depicted in Figure 4. The XPS spectra The surface composition of TiO 2 NTs and the oxidation states of Ag on the surface of TiO 2 NTs were investigated by XPS. Figure 4a shows the wide-scan XPS spectra in which titanium, oxygen, carbon, nitrogen and silver (in TiO 2 NTs/Ag) are detected. That is, we investigate the local structure of all native ions, the Ag dopants, and additional species present due to adventitious contamination on the surface. A tiny amount of nitrogen may come from electrolyte (ammonium fluoride solution). The high-resolution XPS spectra of Ag 3d, Ti 2p and O 1s are depicted in Figure 4. The XPS spectra of Ag 3d in Figure 4c reveal two prominent peaks with 6 eV spin-orbit separation that are centered around 374.2 eV and 368.2 eV. These correspond to Ag 3d 5/2 and Ag 3d 3/2 , respectively. This is the characteristic of metallic silver (Ag 0 ) [41] indicating that the Ag + ions are reduced to metallic silver (Ag 0 ) on TiO 2 NTs during the UV light irradiation. The intensities of the Ag 3d peaks understandably increased with increasing Ag concentration. In the XPS spectra of Ti 2p and O 1s in Figure 4a,b, two peaks of Ti 2p 3/2 , Ti 2p 1/2 and one peak of O 1s are observed at the binding energy of 458.5 eV, 464.3 eV and 530.0 eV respectively. The difference in binding energies of Ti 2p 3/2 and Ti 2p 1/2 is 5.8 eV, implying that Ti has a tetravalent state in TiO 2 NTs [21,23]. The measured binding energies of Ti 2p 3/2 peak, Ti 2p 1/2 peak and O 1s peak are in good agreement with the usual results for anatase TiO 2 [42].
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 13 of Ag 3d in Figure 4c reveal two prominent peaks with 6 eV spin-orbit separation that are centered around 374.2 eV and 368.2 eV. These correspond to Ag 3d5/2 and Ag 3d3/2, respectively. This is the characteristic of metallic silver (Ag 0 ) [41] indicating that the Ag + ions are reduced to metallic silver (Ag 0 ) on TiO2NTs during the UV light irradiation. The intensities of the Ag 3d peaks understandably increased with increasing Ag concentration. In the XPS spectra of Ti 2p and O 1s in Figure 4a,b, two peaks of Ti 2p3/2, Ti 2p1/2 and one peak of O 1s are observed at the binding energy of 458.5 eV, 464.3 eV and 530.0 eV respectively. The difference in binding energies of Ti 2p3/2 and Ti 2p1/2 is 5.8 eV, implying that Ti has a tetravalent state in TiO2NTs [21,23]. The measured binding energies of Ti 2p3/2 peak, Ti 2p1/2 peak and O 1s peak are in good agreement with the usual results for anatase TiO2 [42].  Table 1 lists the surface Ti, O, C, N and Ag atomic concentrations determined by XPS peak intensities for all samples. The XPS results verified the presence of Ag on the surface of all Ag-loaded TiO2NTs samples. Indeed, the amount of Ag on the surface, as indicated by XPS, is approximately proportional to the total concentration of Ag throughout the sample. This means that Ag is always present on the nanotube surfaces, but that at lower concentrations it does not modify structural and optical characteristics of the samples. This is again consistent with our claim that the qualitative difference in optical absorbance between samples with 0.1 M Ag compared to those of higher Ag content indicates a change in Ag morphology, with AgNPs present in the latter.   Table 1 lists the surface Ti, O, C, N and Ag atomic concentrations determined by XPS peak intensities for all samples. The XPS results verified the presence of Ag on the surface of all Ag-loaded TiO 2 NTs samples. Indeed, the amount of Ag on the surface, as indicated by XPS, is approximately proportional to the total concentration of Ag throughout the sample. This means that Ag is always present on the nanotube surfaces, but that at lower concentrations it does not modify structural and optical characteristics of the samples. This is again consistent with our claim that the qualitative difference in optical absorbance between samples with 0.1 M Ag compared to those of higher Ag content indicates a change in Ag morphology, with AgNPs present in the latter.

