Journal of Magnetism and Magnetic Materials 324 (2012) 2932–2935

Contents lists available at SciVerse ScienceDirect

Journal of Magnetism and Magnetic Materials

0304-88

http://d

n Corr

E-m

journal homepage: www.elsevier.com/locate/jmmm

Effect of vacuum-annealing on the d0 ferromagnetism of undoped In2O3 films

Shaohua Sun, Ping Wu, Pengfei Xing n

Department of Applied Physics, Institute of Advanced Materials Physics, Tianjin Key Laboratory of Low Dimensional Materials Physics and Preparing Technology,

Faculty of Science, Tianjin University, Tianjin 300072, PR China

a r t i c l e i n f o

Article history:

Received 14 December 2011

Received in revised form

23 April 2012Available online 8 May 2012

Keywords:

d0 ferromagnetism

Indium oxide

Vacuum-annealing

Oxygen vacancy

Radiofrequency magnetron sputtering

53/$ - see front matter & 2012 Elsevier B.V. A

x.doi.org/10.1016/j.jmmm.2012.04.050

esponding author. Tel.: þ86 22 27403488; fa

ail address: pfxing@tju.edu.cn (P. Xing).

a b s t r a c t

Vacuum-annealing was carried out on the pure indium oxide films deposited on Si (100) substrates by

radiofrequency magnetron sputtering. Oxygen-deficiency states and room temperature d0 ferromag-

netism were both detected in the as-grown and vacuum-annealed films. With more oxygen vacancies

appeared through vacuum-annealing, the saturation magnetization increased rapidly from 0.5 to

5.5 emu/cm3. The connection between the highly oxygen-deficiency states and the strong magnetic

moment suggests that oxygen vacancies play a crucial role in mediating the ferromagnetism in In2O3

films. We think that this d0 ferromagnetism mainly stems from Vþ0 and oxygen vacancy clusters in the

interfaces or grain boundaries.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

Since ferromagnetism with TC above 300 K in Mn-doped ZnO wastheoretically predicted by Dietl et al. [1], diluted magnetic oxidesemiconductors (DMOSs) with possibility to simultaneously controlspin and charge have gained much attention due to their potentialapplications in spintronics [2–5]. Room-temperature ferromagnet-ism (RTFM) has been discovered by some groups in transition-metal(TM)-doped oxide semiconductors, such as ZnO, In2O3, SnO2, etc.But the probability that the ferromagnetism originates from ferro-magnetic clusters or secondary phases [6] is hard to be removed dueto the low solubility of the TM elements in the oxide lattice. Recently,the discovery of d0 ferromagnetism in pure HfO2 opens up a newway for magnetic semiconductors [7]. As there is no magnetic-doping in the d0 system, the probable impact induced by TM-dopingcan be effectively ruled out. d0 ferromagnetism has also beenobserved in pure TiO2 [8], CeO2 [9], In2O3 [10,11], nonstoichiometricCaB6 [12] and ZnO powders [13], nanoparticles [14,15], nanowires[5,16], thin films [3,4], etc. There are suggestions that point defectsserve as possible origins for the d0 ferromagnetism [17]. Experimen-tally, Banerjee et al. [13] and Xing et al. [16] attribute the ferromag-netism obtained in pure ZnO powders and nanostructures to oxygenvacancy (VO). Singhal et al. [8] attribute the ferromagnetism in TiO2

to three factors: Ti 3d-O 2p hybridization, VþO (the electrons in singlyoccupied oxygen vacancies) and oxygen vacancy assisted fragmenta-tion of grains. Theoretically, Kim et al. [18] predict the sizablemagnetic moment in rutile TiO2 stems from the lattice distortiondue to oxygen vacancies. The ab initio calculations performed by

ll rights reserved.

x: þ86 22 27406852.

Rahman et al. [19] on SnO2 and Pemmaraju and Sanvito [20] on HfO2

indicate that cation vacancies can induce local moments in sur-rounded O atoms, while calculations to the Hubbard U of defectscarried out by Chakrabarty and Patterson [21] attribute the ariseof ferromagnetism in ZnO films to V�ZnO and V�Zn. Although theferromagnetism observed in undoped oxides mentioned above isconsistently attributed to the existence of native defects (oxygenvacancy, cation vacancy and other complex structures of defects),further and systematic researches are still needed to define when aspecific kind of defect plays a role and how the coupling happens.

