Photocatalytic Activity of SrTiO3 Codoped with Nitrogen and

Dec 3, 2003 - R & D Center, TOTO, Ltd., 2-8-1 Honson, Chigasaki-shi, Kanagawa 253-8577, Japan. Received July 19, 2003. In Final Form: October 13, 2003...
1 downloads 0 Views 208KB Size
Download PDF
232

Langmuir 2004, 20, 232-236

Photocatalytic Activity of SrTiO3 Codoped with Nitrogen and Lanthanum under Visible Light Illumination Masahiro Miyauchi,* Minoru Takashio, and Hiroki Tobimatsu R & D Center, TOTO, Ltd., 2-8-1 Honson, Chigasaki-shi, Kanagawa 253-8577, Japan Received July 19, 2003. In Final Form: October 13, 2003 Yellow SrTiO3 powders codoped with nitrogen and lanthanum (STO:N,La) were studied as visible light photocatalysts. The crystal phase of STO:N,La exhibited a pure perovskite phase, and O and Sr sites atoms were substitutionally doped with N and La atoms, respectively. The first principle calculation of STO:N,La indicated that the edge of the N(2p) band is situated above the valence band, which consisted of O(2p) orbitals, and the La orbitals did not exist in the band gap of SrTiO3. STO:N,La exhibited a higher oxidation activity of gaseous 2-propanol under vis illumination than SrTiO3 doped only with nitrogen (STO:N). The high activity of STO:N,La was due to the decrease in the oxygen vacancies, which acted as electronhole recombination centers, because codoping with La3+ and N3- ions maintained the charge balance. The optimum doping density of N and La for visible light activity was 0.5%, and STO:N,La(0.5%) had an activity under UV illumination similar to pure SrTiO3.

1. Introduction Titanium dioxide (TiO2) is a well-known, efficient photocatalyst1-3 and has various industrial applications such as water or air purification,4-6 antibacterial agents,7 and self-cleaning surfaces.3,8 In addition to a TiO2 photocatalyst, the photocatalytic activities of other metal oxides such as SnO2, ZnO, and SrTiO3 have been reported.9 These efficient photocatalysts are generally wide-gap semiconductors because the decomposition of organic compounds requires a high oxidation power of the holes in the valence band and enough reduction power of the electrons in the conduction band. The photocatalytic reactions in wide-gap semiconductors require UV illumination to proceed. Numerous studies have attempted to extend the photosensitivity of semiconductors toward the visible (vis) light region to use solar or indoor illumination as the light source.10-13 Recently, visible light responses were reported by nitrogen doping into a TiO2 lattice. Asahi et al. reported that substitutionally doping nitrogen into a TiO2 lattice is effective because the N(2p) states mix with the O(2p) states and narrow the band gap.14 They reported that annealing TiO2 powders in gaseous NH3 or by the * Author to whom correspondence should be addressed. Tel.: +81-467-54-3469. Fax: +81-467-54-1185. E-mail: masahiro. [email protected]. (1) Heller, A. Acc. Chem. Res. 1995, 28, 141. (2) Linsebigler, A. L.; Lu, G. Q.; Yates J. T. Chem. Rev. 1995, 95, 735. (3) Fujishima, A.; Hashimoto, K.; Watanabe, T. TiO2 Photocatalysis, Fundamentals and Applications; BKC, Inc.: Tokyo, 1999. (4) Photocatalysis: Fundamentals and Applications; Serpone, N., Pelizzetti, E., Eds; Wiley-Interscience: Amsterdam, 1989. (5) Photocatalytic Purification and Treatment of Water and Air; Ollis, D., Al-Ekabi, E., Eds.; Elsvier: Amsterdam, 1993. (6) Anderson, C.; Bard, A. J. J. Phys. Chem. 1995, 99, 9882. (7) Kikuchi, Y.; Sunada, K.; Iyoda, T.; Hashimoto, K.; Fujishima, A. J. Photochem. Photobiol., A 1997, 106, 51. (8) Heller, A. Acc. Chem. Res. 1995, 28, 503. (9) Miyauchi, M.; Nakajima, A.; Watanabe, T.; Hashimoto, K. Chem. Mater. 2002, 14, 2815. (10) Lehn, J. M.; Sauvage, J. P.; Ziessel, R.; Hilaire, L. Isr. J. Chem. 1982, 22, 168. (11) Serpone, N.; Maruthamuthu, P.; Pichat, P.; Pelizzetti, E.; Hidaka, H. J. Photochem. Photobiol., A 1995, 85, 247. (12) Borgarello, E.; Kiwi, J.; Gratzel, M.; Pelizzetti, E.; Visca, M. J. Am. Chem. Soc. 1982, 104, 2996. (13) Anpo, M.; Ichihashi, Y.; Takeuchi, M.; Yamashita, H. Res. Chem. Intermed. 1998, 24, 143.

