Structural and room-temperature ferromagnetic properties of Fe-doped CuO nanocrystals

JOURNAL OF APPLIED PHYSICS 107, 113908 共2010兲

Structural and room-temperature ferromagnetic properties of Fe-doped CuO nanocrystals Youxia Li, Mei Xu, Liqing Pan,a兲 Yaping Zhang, Zhengang Guo, and Chong Bi Department of Physics, University of Science and Technology Beijing, Beijing 100083, China

共Received 6 January 2010; accepted 26 April 2010; published online 3 June 2010兲 Fe-doped CuO 共Cu1−xFexO兲 nanocrystals 共NCs兲 共x = 0, 0.02, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3兲 are prepared by using the urea nitrate combustion method. X-ray diffraction 共XRD兲 analysis confirmed the monoclinic structure of CuO. Single-phase structure is obtained for the 0%–20% Fe-doped CuO, whereas for the 25% and 30% Fe-doped CuO material, secondary phase, ␣-Fe2O3, is presented. Rietveld refinements of XRD data revealed that with an increase in Fe doping level, there is a monotonic increase in cation vacancies in the Fe-doped samples. X-ray photoelectron spectroscopy measurements on the Cu0.98Fe0.02O sample revealed that the Cu2+ sites are partly substituted by Fe3+ ions. The microstructure is investigated by high-resolution transmission electron microscopy. The magnetic hysteresis loops and the temperature dependence of magnetization of the samples indicated that the samples are mictomagnetic of ferromagnetic domains originated from ferromagnetic coupling between the doping Fe ions in Cu1−xFexO NCs randomly distributed in the antiferromagnetic CuO matrix. The Curie temperature of the ferromagnetic phase is higher than 400 K for all Fe-doped CuO samples. The ferromagnetic behavior of the samples is discussed. © 2010 American Institute of Physics. 关doi:10.1063/1.3436573兴 I. INTRODUCTION

ferromagnetism in CuO by Fe doping is reported and the possible origins of ferromagnetism are discussed.

In the past decades, oxide-based diluted magnetic semiconductors 共DMS兲 have attracted immense interest due to their potential applications in spintronics,1 such as spin transistors and nonvolatile storage devices, in which we can control spin injection by weak magnetic field.2 As an important property of DMS, ferromagnetism at room-temperature 共RT兲 is under extensive study so that the origins of the magnetism can be drawn. Recently, considerable interests have focused on the magnetic properties of different elements 共Mn, Zn, Ga, Fe, Ni, and Co兲 doped cupric oxide 共CuO兲 materials3–9 due to the obvious implications with the physics of high critical-temperature superconducting materials whose basic units are Cu–O chains or layers, high-Curie-temperature induced-multiferroics and ferromagnetism of CuO,10 but there is no common conclusion about the magnetism for the CuO samples doped with different ions and synthesized by different methods. Compared with their bulk counterparts, CuO nanocrystals 共NCs兲 exhibit unique properties, such as superparamagnetism and increasing susceptibility at low temperatures due to uncompensated surface spins.11,12 Hence, it is meaningful to investigate the magnetic characteristics of nanoscale CuO materials. Dietl et al.13 predicted that transition-metal-ion doped wide band-gap semiconductors could exhibit RT ferromagnetism. However, RT ferromagnetism is also observed in CuO, a narrow band-gap semiconductor, doped with magnetic elements. In this work, Fe-doped CuO NCs are prepared by combustion synthesis method. A realization of RT a兲

Author to whom correspondence should be addressed: Electronic mail: [email protected].

