APPLIED PHYSICS LETTERS 90, 171909 共2007兲
First-principles calculation of N:H codoping effect on energy gap narrowing of TiO2 Lan Mi, Peng Xu, Hong Shen, and Pei-Nan Wanga兲 State Key Lab for Advanced Photonic Materials and Devices, Department of Optical Science and Engineering, Fudan University, Shanghai 200433, China
Weidian Shen Department of Physics and Astronomy, Eastern Michigan University, Ypsilanti, Michigan 48197
共Received 4 January 2007; accepted 27 March 2007; published online 24 April 2007兲 The energy band structures and density of states for N-doped and N:H-doped anatase TiO2 are calculated based on the first-principles density-functional theory. For N-doped TiO2, there appear two isolated states above the top of the valence band and the band gap narrowing is very small. With the same nitrogen dopant concentration, N:H doping yields a significant band gap narrowing. The calculated results support our experimental data that N:H-doped TiO2 exhibited higher visible-light photocatalytic efficiency than the N-doped one. © 2007 American Institute of Physics. 关DOI: 10.1063/1.2731707兴 Nanostructured TiO2 has attracted worldwide attention as a promising material in photocatalytic and photoelectrochemical field for years.1–3 However, with its absorption edge below 380 nm, TiO2 has photoactivity only under ultraviolet 共UV兲 light which possesses only a small fraction 共about 8%兲 of the solar energy.4 So, shift in the optical absorption of TiO2 from UV to the visible will impose a profound positive effect on enhancing the photocatalytic efficiency of this material using sunlight as the energy source. Great efforts have been made to improve its visible-light absorption by doping nitrogen into anatase TiO2 and visiblelight photocatalyses have been reported by many research groups.4–12 However, the origin of the redshift of absorption edge and the mechanism of photocatalysis in the visible region are still somewhat under debate. Different chemical species such as NOx,13–18 substitutional N,4,19,20 or NHx5 have been proposed as responsible for this effect. An outstanding issue is which one plays the key role in the visiblelight photoactivity of the material. To explain the beneficial doping effect on the visiblelight photocatalyses in experiments, a variety of theoretical computations have been conducted. Asahi et al.4 attributed the redshift of absorption edge to band gap narrowing driven by strong mixing of N 2p states with O 2p states. However, Di Valentin et al. concluded according to their improved calculation that both N-doping configurations 共substitutional and interstitial兲 induce the formation of localized occupied states in the energy gap. Substitutional nitrogen states lie just above the top of the valence band 共VB兲 共0.14 eV兲, while interstitial nitrogen states lie higher in the band gap 共0.73 eV兲.17,21 Recently, in their computations, Lee et al.22 and Long et al.23 found that the mixing of N with O 2p states is too weak to produce a significant band narrowing and the photocatalytic activity for visible light was due to the presence of isolated N 2p states just above the valence-band maximum 共VBM兲 of TiO2. So far, almost all the theoretical calculations have been focused on the band gap narrowing due to nitrogen doping. The effect of N:H codoping on narrowing the band gap of TiO2 has been neglected although a兲
Author to whom correspondence should be addressed; electronic mail:
[email protected]
this effect was postulated by Diwald et al. in their experimental report.5 However, it should be noted that in many cases, the N-doped TiO2 materials, which showed a remarkable redshift in absorption, were processed or produced with NH3 gas,4–8 NH3 aqueous solution,9–11 or H2.12 Therefore, in this letter, we calculated the band structure and the density of states of N:H-doped anatase TiO2 and compared them with that of undoped and N-doped TiO2, in order to clarify how these impurities impact the band gap of anatase TiO2. To obtain microscopic insight into the impacts of N doping and N:H doping on the photoactivities of anatase TiO2, calculations were performed using the CASTEP code based on the first-principles density-functional theory 共DFT兲.24 The exchange and correlation interactions were modeled using the generalized gradient approximation and the PerdewBurke-Ernzerhof. The wave functions of the valence electrons were expanded using a plane wave basis set within a specified cutoff energy of 340 eV. The core electrons were replaced by the ultrasoft core potentials, and