sputtering process under an Ar/N2 gas mixture yielded nitrogen-doped TiO2. These processes, however, are very reductive and are presumed to create the electron hole recombination centers such as oxygen vacancies.15 Irie et al. reported that as a result of the oxygen vacancies the quantum yield of nitrogen-doped TiO2 decreased with the doping density.16 In addition to the nitrogen-doped TiO2, Kasahara et al. reported a perovskite-type LaTiO2N as a vis-light-driven photocatalyst for splitting water.17 It has also been reported that SrTiO3 codoped with antimony and chromium could decompose water into hydrogen in the presence of methanol when illuminating with vis light.18 The present paper focuses on SrTiO3 codoped with nitrogen and lanthanum (STO:N,La) as a vis-light-driven photocatalyst. The crystal phase of SrTiO3 is perovskite type (ABO3), and the charge balance in STO:N,La is expected to be maintained without forming oxygen vacancies. In the present paper, yellow powders of Sr1-xLaxTiO3-xNx (STO:N,La) with small values of x (∼10%) were synthesized by the sol-gel process and their photocatalytic oxidation activities of gaseous 2-propanol to acetone were evaluated under UV or vis illumination. On the basis of the first principle calculations and experimental results, the band structure of STO:N,La is discussed. 2. Experimental Section 2.1. Electronic Structure Calculation. The first principle method, which is based on the density functional theory, calculated the band structure within the generalized gradient approximation (CASTEP, Accelyls, Inc., San Diego, U.S.A.). The core orbitals were replaced by ultrasoft pseudopotentials, and the kinetic energy cutoff was 300 eV. Three different models were used for the calculations. Figure 1 shows the unit cells for the three models: A, SrTiO3; B, SrTiO3-2xNx (x ) 12.5%); and C, Sr1-xLaxTiO3-xNx (x ) 12.5%). Model A is the stoichiometric (14) Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269. (15) Takeda, S.; Suzuki, S.; Okada, H.; Hosono, H. Thin Solid Films 2001, 392, 338. (16) Irie, H.; Watanabe, Y.; Hashimoto, K. J. Phys. Chem. B 2003, 107, 5483. (17) Kasahara, A.; Nokumizu, K.; Takata, T.; Kondo, J. N.; Hara, M.; Kobayashi, H.; Domen, K. J. Phys. Chem. B 2003, 107, 791. (18) Kato, H.; Kudo, A. J. Phys. Chem. B 2002, 106, 5029.

10.1021/la0353125 CCC: $27.50 © 2004 American Chemical Society Published on Web 12/03/2003

Photocatalytic Activity of SrTiO3

Langmuir, Vol. 20, No. 1, 2004 233 surface area. X-ray photoelectron spectroscopy (XPS) with Mg KR X-rays (model AXIS-HS, Kratos, Manchester, U.K.) evaluated the amount and state of each element in these powders. The photoelectrons of the Sr(3d), La(4d), Ti(2p), N(1s), O(1s), and C(1s) orbitals were recorded with a takeoff angle of 45°. The C(1s), hydrocarbon contamination peak at 284.8 eV was the internal reference for the absolute binding energy. Prior to the XPS measurements, the surface of the powders was etched by Ar+ ion sputtering for 1 min, which eliminated physisorbed species on the surface. The integrated peak areas of each orbital were calculated after subtracting the nonlinear background. Quantitative analysis was based on the peak area multiplied by sensitivity factors supplied by Kratos, which accounted for the geometric configuration of the apparatus. The UV-vis absorption spectra of the powders were recorded on a spectrophotometer (UV-3100, Shimadzu Co., Kyoto, Japan). The absorption coefficient (R %) was obtained by measuring by the diffuse reflectance method, where R ) 100 - reflectance. The number of absorbed photons in each powder was estimated by eq 1.

