Strongly Enhancing Photocatalytic Activity of TiO2 Thin Films by Multi-Heterojunction Technique

The photocatalysts of immobilized TiO2 film suffer from high carrier recombination loss when compared to its powder form. Although the TiO2 with rutile-anatase mixed phases has higher carrier separation efficiency than those with pure anatase or rutile phase, the single junction of anatase/rutile cannot avoid the recombination of separated carriers at the interface. In this study, we propose a TiO2/SnO2/Ni multi-heterojunction structure which incorporates both Schottky contact and staggered band alignment to reduce the carrier recombination loss. The low carrier recombination rate of TiO2 film in TiO2/SnO2/Ni multi-heterojunction structure was verified by its low photoluminescence intensity. The faster degradation of methylene blue for TiO2/SnO2/Ni multi-junctions than for the other fabricated structures, which means that the TiO2 films grown on the SnO2/Ni/Ti coated glass have a much higher photocatalytic activity than those grown on the blank glass, SnO2-coated and Ni/Ti-coated glasses, demonstrated its higher performance of photogenerated carrier separation.


Introduction
Titanium dioxide, TiO 2 , is an inexpensive, non-toxic and chemically stable material. Its high refractive index and transparency for visible light have made it a widely used painting material, and its biological properties have made it a promising material for biomedical applications [1][2][3]. Recently, its strong oxidizing power under ultraviolet (UV) irradiation attracts great attention due to the applications of antibacterial, deodorizing, and remediation of environmental pollutions [4][5][6][7][8][9][10]. TiO 2 has a conduction band edge potential lower than that of hydrogen revolution and valence band edge potential higher than that of oxygen revolution. Therefore, in addition to decomposing the organic pollutions, the photogenerated electrons and holes in TiO 2 can be used to split water into hydrogen and oxygen directly [11]. A lot of studies on the properties of TiO 2 photocatalysts for water treatment or water splitting have been conducted [12][13][14][15][16][17][18]. Generally, TiO 2 in particulate form possesses higher photocatalytic performance due to its large surface area. It was found that the TiO 2 particulates with mixed-phase structure of anatase and rutile exhibit the best photocatalytic activity, followed by with the pure anatase phase, and then with the pure rutile phase. A high photocatalytic efficiency photocatalyst of TiO 2 powder called P-25 (Degussa) has been developed based on the mixture of anatase and rutile phases [19]. Unfortunately, precipitating and recovering the TiO 2 particulates from water limits its widespread use. In contrast, immobilized TiO 2 film is more practical because of its controllability. Therefore, the way to improve the photocatalytic performance of immobilized TiO 2 films is urgently pursued.
The problem with the immobilized TiO 2 films is the high carrier recombination rate of TiO 2, which results in a low effective thickness for converting photon energy into chemical energy. Luttrell et al. [20] pointed out that the bulk transport ability of excitons to the surface dominates the photocatalytic activity of TiO 2 films. Their studies on the photodegradation of methyl orange demonstrated that the photocatalytic activity of TiO 2 films increases with the film thickness but reaches a maximum at 2.5 nm for rutile and~5 nm for anatase. This means that the carriers generated deeper than 2.5 nm and 5 nm for rutile and anatase, respectively, contribute little to the photodegradation of methyl orange. However, the thickness of 5 nm is too small when absorbing the incident light. The optical absorption coefficient of anatase TiO 2 in the near-UV region is rather low due to its indirect bandgap nature. The thickness required for anatase film to absorb 50% of the incident light with a wavelength of 340 nm has been reported to be around 500 nm [21]. Therefore, most of the incident lights pass through the 5 nm anatase film. Even if we thicken the film to absorb more incident photons, the carriers generated deeper than 5 nm from the surface cannot migrate to the surface for utilization. Therefore, the immobilized TiO 2 films suffer from either low photon absorption efficiency or high carrier recombination loss. A variety of strategies have been adopted to enhance the photocatalytic activity of immobilized TiO 2 films. The most frequently investigated strategy involves growing a porous structure like a nanotube, nanopillar, nanorod, nanosheet, nanoflake or nanobelt arrays [22][23][24][25][26][27][28][29][30]. Although a porous structure can provide more reaction surface area for photocatalysis, it still cannot overcome the issue of low carrier transport ability. An alternative method which involves loading metal nanoparticles, e.g., platinum, gold or silver [31][32][33][34][35][36][37], on TiO 2 can create an electric field across the interface to facilitate the separation of photogenerated electron-hole pairs. Thus, the carrier transport in TiO 2 is improved by the prolonged carrier lifetime. As a consequence, the photocatalytic activity of TiO 2 films is enhanced. In our previous study [38], we grew TiO 2 films on Ni, Ta, and Ti coated glass substrates and found that the TiO 2 films on the Ni-coated substrate performs the highest photocatalytic activity, followed by on the Ti-coated substrate, and then on the Ta-coated substrate, the same as the sequence of their electron work function of Ni~5.04-5.35 eV, Ti 4.33 eV, and Ta 4.00-4. 15 eV [39]. This is because the high work function metals like platinum, gold, silver or nickel can attract the photogenerated electrons from TiO 2 as they come into contact with TiO 2 , leading to a decrease of carrier recombination loss.