Photodegradation Efficiency of TiO 2 NTs and TiO 2 NTs/Ag Photocatalysts
The propensity for the TiO 2 NTs and TiO 2 NTs/Ag to degrade the organic pollutant, methylene blue (MB), is considered next. Figure 5a shows the photodegradation efficiency (PDE) of TiO 2 NTs and TiO 2 NTs/Ag over the UV irradiation time of 1 to 5 h. The PDE is enhanced by the addition of Ag, and begins to increase immediately. That is, the TiO 2 NTs/Ag-0.1 M sample exhibits a significantly larger PDE than the pure TiO 2 . With increasing Ag loading, the PDE first increases, being largest for TiO2NTs/Ag-0.5 M, then decreases. This limited data set indicates that the optimal Ag concentration for ultraviolet photodegradation of MB lies between 0.1 and 1.0 M. For a given irradiation time of 5 h, the PDE of TiO 2 NTs/Ag-0.1 M, -0.5 M and -1.0 M are respectively 89%, 98% and 87% as compared to a value of 83% for pure TiO 2 NTs. It is noted that when no photocatalyst was used, blanc experiments, the measured photodegradation efficiency was relatively low (4%, 7%, 12%, 15% and 17% for 1-, 2-, 3-, 4-and 5-h of UV irradiation.) Catalysts 2020, 10, x FOR PEER REVIEW 6 of 13

Photodegradation Efficiency of TiO2NTs and TiO2NTs/Ag Photocatalysts
The propensity for the TiO2NTs and TiO2NTs/Ag to degrade the organic pollutant, methylene blue (MB), is considered next. Figure 5a shows the photodegradation efficiency (PDE) of TiO2NTs and TiO2NTs/Ag over the UV irradiation time of 1 to 5 h. The PDE is enhanced by the addition of Ag, and begins to increase immediately. That is, the TiO2NTs/Ag-0.1 M sample exhibits a significantly larger PDE than the pure TiO2. With increasing Ag loading, the PDE first increases, being largest for TiO2NTs/Ag-0.5 M, then decreases. This limited data set indicates that the optimal Ag concentration for ultraviolet photodegradation of MB lies between 0.1 and 1.0 M. For a given irradiation time of 5 hr, the PDE of TiO2NTs/Ag-0.1 M, -0.5 M and -1.0 M are respectively 89%, 98% and 87% as compared to a value of 83% for pure TiO2NTs. It is noted that when no photocatalyst was used, blanc experiments, the measured photodegradation efficiency was relatively low (4%, 7%, 12%, 15% and 17% for 1-, 2-, 3-, 4-and 5-h of UV irradiation.) Since the experiments were carried out under UV irradiation, the enhanced absorption of visible light, due to the LSPR, is not responsible for the improved PDE. Indeed, the increase in PDE occurred even for the AgNPs-0.1 M sample, which showed no evidence for silver nanoparticles and no difference in optical properties from those of pure TiO2. Rather, the improved PDE may be attributed to reduced charge recombination rate occurring in the presence of Ag loading [43][44][45]. A decrease in PDE with a further increase of Ag (1.0 M) is ascribed to an increased agglomeration of Ag nanoparticles on the TiO2 surface which blocks the UV light (less photoactivity) and decreases an effective area for photocatalytic reaction [46,47]. The plots of ln(Ct/C0) versus time, shown in Figure  5b, exhibit approximately linear behavior, indicating first-order kinetics. The degradation rate constant (k) is found to be 0.0060 min −1 , 0.0081 min −1 , 0.0112 min −1 and 0.0075 min −1 for TiO2NTs, TiO2NTs/Ag-0.1 M, -0.5 M and -1.0 M, respectively. It is noted that the degradation rate constant for TiO2NTs/Ag-0.5 M was determined for the UV irradiation only up to 4 h.