As a transparent and wide gap oxide, In2O3 has a varietyof practical applications in transparent conducting films, solarcells, flat-panel displays, etc. Meanwhile, as a base-material formagnetic semiconductors, it is also expected to be made as amulti-functional material with properties of magnetism, opticsand conduction together. However, till now, only a few works[10,11,22,23] have been proceeded on the d0 ferromagnetism ofpure In2O3. The corresponding d0 ferromagnetism origin and itsunderlying physics in this system are poorly understood. In thispaper, we report an enhancement on the d0 ferromagnetism ofpure In2O3 films by vacuum-annealing. The connection betweenthe highly oxygen-deficiency states and the enhanced ferromag-netism indicates that oxygen vacancies play a crucial role inmediating the d0 ferromagnetism in the pure In2O3 films.

2. Experimental details

In2O3 films were deposited on (100) single crystal siliconsubstrates by radiofrequency magnetron sputtering using a cera-mic In2O3 target (99.99%, F60 mm). The chamber was firstly

S. Sun et al. / Journal of Magnetism and Magnetic Materials 324 (2012) 2932–2935 2933

evacuated to a base pressure of 8.7�10�5 Pa. During deposition,the sputtering pressure and substrate temperature were main-tained at 1 Pa and 700 1C. High purity O2 and Ar with a constantflux of 5 sccm and 25 sccm were introduced, serving as reactiveand sputtering gases, respectively. After deposition for 1 h, we gotthe as-grown (AG) In2O3 film with a thickness of 110 nm. There-after, by annealing the as-grown film in vacuum at 650 1C for 6 h,we got a vacuum-annealed (VA) film, and by further annealing theVA film in air, we got an air-annealed (AA) film. The relativelylower annealing temperature is to avoid any significant changesin the crystal structure.

The surface topography and grown quality were investigatedby a scanning electron microscope (SEM). The crystal structureand phase identification were studied by X-ray diffraction(XRD) using a diffractometer D/max-2500 with Cu Ka radiation.The valence and effective concentration of elements were char-acterized with an X-ray photoelectron spectroscopy (XPS). Andthe magnetization measurements were conducted using a highsensitivity Quantum Design MPMS magnetometer.

Fig. 1. (a) SEM pattern of the as-grown In2O3 film. (b) XRD spectra for the as-

grown and vacuum-annealed In2O3 films. The inset shows the magnified part of

(222) peaks.

Fig. 2. XPS survey spectrum of the as-grown In2O3 film. Similar spectrum was also

obtained for the vacuum-annealed In2O3 film.

3. Results and discussion

Fig. 1(a) shows the SEM pattern of the AG In2O3 film, from whichwe can see that the as-grown film has a neat and clean surfacewithout any impurity precipitates. Cubic-grown grains are well-distributed in the film with the grain size of about 25 nm. XRDpatterns for the films before and after vacuum-annealing are shownin Fig. 1(b). From the patterns, we can see that all diffraction peaks areindexed well to a cubic bixbyite In2O3 structure (space group Ia3(206)). No peaks of impurities or other indium-related secondaryphases are detected within the detection limit. Both the two filmshave a strong diffraction peak along (222), indicating that films beforeand after vacuum-annealing both have a strong texture structurealong (222) direction. As a result of the vacuum-annealing, theposition of the (222) peak moves from 30.5801 to 30.6611 (inset inFig. 1). The corresponding lattice constant decreases from 10.092 A to10.089 A. This is mainly caused by the oxygen atoms escaping fromthe lattice or the interstitial positions during vacuum annealing.Meanwhile, the full width at half maximum (FWHM) of the (222)peak decreases from 0.3641 to 0.3021. The corresponding grain sizecalculated by the Scherrer formulation increases from 22.4 nm to27.0 nm. That’s to say phase transformation does not occur during thevacuum-annealing process, but the crystal quality is improved tosome extent, with the lattice more compact and the grain size a littlebigger.

Fig. 2 shows the XPS spectrum of the AG In2O3 film depicting afull range scan from 0 to 1200 eV. Similar spectrum is also obtainedfor the VA In2O3 film which is not shown here. Peaks such as O 1s,In 3s, In 3d, In 3p, In 4s, In 4p, In 4d and C are detected. No othermagnetic impurity peaks are detected in the detection limit,indicating that both our AG and the VA samples are clean withoutany contamination. Fig. 3 shows the In 3d spectra for the AG andVA In2O3 films. The peaks at 444.2 eV and 451.8 eV represent thecore levels of In 3d5/2 and In 3d3/2, respectively. The energydifference between the two peaks is coincident with the standardreference value of In2O3, indicating that the In element is in a þ3valence state. For the VA film, the peaks have a tiny shift towardlower energy and become sharper after vacuum-annealing, whichmanifest the evidence that the surroundings of In have beenchanged due to the creation of oxygen vacancies [3]. However,from the tiny change in the intensity of In 3d spectrum, it is hard todetermine the presence of In vacancies.

