Preparation and spectroscopic characterisation of nitrogen doped titanium dioxide

Studies in Surface Science and Catalysis 155 A. Gamba, C. Colella and S. Coluccia (Editors) 9 2005 Elsevier B.V. All rights reserved

375

Preparation and Spectroscopic Characterisation of Nitrogen Doped Titanium Dioxide Stefano Livraghi, Annamaria Votta, Maria Cristina Paganini*, Elio Giamello Dipartimento di Chimica IFM, Universit/ldi Torino and NIS, Nanostructured Interfaces and Surfaces Centre of Excellence, Via P. Giuria 7, I - 10125 Torino (Italy) Photocatalytic oxidation constitutes one of the most promising methods for indoor and outdoor air purification. Solar energy contains only about 5% UV light and much of the rest is visible light. In order to utilize solar energy efficiently in photocatalytic reactions, it is necessary to develop a visible light reactive photocatalyst having smaller band gaps than TiO2 rutile and anatase. In 1986 Sato reported that calcinations of NH4C1 containing titanium hydroxide caused the photocatalytic sensitization of TiO2 into the visible light region. The author proposed that the powder prepared according to the described method are actually NO• TiO2 and that the sensitization of these materials is due to the presence of NOx impurity. Several years after it has been reported that nitrogen-doped titania with a particularly high visible light photocatalytic activity had been prepared. The aim of the present work was to synthesize and characterize new materials based on N-doped TiO2. The materials N/TiO2 were prepared via sol-gel technique using solutions containing various kind of nitrogen compounds. UV-Vis diffuse reflectance spectroscopy and Electron Paramagnetic Resonance were the main experimental techniques used to characterized the materials. The latter technique was adopted to verify the presence of paramagnetic entities formed during the synthesis. In most cases the samples prepared via sol-gel reaction exhibit a pale yellow color. 1. INTRODUCTION The search for a good titanium dioxide based photocatalytic system active with visible light has been the object of a great deal of research in recent years. Bare titanium oxide in fact is the most important photocatalyst for the decomposition of pollutants either in gas or in liquid phase because of its high activity and chemical stability. A limit of bare TiO2 however consists in the amplitude of its band gap (3.2 eV for Anatase, the most catalytically active phase) which requires the use of UV electromagnetic radiation to produce the electron-hole separation. Among the materials based on TiO2 a considerable interest was raised by those doped with non metal atoms. In particular, the most promising system seems to be that obtained doping the oxide with nitrogen. This reduces the amplitude of the gap between the valence and the conduction band [ 1]. In 1986 S. Sato [2] reported the preparation of a yellow material obtained by calcination of titanium hydroxide in the presence of ammonium chloride. The optical properties of this solid were attributed to rather undefined NOx species supposed to produce impurity levels in the band gap of TiO2. Later, in the nineties, several papers reported the preparation of N doped TiO2 (N-TiO2) according to various experimental procedures including mechanochemistry [3], DC magnetron sputtering [4], sol-gel synthesis [5], high

376 temperature treatments in N2 or NH3 atmosphere [6] or oxidation of a precursors such as TiN [7]. However, despite the great deal of activity in the field, the typical features of N-TiO2 causing its peculiar catalytic activity are not yet fully clarified. While Sato [2] points to the presence of rather undetermined nitrogen oxospecies (NOx), Asahi et al. [1] suggested (based on theoretical calculations) the presence of nitrogen impurities in substitutional sites of the TiO2 matrix as responsible of the band gap narrowing (caused by mixing of N and O 2p states) necessary for the onset of reactivity under visible light. More recently the presence of defective sites like oxygen vacancies and reduced centres [5,6] has been also suggested. In parallel with the debate on the reasons of the photocatalytic properties of N-TiO2 an intense activity grew up aimed to characterize the typical features of this relatively new solid system. A controversial point of this debate concerns the presence and the role of a typical XPS feature at 396eV observed in the case of N-TiO2 prepared by magnetron sputtering and assigned by the authors to N atoms present in the TiO2 lattice in substitutional position. As a matter of fact such a feature was not observed in the case of other materials also active in visible light photocatalysis. Minor emphasis was devoted in the literature to the presence of paramagnetic defect species except for few poorly assigned signals reported by Sakatani et al. [8] and despite the fact that radical centres often are important defects of the solid state connected to important physical properties. The role of EPR in the characterisation of defects and impurities in the solid state is well known as, in many cases, the defective centres (dangling bonds, trapped electrons, atomic impurities) are paramagnetic [9]. The present paper describes the features of a series of N-TiO2 materials prepared via sol-gel synthesis or alternatively via mechanochemical activation.