Ia )

Figure 1. Unit cells for calculations. A, SrTiO3; B, SrTiO3-2xNx (x ) 12.5%); and C, Sr1-xLaxTiO3-xNx (x ) 12.5%). SrTiO3. In model B, 12.5% of the oxygen sites were doped with nitrogen atoms and the oxygen vacancies existed with the same density as nitrogen. In model C, 12.5% of the oxygen sites were doped with nitrogen atoms and 12.5% of the strontium sites were doped with lanthanum atoms. The sites for N, La, and oxygen vacancy in models B or C were determined so that the crystal had the lowest total energy. 2.2. Synthesis of the Powders. A titanium tetraisopropoxide, a strontium 2-ethylhexanoate toluene solution, and a lanthanum 2-ethylhexanoate toluene solution (Wako Pure Chemical Co., Osaka, Japan) were dissolved into 2-propanol. The molar ratio of (Sr + La)/Ti was 1.0, and the La/Ti ratio varied (0.1-50.0%). Acetylacetone (Wako Pure Chemical Co., Osaka, Japan) was added into these solutions to moderate the hydrolysis reaction, and the concentration of titanium ions in these solutions was 2.0 wt %. These solutions were dried in air at 90 °C for 2 days and resulted in amorphous powders. These powders were calcined at 850 °C for 1 h to yield white crystalline La-doped SrTiO3 powders (STO:La). These white powders were annealed in gaseous ammonia (NH3) at 750 °C for 1 h, and the white samples turned green. Afterward, they were annealed in air at 600 °C for 1 h to yield yellowish powders (STO:N,La). In addition, a pure SrTiO3 powder (STO) and a SrTiO3 doped only with nitrogen powder (STO:N) were synthesized using the same annealing temperature mentioned previously. 2.3. Characterization. The crystal phases of the powders were evaluated by X-ray diffraction (XRD) with Cu KR rays (model RINT-2100, Rigaku Co., Tokyo, Japan). The morphologies of particles were examined by a scanning electron microscope (SEM; model S-4200, Hitachi Co., Tokyo, Japan). The BET method (ASAP2000, Shimadzu Co., Kyoto, Japan) determined the specific

∫I R dλ 0

(1)

where I0 and R are the incident photon flux and the absorption coefficient, respectively. The incident photon flux (I0) was measured by a spectro-radiometer (USR-40D, Ushio Co., Tokyo, Japan). 2.4. Photocatalytic Activities. Monitoring the decomposition of gaseous 2-propanol in air (25 °C) evaluated the photocatalytic activities. A total of 0.3 g of the powder was placed in a 0.8 dm3 Pyrex glass vessel. An O2 (20%)-N2 (80%) gaseous mixture, which was passed through a 30 °C water humidifier to adjust the relative humidity to 50%, filled the vessel. The headspace (20 cm3) of the glass bottle reservoir, which contained liquid 2-propanol, was injected into the Pyrex glass vessel using a syringe. The initial concentration of 2-propanol was 500 ppm. The photocatalytic decomposition of gaseous 2-propanol was evaluated while illuminating with UV and vis light. A 10-W black light bulb (Toshiba Co., Tokyo, Japan) provided the UV illumination. The source for vis illumination was a 250-W xenon lamp (LA-250Xe, Hayashi Watch-works Co., Ltd., Tokyo, Japan), which was used in conjunction with an optical fiber coupler, an UV cutoff filter (Y-43, Toshiba Co., Ltd., Tokyo, Japan), and an IR cutoff filter (V-40, Toshiba Co., Ltd., Tokyo, Japan). The vis wavelengths ranged between 410 and 500 nm. A spectro-radiometer (USR40D, Ushio Co., Tokyo, Japan) measured the light intensities, which were adjusted to 0.6 and 6.0 mW/cm2 for UV and vis light, respectively. Once the gaseous and absorbed 2-propanol equilibrated, then illumination was initiated. The concentrations of 2-propanol, acetone, and carbon dioxide (CO2) were measured by a photoacoustic multigas monitor (model 1312, Innova, Ballerup, Denmark), which had a detection limit of approximately 0.1 ppm.