Nevertheless, the photogenerated holes in TiO 2 adjacent to the filled states of metal can also possibly cross into the metal to recombine with the electrons. Hence, introducing a semiconductor layer with an appropriate energy band structure in between the metal and TiO 2 is considered to be a feasible method for alleviating the recombination issue because the built energy barrier from the heterojunction between semiconductor and TiO 2 can block the holes in TiO 2 from entering the semiconductor and the metal. A similar concept of multi-heterojuction has been applied on the photo-induced hydrophilic conversion for TiO 2 /WO 3 systems by Miyauchi et al [40]. The rutile-anatase mixed-phase TiO 2 with higher photocatalytic activity than the pure anatase or pure rutile phase TiO 2 has been attributed to the staggered energy band alignment at the anatase/rutile interface [41]. Similarly, SnO 2 with suitable conduction and valence band edge potentials can form a staggered energy band alignment with anatase or rutile TiO 2 . It is expected that placing a thin SnO 2 layer in between the TiO 2 and high work function metal can effectively improve the separation of the photogenerated electron-hole pairs. In this study, a thin SnO 2 layer was placed in between the TiO 2 and Ni metal films to alleviate the recombination issue. The results show that with the appropriate band alignment of the heterojunction TiO 2 /SnO 2 /Ni, the photocatalytic activity of TiO 2 films has been highly improved.

Results and Discussion
Besides the heterojunction, the structure and surface roughness also affect the photocatalytic activity of TiO 2 films. Therefore, we first describe the structure and surface roughness of TiO 2 films grown on the chosen substrates of blank glass, Ni/Ti coated glass, SnO 2 coated glass and SnO 2 /Ni/Ti coated glass. Then, the measured photocatalytic activity of TiO 2 films is rendered and the junctions of TiO 2 /Ni, TiO 2 /SnO 2 and TiO 2 /SnO 2 /Ni on the photocatalytic activity of TiO 2 films are discussed. Figure 1a,b show the XRD patterns of TiO 2 films grown on the blank glass, Ni/Ti coated glass, SnO 2 coated glass and the SnO 2 /Ni/Ti coated glass at 250 and 350 • C, respectively. For the deposition temperature of 250 • C the films grown on the blank glass, Ni/Ti coated glass and SnO 2 coated glass are all crystallized in anatase form, but those grown on SnO 2 /Ni/Ti coated glass are a mixture of anatase and rutile. As the deposition temperature is raised to 350 • C, the structure of TiO 2 films grown on the blank glass and SnO 2 coated glass become a mixture of anatase and rutile, but the structure of TiO 2 films grown on the Ni/Ti coated glass become an almost pure rutile. The TiO 2 films grown on the SnO 2 /Ni/Ti coated glass, however, still maintain the anatase-rutile mixed structure. Apparently, the structure of TiO 2 films is related to both the deposition temperature and the substrate material. The SnO 2 /Ni/Ti composite layer tends to enhance the formation of the rutile phase at a low temperature, but the Ni/Ti composite layer facilitates to the formation of pure rutile film at a relatively higher deposition temperature. The values of the surface roughness of these TiO 2 films are summarized in Table 1. For a given deposition temperature, the films grown on SnO 2 /Ni/Ti coated glass have the roughest surface, followed by the films grown on blank glass, then the films grown on SnO 2 coated glass, and then the films grown on Ni/Ti coated glass. For a selected substrate, however, the films grown at 350 • C are rougher than the films grown at 250 • C. The photocatalytic activities of the TiO 2 films grown at 250 • C are shown in Figure 2. As expected, the SnO 2 /Ni underlying layer performs far better than the SnO 2 and Ni underlying layers for improving the photocatalytic activity of TiO 2 films. However, it is interesting that the SnO 2 underlying layer has similar ability to the Ni underlying layer for improving the photocatalytic activity of TiO 2 films. Because the TiO 2 films on the Ni/Ti and SnO 2 coated glass have the same structure of anatase form as those on the blank glass but with a lower surface roughness, the mechanism of improvement in the photocatalytic activity of TiO 2 films by the Ni and SnO 2 underlying layers can be concluded to be the heterojunction of TiO 2 /Ni and TiO 2 /SnO 2 . Without the effect of heterojunction, the TiO 2 films on the Ni or SnO 2 should have lower photocatalytic activity than on blank glass because of their lower surface roughness.