2. E X P E R I M E N T A L .

N-TiO2 (100% anatase) samples were prepared via two different routes (sol-gel chemistry and mechanochemistry) using Ti compounds and various N containing compounds (NH4C1, N2H4) to obtain N doping. We will report hereafter procedure followed using NH4C1 as nitrogen source. For both preparation routes the N/TiO2 ratio in the starting mixture was the same (5 wt%). However the N content of the N/TiO2 final samples was not determined. In both cases pale yellow samples were obtained with the typical optical absorption centred at about 450 nm. The solids show catalytic activity in visible light which was monitored according to the method proposed by Burda et al. [10] based on the methylene blue degradation under visible irradiation followed by U.V.-Vis. Spectrometer. Sol-gel N-TiO2 samples were prepared mixing a solution of titanium (IV) isopropoxide in isopropilic alcohol with a solution of water and keeping the mixture under constant stirring at room temperature until completed hydrolysis. The gel was left ageing for 15 hours at room temperature and subsequently dried at 70~ for 2 hours. The dried material was calcined in air at 770K from a minimum of one hour up to a maximum of 15 hours. After calcination the solid is pale yellow. The mechanochemical preparation of N-TiO2 was performed starting from an intimate mixture of powders of TiO2 (synthesised via sol-gel method before described) and NH4C1 which was put in a mechanical mill and ground using corundum balls for 1 hours. After 1 hour calcinations in air at 770K the activated material was put in the glass manifold for spectroscopic investigations. For a better understanding of EPR spectra, isotopically label materials were also prepared using 70% enriched 15NH4C1 and following the same procedures described above. A

377 reference sample of bare TiO2 was also prepared via hydrolysis in pure water. XRD analysis of the sol-gel material after calcination indicates the exclusive formation of the anatase phase of TiO2 with particle size of about 50 nm while in the case of the mechanically activated material a fraction of rutile wad detected by XRD whose formation has to be ascribed to the high energies developed during the mechanical milling. The UV-Vis measurement were performed by means of a Varian Cary 5 spectrometer using a Cary win-U.V./scan software for data handling. EPR spectra were run in the range between 77K and room temperature by a Bruker EMX spectrometer working in the X-band and equipped whit a cylindrical cavity. Computer simulation of the EPR spectra were obtained by the SIM32 program elaborated by Prof. Z. Sojka (Jagellonian University- Cracow). 3. RESULTS AND DISCUSSION. The diffuse reflectance (DR) UV-Vis spectrum of a N-TiO2 sample prepared via sol-gel is reported in Fig. 1 where it is compared with that of bare TiO2. Similar spectra (also in terms of adsorption intensity) were obtained using samples get by mechanochemical treatment. Fig. 1 shows that the N doped sample is characterised by an absorption in the visible region at about 450 nm and by a band gap narrowing of about 0.08 eV (derived from the plot of the square root of the Kubelka-Munk function [F( R )E] 1/2 against the photon energy E) [ 11].

%R a.u.

/ ./....."......'.....~............., /

:y

~60

,~0

9

N-dopedAnatase

500 '

~60 nm

~60

800 '

Figure 1. Diffuse Reflectance U.V.-Vis. spectra of pure and nitrogen doped TiO2 All the samples exhibiting the described absorption in the visible also show a catalytic activity in the photodecomposition of methylene blue using visible light. The activity (details will be reported in a forthcoming paper) was measured by comparison with the activity of a sample of pure TiO2 in the same experimental conditions. A second important feature of N-TiO2 is the presence of EPR signals not observed in the bare oxide. All the observed signals are due to N containing species and are thus related to the doping process. Two main EPR signals were observed after calcinations in air of the xerogel precursor (prepared using NHaC1 as doping agent).