Physical Origin of Diminished Efficiency in Recycled TiO2NTs and TiO2NTs/Ag Photocatalysts
Recycling experiments, in which several photodegradation trials were performed successively, were carried out for these systems. For each trial, an initial concentration of 12mg/l of MB with pH = 1 was used. Figure 6 shows the photodegradation efficiencies of TiO2NTs and TiO2NTs/Ag-0.5 M for these recycling runs. It is found that the photodegradation efficiency of TiO2NTs dropped from 83% to 76% after eight cycles while that of the TiO2NTs/Ag-0.5 M decreased from 98% to 80% after six cycles. Two plausible explanations for the diminishing photodegradation efficiency upon recycling are (i) carbon enrichment on the surface of the photocatalysts and (ii) Ag depletion. Central to the present work is an effort to identify which mechanism is more important. Since XPS is a surfacesensitive technique for chemical analyses, it is ideal for addressing this question. Since the experiments were carried out under UV irradiation, the enhanced absorption of visible light, due to the LSPR, is not responsible for the improved PDE. Indeed, the increase in PDE occurred even for the AgNPs-0.1 M sample, which showed no evidence for silver nanoparticles and no difference in optical properties from those of pure TiO 2 . Rather, the improved PDE may be attributed to reduced charge recombination rate occurring in the presence of Ag loading [43][44][45]. A decrease in PDE with a further increase of Ag (1.0 M) is ascribed to an increased agglomeration of Ag nanoparticles on the TiO 2 surface which blocks the UV light (less photoactivity) and decreases an effective area for photocatalytic reaction [46,47]. The plots of ln(C t /C 0 ) versus time, shown in Figure 5b, exhibit approximately linear behavior, indicating first-order kinetics. The degradation rate constant (k) is found to be 0.0060 min −1 , 0.0081 min −1 , 0.0112 min −1 and 0.0075 min −1 for TiO 2 NTs, TiO 2 NTs/Ag-0.1 M, -0.5 M and -1.0 M, respectively. It is noted that the degradation rate constant for TiO 2 NTs/Ag-0.5 M was determined for the UV irradiation only up to 4 h.

Physical Origin of Diminished Efficiency in Recycled TiO 2 NTs and TiO 2 NTs/Ag Photocatalysts
Recycling experiments, in which several photodegradation trials were performed successively, were carried out for these systems. For each trial, an initial concentration of 12mg/l of MB with pH = 1 was used. Figure 6 shows the photodegradation efficiencies of TiO 2 NTs and TiO 2 NTs/Ag-0.5 M for these recycling runs. It is found that the photodegradation efficiency of TiO 2 NTs dropped from 83% to 76% after eight cycles while that of the TiO 2 NTs/Ag-0.5 M decreased from 98% to 80% after six cycles. Two plausible explanations for the diminishing photodegradation efficiency upon recycling are (i) carbon enrichment on the surface of the photocatalysts and (ii) Ag depletion. Central to the present work is an effort to identify which mechanism is more important. Since XPS is a surface-sensitive technique for chemical analyses, it is ideal for addressing this question. Catalysts 2020, 10, x FOR PEER REVIEW 7 of 13   Figure 7d. It is noted that a small amount of nitrogen, less than a few at %, was also detected and not shown in the figure. As shown in Figure 7d, the atomic ratio between Ti and O is not 1:2. There is additional oxygen coming from surface contaminants. The obvious changes from the fresh to recycled TiO2NT samples are observed in the XPS spectra of C 1s. The peak associated with C increases rapidly with cycle count. It is well-known that the XPS C1s spectrum for organic contaminants typically has C-C, C-O-C and O-C=O components with peak positions at 284.8, 286.0 and 288.5 eV, respectively.    Figure 7d. It is noted that a small amount of nitrogen, less than a few at %, was also detected and not shown in the figure. As shown in Figure 7d, the atomic ratio between Ti and O is not 1:2. There is additional oxygen coming from surface contaminants. The obvious changes from the fresh to recycled TiO 2 NT samples are observed in the XPS spectra of C 1s. The peak associated with C increases rapidly with cycle count. It is well-known that the XPS C1s spectrum for organic contaminants typically has C-C, C-O-C and O-C=O components with peak positions at 284.8, 286.0 and 288.5 eV, respectively.   Figure 7d. It is noted that a small amount of nitrogen, less than a few at %, was also detected and not shown in the figure. As shown in Figure 7d, the atomic ratio between Ti and O is not 1:2. There is additional oxygen coming from surface contaminants. The obvious changes from the fresh to recycled TiO2NT samples are observed in the XPS spectra of C 1s. The peak associated with C increases rapidly with cycle count. It is well-known that the XPS C1s spectrum for organic contaminants typically has C-C, C-O-C and O-C=O components with peak positions at 284.8, 286.0 and 288.5 eV, respectively.   