The improvement in photocatalytic activity of TiO 2 films by the Ni underlying layer can be described by the mechanism of Schottky-contact assisted carrier separation. The Ni has a work function of~5.04-5.35 eV, and the TiO 2 has a conduction band minimum of~−4.21 eV [42]. The junction of Ni with TiO 2 will exist a Schottky barrier. Figure 3 shows the I-V characteristics of the ALD TiO 2 on Ni. The diode behavior verifies the Schottky junction of Ni with TiO 2 . As the band diagram of TiO 2 /Ni shown in Figure 4, when the TiO 2 is irradiated with an intense UV light, a large amount of electron-hole pairs will be created and then the electron carriers in TiO 2 will flow to the Ni layer as depicted in Figure 4. As the photogenerated electron-hole pairs are separated, the Schottky barrier at the interface will block the photogenerated electrons from backing to the TiO 2 , leaving the photogenerated holes, which have been considered to be the rate limiting carrier for methylene blue photooxidation in TiO 2 . It is why the TiO 2 /Ni heterojunction can improve the photocatalytic activity of TiO 2 films. The photocatalytic activities of the TiO2 films grown at 250 ˚C are shown in Figure 2. As expected, the SnO2/Ni underlying layer performs far better than the SnO2 and Ni underlying layers for improving the photocatalytic activity of TiO2 films. However, it is interesting that the SnO2 underlying layer has similar ability to the Ni underlying layer for improving the photocatalytic activity of TiO2 films. Because the TiO2 films on the Ni/Ti and SnO2 coated glass have the same structure of anatase form as those on the blank glass but with a lower surface roughness, the mechanism of improvement in the photocatalytic activity of TiO2 films by the Ni and SnO2 underlying layers can be concluded to be the heterojunction of TiO2/Ni and TiO2/SnO2. Without the effect of heterojunction, the TiO2 films on the Ni or SnO2 should have lower photocatalytic activity than on blank glass because of their lower surface roughness. The improvement in photocatalytic activity of TiO2 films by the Ni underlying layer can be described by the mechanism of Schottky-contact assisted carrier separation. The Ni has a work function of 5.04-5.35 eV, and the TiO2 has a conduction band minimum of −4.21 eV [42]. The barrier at the interface will block the photogenerated electrons from backing to the TiO2, leaving the photogenerated holes, which have been considered to be the rate limiting carrier for methylene blue photooxidation in TiO2. It is why the TiO2/Ni heterojunction can improve the photocatalytic activity of TiO2 films.   The mechanism of TiO2/Ni Schottky-contact assisted carrier separation was further verified by the dependence of the photocatalytic activity of TiO2 films on the thickness of Ni underlying layer. Figure 5 shows the photocatalytic activities of TiO2 films on the Ni/Ti coated glass with Ni thicknesses of 25, 50 and 100 nm, and Figure 6 is the extracted MB decay constants from the curves in Figure 5. The photocatalytic activity of TiO2 film increases with the increase of Ni layer thickness but becomes saturated at a certain thickness. This result coincides with the mechanism described above. The thicker the Ni layer, the more low energy states in Ni for receiving the photogenerated electrons from TiO2, leading to higher carrier separation efficiency. However, as the Ni layer is thick enough to receive most of the photogenerated electrons in TiO2, the photocatalytic activity of TiO2 film would become less dependent on the thickness of Ni layer.