378

[

a

~'~

'

~'00

'

B/Gauss

~'~

Figure 2. EPR spectra recorded at 298 K of species B present in the N-TiO2 matrix, a: spectrum obtained using 14N compounds; b: Spectrum obtained using 15N enriched compounds.

The first one (species A) is observed at 77K and disappears rising the temperature at about 170K revealing the presence of a second signal (species B) which is observed also at room temperature. Both species were obtained using either 14N or 15N enriched (70%) compounds (Fig. 2 and 3). The EPR spectrum of species B (Fig.2) is characterised by a g tensor with the three principal values extremely close one to the other (gl = 2.003, g2= 2.004, g3= 2.005) and with a hyperfine structure dominated by a triplet of lines (the nuclear spin is I=l for 14N and a triplet of lines is expected for the hyperfine structure of a single N atom in the species) with 32 G separation. The whole tensor is as follows: A~=32 G, A2~0 G, and A3~5 G. The second triplet at 5 G is not fully resolved in the experimental spectrum (Fig. 2a). These features are fully confirmed by the spectrum generated using 15N enriched compounds (Fig. 2b, about 70% of enrichment). Since the nuclear spin of 15N is 89the main triplet is now a pair of hyperfine lines separated by 45 G as expected on the basis of the ratio of the two g nuclear factors of 14N and 15N which is 1.41. It has to be noticed that Fig.2b still shows the 14N trace which contributes for some 30% to the whole spectrum. The illustrated features are typical of a species containing one N atom with half of the total spin density concentrated in the p orbital of this atom. Species A signal shows up by cooling the system at 77K. As easily deduced from the spectra (Fig. 3) also A contains one nitrogen atom and is thus related to the doping process of the solid. Fig. 3 reports the EPR spectra of species A respectively obtained employing in the synthesis 14NH4C1 (3a) and 15NH4C1 (3b). The spectrum in Fig. 3a, whose main feature is a 14N hyperfine triplet centered at about g = 2.0 is, differently from the case of species B, characterized by a rhombic g tensor (gl = 2.001, g2 =1.998, g3 = 1.927 ) with one value (g3) pronouncedly lower than the free spin value (2.0023). Also the A tensor is anisotropic with A~ 0, A2 = 32.12 G, A3 "- 10 G. The latter value is not completely resolved in the spectrum and has been deduced by comparison of the two experimental spectra (3a and 3b). In the case of 15N signal each hyperfine triplet is substituted by a doublet whose separation is again that expected on the basis of the ratio of the two ~ nuclear factors.

379

~2 I

g3

j1~

S lN

Y ~'00

'

~'~

BG / auss '

~.'~0

'

~'~s

'

Figure 3. EPR spectra (recorded at 77K) of species A, trapped into the N-TiO2 matrix, a: spectrum obtained with 14N compounds, b: Spectrum obtained with 15N enriched (70%) compounds. The structure of species A signal is in agreement with what expected for a 1 l e diatomic n radical and is assigned to nitric oxide (NO). The ground state configuration of this radical is: 1(5"2 2cy2 3cy2 4cy2 lit 4 5cy2 2n 1 and its spectrum is not observed in the spectral region of the free electron unless the degeneracy between the two 2n antibonding orbitals is lift by the effect of some asymmetric electric field [12,13,14]. In this case the unpaired electron is confined, at first approximation, in the n orbital with lower energy. This leads to a rhombic g tensor whose elements are, at first order, as follows: gxx =

g l = ge

gyy = g2 = g e - 2

L/E

(1)

gzz = g3 = ge- 2;~/A where )~ is the the spin orbit coupling parameter associated to the investigated radical while A is the 2ggx - 2ggy separation and E is the 2ggx - 5~g separation. The gzz (g3) component (z is now the direction of the internuclear axis) is a measure of the value of A and indirectly of the weak field perturbing the NO orbitals. The fact that both A and B species are not affected by outgassing up to about 470K and the absence of dipolar broadening when the EPR spectra are recorded in the presence of molecular oxygen indicates that the two N containing species here described are located into the bulk of N-TiO2 microcrystals. The NO molecule is probably segregated in some morphological imperfection of the structure (microvoids) formed along the preparation of the solid. The assignment of species B is not straightforward. The electronic structure and the g values of the species described above are compatible with the properties of an oxo-nitrogen radical ion like NO22, a 19-electrons bent species with the unpaired electron confined into a 2bl antibonding orbital mainly involving a p orbital of the central nitrogen atom and two parallel p orbitals of the oxygen atoms [ 15]. Formation of NOe 2 indicates a deep chemical