Fig. 4 shows the spectra of O 1s of the AG and VA In2O3 films.The peaks are asymmetric in shape, indicating the presence ofsurface contamination [8,14,15], and they can be fitted with two

Fig. 3. In 3d XPS spectra of the as-grown and vacuum-annealed In2O3 films.

Fig. 4. O 1s spectra of the as-grown (square) and the vacuum-annealed (circle)

In2O3 films. The detailed shapes were fitted to two components using Gaussian

fits. The peak at lower energy is attributed to the In–O bond, and the other one is

assigned to contamination on the surface in In2O3.

Fig. 5. Magnetic hysteresis loops of the as-grown (square), vacuum-annealed

(circle) and air-annealed (diamond) In2O3 films measured at 300 K. The inset

shows the expanded low field region of the M–H loops.

S. Sun et al. / Journal of Magnetism and Magnetic Materials 324 (2012) 2932–29352934

Gaussians to separate out the bulk oxygen content. The main peakat 529.7 eV is concerned with the In2O3 crystal oxygen attributedto the strong In–O bond, while the peak at a higher energy(531.5 eV) is associated with oxygen interstitials in the In2O3

matrix, as well as the chemisorbed oxygen of the surface hydroxyl,–CO3, absorbed H2O or absorbed O2 [15]. After vacuum-annealing,the peak intensities of both two components decrease rapidly,indicating that oxygen is escaping from the films due to vacuum-annealing. Therefore, oxygen vacancies will appear. In order toestimate the concentration of the oxygen vacancies, the atomicratio of O/In is determined from the areas under the peak for therespective element. For the AG film, the atomic ratio of O/In is1.78. By eliminating the effect of the surface contamination, anatomic ratio of 1.22 for the crystal O and In is obtained. This ratio issmaller than the stoichiometric ratio of 1.5 for In2O3, indicatingthat the AG film is nonstoichiometric, with some content of oxygenvacancies existing. After vacuum-annealing, a further decreaseof the bulk oxygen content by about 13.7% was detected. That’sto say, oxygen vacancies exist in both the AG and VA films, andafter vacuum-annealing, more oxygen vacancies appear, leading toa highly oxygen-deficiency state.

Fig. 5 shows the hysteresis loops of the AG, VA and AA In2O3

films, which were measured at room temperature with theapplied field parallel to the film plane. The diamagnetic signalshave been subtracted linearly by the high-field magnetization.From the figure, we can see that all the three films show roomtemperature ferromagnetism. The saturation magnetization forthe AG film is 0.5 emu/cm3. After vacuum-annealing, the satura-tion magnetization increases rapidly to 5.5 emu/cm3. Similarenhancement has been reported by Panguluri et al. [10] andSudakar et al. [11], with much smaller values of 0.3 emu/cm3 and0.5 emu/cm3, respectively. After further air-annealing, the mag-netic moment decreases again to the original value. Furthermore,the inset shows the enlarged low-field loops, from which wecan see that for the VA film, the remnant magnetization andcoercivity are 0.5 emu/cm3 and 95 Oe, respectively. Both therelatively low coercivity and the ferromagnetism dependence onthe annealing conditions indicate that the observed ferromagnet-ism stems from defects (oxygen vacancies), which is consistentwith the highly oxygen-deficiency state detected by XPS.