The carbon measured by XPS from the fresh TiO 2 NTs sample was found to be about 8 at %. This carbon is the typical contamination on the surface of the sample. After two runs of the photocatalysis reaction, the amount of detected carbon increases by a factor of three. It should be noted that the sample was rinsed in water and dried under air blow after each photocatalysis trial. The amount of carbon did not change significantly after the first few runs: the amount of C was approximately the same after four and eight runs than detected after two runs. The additional carbon detected after two photodegradation cycles is likely that of CO 2 and other products of a complete photocatalytic oxidation process. This explains the increase of the intensity of C 1s photoelectrons at binding energies of 286.0 and 288.0 eV, due to C-O-C and O-C=O components, respectively. Given that the large increase in C contaminants, seen during the first two photodegradation cycles, coincides with the significant drop in PDE, there is an indication that the decline in the performance of TiO 2 NTs as a photocatalyst results from carbon contamination on the surface. It is speculated that among several possible sources of carbon contamination, hydrocarbon adsorption (getting air-exposed) within recycling experiments and carbonate precipitation/MB sedimentation during photocatalytic reactions most likely contributed to surface carbon contamination in our samples. Figure 8 shows the XPS spectra of Ti 2p, O 1s, Ag 3d and C 1s for fresh and recycled TiO 2 NTs/Ag-0.5 M photocatalysts. There are no observable changes in the oxidation state of Ti and O over serval recycling runs. From the XPS spectra, the surface atomic concentrations of Ti, O, Ag and C in fresh and recycled TiO 2 NTs/Ag photocatalysts are calculated and plotted in Figure 9. A striking feature is the rapid reduction of Ag concentration. After the 6th run, AgNPs could not be detected. The diminishing Ag concentration may be due to leaching of AgNPs from the surface of TiO 2 NTs since the samples were rinsed with water and dried under air blow after each cycle. It is interesting to point out that the increase in carbon did not begin until the AgNPs were stripped from the samples. Once the Ag was removed, the bare TiO 2 NTs experienced the same rapid accumulation of carbon contaminants that was seen in Figure 7. Obviously, as the Ag was stripped from the samples, the photodegradation capability of TiO 2 NTs/Ag approached that of TiO 2 NTs, which is seen in Figure 6. The carbon measured by XPS from the fresh TiO2NTs sample was found to be about 8 at %. This carbon is the typical contamination on the surface of the sample. After two runs of the photocatalysis reaction, the amount of detected carbon increases by a factor of three. It should be noted that the sample was rinsed in water and dried under air blow after each photocatalysis trial. The amount of carbon did not change significantly after the first few runs: the amount of C was approximately the same after four and eight runs than detected after two runs. The additional carbon detected after two photodegradation cycles is likely that of CO2 and other products of a complete photocatalytic oxidation process. This explains the increase of the intensity of C 1s photoelectrons at binding energies of 286.0 and 288.0 eV, due to C-O-C and O-C=O components, respectively. Given that the large increase in C contaminants, seen during the first two photodegradation cycles, coincides with the significant drop in PDE, there is an indication that the decline in the performance of TiO2NTs as a photocatalyst results from carbon contamination on the surface. It is speculated that among several possible sources of carbon contamination, hydrocarbon adsorption (getting air-exposed) within recycling experiments and carbonate precipitation/MB sedimentation during photocatalytic reactions most likely contributed to surface carbon contamination in our samples. Figure 8 shows the XPS spectra of Ti 2p, O 1s, Ag 3d and C 1s for fresh and recycled TiO2NTs/Ag-0.5 M photocatalysts. There are no observable changes in the oxidation state of Ti and O over serval recycling runs. From the XPS spectra, the surface atomic concentrations of Ti, O, Ag and C in fresh and recycled TiO2NTs/Ag photocatalysts are calculated and plotted in Figure 9. A striking feature is the rapid reduction of Ag concentration. After the 6th run, AgNPs could not be detected. The diminishing Ag concentration may be due to leaching of AgNPs from the surface of TiO2NTs since the samples were rinsed with water and dried under air blow after each cycle. It is interesting to point out that the increase in carbon did not begin until the AgNPs were stripped from the samples. Once the Ag was removed, the bare TiO2NTs experienced the same rapid accumulation of carbon contaminants that was seen in Figure 7. Obviously, as the Ag was stripped from the samples, the photodegradation capability of TiO2NTs/Ag approached that of TiO2NTs, which is seen in Figure 6.