380 interaction between the ammonium ions employed in the preparation and the oxide phase. However the features of species B are compatible also with other nitrogen species deeply interacting with the oxide lattice. In particular recent calculations [ 16] on the state of the N impurity in the TiO2 anatase lattice point to the formation of an isolated paramagnetic state in the band gap close to the limit of the valence band edge. A similar isolated N center should carry, as indicated by preliminarily calculations, a spin density of about 0.5 with the remaining spin density distributed on the nearby oxygen and titanium ions. Such a results is in nice agreement with the features of species B. Further experiments (possibly EPR of single crystal) and elaborations are needed to reach a complete understanding of the nature and structure of this center. Whatever the exact nature of the B species the fact remains that the latter one is placed in the bulk of the system and is the result of a deep interaction with the matrix. In conclusion we have shown that N-TiO2, an intriguing photocatalytic system active in visible light, contains in its bulk two types of radical centers. These are formed during the synthesis of the material (performed by two different methods) by interaction of an oxidic phase with nitrogen containing compounds. The two centers are observed after calcination in oxygen up to 770K indicating that we are dealing with stable non transient species well stabilized into the oxide matrix. The first one is molecular NO permanently trapped into the crystal and the second one is a nitrogen based paramagnetic center deeply interacting with the oxide and characterized by a remarkable spin density in a p orbital of the nitrogen atom. The connection between one or both these species with the photocatalytic activity of N-TiO2 is currently under investigation in our laboratory.

REFERENCES [ 1] R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, Y. Taga, Science, 293 (2001) 269. [2] S. Sato, Chem. Phys. Lett., 123 (1986) 126. [3] S. Yin, Q. Zhang, F. Saito, T. Sato, Chem. Lett., 32 (2003) 358. [4] T. Lindgren, J.M. Mwabora, E. Avendano, J. Jonsson, A. Hoel, C. Granqvist, S. Lindquist, J. Phys. Chem B, 107 (2003) 5709. [5] a) S. Sakthivel, H. Kisch, ChemPhysChem ,4 (2003) 487. b) T. Ihara, M. Miyoshi, Y. Iriyama, O. Matsumoto, S. Sugihara, Applied Catal. B: Envir., 42 (2003) 403. [6] H. Irie, Y. Watanabe, K. Hashimoto, J. Phys. Chem. B, 107 (2003) 5483. [7] T. Morikawa. R. Asahi, T. Ohwaki, K. Aoki, Y. Taga, Japn. J. Appl. Phys., 40 (2001) L561. [8] Y. Sakatani, J. Nunoshige, H. Ando, K. Okusako, H. Koike, T. Takata, J.N. Kondo, M. Hara, K. Domen, Chem. Lett., 32(12) (2003) 1156. [9] F. Agullo Lopez, C.R.A. Catlow, P.D. Towsend, Point defects in materials, Academic Press pp 1012. (1988). [ 10] C. Burda, Y. Lou, X. Chen, A.C.S. Samia, J. Stout, J. L. Gole, Nanoletters, 3 (2003) 1049. [ 11] D. G. Barton, M. Shtein, R. D. Wilson, S. L. Soled, E. Iglesia, J. Phys. Chem. B 103 (1999) 630. [12] C. Di Valentin, G. Pacchioni, M. Chiesa, E. Giamello, S. Abbet, H. Heiz, J. Phys. Chem. B, 106 (2002) 1637. [ 13] J. H. Lunsford, J. Phys. Chem., 72 (1968) 2141. [ 14] M. Primet, M. Che, C. Naccache, M. V. Mathieu, B. Imelik, J. Chim. Phys., 67 (1970) 1629. [ 15] P.W. Atkins, M. C. R. Symons, The Structure of Inorganic Radicals, Elsevier (Amsterdam) pp 148(1967). [16] C. Di Valentin, G. Pacchioni, A. Selloni, Phys. Rev. B, 70 (8) (2004) 85116.