APPLIED PHYSICS LETTERS 98, 092510 2011

Evidence of defect-induced ferromagnetism and its switch action in pristine bulk TiO2
R. K. Singhal,1,a Sudhish Kumar,2 P. Kumari,1 Y. T. Xing,3 and E. Saitovitch3
1 2

Department of Physics, University of Rajasthan, Jaipur 302055, India Department of Physics, M. L. Sukhadia University, Udaipur 313 002, India 3 CBPF, Rua Dr. Xavier Sigaud 150, Urca, Rio de Janeiro 22290-180, Brazil

Received 11 February 2011; accepted 14 February 2011; published online 3 March 2011 In this work, we demonstrate that room temperature ferromagnetism can be induced in pristine anatase TiO2 paramagnetic bulk powder through extended hydrogenation. Defect complexes, Ti3+ VO Ti3+ ions accompanied by oxygen vacancies are clearly identied in hydrogenated TiO2 by combining x-ray diffraction and photoemission spectroscopy. The observed ferromagnetic ordering is reversible that can be switched between on and off by inducing or removing, respectively, these defect complex. We convincingly elucidate that the factors i Ti 3d O 2p hybridization iii F+ centers the electrons in singly occupied oxygen vacancies , and iii oxygen vacancy assisted fragmentation of grains, compositely contribute to the ferromagnetic ordering. 2011 American Institute of Physics. doi:10.1063/1.3562328 In current materials science research, the ferromagnetism observed in defective oxides, including TiO2,16 challenges the conventional understanding of ferromagnetism, which is rather due to spin-split states. Intensive theoretical efforts have been made to understand the ferromagnetism in these defective oxides7,8 which claim that triplet states of p-like electrons, located at cation or oxygen vacancies, yield local moments, leading to ferromagnetic FM ordering without involving 3d electrons. Experimentally, the ferromagnetism in TiO2 has been established to have close link with oxygen vacancies VO ;24 however, the exact mechanism of FM exchange remains illusive.1 Therefore, more systematic experimental work is needed to clarify the fundamental questions: where are the moments located, and what other factors, if any, might hold the key in this mechanism? Lot of research has been performed on transition metal TM doped semiconductors of ZnO, TiO2, In2O3, etc., however, the reports on pristine systems are still rare. The ferromagnetism in absence of TM ions was reported in laserablated TiO2 lms2,3 where Zener double exchange was claimed to generate the ordering. The ferromagnetism in vacuum-annealed TiO2 lms prepared by sputter-deposition was claimed to be mediated by surface defects or interfacial effects.9 In TiO2 single crystals the Ti3+ ions, generated due to oxygen ion irradiation, are reported to provide local 3d moments related with the ferromagnetism.5 On the contrary, there are hardly any such reports on pristine bulk specimens of TiO2. While the reports of ferromagnetism in TM doped oxides are suspected in terms of cluster formation and evolution of secondary phases, the pristine systems are, to some extent, free from such fears. Hence, the selection of highly pure TiO2 virgin samples, in the present research, is expected to exclude effect of magnetic impurities and enable us to focus on the questions outlines above. Argon sealed anatase TiO2 powder purity 99.999% was used in this study. One part of sample was transformed into pellets and other was annealed in high ux of hydrogen for 12 h at 600 C hereafter referred to as TiO2 : H and then
a

transformed into pellets. The crystal structures were investigated by analyzing x-ray powder diffraction patterns recorded using Cu K radiation and Rietveld prole renement were done using FULLPROF program.10 Field dependent magnetization measurements were performed on superconducting quantum interference device SQUID and resistances were measured using Keithley Electrometer. X-ray photoemission spectra XPS were recorded using VG3000 spectrometer having resolution of 0.8 eV. The structural and magnetic characterizations performed in identical conditions conrmed that hydrogenation neither altered the structure nor generated any impurity phase. Figure 1 displays the tted x-ray diffraction XRD patterns of pristine and the hydrogenated TiO2. The Rietveld analysis conrmed that both the samples are in anatase single phase. The obtained lattice and structural parameters are in good agreement with those reported in literature. Hydrogenation of TiO2 leads to small expansion in the cell volume and nearly 5.8% depletion in the oxygen occupancy.

Electronic mail: singhal46@yahoo.co.in. Tel.:

91 141 2545931.