Ni
TiO 2 ~5 eV ~4.21 eV The mechanism of TiO 2 /Ni Schottky-contact assisted carrier separation was further verified by the dependence of the photocatalytic activity of TiO 2 films on the thickness of Ni underlying layer. Figure 5 shows the photocatalytic activities of TiO 2 films on the Ni/Ti coated glass with Ni thicknesses of 25, 50 and 100 nm, and Figure 6 is the extracted MB decay constants from the curves in Figure 5. The photocatalytic activity of TiO 2 film increases with the increase of Ni layer thickness but becomes saturated at a certain thickness. This result coincides with the mechanism described above. The thicker the Ni layer, the more low energy states in Ni for receiving the photogenerated electrons from TiO 2 , leading to higher carrier separation efficiency. However, as the Ni layer is thick enough to receive most of the photogenerated electrons in TiO 2 , the photocatalytic activity of TiO 2 film would become less dependent on the thickness of Ni layer.  The improvement in the photocatalytic activity of TiO2 films by the SnO2 underlying layer is, however, due to the carrier separation assisted by staggered band alignment. SnO2 is also an inherent n-type semiconductor with a conduction band minimum of -4.50 eV and a band gap of 3.67 eV [43]. In this work the carrier concentration in the SnO2 measured by 4-point probe is as high as ~10 21 cm −3 which makes the Fermi level close to the conduction band minimum. Like the TiO2/Ni junction, the TiO2/SnO2 junction is also favorable for the separation of photogenerated electron-hole pairs as shown in Figure 7. Despite the same features of TiO2/Ni and TiO2/SnO2 junctions for carrier separation, two differences exist between them. One is that the available low energy states in SnO2 are less than in Ni for receiving the photogenerated electrons from TiO2; that is, the Ni underlying layer is more conductive than the SnO2 underlying layer to the separation of photogenerated carriers in TiO2. The other is that the TiO2/SnO2 junction has a higher barrier to hinder the diffusion of photogenerated holes from TiO2 into SnO2; that is, the SnO2 underlying layer is more able than the  The improvement in the photocatalytic activity of TiO2 films by the SnO2 underlying layer is, however, due to the carrier separation assisted by staggered band alignment. SnO2 is also an inherent n-type semiconductor with a conduction band minimum of -4.50 eV and a band gap of 3.67 eV [43]. In this work the carrier concentration in the SnO2 measured by 4-point probe is as high as ~10 21 cm −3 which makes the Fermi level close to the conduction band minimum. Like the TiO2/Ni junction, the TiO2/SnO2 junction is also favorable for the separation of photogenerated electron-hole pairs as shown in Figure 7. Despite the same features of TiO2/Ni and TiO2/SnO2 junctions for carrier separation, two differences exist between them. One is that the available low energy states in SnO2 are less than in Ni for receiving the photogenerated electrons from TiO2; that is, the Ni underlying layer is more conductive than the SnO2 underlying layer to the separation of photogenerated carriers in TiO2. The other is that the TiO2/SnO2 junction has a higher barrier to hinder the diffusion of photogenerated holes from TiO2 into SnO2; that is, the SnO2 underlying layer is more able than the The improvement in the photocatalytic activity of TiO 2 films by the SnO 2 underlying layer is, however, due to the carrier separation assisted by staggered band alignment. SnO 2 is also an inherent n-type semiconductor with a conduction band minimum of~−4.50 eV and a band gap of~3.67 eV [43]. In this work the carrier concentration in the SnO 2 measured by 4-point probe is as high as~10 21 cm −3 which makes the Fermi level close to the conduction band minimum. Like the TiO 2 /Ni junction, the TiO 2 /SnO 2 junction is also favorable for the separation of photogenerated electron-hole pairs as shown in Figure 7. Despite the same features of TiO 2 /Ni and TiO 2 /SnO 2 junctions for carrier separation, two differences exist between them. One is that the available low energy states in SnO 2 are less than in Ni for receiving the photogenerated electrons from TiO 2 ; that is, the Ni underlying layer is more conductive than the SnO 2 underlying