the valence atomic configurations were 3s23p63d24s2 for Ti, 2s22p4 for O, 2s22p3 for N, and 1s2 for H. A conventional unit cell of TiO2 has 12 atoms: four Ti and eight O. It has a body centered tetragonal structure and belongs to the I41 / amd space group with local symmetry D2d. Since substitutional N was detected in many N-doping experiments for TiO2 and was considered to be responsible for the visible-light catalysis by many researchers,4,6,25 we chose substitutional N for the calculation of N-doping and N:H codoping effects in this work. In the calculations, supercells with two unit cells were used in constructing the N-doped and N:H-doped structures, where one O atom is replaced by one N atom or a N–H species, as shown in Fig. 1. One supercell consists of 8 titanium and 16 oxygen atoms. Therefore, 6.25% of the O sites is replaced by N or N–H corresponding to a N concentration of 4.17%, which is comparable with that obtained in our previous experiment.26 It was demonstrated in Di Valentin’s calculations that up to 5% of N sites, the structural modification is still small and similar to that observed at 2%.21 Thus, we conducted the calculation in a N concentration of 4% and focused on the differences between N-doping and N:H codoping effects. The lattice parameters and the atomic coordinates of the super-
0003-6951/2007/90共17兲/171909/3/$23.00 90, 171909-1 © 2007 American Institute of Physics Downloaded 24 Apr 2007 to 202.120.224.18. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
171909-2
Mi et al.
Appl. Phys. Lett. 90, 171909 共2007兲
FIG. 1. 共Color online兲 Partial geometry of the model for N:H-doped anatase TiO2. The N, H, O, and Ti atoms are represented by blue, green, red, and gray spheres, respectively.
cells including nitrogen impurities were set as the same values corresponding to a perfect cell. The crystal structure was determined in the computation by minimizing the total energy, and all atoms were relaxed to have a stable configuration. The optimized bond length of N–H was 1.033 Å, which is comparable with that in a NH3 molecule 共1.05 Å兲. A doped nitrogen atom in the O site introduces an unpaired electron which can contribute to the N–H bonding. The calculation shows that the formation of a pair of N and H is energetically favorable. After the crystalline structures of the undoped and doped anatase TiO2 were geometrically optimized by computation, the energy band structures and the total density of states 共TDOSs兲 were then calculated to explore the origin of visible-light photocatalysis. For further analysis of the constitution of VB and conduction band, calculation of partial density of states 共PDOSs兲 was also carried out. The calculated band gap of undoped anatase TiO2 was 2.42 eV, as shown in Fig. 2共a兲, which is similar to that reported,21–23 but underestimated compared with the experimental Eg = 3.2 eV due to the well-known limitation of DFT. The calculated band structures of N-doped TiO2 and N:Hdoped TiO2 are displayed in Figs. 2共b兲 and 2共c兲, respectively. For N-doped TiO2, two isolated energy states are located just above the top of the VB of TiO2 and the resulted band gap is 2.38 eV, as shown in Fig. 2共b兲, which is very close to that of undoped TiO2 共2.42 eV兲. Even with a high N-doping concentration of 4%, the narrowing of the band gap is only 0.04 eV, which is similar to the results obtained by Lee et al.22 and Long et al.23 with a doping concentration of 2.08%. They found that the charge character of the N 2p states showed a very weak interaction with the neighboring Ti and O atoms.22 For N:H-doped TiO2, no isolated energy states appear in the band gap, as shown in Fig. 2共c兲. In this case, a continuum of states is formed and the band gap is narrowed to 2.16 eV. The narrowing of the band gap is about 0.26 eV, much larger than that of N-doped TiO2 共0.04 eV兲. To further understand the origin of the band gap narrowing due to the N doping and N:H codoping, the calculated TDOS and PDOS 共as depicted in Fig. 3兲 are inspected. Two isolated N 2p states above the VB can be clearly seen for N-doped TiO2, as shown in Figs. 3共a兲 and 3共b兲. Whereas for N:H-doped TiO2, the N 2p states are slightly above the VB of TiO2, which sufficiently mixed with the O 2p states in the VB 关Figs. 3共c兲 and 3共d兲兴. No isolated N 2p states are observed. The continuum mixed states shift the VBM upwards. The effect of codoping in semiconductors has been studied previously.5,27 As postulated by Diwald, a hybridized state 共N p orbitals and H s orbitals兲 is formed in N:H-doped TiO2.