Development of TiO2NTs and TiO2NTs/Ag Samples
Titanium foils (area-2.5 cm × 2.0 cm, thickness-0.02 cm, purity 99.7%) were ultrasonically cleaned with acetone, ethanol and isopropanol for 10 min each. They were rinsed with distilled water and dried under air blow. The titanium dioxide nanotube (TiO2NT) arrays were developed on the Ti base by electrochemical anodization. The anodization process was carried out with magnetic agitation at room temperature in a two-electrode system with the Ti foil acting as the working electrode and a Pt rod as the counter electrode, which were connected to the positive and negative terminals of a dc power supply respectively. The distance between the two electrodes was fixed at approximately 2 cm. The electrolyte for anodization was a mixture of ethylene glycol and water (90 wt %:10 wt %) with 0.5 wt % of ammonium fluoride. The anodization was carried out with the applied voltage of 50 V for 3 h. When the anodization process was completed, the samples were removed from the electrolyte, then rinsed with distilled water and dried under air blow. They were subsequently annealed in a muffle furnace at 500 °C for 3 h for crystallization. Silver was loaded on the surface of TiO2NTs by the photodeposition method to achieve the TiO2NTs/Ag. In this process, the TiO2NTs samples were soaked into silver nitrate aqueous solutions with three different concentrations (0.1 M, 0.5 M and 1.0 M) for 2 h followed by UV-illumination for 2 h. Each sample is named according to the concentration of silver nitrate solution, i.e., TiO2NTs/Ag-0.1 M, TiO2NTs/Ag-0.5 M and TiO2NTs/Ag-1.0 M. The amount of the photocatalyst for each experiment may be estimated from the volume of the photocatalyst, which is the multiplication of the substrate dimensions and the film thickness (2 cm × 2 cm × 2 µm).

Physicochemical Characterization
The physicochemical properties such as size, size distribution, surface morphology, crystalline phase, optical behavior of TiO2NTs and TiO2NTs/Ag were investigated by field emission scanning electron microscopy (FESEM) (AURIGA CrossBeam Workstation, Carl Zeiss, Germany), X-ray diffraction (XRD) (SmartLab X-ray diffractometer with Cu Kα radiation, Rigaku, Japan), and Ultraviolet-visible diffuse reflectance spectroscopy (UV-DRS) (UV-2600 UV-VIS Spectrophotometer, Shimadzu, Japan). The surface properties and chemical states of TiO2NTs and TiO2NTs/Ag photocatalysts before and after the photocatalysis reactions were carried out were studied using Xray photoelectron spectroscopy (XPS) (PHI5000 Versa Probe II, ULVAC-PHI, Japan) at the SUT-NANOTEC-SLRI joint research facility, Synchrotron Light Research Institute. The monochromatic Al Kα X-ray (1486.6 eV) was used.

Development of TiO 2 NTs and TiO 2 NTs/Ag Samples
Titanium foils (area-2.5 cm × 2.0 cm, thickness-0.02 cm, purity 99.7%) were ultrasonically cleaned with acetone, ethanol and isopropanol for 10 min each. They were rinsed with distilled water and dried under air blow. The titanium dioxide nanotube (TiO 2 NT) arrays were developed on the Ti base by electrochemical anodization. The anodization process was carried out with magnetic agitation at room temperature in a two-electrode system with the Ti foil acting as the working electrode and a Pt rod as the counter electrode, which were connected to the positive and negative terminals of a dc power supply respectively. The distance between the two electrodes was fixed at approximately 2 cm. The electrolyte for anodization was a mixture of ethylene glycol and water (90 wt %:10 wt %) with 0.5 wt % of ammonium fluoride. The anodization was carried out with the applied voltage of 50 V for 3 h. When the anodization process was completed, the samples were removed from the electrolyte, then rinsed with distilled water and dried under air blow. They were subsequently annealed in a muffle furnace at 500 • C for 3 h for crystallization. Silver was loaded on the surface of TiO 2 NTs by the photodeposition method to achieve the TiO 2 NTs/Ag. In this process, the TiO 2 NTs samples were soaked into silver nitrate aqueous solutions with three different concentrations (0.1 M, 0.5 M and 1.0 M) for 2 h followed by UV-illumination for 2 h. Each sample is named according to the concentration of silver nitrate solution, i.e., TiO 2 NTs/Ag-0.1 M, TiO 2 NTs/Ag-0.5 M and TiO 2 NTs/Ag-1.0 M. The amount of the photocatalyst for each experiment may be estimated from the volume of the photocatalyst, which is the multiplication of the substrate dimensions and the film thickness (2 cm × 2 cm × 2 µm).