layer to the separation of photogenerated carriers in TiO 2 . The other is that the TiO 2 /SnO 2 junction has a higher barrier to hinder the diffusion of photogenerated holes from TiO 2 into SnO 2 ; that is, the SnO 2 underlying layer is more able than the Ni underlying layer to avoid recombination of the separated carriers. The combination of the two differences may be the reason for the similar ability to improve the photocatalytic activity of TiO 2 films for both SnO 2 and Ni underlying layers. Ni underlying layer to avoid recombination of the separated carriers. The combination of the two differences may be the reason for the similar ability to improve the photocatalytic activity of TiO2 films for both SnO2 and Ni underlying layers. The advantages of the high Schottky barrier of the TiO2/Ni junction and the high hole diffusion barrier of TiO2/SnO2 junction can be incorporated together by inserting a thin SnO2 layer in between TiO2 and Ni. Figure 8 shows that the energy band diagram of TiO2/SnO2/Ni multi-junctions. The multi-junctions result in a staggered valence band alignment to block the holes from diffusing into the Ni layer and simultaneously keep the band bending structure for driving the electrons to the Ni layer. Thus the photogenerated electron-hole pairs are efficiently separated, and the separated carriers are isolated by a distance of SnO2 layer from recombination at the interface. The low carrier recombination rate for TiO2/SnO2/Ni multi-heterojunctions is verified by the PL spectra shown in Figure 9. The PL intensity of TiO2 film on SnO2/Ni is less than one-fifteenth of that on Ni, indicating that placing the SnO2 layer in between TiO2 and Ni layers highly reduces the recombination of photogenerated carriers in TiO2. It is considered to be the reason for the TiO2 films grown on the SnO2/Ni/Ti underlying layer with the best photocatalytic activity. However, the photocatalytic activity is also a function of the crystalline structure and surface roughness. The TiO2 films on the SnO2/Ni/Ti coated glass have a mixture of anatase and rutile structures, which is different from the pure anatase structure of TiO2 films on SnO2 or Ni/Ti coated glass. Moreover, the TiO2 films on the SnO2/Ni/Ti coated glass have a higher surface roughness than those on the SnO2 or Ni/Ti coated glass. Both factors beneficial to the photocatalytic activity of TiO2 films have been reported. Therefore, to identify this conclusion, it is necessary to further investigate the photocatalytic activity of TiO2 films with the same structure and the same surface roughness on these underlying layers. Unfortunately, it is difficult to get TiO2 films with the same structure and the same surface roughness on the SnO2/Ni/Ti coated glass and on the SnO2 or Ni/Ti coated glass. Out of compromise, the films grown at 350 ˚C were investigated to further understand the effects of mixed structure and roughness on the photocatalytic activity of TiO2 films. The advantages of the high Schottky barrier of the TiO 2 /Ni junction and the high hole diffusion barrier of TiO 2 /SnO 2 junction can be incorporated together by inserting a thin SnO 2 layer in between TiO 2 and Ni. Figure 8 shows that the energy band diagram of TiO 2 /SnO 2 /Ni multi-junctions. The multi-junctions result in a staggered valence band alignment to block the holes from diffusing into the Ni layer and simultaneously keep the band bending structure for driving the electrons to the Ni layer. Thus the photogenerated electron-hole pairs are efficiently separated, and the separated carriers are isolated by a distance of SnO 2 layer from recombination at the interface. The low carrier recombination rate for TiO 2 /SnO 2 /Ni multi-heterojunctions is verified by the PL spectra shown in Figure 9. The PL intensity of TiO 2 film on SnO 2 /Ni is less than one-fifteenth of that on Ni, indicating that placing the SnO 2 layer in between TiO 2 and Ni layers highly reduces the recombination of photogenerated carriers in TiO 2 . It is considered to be the reason for the TiO 2 films grown on the SnO 2 /Ni/Ti underlying layer with the best photocatalytic activity. However, the photocatalytic activity is also a function of the crystalline structure and surface roughness. The TiO 2 films on the SnO 2 /Ni/Ti coated