3. Results and Discussion The band structures were calculated for the pure SrTiO3, the nitrogen-doped SrTiO3 with oxygen vacancies (SrTiO3-2xNx), and the SrTiO3 codoped with nitrogen and lanthanum (Sr1-xLaxTiO3-xNx) to determine if codoping effectively reduced the oxygen vacancies. Figure 2 shows the calculated band structures of each crystal. The valence and conduction bands for the pure STO consisted of O(2p) and Ti(3d) orbitals, respectively (Figure 2A). The calculated band gap of STO was 1.6 eV, which underestimated the experimental value and is typical of the generalized gradient approximation.19 If the nitrogen ions were doped into oxygen sites substitutionally without forming any oxygen vacancies, it would display the p-type conductivity to maintain the charge balance in the crystal. However, it is known that the p-type conductivity in the wide band gap oxide semiconductor is very difficult to obtain because the upper edge of the valence band is deep, and the acceptor (19) Luo, W.; Beigi, S. I.; Cohen, M. L.; Louie, S. G. Phys. Rev. B 2002, 66, 195215.

234

Langmuir, Vol. 20, No. 1, 2004

Miyauchi et al.

Figure 3. XRD patterns for powders analyzed by the step scan method. 2θ was set at the diffraction for the (220) face of the perovskite phase and ranged from 66.5-69.0°. a, STO; b, STO:La(0.5%); c, STO:N; d, STO:N,La(0.5%); and e, STO:N,La(5.0%). Table 1. Doping Density of Nitrogen and Lanthanum Analyzed by XPS STO STO:La(0.5%) STO:N STO:N,La(0.5%) STO:N,La(5%)

Figure 2. Calculated DOS for each element. A, SrTiO3; B, SrTiO3-2xNx (x ) 12.5%); and C, Sr1-xLaxTiO3-xNx (x ) 12.5%).

level is thermodynamically unstable. Thus, most nitrogendoped oxides are supposed to contain oxygen vacancies to maintain the charge balance in the crystal, as shown in model B. As for model B (SrTiO3-2xNx), the isolated band, which consisted of Ti(3d) orbitals, however, was observed between the band gap of the SrTiO3-2xNx and was mainly due to the localized electrons of Ti3+ ions. A previous study reported that the Ti3+ bands are located between 0.75 and 1.18 eV, which is below the minimum level of the conduction band.20 These defective sites act as electronhole recombination centers15 and photocatalytic reactions should be suppressed. Consequently, Irie et al. reported that the quantum yield of nitrogen-doped TiO2 decreased with the doping density because of the oxygen vacancies.16 On the other hand, the Sr1-xLaxTiO3-xNx (STO:N,La) did not display isolated orbitals attributed to lanthanum or oxygen vacancies in the band gap. The N(2p) orbitals of STO:N,La were situated above the O(2p) orbitals, which narrowed the band gap. The calculated band gap energy of STO:N,La (1.4 eV) was lower than that of the pure STO (1.6 eV). On the basis of the XRD results, except for a small shift, the crystal phase of STO:N,La was the same as that of pure STO and indicated that the Sr2+ sites were substi(20) Cronemeyer, D. D. Phys. Rev. 1959, 113, 1222.

La/Ti (atom %)

N/Ti (atom %)