0021-8979/2010/107共11兲/113908/6/$30.00

II. EXPERIMENTAL DETAILS

Fe-doped CuO NCs are prepared by using combustion synthesis method.14 Briefly, the combustion synthesis technique comprises several steps: first, to heat a mixture aqueous solution of suitable metal salts and organic fuel to boil, then to wait until the mixture ignites and a self-propagating combustion reaction proceeds, with large amount of gases giving off, and finally resulting in a dry, usually crystalline, fine oxide powder.15 In our work, the initial metal salts are Cu共NO3兲2 共AR兲 and Fe共NO3兲2 共analytical reagent兲, while the fuel is urea 共AR兲, which are all dissolved in deionized water, respectively. The reaction is as follows: 共1 − x兲Cu共NO3兲2 + xFe共NO3兲2 + CH4N2O + O2 → Cu1−xFexO + H2O + CO2 + NO2 , where x = 0, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3, hence the dependence of physical properties on Fe doping level can be investigated. As for the molar ratio of urea:NO−3 , for all x value we keep it at 5:8, which is found to be the most proper value to get monoclinic CuO structure after a series of trial on the urea-to-NO−3 ratio. Structural and magnetic properties of aforementioned samples are investigated by several measurements. The crystalline structures of the samples are analyzed by means of an x-ray powder diffractometer 共D/max-250兲 employing Cu K␣ radiation at RT with a scanning step of 0.02°. The x-ray is operated at 45 kV and 200 mA. The Rietveld refinement studies of the powder x-ray diffraction 共XRD兲 profiles are performed by using the GSAS/EXPUI refinement program.16 X-ray photoelectron spectra 共XPS兲 of the samples are re-

107, 113908-1

© 2010 American Institute of Physics

113908-2

J. Appl. Phys. 107, 113908 共2010兲

Li et al.

FIG. 1. 共Color online兲 XRD patterns of the Cu1−xFexO 共x = 0.02, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3兲 NC samples. The vertical lines at the bottom are the standard JCPDS data of monoclinic CuO 共card No.89–5898兲. The arrows ↓ denotes the peaks of the miscellaneous phase ␣-Fe2O3.

corded by PHI Quantera SXM analyzer under a vacuum of 3.4⫻ 10−9 Torr, using Mg K␣ radiation. The binding energy values are charge-corrected to the C 1s signal 共284.6 eV兲. The microstructures and morphologies of the samples are studied by a JEOL-2010 high-resolution transmission electron microscope 共HRTEM兲. The magnetic hysteresis curves at 300 and 60 K in the field range of −2.9 T ⱕ H ⱕ 2.9 T and the magnetization curves as a function of temperature from 55 to 400 K under a magnetic field of 200 Oe are measured on a vibrating sample magnetometer 共Versalab, Quantum Design Co.兲 and the magnetic measurements are also performed on a physical properties measurement system 共PPMS, Quantum Design Co.兲 in the temperature range of 5–300 K. III. RESULTS AND DISCUSSIONS A. Structural analyses by XRD

XRD patterns of Cu1−xFexO samples are shown in Fig. 1. Compared with the standard diffraction data 共JCPDS No. 89–5898兲, the Cu1−xFexO samples of low doping level 共x = 0, 0.05, 0.1, 0.15, and 0.2兲 are of pure monoclinic CuO structure. When the Fe concentration increases, there is a tendency for Fe ions to form oxide with O2− directly instead of substituting Cu ions in the crystal lattice. As a result, weak diffraction peaks 共marked with “↓” in Fig. 1兲 corresponding to rhombohedral ␣-Fe2O3 共JCPDS No. 89–2810兲 appear for high doping level of x = 0.25 and 0.30.

FIG. 2. 共Color online兲 Representative XRD refinement patterns for Cu1−xFexO 共x = 0.05, 0.1, and 0.2兲 samples. The circle signs 共䊊兲 are the experimental XRD data and the solid lines almost at the same position of each x value are the calculated profile, below which is the difference between experimental and calculated data. The vertical lines at the bottom indicate the diffraction angles of allowed Bragg reflections.