glass have a mixture of anatase and rutile structures, which is different from the pure anatase structure of TiO 2 films on SnO 2 or Ni/Ti coated glass. Moreover, the TiO 2 films on the SnO 2 /Ni/Ti coated glass have a higher surface roughness than those on the SnO 2 or Ni/Ti coated glass. Both factors beneficial to the photocatalytic activity of TiO 2 films have been reported. Therefore, to identify this conclusion, it is necessary to further investigate the photocatalytic activity of TiO 2 films with the same structure and the same surface roughness on these underlying layers. Unfortunately, it is difficult to get TiO 2 films with the same structure and the same surface roughness on the SnO 2 /Ni/Ti coated glass and on the SnO 2 or Ni/Ti coated glass. Out of compromise, the films grown at 350 • C were investigated to further understand the effects of mixed structure and roughness on the photocatalytic activity of TiO 2 films.  Figure 10 shows the photocatalytic activities of TiO2 films grown on those underlying layers at 350 ˚C. Interestingly, compared with 250 ˚C the photocatalytic activity of the TiO2 films grown on the SnO2/Ni/Ti coated glass increased slightly, but the photocatalytic activity of the TiO2 films grown on the blank glass decreased slightly. Moreover, the photocatalytic activity of the TiO2 films grown on the SnO2 coated glass and the Ni/Ti coated glass drops dramatically. From the results of XRD in Figure 1, the structure of TiO2 films grown on the blank glass and SnO2 coated glass changes from pure anatase to a mixture of anatase and rutile as the deposition temperature increases from 250 to 350 ˚C The results in Figure 10 seem contradictory to the report that the TiO2 with rutile-anatase mixed phases has higher photocatalytic activity than those with pure anatase or rutile phase. The contradiction was also found in our previous studies of TiO2 films grown on Ni and Ta underlying  Figure 10 shows the photocatalytic activities of TiO2 films grown on those underlying layers at 350 ˚C. Interestingly, compared with 250 ˚C the photocatalytic activity of the TiO2 films grown on the SnO2/Ni/Ti coated glass increased slightly, but the photocatalytic activity of the TiO2 films grown on the blank glass decreased slightly. Moreover, the photocatalytic activity of the TiO2 films grown on the SnO2 coated glass and the Ni/Ti coated glass drops dramatically. From the results of XRD in Figure 1, the structure of TiO2 films grown on the blank glass and SnO2 coated glass changes from pure anatase to a mixture of anatase and rutile as the deposition temperature increases from 250 to 350 ˚C The results in Figure 10 seem contradictory to the report that the TiO2 with rutile-anatase mixed phases has higher photocatalytic activity than those with pure anatase or rutile phase. The contradiction was also found in our previous studies of TiO2 films grown on Ni and Ta underlying  Figure 10 shows the photocatalytic activities of TiO 2 films grown on those underlying layers at 350 • C. Interestingly, compared with 250 • C the photocatalytic activity of the TiO 2 films grown on the SnO 2 /Ni/Ti coated glass increased slightly, but the photocatalytic activity of the TiO 2 films grown on the blank glass decreased slightly. Moreover, the photocatalytic activity of the TiO 2 films grown on the SnO 2 coated glass and the Ni/Ti coated glass drops dramatically. From the results of XRD in Figure 1, the structure of TiO 2 films grown on the blank glass and SnO 2 coated glass changes from pure anatase to a mixture of anatase and rutile as the deposition temperature increases from 250 to 350 • C. The results in Figure 10 seem contradictory to the report that the TiO 2 with rutile-anatase mixed phases has higher photocatalytic activity than those with pure anatase or rutile phase. The contradiction was also found in our previous studies of TiO 2 films grown on Ni and Ta underlying layers that the photocatalytic activity of TiO 2 films decreases when the film structure changes from pure anatase to a mixture of anatase and rutile [38]. Although the fundamentals are still under investigation, this phenomenon indicates a high carrier recombination occurring in our rutile-anatase mixed-phase films. The most plausible reason for the decrease is