0 0.6 0 0.6 5.4

0 0 0.3 0.7 1.3

tuted by lanthanum ions. Figure 3 shows the XRD patterns for a (220) face of a perovskite phase analyzed by a step scan method. Compared to the other samples, the peak of STO:N,La(5%) was broader and shifted toward a lower 2θ value. These results indicate that crystals with a high doping density are distorted and contain internal strain. The images revealed that all particles exhibited nearly identical morphologies, regardless of the doping density. The particle size of the powders ranged from 100 to 200 nm. The BET surface area ranged from 7-8 m2/g for all powders. Doping densities and states of nitrogen and lanthanum were investigated by the XPS measurement. The N(1s) orbital had a binding energy of 396 eV in STO:N and STO:N,La. The peak at 396 eV was assigned to Ti-N bonding,21 indicating that the oxygen sites were substituted by nitrogen atoms. Table 1 lists the doping densities of nitrogen and lanthanum analyzed by XPS. The doping densities of nitrogen and lanthanum were nearly identical in STO:N,La(0.5%) powder. As for STO:N,La(5%), the doping density of nitrogen (1.3%) was less than that of lanthanum (5.4%). Figure 4 shows the absorption spectra for the powders. Both STO and STO:La did not absorb the vis light. STO:N,La powders had two types of absorption in the vis light region, shoulder peaks around 400-500 nm and a broad absorption above 500 nm. On the basis of previous studies and our electronic structure calculations, the shoulder peak around 400-500 nm is supposed to be isolated N(2p) orbital levels. This shoulder peak of STO:N was lower than that of STO:N,La because the doping density of nitrogen for STO:N (0.3%) was less than that of STO:N,La (0.5%). The broad absorption band above 500 nm was assigned to the oxygen vacancy states, which were between 0.75 and 1.18 eV and below the minimum level of the conduction band.20 Therefore, the STO with (21) Saha, M. C.; Tompkins, H. G. J. Appl. Phys. 1999, 72, 3072.

Photocatalytic Activity of SrTiO3

Figure 4. UV-vis spectra for powders measured by the diffuse reflectance method. The absorption coefficient (R%) was obtained by the equation R ) 100 - R, where R is the reflectance (%). a, STO; b, STO:La(0.5%); c, STO:N; d, STO:N,La(0.5%); and e, STO:N,La(5.0%).

Langmuir, Vol. 20, No. 1, 2004 235

Figure 6. Acetone generation versus La content in the first 5 h under vis illumination.

Figure 7. Acetone generation versus La content in the first 50 min under UV illumination. Figure 5. Photocatalytic oxidation of gaseous 2-propanol to acetone. a, STO; b, STO:La(0.5%); c, STO:N; d, STO:N,La(0.5%); and e, STO:N,La(5.0%).

oxygen vacancies can absorb a broad range of vis light above 500 nm. It is noteworthy that the broad absorption above 500 nm of STO:N,La(0.5%) was less than that of STO:N, which implied that the creation of oxygen vacancies was suppressed by codoping with nitrogen and lanthanum. The photocatalytic oxidation of gaseous 2-propanol to acetone under vis illumination was evaluated. Figure 5 shows the acetone generation under vis illumination (410-500 nm). The photocatalytic decomposition path of gaseous 2-propanol has already been reported. It is oxidized directly to carbon dioxide, or it is oxidized through intermediate acetone to carbon dioxide.22 Both STO and STO:La did not display activity when illuminating with vis light, whereas STO:N and STO:N,La exhibited vis light activity. STO:N,La(0.5%) exhibited the highest rate for acetone generation. We also evaluated the generated CO2, and gaseous 2-propanol was completely decomposed to CO2 (1500 ppm) when exposed to vis light for a long time (250 h). The changes of the acetone concentration under dark conditions for these samples were measured, and they did not display any activities. Moreover, STO:N,La(0.5%) did not display activity under vis illumination above 500 nm, indicating that the (22) Ohko, Y.; Hashimoto, K.; Fujishima, A. J. Phys. Chem. A 1997, 101, 8057.

absorption shoulder peak around 400-500 nm of the N(2p) orbital is indispensable for the vis light activity. There was an optimum amount of doping density of La for the vis light activity. Figure 6 shows the generated acetone under vis illumination versus the La content in STO:N,La. The rate for acetone generation was highest with a La content of 0.5%, which corresponded to an equivalent doping density of N and La as analyzed by XPS. When the doping density of La is higher than 0.5%, the vis light activity was decreased as a result of creating defects or lattice distortion. Figure 7 shows the photocatalytic activities under UV illumination. The acetone generation of STO:N was less than that of STO because the oxygen vacancies of STO:N act as electron-hole recombination centers. STO:N,La(0.5%), which was the optimum doping density for the vis light activity, also exhibited a high activity under UV illumination. STO:N,La(0.5%) generated 0.49 and 0.07 (ppm/min) acetone under UV and vis illumination, respectively. The light intensities of UV and vis light were 0.6 and 6.0 mW/cm2, and absorbed photons in STO:N,La under UV and vis illumination were 8.6 × 1014 and 2.8 × 1015 (quanta/(cm2 s), respectively. Although more photons were absorbed for vis illumination than for UV illumination, less acetone was produced under vis light. If the N(2p) levels mixed with the O(2p) orbital and the band gap narrowed, then the photogenerated holes in the valence band would diffuse to the upper edge of the valence band by a radiative process. Under these conditions, the oxidation power of the photogenerated holes was identical