For the VA film, the in-plane and out-of-plane hysteresis loopsat different temperatures (5–300 K) were measured in detail.The in-plane loops at 5 K and 300 K, and the out-of-plane loopat 300 K were selected as representatives, as shown in Fig. 6. Fromthe figure, we can see that the practically anhysteretic loops showno temperature-dependence in the range from 5 to 300 K, and noobvious magnetocrystalline anisotropy exists. These characteris-tic features are quite similar to Coey’s related results [24], fromwhich we can further discuss how the ferromagnetism distributesin the film. The loop can be fitted to the empirical expression[24–26]

M¼Ms tan h ðH=H0Þ

where Ms is the saturation magnetization, and H0 is an effectivesaturation field. Quantitative results were obtained from thefitted line (inset in Fig. 6). The saturation magnetization Ms isapproximately 5.25 emu/cm3. The field H0 is 1971 Oe, which isbound up with the local spontaneous magnetization M0 of theferromagnetic region by the effective demagnetizing factorH0¼NM0. As there is little difference in H0 when the field isapplied parallel and perpendicular to the plane of the film, so theratio H0/M0 is unlikely to exceed 0.33. Assuming N¼0.33, themagnetization M0 is obtained to be 473.1 emu/cm3 and that onlyabout Ms/M0¼1.11% volume fraction of the In2O3 film is

Fig. 6. Magnetic hysteresis loops of the vacuum-annealed In2O3 films, including

the in-plane loops at 5 K (diamond) and 300 K (circle), and the out-of-plane loop

at 300 K (square). Inset shows the fitted curve of the out-of-plane loop.

S. Sun et al. / Journal of Magnetism and Magnetic Materials 324 (2012) 2932–2935 2935

magnetically ordered. This value is consistent with a model wherethe ferromagnetism is associated with grain boundaries or inter-faces [24–26]. That’s to say, the d0 ferromagnetism in our VAIn2O3 film is not a bulk effect, and it distributes unevenly in thegrain boundaries or interfaces.

As demonstrated above, a systematic study on the structuraland magnetic properties of the pure In2O3 films indicate a directconnection between the oxygen vacancies and the d0 ferromag-netism. As we know, during vacuum-annealing process, oxygenvacancies (VþO and V2þ

O ) can be more easily formed at grainboundaries or interfaces due to the defective alignment of atoms[27]. Theoretical calculations [28] show that oxygen vacancy caninduce a defect-related hybridization at the Femi level andgenerate a long-range ferromagnetic ordering. Since V2þ

O can bepartially transformed to VþO by vacuum-annealing, and the resultof ab initio calculations [15] show a net moment of 0.98 mB for perVþO and 0 mB for V2þ

O , the d0 ferromagnetism observed in oursamples should have some connections with VþO . For the AG films,in which the oxygen vacancy concentration is relatively low, thetiny magnetization of 0.5 emu/cm3 may be attributed to thescattering VþO . After vacuum-annealing, the oxygen vacancy con-centration increases rapidly, long-range ferromagnetic couplingmay be achieved through the ferromagnetic aligned VþO . More-over, high concentration of VþO can form anionic vacancy clusters[13], and the ferromagnetic coupling may also be achievedthrough the exchange interaction between the vacancy clusters,or between the VþO and the vacancy cluster. That’s to say, thestronger magnetization of 5.5 emu/cm3 may stem from VþO ,oxygen vacancy clusters or the coupling between them. By furtherair-annealing, oxygen vacancies generated during vacuum-annealing were recovered, the magnetic moment decreased againto the original value. However, the effect of other defects (likeindium vacancy) cannot be completely ruled out because ourfindings show no evidence of their absence. Therefore, the originof the d0 ferromagnetism in the oxygen-deficiency oxides stillneeds further exploring both experiments to give an evidence andin theories to establish a thorough interpretation.

4. Conclusion

In summary, the effect of vacuum-annealing on the structural,electronic and magnetic properties of In2O3 films has been

studied. Room temperature d0 ferromagnetism and oxygen-deficiency states were both observed in the as-grown andvacuum-annealed In2O3 films. A stronger ferromagnetism with asaturation magnetization of 5.5 emu/cm3 was obtained with moreoxygen vacancies created upon vacuum-annealing. The connec-tion between the induced oxygen vacancies and the enhancedferromagnetism implies that the oxygen vacancy is the key factorregulating the ferromagnetic order in In2O3 films. We think thatthis d0 ferromagnetism may originate from VþO and oxygenvacancy clusters in the interfaces or grain boundaries.

Acknowledgments

This work is supported by the National Natural Science Founda-tion of China Nos 11004149 and 11004148, the Tianjin NaturalScience Foundation (11JCZDJC22100) and the Seed Foundation ofTianjin University.

References

[1] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, D. Ferrand, Science 287 (2000) 1019.[2] H. Ohno, Science 281 (1998) 951.[3] Qingyu Xu, Zheng Wen, Liguo Xu, Jinlong Gao, Di Wu, Kai Shen, Teng Qiu,

Shaolong Tang, Mingxiang Xu, Physica B 406 (2011) 19–23.[4] Qingyu Xu, Heidemarie Schmidt, Shengqiang Zhou, Kay Potzger,

Manfred Helm, Holger Hochmuth, Michael Lorenz, Annette Setzer,Pablo Esquinazi, Christoph Meinecke, Marius Grundmann, Applied PhysicsLetters 92 (2008) 082508.

[5] S. Ghosh, Gobinda Gopal Khan, Bipul Das, K. Mandal, Journal of AppliedPhysics 109 (2011) 123927.

[6] K. Rode, R. Mattana, A. Anane, V. Cros, E. Jacquet, J.-P. Contour, F. Petroff,A. Fert, M.-A. Arrio, Ph. Sainctavit, P. Bencok, F. Wilhelm, N.B. Brookes,A. Rogalev, Applied Physics Letters 92 (2008) 012509.

[7] M. Venkatesan, C.B. Fitzgerald, J.M.D. Coey, Nature 430 (2004) 630.[8] R.K. Singhal, Sudhish Kumar, P. Kumari, Y.T. Xing, E. Saitovitch, Applied

Physics Letters 98 (2011) 092510.[9] R.K. Singhal, P. Kumari, A. Samariya, Sudhish Kumar, S.C. Sharma, Y.T. Xing,

Elisa B. Saitovitch, Applied Physics Letters 97 (2010) 172503.[10] Raghava P. Panguluri, P. Kharel, C. Sudakar, R. Naik, R. Suryanarayanan,

V.M. Naik, A.G. Petukhov, B. Nadgorny, G. Lawes, Physical Review B 79 (2009)165208.

[11] C. Sudakar, A. Dixit, Sanjiv Kumar, M.B. Sahana, G. Lawes, R. Naik, V.M. Naik,Scripta Materialia 62 (2010) 63–66.

[12] J.M.D. Coey, Solid State Sciences 7 (2005) 660–667.[13] S. Banerjee, M. Mandal, N. Gayathri, M. Sardar, Applied Physics Letters 91

(2007) 182501.[14] R. Escudero, R. Escamilla, Solid State Communications 151 (2011) 97–101.[15] Weijing Liu, Wenwu Li, Zhigao Hu, Zheng Tang, Xiaodong Tang, Journal of

Applied Physics 110 (2011) 013901.[16] Guozhong Xing, Dandan Wang, Jiabao Yi, Lili Yang, Ming Gao, Mi He,

Jinghai Yang, Jun Ding, Tze Chien Sum, Tom Wu, Applied Physics Letters 96(2010) 112511.

[17] Nguyen Hoa Hong, Joe Sakai, Nathalie Poirot, Virginie Brize, Physical ReviewB 73 (2006) 132404.

[18] Dongyoo Kim, Jisang Hong, Young Ran Park, Kwang Joo Kim, Journal ofPhysics: Condensed Matter 21 (2009) 195405. (4pp).

[19] Gul Rahman, Vıctor M. Garcıa-Suarez, Soon Cheol Hong, Physical Review B 78(2008) 184404.

[20] Chaitanya Das Pemmaraju, S. Sanvito, Physical Review Letters 94 (2005)217205.

[21] Aurab Chakrabarty, Charles H. Patterson, Physical Review B 84 (2011)054441.

[22] L.X. Guan, J.G. Tao, C.H.A. Huan, J.L. Kuo, L. Wang, Applied Physics Letters 95(2009) 012509.

[23] Z.R. Xiao, X.F. Fan, L.X. Guan, C.H.A. Huan, J.L. Kuo, L. Wang, Journal ofPhysics: Condensed Matter 21 (2009) 272202. 6pp.

[24] J.M.D. Coey, Jerome T. Mlack, M. Venkatesan, P. Stamenov, IEEE Transactionson Magnetics 46 (2010) 6.

[25] Karl Ackland, Lorena M.A. Monzon, M. Venkatesan, J.M.D. Coey, IEEETransactions on Magnetics 47 (2011) 10.

[26] K. Ackland, M. Venkatesan, J.M.D. Coey, Journal of Applied Physics 111 (2012)07A322.

[27] X.J. Liu, X.Y. Zhu, J.T. Luo, F. Zeng, F. Pan, Journal of Alloys and Compounds482 (2009) 224–228.

[28] J.M.D. Coey, M. Venkatesan, C.B. Fitzgerald, Nature Materials 4 (2005) 173.