that the separated carriers recombine at the interfaces of rutile/anatase, rutile/SnO 2 and rutile/Ni. Despite the fact that these junctions can separate the photogenerated electron-hole pairs, the separated electrons and holes are able to recombine through the defects at the interface. In addition, the lower oxidizing power of rutile TiO 2 compared to anatase TiO 2 may result in a pileup of hole carriers in rutile and thus a more serious interface recombination. layers that the photocatalytic activity of TiO2 films decreases when the film structure changes from pure anatase to a mixture of anatase and rutile [38]. Although the fundamentals are still under investigation, this phenomenon indicates a high carrier recombination occurring in our rutile-anatase mixed-phase films. The most plausible reason for the decrease is that the separated carriers recombine at the interfaces of rutile/anatase, rutile/SnO2 and rutile/Ni. Despite the fact that these junctions can separate the photogenerated electron-hole pairs, the separated electrons and holes are able to recombine through the defects at the interface. In addition, the lower oxidizing power of rutile TiO2 compared to anatase TiO2 may result in a pileup of hole carriers in rutile and thus a more serious interface recombination. In contrast, the SnO2 interlayer between Ni and TiO2 can separate the electrons from holes by a distance of SnO2 layer to avoid the recombination at the rutile-SnO2 interface. Therefore, the TiO2 films grown on SnO2/Ni/Ti coated glass possess high photocatalytic activity even if they have the anatase-rutile mixed-phase structure. Furthermore, the 350 ˚C-deposited film has a slightly higher surface roughness than the 250 ˚C-deposited film, resulting in a slightly higher photocatalytic activity. The results in Figure 10 verify the conclusion that the high photocatalytic activity of TiO2 films grown on SnO2/Ni/Ti coated glass is due to the TiO2/SnO2/Ni multi-junctions rather than the anatase-rutile mixed-phase structure or the surface roughness. It is worth nothing that besides application for the photocatalysis, the high carrier separation ability of the multi-junctions can also be applied to other devices such as solar cells.

Experimental Section
Corning E2000 glass sheet with dimension of 3 × 2.5 cm 2 was used as the substrate for the deposition of various thin films. Four kinds of film configurations, namely TiO2, TiO2/SnO2, TiO2/Ni/Ti and TiO2/SnO2/Ni/Ti, were adopted in this study. The use of the Ti layer with a fixed thickness of ~30 nm was to improve the adhesion of the Ni layer to the glass substrate. The adoption of Ni layer is due to its high work function and low cost compared to the noble metals such as Ag, Au and Pt. The thickness of Ni layer was fixed at 50 nm except for the samples to evaluate the effect of TiO2/Ni heterojunction on the separation of photogenerated carriers in TiO2. The adoption of the SnO2 layer was because of its appropriate energy band structure which has higher valence band edge potential than TiO2 for blocking the holes from entering into the Ni layer and higher electron affinity to catch electrons from TiO2 [41]. Before the film deposition, the glass substrates were ultrasonically cleaned by acetone, methanol and DI water for 5 min in each step, and then dried by nitrogen purge In contrast, the SnO 2 interlayer between Ni and TiO 2 can separate the electrons from holes by a distance of SnO 2 layer to avoid the recombination at the rutile-SnO 2 interface. Therefore, the TiO 2 films grown on SnO 2 /Ni/Ti coated glass possess high photocatalytic activity even if they have the anatase-rutile mixed-phase structure. Furthermore, the 350 • C-deposited film has a slightly higher surface roughness than the 250 • C-deposited film, resulting in a slightly higher photocatalytic activity. The results in Figure 10 verify the conclusion that the high photocatalytic activity of TiO 2 films grown on SnO 2 /Ni/Ti coated glass is due to the TiO 2 /SnO 2 /Ni multi-junctions rather than the anatase-rutile mixed-phase structure or the surface roughness. It is worth nothing that besides application for the photocatalysis, the high carrier separation ability of the multi-junctions can also be applied to other devices such as solar cells.