FIG. 1. Color online XRD patterns with Rietveld renement for pristine TiO2 and TiO2 : H. The observed intensity are shown by open circle and the solid line is the calculated intensity; the vertical lines are position of Bragg reections and the lower curve is difference between observed and calculated intensities. 2011 American Institute of Physics

0003-6951/2011/98 9 /092510/3/$30.00

98, 092510-1

Author complimentary copy. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

092510-2

Singhal et al.

Appl. Phys. Lett. 98, 092510 2011

FIG. 2. Color online a Ti 2p XPS spectra for pristine TiO2, hydrogenated TiO2 : H, air-heated TiO2 : H : Htd b Gaussian ttings to Ti 2p3 / 2 peak for TiO2 c Gaussian ttings to Ti 2p3 / 2 peak for TiO2 : H.

FIG. 3. Color online a O 1s spectra for pristine TiO2, hydrogenated TiO2 : H and the air-heated TiO2 : H : Htd b O 1s XPS spectra for TiO2 with Gaussian ts. c Bulk oxygen peaks for pristine TiO2, hydrogenated TiO2 : H, and the air-heated TiO2 : H : Htd d magnied view of the strongest Bragg reected 101 peak for pristine and hydrogenated TiO2.

Figure 2 a shows the Ti 2p XPS spectra for pristine TiO2, hydrogenated TiO2 : H and reheated TiO2 : H : Htd. The spectra of pristine TiO2 shows the core levels of Ti 2p3 / 2 and Ti 2p1 / 2 at 458.85 eV A1 and 464.56 eV A2 , respectively, assigned to Ti4+ in anatase TiO2.4,11 However, the spectrum of TiO2 : H shows interesting changes. The energies of Ti 2p peaks shift 0.36 eV to lower side than the pristine sample and the feature P corresponding to Ti3+ state, not visible in the spectrum of pristine TiO2, appears dominantly in TiO2 : H. Also the peak P, corresponding to Ti4+, gets suppressed in TiO2 : H. These observations demonstrate that the Ti ions transform from 4+ to 3+ state, upon hydrogenation due to deoxidation. This is consistent with Rietveld analysis which revealed that unit cell expands upon hydrogenation, owing to bigger ionic size of Ti3+ 0.67 than that of Ti4+ 0.605 . Notably, the Ti3+ features smear out when the TiO2 : H is reannealed in air for 20 h at 600 C . While a single Gaussian was nicely tted to Ti 2p3 / 2 peak of pristine TiO2 Fig. 2 b , the TiO2 : H spectra was tted by two Gaussians Fig. 2 c . The small Gaussian on lower side clearly indicates the existence of Ti3+ ions which are 9% of the Ti4+ ions the higher side huge Gaussian . Interestingly, the energy of O 1s peak for TiO2 : H, as compared to pristine sample, shifts to higher side i.e., opposite to shift direction of Ti 2p peaks Fig. 3 a . Further, all the O 1s spectra are asymmetric in shape, indicative of presence of surface contamination.1214 Therefore, to get quantitative information on the bulk oxygen content, the spectra were tted with three Gaussians Fig. 3 b . The intense rst peak at 531.3 eV is due to bulk oxygen ensuring that the main signal is eminent from the bulk. The high energy small peaks are due to the surface contamination.1214 The rst Gaussians O 1s , whose intensity is direct measure of bulk oxygen, are separately shown in Fig. 3 c . The hydrogenated sample shows remarkable oxygen depletion 8% . The Ti 2p and O 1s spectra for reheated sample TiO2 : H : Htd were replica of those of pristine TiO2 indicating that