The average grain sizes of Cu1−xFexO, x = 0, 0.02, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3 are found to be 19 nm, 22 nm, 17 nm, 16 nm, 15 nm, 16 nm, 16 nm, and 16 nm, respectively, estimated from the full width at half maximum of XRD peaks by Scherrer formula.17 In other words, nanocrystalline Cu1−xFexO particles can be prepared by combustion synthesis method, using Cu共NO3兲2, Fe共NO3兲3, and urea as reactants. Based on the XRD results, Rietveld refinement is used to analyze the structure properties of the samples as a further method. For the refinements, experimental data in the diffraction angle 共2␪兲 range of 10°–130° are used. The space group of monoclinic CuO is C2 / c with Cu2+ ions occupying the symmetry site 4共c兲 that contains no adjustable parameters: 共1/4, 1/4, 0兲, 共3/4, 1/4, 1/2兲, 共3/4, 3/4, 0兲, and 共1/4, 3/4, 1/2兲, meanwhile O2− ions are at 4共e兲, 共0 , y , 1 / 4兲, 共1 / 2 , 1 / 2 + y , 1 / 4兲, 共0 , − y , 3 / 4兲, and 共1 / 2 , 1 / 2 − y , 3 / 4兲, which may be adjusted by changing “y” to give the correct intensity distribution in a diffraction data. In our work, we choose an initial value for y = 0.4184 共Ref. 18兲 and hypothesized that Fe ions substitute Cu ions in CuO lattice. Figure 2 shows the typical refinement plots of Cu1−xFexO 共x = 0.05, 0.1, and 0.2兲 samples and the refinement results of all the samples with x ⱕ 0.2 are listed in Table I. Obviously, the fitted patterns are in agreement with the respective experimental data, noticed by the Rwp and Rp factors 共shown in Table I兲, which present values of 2.8%–4.1% and 1.8%– 3.2%, respectively, indicating a fine refinement.

TABLE I. XRD refinement results of Cu1−xFexO samples. Refine results

Lattice parameters

R factors

Cu1−xFexO samples

Cu 共%兲

Fe 共%兲

Vacancies 共%兲

a 共Å兲

b 共Å兲

c 共Å兲

␤ 共deg兲

y

RWP 共%兲

RP 共%兲

x = 0.00 x = 0.02 x = 0.05 x = 0.10 x = 0.15 x = 0.20

89.3 97.0 93.6 84.8 63.6 58.4

0.00 1.98 4.96 9.51 11.2 14.2

10.7 1.03 1.49 5.67 25.2 27.4

4.683 4.689 4.685 4.686 4.686 4.689

3.423 3.423 3.424 3.425 3.423 3.422

5.129 5.132 5.130 5.132 5.132 5.132

99.40 99.43 99.42 99.42 99.43 99.46

0.431 0.423 0.428 0.433 0.446 0.448

4.1 2.9 2.8 2.8 3.0 3.1

3.2 1.8 2.1 2.1 2.3 2.4

113908-3

Li et al.

FIG. 3. 共Color online兲 The results obtained from XRD refinement of Cu1−xFexO 共x = 0, 0.02, 0.05, 0.1, 0.15, 0.2兲 samples. 共a兲 The actual Fe concentration and 共b兲 the vacancy concentration.