Experimental Section
Corning E2000 glass sheet with dimension of 3 × 2.5 cm 2 was used as the substrate for the deposition of various thin films. Four kinds of film configurations, namely TiO 2 , TiO 2 /SnO 2 , TiO 2 /Ni/Ti and TiO 2 /SnO 2 /Ni/Ti, were adopted in this study. The use of the Ti layer with a fixed thickness of~30 nm was to improve the adhesion of the Ni layer to the glass substrate. The adoption of Ni layer is due to its high work function and low cost compared to the noble metals such as Ag, Au and Pt. The thickness of Ni layer was fixed at~50 nm except for the samples to evaluate the effect of TiO 2 /Ni heterojunction on the separation of photogenerated carriers in TiO 2 . The adoption of the SnO 2 layer was because of its appropriate energy band structure which has higher valence band edge potential than TiO 2 for blocking the holes from entering into the Ni layer and higher electron affinity to catch electrons from TiO 2 [41]. Before the film deposition, the glass substrates were ultrasonically cleaned by acetone, methanol and DI water for 5 min in each step, and then dried by nitrogen purge gas. The Ti and Ni layers were grown by E-beam evaporator, and the SnO 2 and TiO 2 layers were grown by atomic layer deposition (ALD). SnCl 4 and TiCl 4 were used as the precursors of Sn and Ti for the ALD SnO 2 and ALD TiO 2 , respectively, H 2 O was used as the oxygen source and Ar gas as the purge gas. Each cycle of ALD SnO 2 (TiO 2 ) includes four steps of SnCl 4 (TiCl 4 ) pulse with 1 s, Ar purge with 2 s, H 2 O pulse with 1 s and Ar purge with 2 s. The SnO 2 films were deposited at 300 • C, and the TiO 2 films were grown at 250 and 350 • C. The deposition cycle for both the ALD SnO 2 and ALD TiO 2 was 1000, yielding a film thickness of~52 nm for SnO 2 and~55 nm for TiO 2 .
The conductivity of Ni/Ti and SnO 2 layers was characterized by 4-point probe. The thickness of Ni/Ti layers was determined by profilometer, and the thickness of SnO 2 and TiO 2 layers was determined by ellipsometer. The surface roughness of films was measured by atomic force microscope. The crystalline structures of deposited films were identified by a grazing incident X-ray diffractometer with a voltage of 40 kV and a current of 40 mA at a wavelength of 1.5418 Å. The photoluminescence (PL) of TiO 2 on Ni and SnO 2 /Ni coated glass was recorded using a He-Cd laser of 325 nm wavelength as the excitation source at room temperature. The photocatalytic activity of TiO 2 films was evaluated by measuring the degradation of methylene blue (MB) under UV-light irradiation at room temperature. Three 10 W of Sankyo Denki blacklight lamps with a center wavelength at 352 nm in parallel were used as the UV light sources. The samples were placed at the bottom of the glass cells (50 × 40 × 50 mm 3 internal dimensions) filled with the MB solution of concentrations of 10 −5 mol L −1 with the height of 10 mm. The measured irradiation intensity at the film surface was 0.59 mW cm −2 , which is relatively low compared with others [44,45] in order to prevent the MB diffusion in the solution from becoming a limitation factor for the photocatalysis experiment. According to the Beer-Lambert law, the absorbance peak intensity of MB solution at 668 nm is proportional to the MB concentration, so that can be used to monitor the degradation of MB solution. The decrease of the absorbance of MB solutions was measured by a spectrometer at fixed intervals, and the residual MB concentration (C/C 0 ) was extracted by the change of absorbance at 668 nm. The photocatalytic degradation of MB can be described by an exponential decay function where C 0 and C(t) is the MB concentration of initial and after exposure time t, and k is the exponential decay constant or photocatalytic activity.

Conclusions
TiO 2 films with thickness of~55 nm were grown on blank glass, Ni/Ti coated glass, SnO 2 coated glass and SnO 2 /Ni/Ti coated glass. The photocatalytic activity of these TiO 2 films was evaluated by measuring the MB degradation rate under irradiation of 352 nm UV light at room temperature. The results demonstrate that all the underlying layers of Ni, SnO 2 and SnO 2 /Ni can improve the photocatalytic activity of the deposited TiO 2 films. Among them, the SnO 2 /Ni underlying layer performs the best. The photocatalytic activity of TiO 2 films improved by Ni underlying layer and SnO 2 underlying layer is due to the Schottky barrier and staggered band alignment at the TiO 2 /Ni and TiO 2 /SnO 2 interfaces, respectively. However, the single junction of TiO 2 /Ni or TiO 2 /SnO 2 cannot avoid the carrier recombination at the interface. The multi-junctions of TiO 2 /SnO 2 /Ni can further separate the photogenerated electrons from holes by a distance of SnO 2 layer to avoid the recombination of separated carriers, and thus the photocatalytic activity of TiO 2 films is highly improved.