H-induced modications in Ti and O states are reversed upon extended reheating. Above observations can be substantiated as follows: First, the observed shifts of Ti 2p and O 1s XPS peaks, in the directions opposite to each other, yield strong evidence of Ti 3d O 2p hybridization and/or a charge transfer in the hydrogenated sample. Second, there is clear evidence of H-induced formation of O vacancies in the host TiO2 lattice and a subsequent reduction of Ti valence from 4+ to 3+ in accordance with charge neutrality. The observed O deciency 8% is quantitatively consistent with creation of 9% Ti3+ ions in TiO2 : H. Therefore, we afrm that upon hydrogenation the defect complexes Ti3+ VO are originated in TiO2, maintaining the anatese crystalline order. Similar defect complexes were observed in crystals of TiO2 upon oxygen ion irradiation.5 The irradiation was claimed to cause local distortion of the overall TiO6 co-ordination giving rise to a local TiO6x stretching mode in Raman studies.5 Now the question is how are the Ti3+ VO complexes generated upon hydrogenation? When the lattice oxygen is removed from TiO2 upon hydrogenation, the excess electrons which get unpaired,15 tend to occupy the nearby localized Ti 3d orbit and convert Ti4+ ions to Ti3+ state. Therefore, each of H-induced oxygen vacancies leaves two electrons and by converting neighboring Ti4+ to Ti3+ ions, form the Ti3+ VO defect complexes. In rutile TiO2, similar defect complex, Ti3+ VO, has been established by rstprinciples calculations.16 The SQUID measurements for pristine TiO2 and hydrogenated sample TiO2 : H are shown in Figs. 4 a and 4 b . They show weak paramagnetism for pristine TiO2 but a clear ferromagnetic behavior of hydrogenated TiO2 : H coercive eld 171 Oe and saturation magnetization 0.066 emu/ g . The TC estimated from M-T curve in external eld 500 Oe is found to be 450 K Inset Fig. 4 b . The induced FM disappears upon annealing the sample in air, as evidenced by the linear M-H curve Fig. 4 a . For TiO2 : H, we also measured hysteresis loops at different tem-

Author complimentary copy. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

092510-3

Singhal et al.

Appl. Phys. Lett. 98, 092510 2011

FIG. 4. Color online M-H curves for a pristine TiO2, and b hydrogenated TiO2 : H inset shows M-T curve c resistance vs sample states d oxygen content vs sample states. The M-H curves for air-heated TiO2 : H : Htd is also shown in Fig. 4 a .

ferromagnetism.13 In this mechanism, the F+ centers, basically the electrons in the singly occupied oxygen vacancies lying deep in the gap, can favor a ferromagnetic ordering. In pristine TiO2 the F+ centers are likely to be strongly localized; obviously, the pristine sample is likely to be paramagnetic, in agreement with our ndings. On the contrary, a good density of the VOs is created in TiO2 : H which form signicant density of F+ center, that eases the mediation of the observed ferromagnetic coupling. In summary, through simple hydrogenation, it is possible to induce room temperature ferromagnetism in pristine TiO2 bulk powder with paramagnetic background. The induced ferromagnetism bounces back upon reannealing the sample in air. Conclusively, we have identied following factors to be important for the induced ferromagnetism: i Ti3+ VO defect complexes generated due to hydrogenation which provide local 3d moments, ii oxygen vacancy assisted fragmentation of the grains indicative of grain boundary enhancement, iii F+ centers, i.e., the electrons in singly occupied oxygen vacancies, and iv Ti 3d O 2p hybridization as evidenced by photoemission spectroscopy. This work is dedicated to our beloved Mr. G. K. Singhal.
J. M. D. Coey, M. Venkatesan, P. Stamenov, C. B. Fitzgerald, and L. S. Dorneles, Phys. Rev. B 72, 024450 2005 . 2 S. D. Yoon, Y. Chen, A. Yang, T. L. Goodrich, X. Zuo, D. A. Arena, K. Ziemer, C. Vittoria, and V. G. Harris, J. Phys. Condens. Matter 18, L355 2006 . 3 N. H. Hong, J. Sakai, N. Poirot, and V. Briz, Phys. Rev. B 73, 132404 2006 . 4 R. K. Singhal, A. Samariya, S. Kumar, Y. T. Xing, D. C. Jain, S. N. Dolia, U. P. Deshpande, T. Shripathi, and E. Saitovitch, J. Appl. Phys. 107, 113916 2010 . 5 S. Zhou, E. imr, K. Potzger, M. Krause, G. Talut, M. Helm, J. Fassbender, S. A. Zvyagin, J. Wosnitza, and H. Schmidt, Phys. Rev. B 79, 113201 2009 . 6 S. M. Koohpayeh, A. J. Williams, J. S. Abell, J. Lim, and E. Blackburn, J. Appl. Phys. 108, 073919 2010 . 7 J. Osorio-Guilln, S. Lany, S. V. Barabash, and A. Zunger, Phys. Rev. Lett. 96, 107203 2006 . 8 T. Chanier, I. Opahle, M. Sargolzaei, R. Hayn, and M. Lannoo, Phys. Rev. Lett. 100, 026405 2008 . 9 C. Sudakar, P. Kharel, R. Suryanarayanan, J. S. Thakur, V. M. Naik, R. Naik, and G. Lawes, J. Magn. Magn. Mater. 320, L31 2008 . 10 J. Rodriguez-Carvajal, FULLPROF version 3.0.0. Laboratorie Leon Brillouin, CEA-CNRS, France, 2003. 11 A. Manivannan, M. S. Seehra, S. B. Majumder, and R. S. Katiyar, Appl. Phys. Lett. 83, 111 2003 . 12 A. Samariya, R. K. Singhal, S. Kumar, Y. T. Xing, S. C. Sharma, P. Kumari, D. C. Jain, S. N. Dolia, U. P. Deshpande, T. Shripathi, and E. Saitovitch, Appl. Surf. Sci. 257, 585 2010 . 13 R. K. Singhal, P. Kumari, A. Samariya, S. Kumar, S. C. Sharma, Y. T. Xing, and E. Saitovitch, Appl. Phys. Lett. 97, 172503 2010 . 14 R. K. Singhal, A. Samariya, S. Kumar, Y. T. Xing, U. P. Deshpande, T. Shripathi, and E. Saitovitch, Solid State Commun. 150, 1154 2010 . 15 G. U. von Oertzen and A. R. Gerson, J. Phys. Chem. Solids 68, 324 2007 . 16 G. U. Von Oertzen and A. R. Gerson, Int. J. Quantum Chem. 106, 2054 2006 . 17 G. Kopnov, Z. Vager, and R. Naaman, Adv. Mater. 19, 925 2007 . 18 M. D. McCluskey and S. J. Jokela, J. Appl. Phys. 106, 071101 2009 . 19 B. B. Straumal, A. A. Mazilkin, S. G. Protasova, A. A. Myatiev, P. B. Straumal, G. Schtz, P. A. van Aken, E. Goering, and B. Baretzky, Phys. Rev. B 79, 205206 2009 . 20 Y. K. Vijay, Int. J. Hydrogen Energy 33, 340 2008 . 21 J. M. D. Coey, M. Venkatesan, and C. B. Fitzgerald, Nature Mater. 4, 173 2005 .
1