FIG. 2. 共Color online兲 Calculated band structure of 共a兲 undoped, 共b兲 N-doped, and 共c兲 N:H-doped TiO2 along the symmetry lines of the first Brillouin zone.
The donors 共hydrogen兲 may contribute to the lowering of the energy levels of the acceptors 共nitrogen兲, bringing the N states closer to the VB, therefore enhancing the mixing of N 2p states with the O 2p states in VB. In this case, the isolated states disappear and a continuum is formed as demonstrated in our computation. The mixing of the O states in the VB with the N states above the VBM leads to a real band gap narrowing and consequently a redshift of the optical absorption edge. Although the isolated states are not contributing to the narrowing of band gap, namely, the redshift of absorption edge, they are still propitious to the absorption in longer wavelength in a certain extent and favor the visible-light catalytic activities. The N-doping and N:H-doping effects on photocatalytic activity were also studied by experiment. The bare TiO2,
Downloaded 24 Apr 2007 to 202.120.224.18. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
171909-3
Appl. Phys. Lett. 90, 171909 共2007兲
Mi et al.
FIG. 3. 共Color online兲 共a兲 TDOS of N-doped TiO2, 共b兲 PDOS of N 2p in N-doped TiO2, 共c兲 TDOS of N:H-doped TiO2, and 共d兲 PDOS of H 1s and N 2p in N:H-doped TiO2.
N-doped, and N:H codoped TiO2 films in anatase phase were prepared on glass substrates by laser ablation of titanium target under the same experimental conditions but different environment atmospheres of O2, N2 / O2, and NH3 / N2 / O2, respectively. The experimental details can be found in our previous paper.26 The visible-light photocatalytic ability was evaluated by studying the decomposition of methylene blue 共MB兲 in aqueous solution under the 405 nm illumination on the catalyst dipped in the solution. The results are demonstrated in Fig. 4. The lower absorbance of the solution corresponds to more decompositions of MB. The curves were obtained by measuring the time-dependent absorbances of aqueous MB solutions within 24 h. As an example, a time evolution of the absorption spectra is shown in the inset of Fig. 4. It can be seen that very little change is observed for the absorbance of the solution with bare TiO2, indicating that bare TiO2 does not have significant photocatalytic ability in the visible region and the possibility of direct photolysis of MB at this wavelength can also be ruled out. Obviously, the film prepared in NH3 / N2 / O2 exhibited higher photocatalytic ability than that prepared in N2 / O2. The visible-light catalytic ability of N-doped TiO2 might be due to the isolated N 2p states above the VBM, while for N:H-doped TiO2, a higher photocatalytic ability was achieved due to the band
FIG. 4. 共Color online兲 Decomposition of MB in solutions under 405 nm illumination with catalysts of bare 共square兲, N-doped 共circle兲, and N:Hdoped 共triangle兲 TiO2.
gap narrowing. The experimental data seem to support our theoretical conclusion to a certain extent. However, the doping situation varies with different preparation methods and experimental conditions. The photocatalysis reaction is a very complicated phenomenon including light absorption by semiconductor band gap, excitation and migration of charge carriers, redox reaction on the surface with chemical species, and so on. More studies should be conducted to provide further evidences for the origin of the visible-light photocatalysis of doped TiO2. In summary, by means of first-principles DFT calculation, at a nitrogen dopant concentration of 4%, N:H doping yields a significant band gap narrowing of 0.26 eV while it is only 0.04 eV resulted from N doping. These calculation results provide an explanation for our experimental data. It is most likely that N:H codoping can effectively redshift the absorption edge and result in a significant improvement in the visible-light photocatalytic activity of TiO2. Financial support from the National Natural Science Foundation of China 共60638010兲 is gratefully acknowledged. The authors are grateful to Kangnian Fan of Department of Chemistry, Fudan University for his help in the computational calculation. 1
I. Nakamura, N. Negishi, S. Kutsuna, T. Ihara, S. Sugihara, and E. Takeuchi, J. Mol. Catal. A: Chem. 161, 205 共2000兲. A. L. Linsebigler, G. Lu, and J. T. Yates, Jr., Chem. Rev. 共Washington, D.C.兲 95, 735 共1995兲. 3 M. R. Hoffmann, S. T. Martin, and D. W. Bahneman, Chem. Rev. 共Washington, D.C.兲 95, 69 共1995兲. 4 R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, and Y. Taga, Science 293, 269 共2001兲. 5 O. Diwald, T. L. Thompson, T. Zubkov, E. G. Goralski, S. D. Walck, and J. T. Yates, J. Phys. Chem. B 108, 6004 共2004兲. 6 H. Irie, Y. Watanabe, and K. Hashimoto, J. Phys. Chem. B 107, 5483 共2003兲. 7 S. Mozia, M. Tomaszewska, B. Kosowska, B. Grzmil, A. W. Morawski, and K. Kalucki, Appl. Catal., B 55, 195 共2005兲. 8 B. Wawrzyniak and A. W. Morawski, Appl. Catal., B 62, 150 共2006兲. 9 T. Ihara, M. Miyoshi, Y. Iriyama, O. Matsumoto, and S. Sugihara, Appl. Catal., B 42, 403 共2003兲. 10 H. Li, J. Li, and Y. Huo, J. Phys. Chem. B 110, 1559 共2006兲. 11 Y. Kuroda, T. Mori, K. Yagi, N. Makihata, Y. Kawahara, M. Nagao, and S. Kittaka, Langmuir 21, 8026 共2005兲. 12 L. Miao, S. Tanemura, H. Watanabe, Y. Mori, K. Kaneko, and S. Toh, J. Cryst. Growth 260, 118 共2004兲. 13 S. Sato, R. Nakamura, and S. Abe, Appl. Catal., B 284, 131 共2005兲. 14 S. Sato, Chem. Phys. Lett. 123, 126 共1986兲. 15 S. Sakthivel, M. Janczarek, and H. Kisch, J. Phys. Chem. B 108, 19384 共2004兲. 16 Y. Sakatani, J. Nunoshige, H. Ando, K. Okusako, H. Koike, T. Takata, J. N. Kondo, M. Hara, and K. Domen, Chem. Lett. 32, 1156 共2003兲. 17 C. Di Valentin, G. Pacchioni, A. Selloni, S. Livraghi, and E. Giamello, J. Phys. Chem. B 109, 11414 共2005兲. 18 X. B. Chen and C. Burda, J. Phys. Chem. B 108, 15446 共2004兲. 19 M. Miyauchi, A. Ikezawa, H. Tobimatsu, H. Irie, and K. Hashimoto, Phys. Chem. Chem. Phys. 6, 865 共2004兲. 20 Z. Lin, A. Orlov, R. M. Lambert, and M. C. Payne, J. Phys. Chem. B 109, 20948 共2005兲. 21 C. Di Valentin, G. Pacchioni, and A. Selloni, Phys. Rev. B 70, 085116 共2004兲. 22 J.-Y. Lee, J. Park, and J.-H. Cho, Appl. Phys. Lett. 87, 011904 共2005兲. 23 M. Long, W. Cai, Z. Wang, and G. Liu, Chem. Phys. Lett. 420, 71 共2006兲. 24 M. D. Segall, P. J. D. Lindan, M. J. Probert, C. J. Pickard, P. J. Hasnip, S. J. Clark, and M. C. Payne, J. Phys.: Condens. Matter 14, 2717 共2002兲. 25 K. Shankar, K. C. Tep, G. K. Mor, and C. A. Grimes, J. Phys. D 39, 2361 共2006兲. 26 P. Xu, L. Mi, and P.-N. Wang, J. Cryst. Growth 289, 433 共2006兲. 27 T. Yamamoto, Thin Solid Films 420-421, 100 共2002兲. 2
Downloaded 24 Apr 2007 to 202.120.224.18. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