236

Langmuir, Vol. 20, No. 1, 2004

Figure 8. Assumed band structures of STO, STO:N, and STO: N,La.

for both UV and vis illumination. However, our experimental data demonstrated that the photocatalytic activity of STO:N,La under vis illumination was much lower than that under UV illumination. These results indicated that the photogenerated holes produced by the vis light in STO:N,La were localized in the isolated levels between the valence and conduction bands. The photogenerated holes in the isolated band were localized and have slower mobility than those in the valence band. On the basis of our experimental results, Figure 8 shows the assumed band structures for STO, STO:N, and STO:N,La. The pure STO had a large band gap; thus, it had the photocatalytic activity under UV illumination but not under vis light. In contrast, STO:N displayed vis light activity, which was very low. The origin of the vis light activity in this sample is the nitrogen orbital; however, the oxygen vacancies act as electron-hole recombination centers, and, thus, its activities under UV as well as vis illumination are very low. The state of the oxygen vacancies is located deep below the conduction band minimum; thus, the excited electrons in this level would have low mobility because of the localization. As for STO:N,La, the Sr and O sites were substitutionally doped with La and N atoms and charge balance was maintained without forming oxygen vacancies. Therefore, STO:N,La had a much higher vis light activity than STO:N. However, the isolated N(2p) orbitals were situated above the O(2p) orbitals and the holes generated by vis illumination had a slower mobility than those generated by UV illumination. Our electronic structure calculations were conducted under the higher doping densities (12.5%) and indicated that the N(2p) orbitals were mixing with the O(2p) orbitals. The density of states (DOS) for the N(2p) orbitals depends

Miyauchi et al.

on the doping density. It is assumed that the lower doping density causes the degeneration, and the N(2p) orbitals are supposed to be localized. The calculation for the lower doping density requires the larger unit cell volume; thus, it requires the higher performance of a computer. At present, calculations were conducted under doping densities higher than those in experimental observation; however, these calculations show very important information for the effect of codoping. It must be noted that the final annealing procedure in air at 600 °C is very important for the vis light activity. Annealing in gaseous NH3 resulted in a green STO:N,La powder, which did not display photocatalytic activity under either UV or vis illumination. The final annealing procedure in air can replenish the oxygen atoms into the vacant sites. This procedure, however, oxidizes nitrogen sites, and the doped nitrogen atoms are released into air. The optimum annealing temperature in air was 600 °C,23 and under this condition, the optimum doping density of La was 0.5%. the STO:N,La powder with the optimum doping density was stable, and its activity did not decrease under exposure to the sunlight for a long term in air as well as in water. The active wavelength of STO:N,La promises a wide range of applications because it absorbs a lot of photons under the sunlight or the indoor lighting. 4. Conclusions STO:N,La exhibited high photocatalytic activities under vis illumination, which is due to the decrease in the oxygen vacancies because codoping maintains the charge balance. The optimum doping density of N and La (0.5%) for STO:N,La exhibits high vis activity without decreasing the UV activity. The present study is quite valuable because it demonstrates that STO:N,La can decompose organic compounds adsorbed on a surface. Therefore, it is possible that this material will be used for various applications such as selfcleaning, water, or air purification. Acknowledgment. The authors would like to thank Mr. Y. Tsuru at TOTO, Ltd., for calculating the band structures. LA0353125 (23) Miyauchi, M.; Takashio, M.; Tobimatsu, H. 70th Conference of The Electrochemical Society of Japan; Tokyo, 2003.