peratures not shown which are weakly temperature dependent, a feature of the defect-induced ferromagnetism.17 The magnitude of resistance was measured on identical bricks 4.0 2.0 0.5 mm3 for these samples. Resistance was found to be 450 G for pure TiO2 which reduces to 12 G for TiO2 : H. However, the order of magnitude, in gigaohm range, indicates low concentration of carriers, despite hydrogenation. Resistance of the reheated sample TiO2 : H : Htd again increases to 413 G Fig. 4 c . The creation of VO is akin to creating conduction electrons consistent with depression in the resistance revealing that the carrier concentration in TiO2 is affected by O vacancies. Figure 4 d shows the behavior of VO for different sample states. To ensure that the O deciency and the associated moments are not conned to the surface layers we etched as well as broke the hydrogenated pellet and performed the SQUID and XPS measurements which show consistent results. In light of the claims18,19 that the grain boundaries and the vacancy defects are important causes of intrinsic ferromagnetism, we closely re-examined x-ray diffractograms. Figure 3 d compares magnied view of the strongest Bragg reected 101 peak for pristine, hydrogenated, and the reheated TiO2. A remarkable broadening in the peak is observed for hydrogenated sample. This can be attributed to the oxygen vacancy assisted fragmentation of the grains. Since the hydrogen in metals and their oxides resides at the interstitial sites, i.e., the grain boundaries,20 we elucidate that the oxygen depletion takes place at the grain boundaries of TiO2. This in turn is likely to cause migration of oxygen from lattice to the grain boundary side, therefore, generating the oxygen vacancy assisted grain boundary. This would result in reduction in average grain size as indicated by the XRD peak broadening. These ndings show qualitative agreement with the ndings which claim grain boundaries to be crucial for ferromagnetic ordering.19 Interestingly, on deloading the hydrogen the tails of XRD peaks again sharpen. The F-center exchange21 mechanism, a subcategory of the bound magnetic polaron model, is very useful for the relevance of oxygen vacancies in the induced

Author complimentary copy. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp