4096 IEEE TRANSACTIONS ON MAGNETICS, VOL. 45, NO. 10, OCTOBER 2009

Magnetic Characteristics of Carbon-Doped Nanocrystalline TiO�

Qi-Ye Wen�, Huai-Wu Zhang�, Qing-Hui Yang�, De-En Gu�, Yuan-Xun Li�, Ying-Li Liu�, Jian Shen�, andJ. Q. Xiao���

State Key Laboratory of Electronic Films and Integrated Devices, University of Electronic Science and Technology of China,Chengdu 610054, China

Department of Physics and Astronomy, University of Delaware, Newark, DE 19716 USA

Carbon doped TiO� nanocrystal powder samples with various C contents were synthesized by a low-temperature sol-gel method, anda large lattice C contents of 12.7% was obtained. We demonstrated that the C ions were homogenously doped into the TiO� matrix atthe form of substitutional C (Cs-o) and interstitial C ����, without precipitation of TiC and other impurity phase. Our results show thatC-doped TiO� is ferromagnetic with the Curie temperature well above room temperature. The magnetization is mainly related to thecontent of Cs-o, suggesting that the intrinsic ferromagnetism originates from the Ti-C system in the TiO� environment.

Index Terms—Carbon doping, magnetic semiconductors, nanocrystal, sol-gel.

I. INTRODUCTION

T HE PRACTICAL application of diluted magnetic semi-conductors (DMSs) in spintronics requires the DMS to ex-

hibit ferromagnetism (FM) at or above room temperature (RT).One of the most interests in the researches of DMSs is the ferro-magnetic oxides such as transition metal (TM) doped ZnO andTiO [1], [2]. However, controversial explanations for the ob-served FM in DMS are still at the center stage. The FM mightarise from TM metal clusters, defects or ferromagnetic impuri-ties phases, and the mechanism of FM in TM-doped ZnO andTiO remains unclear. It is speculated that if non-TM dopantscan be incorporated into these metal-oxides and induce mag-netism, DMS thus produced would not suffer from problemsrelated to precipitates of dopants since they do not contributeto ferromagnetism. Based on this speculation, copper doping inZnO and GaN have been investigated and it has been confirmedexperimentally that both materials are RT DMSs. [3]–[6]

Very recently, Pan et al. investigated the doping of carbon ionsinto ZnO films by pulsed laser deposition and strong RT FMwas detected [7], which was further confirmed experimentallyin carbon ions-implanted ZnO films [8]. The first-principles cal-culations by Pan et al. predicted this intrinsic magnetism suc-cessfully and revealed that it resulted from carbon substitutionfor oxygen. These theoretical and experimental results demon-strate that carbon is a novel dopant in the class of doped oxideDMS materials. Therefore, it is speculated that RT FM can alsobe obtained in C doped TiO . Actually, a previous theoreticalcalculation by Yang suggested strong ferromagnetic couplingin C-doped anatase TiO [9], but no experimental results havebeen reported so far. In this letter, C-doped TiO nanocrystallinepowders were synthesized by a low-temperature sol-gel methodand the structure and magnetic properties were investigated.Usually, a high-temperature process is required in the fabrica-tion of C-doped anatase TiO by sol-gel method, which leads

Manuscript received March 04, 2009; revised May 08, 2009. Current ver-sion published September 18, 2009. Corresponding author: Q.-Y. Wen (e-mail:qywen@163.com).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TMAG.2009.2024317

to a relatively low content of C ions [10]–[12]. In this work, adoping content as high as 12.7 at% of carbon was obtained withthe low-temperature sol-gel method. Our experiment demon-strated a very simple and effective method to induced high con-tent carbon into metal oxides system, and it clearly shows thatthe C-induced ferromagnetism can also be achieved in the classof TiO .

II. EXPERIMENTS

The sample of C Ti O was prepared by reacting ap-propriate amounts of titanium carbon (TiC), CH CH OH, andNitric Acid as following: the TiC powder was firstly ultrasonicdispersed in anhydrous ethanol for 10 minutes, and then NitricAcid was added into the solution under constant stirring witha glass stirrer. The solution was then dried at 50 C for 48 h toevaporate off the solvent, and the powder obtained was bakedat 120 C for 12 h. It is noted that the reaction is controlled bythe volume of CH CH OH, and the C content in the TiO ma-trix is determined by the molar ratio (R) of CH CH OH toTiC. Samples with , and 50 are prepared to ob-tained different C content (designated as sample CT01, CT02,CT04, CT05). A pure TiO nanopowder was also synthesizedby sol-gel method for a comparison, in which C H O Tiwas used as the raw materials and the powder was also obtainedafter baked at 120 C for 12 h.

The crystal structure was characterized on a Philips X’pertX-ray diffractometer (XRD) operated at 40 kV with Cu-Ka ra-diation. The morphology and crystallinity were further studiedusing high resolution transmission electron microscopies(HRTEM): JEOL-2010. All samples were investigated usingX-ray Photoelectron Spectroscopy (XPS) to check the chem-ical states of Ti, C, O and the corresponding concentration.The XPS measurement was performed on VG MICROLABMKa, with Aluminium Ka radiation as the excitation source( eV). All the binding energies were calibrated bythe C1s peak at 285.0 eV of the adventitious surface carbon.The magnetic properties were measured using an alternatinggradient magnetometer (AGM, Micromag TM 2900) and acommercial superconducting quantum interference device(SQUID).

0018-9464/$26.00 © 2009 IEEE

WEN et al.: MAGNETIC CHARACTERISTICS OF CARBON-DOPED NANOCRYSTALLINE TIO 4097

Fig. 1. XPS survey spectrum for C doped TiO . Inset is the enlarged energyarea from 395 eV to 405 eV.

Fig. 2. XRD patterns for pure and carbon doped TiO powder.

III. RESULTS AND DISCUSSION

The typical XPS survey spectrum of CT04 is plotted inFig. 1. It shows that there are only Ti, O, and C elements inthe sample. Since Nitric Acid was used as reacting materials,the bonding energy area (395 eV–405 eV) for N signalis also carefully examined, as plotted in the inset of Fig. 1.Clearly, no signal for N element is detected. These resultsconfirm that no contaminations were introduced into TiOmatrix in the procedure of sample preparation. Fig. 2 showsthe XRD patterns of pure and C-doped TiO powder samples.According to the standard JCPDS card (PDF No. 84-1286), theobserved diffractive peaks correspond to (101), (004), (200),(105) crystal plane of anatase TiO , respectively. This indi-cates that all the samples are anatase TiO without detectablerutile phase. Also, no characteristic peaks of carbon oxideswere showed, which implies C is incorporated in the lattice ofanatase TiO or carbon oxide is very small and highly dispersed[12]. The broadened peaks confirm the nanocrystalline nature.With Scherrer formula the grain size of C doped samples wascalculated to be 3–4 nm while that for pure TiO powder is

Fig. 3. HRTEM images for (a) pure TiO and (b) CT04 samples. Insets are thecorresponding SADP.

about 4.63 nm. The slight difference of the grain size may beascribed to the carbon doping effect.

Interestingly, the XRD patterns for carbon doped TiO sam-ples are noisier compared to the undoped sample, and with moredoping the pattern become more noisy. One possible reason isthat, with C doping, the crystallographic quality of the samplesbecomes slightly worse than the pure TiO , as indicated by theHRTEM results that we will discuss later. For instance, the in-crease of the interstitial C ions with the doping density mightdistort the crystal lattice and cripple the crystallographic quality.Another possible reason is that the smaller grain sizes of the Cdoped sample gives rise to weaker intensity of the XRD signal.Therefore, the ratio of signal/noise decreased and the XRD spec-trum seems to be nosier.

Fig. 3 shows typical HRTEM images of pure TiO and CT04sample. The insets are the corresponding selected area diffrac-tion patterns (SADP). It shows that the powders are composedof nanocrystals with a grain size of a few nanometers, which isconsistent with the XRD results. From the HRTEM picture, thelattice fringe of anatase TiO is clearly seen and a lattice spaceof the (101) plane was measured to be 0.35 nm, which is veryclose to the typical value for the bulk TiO (0.352 nm). No Cclusters or other phases were observed, indicating that C wasdoped into the TiO matrix homogeneously and no secondaryimpurity phase formed.

Fig. 4(a) and (b) shows the C1s XPS spectra of pure andC doped TiO , respectively. The peak at 284.8 eV appears inthe spectra for all samples, which can be attributed to the “freecarbon.” As compared with the pure TiO powder, two extragroups of peaks in the C1s binding energy of 281.9 eV and288.6 eV were observed in the XPS spectrum of C doped TiOpowder. The peak of 288.6 eV, according to Valentin et al., [13]was derived from the interstitial carbon in the TiO matrix,while the C1s peak at the binding energy of 281.9 eV suggestthe presence of carbon atoms in the carbide form. In a studyon TiC by XPS technology, Zhang et al. also observed an ob-vious peak at 281.9 eV and they ascribed it to the C1s for C-Tibonding [14].

In order to clarify if there are TiC phase formed in the Cdoped TiO , the XPS spectra of Ti2p for TiC powder and Cdoped TiO powder were measured and compared, as plottedin Fig. 3(c) and (d), respectively. The typical spectral peaks forTi2p were fitted using Gaussian method and the resulting peaks

4098 IEEE TRANSACTIONS ON MAGNETICS, VOL. 45, NO. 10, OCTOBER 2009

Fig. 4. C1s XPS spectra for pure TiO (a) and CT04 (b), and Ti2p XPS spectrafor TiC (c) and CT04 (d).

TABLE ICONCENTRATIONS OF LATTICE-CARBON IN C-DOPED TIO

were marked at the spectra. For TiC powder four distinct peakswere observed: the two peaks at 458.6 eV and 464.3 eV corre-spond to the Ti2p and 2p of TiO while other two peaksat 455.3 eV and 461.3 eV resulted from Ti2p and 2p corelevels for C-Ti bonding [13]. However, in the XPS spectra ofC doped TiO only peaks at 458.7 eV and 464.4 eV appearedand the other two peaks were not observed, excluding the pre-cipitation of TiC phase. Therefore, the XPS signal for C1s at281.9 eV indicates the carbon institution for oxygen and forma-tion of Ti-C bonds in the TiO matrix.

Based on the XPS spectrum, the contents of substitutional C(Cs-o), interstitial C and the total lattice C for C dopedTiO were calculated and listed in Table I. The concentration ofC increases with the increasing of R factor and then dropdramatically when reach to 50, while for , the concentra-tion increase rapidly at low and becomes saturated whenis larger than 20. With this low temperature sol-gel method, alattice carbon concentration as high as 12.7 at% was obtainedin the sample CT04, which is fairly larger than any other resultsreported previously. By this low temperature sol-gel method, itis found that the factor has a dominating effect on the C con-centration. Actually, in the initial process of the reaction, mostof the C in TiC was oxidized by nitric-acid and released as COgas; only a small fraction of C was doped into the crystal latticeas substitutional or interstitial Carbon. One function of the CHCH OH is to slow-down and stabilize the speed of reaction be-tween TiC and nitric-acid, and another function is to dilute the

Fig. 5. Temperature dependence of magnetization for (a) undoped TiO , and(b) CT04 with applied field of 1 tesla. The insets show the M-H loops at 5 Kand 300 K.

nitric-acid thus reduce the oxidation of C. Therefore, with theincrease of (CH CH OH), the reaction between TiC and ni-tric-acid is depressed and the lattice C concentration increases.However, if the is beyond a optical value, the reaction timewill be so long that the nitric-acid will oxide the new-formedC:TiO seriously. In our experiment, it seems that is theoptical value. When the increase beyond 50, C in the formedC:TiO was again oxidized into gas by nitric-acid, leading to thedecrease of lattice C.

The dependence of magnetic properties on temperatureand field for all samples were investigated. The mag-

netic loops of undoped TiO and CT04 were plotted and com-pared in Fig. 5 as an illustration. Fig. 5(a) shows typical magne-tization vs temperature (M-T) curve of undoped TiO powder,in which the inset is M-H loops at 5 K and room temperature,and the red solid line is the fitting using Brillouin function to theexperimental data (solid circle). The curve measured at 300 Kshow zero coercive field and no remanent magnetization, whichis the typical behavior of paramagnetism. Only a very weak fer-romagnetic signal was detected at 5 K. The M-T curve agreeswith Curie Weiss’s Law very well. However, Fig. 5(b) showsthat both the M-T curves and M-H loops of C doped-TiO , atlow and room temperature, exhibits strong ferromagnetism. TheM-H curve indicates the coexistence of FM and paramagnetismat 5 K, while at 300 K the FM becomes dominant with a satura-tion magnetization (Ms) of 0.1 emu/g and a coercivity of about130 Oe. From the curve, one can see that such a ferro-magnetic phase is dominant over the whole range of temper-atures, indicating the CT04 sample are clearly room-tempera-ture ferromagnetic. Actually, all C doped TiO samples exhibitthe similar magnetic properties, confirming the C induced roomtemperature ferromagnetism in TiO .

Since the C ions are paramagnetic, the RT FM cannot comefrom the doped ions. Sol-gel is a non-vacuum process occurredat air atmosphere thus the oxygen vacancies or defects usuallygenerated by high-vacuum process are eliminated. Furthermore,when Ti was substituted by C ion, Vo is not necessary producedto keep charge neutrality, since the chemical state of both C andTi are as confirmed by XPS studies. Recently, Sundaresanet al. have reported a universal ferromagnetism in nanoparti-cles of metal oxides [15]. They believe that the origin of FM innanoparticles may be the exchange interactions between local-ized electron spin moments resulting from the oxygen vacancieslocated at the nanoparticle surface. This FM is also found in our

WEN et al.: MAGNETIC CHARACTERISTICS OF CARBON-DOPED NANOCRYSTALLINE TIO 4099

Fig. 6. R dependence of the contents of (a) interstitial C (CI), (b) substitutionalC (Cs-o), and (c) the saturation magnetization.

undoped TiO powder at low temperatue, with a very strongparamagnetism. However, the measured Ms of C doped TiO issignificantly larger than that of the pure TiO nanocrystal pow-ders. Therefore, the observed ferromagnetism should be relatedto the chemical involvement of carbon, as pointed out by Panin the investigation on C doped ZnO system. In order to fur-ther understand the observed ferromagnetism, the measured Cconcentration and RT magnetization with respect to R ratio ofour C doped samples are compared in Fig. 6. There is an ob-vious scaling effect between Cs-o and magnetization. This im-plies that in C doped TiO powder the FM is induced mainly bysubstitutional carbon rather than by interstitial C. Our experi-ments provide a proof to the theoretical assumption that the FMrelated to the doped carbon in the carbide form, i.e., the Ti-Csystem in the TiO environment.

IV. CONCLUSION

In summary, C doped TiO nanocrystal powders were fab-ricated by a low temperature sol-gel method and a very high

doping C content was obtained. It is proved by XRD, XPS andHRTEM that the C ions were homogenously doped into theTiO matrix even the doping C content is as high as 12.7 at%.Room temperature ferromagnetism was clearly observed inC-doped TiO , which is mainly related to the C institution foroxygen in TiO matrix rather than the interstitial C.

ACKNOWLEDGMENT

This work was supported in part by National Science Foun-dation of China under Grant 60801023 and Grant 60721001,in part by the National Basic Research Program of China(973) under Grant 2007CB310407, and in part by the Inter-national S&T Cooperation Program of China under Grant2006DFA53410 and Grant 2007DFR10250.

REFERENCES

[1] W. Prellier, A. Fouchet, and B. Mercey, J. Phys.: Condens. Matter.,vol. 15, pp. 1583–1601, 2003.

[2] R. Janisch, P. Gopal, and N. A. Spaldin, J. Phys.: Condens. Matter.,vol. 17, pp. 657–689, 2005.

[3] R. Q. Wu, G. W. Peng, L. Liu, Y. P. Feng, Z. G. Huang, and Q. Y. Wu,Appl. Phys. Lett., vol. 89, p. 062505, 2006.

[4] L. H. Ye, A. J. Freeman, and B. Delley, Phys. Rev. B, vol. 73, p. 033203,2006.

[5] D. B. Buchholz, R. P. H. Chang, J. H. Song, and J. B. Ketterson, Appl.Phys. Lett., vol. 87, p. 082504, 2005.

[6] J. H. Lee, I. H. Choi, S. W. Shin, S. G. Lee, J. Lee, C. N. Whang, S. C.Lee, and K. R. Lee, Appl. Phys. Lett., vol. 90, p. 032504, 2007.

[7] H. Pan, J. B. Yi, L. Shen, R. Q. Wu, J. H. Yang, J. Y. Jin, Y. P. Feng,J. Ding, L. H. Van, and J. H. Yin, Phys. Rev. Lett., vol. 99, p. 127201,2007.

[8] S. Q. Zhou et al., Appl. Phys. Lett., vol. 93, p. 232507, 2008.[9] K. S. Yang, Y. Dai, B. B. Huang, and M. H. Whangbo, Appl. Phys.

Lett., vol. 93, p. 132507, 2008.[10] S. Sakthivel and H. Kisch, Daylight Photocatalysis by Carbon-Mod-

ified Titanium DioxideAngewandte Chemie International Edition ed.vol. 42, p. 4908, 2003.

[11] Y. Li, D. Hwang, N. H. Lee, and S. J. Kimet, Chem. Phys. Lett., vol.404, p. 25, 2005.

[12] H. Irie, Y. Wananabe, and K. Hashimoto, Chem. Lett., vol. 32, p. 772,2003.

[13] C. D. Valentin, G. Pacchioni, and A. Selloni, Chem. Mater., vol. 17, p.6656, 2005.

[14] L. Zhang and R. V. Koka, Mater. Chem. Phys., vol. 57, no. 23, 1998.[15] A. Sundaresan, R. Bhargavi, N. Rangarajan, U. Siddesh, and C. N. R.

Rao, Phys. Rev. B, vol. 74, p. 161306R, 2006.