Physicochemical Characterization
The physicochemical properties such as size, size distribution, surface morphology, crystalline phase, optical behavior of TiO 2 NTs and TiO 2 NTs/Ag were investigated by field emission scanning electron microscopy (FESEM) (AURIGA CrossBeam Workstation, Carl Zeiss, Germany), X-ray diffraction (XRD) (SmartLab X-ray diffractometer with Cu Kα radiation, Rigaku, Japan), and Ultraviolet-visible diffuse reflectance spectroscopy (UV-DRS) (UV-2600 UV-VIS Spectrophotometer, Shimadzu, Japan). The surface properties and chemical states of TiO 2 NTs and TiO 2 NTs/Ag photocatalysts before and after the photocatalysis reactions were carried out were studied using X-ray photoelectron spectroscopy (XPS) (PHI5000 Versa Probe II, ULVAC-PHI, Japan) at the SUT-NANOTEC-SLRI joint research facility, Synchrotron Light Research Institute. The monochromatic Al Kα X-ray (1486.6 eV) was used.

Photocatalytic and Recycle Experiments
Methylene blue (MB) was used as a representative organic pollutant. It is a traditional dye widely used in textile industries, thus forming a common ingredient in wastewater from such industries. It contaminates the aquatic environment. In addition, it can cause eye burn which might lead to a permanent injury to human eyes. Therefore, the removal or separation of MB from water is considered one of the most environmental challenges during recent years. An MB solution was prepared by mixing 1.2 mg of MB and 100 mL of distilled water at pH-1. The TiO 2 NTs, TiO 2 NTs/Ag (0.1 M, 0.5 M and 1.0 M) were used as photocatalysts for degradation of MB in the aqueous solution. The reaction was carried out under the 24-W UV-A lamp (high-pressure mercury lamp, λ = 365 nm). Every 30 min, 2 mL of MB solution was withdrawn to measure the MB concentration using the UV-1600 spectrophotometer. The photodegradation efficiency, indicating simply the fraction of MB that remains after a given time, is η(%) = [(C 0 − C t )/C 0 ] × 100 where, C 0 and C t are the concentration of MB at t = 0 s and any time t, respectively. To study the changes in the system after multiple photodegradation cycles, we carried out repeated trials under identical conditions. The photocatalysts were rinsed with water and dried under air blow after each trial.

Conclusions
In summary, the electrochemically anodizedTiO 2 NTs and TiO 2 NTs loaded with varying amounts of Ag (TiO 2 NTs/Ag) were used as UV-driven photocatalysts for the degradation of an organic pollutant (MB) in aqueous solution. The TiO 2 NTs possess the anatase phase and the size of TiO 2 NTs are 90-120 nm in diameter and~2 um in length. The photodegradation experiment indicated that theTiO 2 NTs/Ag-0.5 M catalyst provided a photodegradation efficiency of 98% and degradation rate constant of 0.0138 min-1which are higher than those of pristine TiO 2 NTs (83%, 0.0059 min −1 ). This can be attributed to enhanced charge separation and higher reductive power in the presence of Ag, thereby suppressing charge recombination. The recycling experiment revealed that the PDE is diminished by 8.4% after eight runs of TiO 2 NTs and by 18.4% after six runs of TiO 2 NTs/Ag catalyst. The XPS surface analysis demonstrated that the physical origin of diminishing efficiency is a combination of carbon contamination and Ag leaching in recycled TiO 2 NTs/Ag. The interesting finding was that the presence of silver on TiO 2 TNs could reduce the accumulation of carbon contamination on the surface of TiO 2 TNs. This suggests that a proper preparation method is required to decorate silver on TiO 2 NTs to reduce a leasing rate in order to prolong the lifetime of the TiO 2 NTs/Ag photocatalyst.