Lattice parameters of pure monoclinic CuO are a = 4.68 Å, b = 3.42 Å, c = 5.13 Å, and ␤ = 99.40°,18 and the lattice parameters of our samples given by XRD refinement 共the fifth to eighth columns in Table I兲 are almost the same as that of pure CuO. This indicates that doping of Fe ions has little influence on the crystalline structure and the combined action of Fe substitution and cation vacancy is responsible for this result. As shown in Table I, the actual Fe ion concentrations in our samples 共the third column兲 are slightly lower than the nominal ones 共the first column兲, which commonly occurs in synthesis process because of unavoidable loss. This result is plotted in Fig. 3 关curve 共a兲兴. The figure is approximately of linearity and the slope is nearly equal to 1, which means that combustion synthesis ensures an effective doping. The fourth column in Table I gives the vacancy percentage on cation sites. Obviously, the vacancies, in addition to Fe ions, occupy the cation sites too, not only in doped samples but also in undoped sample 共x = 0兲. We found that the undoped sample contains quite a few cation vacancies 共10.7%兲, which consistent with the calculation result by using the LSDA + U method.19 When the samples start to be doped with Fe ions, the vacancy percentage decreases sharply to 1.03% for x = 0.02 and then increases with the increasing concentration of Fe ions, as shown in curve 共b兲 of Fig. 3. For further investigation on the relationship between vacancies and Fe ions, XPS measurement is carried out on a typical sample, Cu0.8Fe0.2O. Figure 4 represents the XPS analysis of Cu 2p and Fe 2p core levels. Due to the spinorbit coupling, the 2p core level splits into 2p1/2 and 2p3/2 components. The Cu 2p spectra 关Fig. 4共a兲兴 peaks at the binding energies of 953.5 and 933.6 eV accompanying with their shakeup satellites, corresponding to Cu 2p1/2 and 2p3/2, respectively. This result agrees with the characteristic of CuO,20 indicating that +2 is the only valence for Cu ions in our samples. In other words, Cu ions are thoroughly oxidized during the reaction process. Meanwhile, Fe ions should be thoroughly oxidized at the same time because they are in the equal status with Cu during reaction. This inference is confirmed by the XPS result for Fe 2p core level. As shown in Fig. 4共b兲, the experiment data is fitted using 20% Gaussian ⫺80% Lorentzian functions. The main peaks are located at 724.1 and 711.2 eV, and a satellite at 719.3 eV, which are in agreement with the XPS data for Fe2O3 in

J. Appl. Phys. 107, 113908 共2010兲

FIG. 4. 共Color online兲 XPS spectra of 共a兲 Cu 2p and 共b兲 Fe 2p for Cu0.8Fe0.2O. The circle signs 共䊊兲 and the solid plots are the experimental data and the simulated data/peaks, respectively, while the dashed lines are the background.

literature.21 This indicates that the chemical valence of Fe ions is predominantly +3. In other words, Fe3+, instead of Fe2+, replaces Cu2+ in the crystal lattice to form Cu1−xFexO samples. Now it is understandable why the concentration of vacancies follows similar trend of Fe ions in Fig. 3. The more Fe3+ substituting for Cu2+, the more charge variation, and hence the higher vacancy percentage in order to fit the charge conservation law. Basically, monoclinic Cu1−xFexO 共x = 0, 0.02, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.3兲 samples are prepared by combustion synthesis method using Cu共NO3兲2, Fe共NO3兲2, and urea as reactants, and the nanocrystalline particle size is around 20 nm. From the structural analyses, Cu2+ in the crystal lattice is replaced by Fe3+, which leads to cation vacancies increasing because of charge variation.

B. Structural analyses by HRTEM

The TEM images of Cu0.8Fe0.2O sample are shown in Fig. 5. The nanoparticles are aggregated and the particle size mainly ranges from 10 to 20 nm, in consistent with the result calculated by Scherrer formula in Sec. III A. The randomly oriented crystal lattice fringes shown in Fig. 5共b兲 indicate that the sample is of polycrystalline. Further illustration about this polycrystalline can be seen in the electron diffraction 共ED兲 pattern 共the inset of Fig. 5兲, in which a set of concentric rings confirm the polycrystalline structure and all the rings can be indexed as CuO structure.

FIG. 5. Typical 共a兲 TEM and 共b兲 HRTEM images of Cu0.8Fe0.2O. The inset in 共b兲 is the corresponding ED pattern.

113908-4

Li et al.

J. Appl. Phys. 107, 113908 共2010兲

FIG. 6. 共Color online兲 The magnetization of ferromagnetic part of the Cu1−xFexO samples 共x = 0.00, 0.02, 0.05, 0.10, 0.15, and 0.20兲 as a function of applied magnetic field, 共a兲 at 300 K and 共b兲 at 60 K. The insets are the zoomed in hysteresis around zero applied magnetic field.

C. Magnetic properties

Figure 6 shows the M-H curves of Cu1−xFexO 共x = 0, 0.02, 0.05, 0.1, 0.15, and 0.2兲 samples at 300 and 60 K. In the inset of Fig. 6共a兲, we can clearly see the open loop around zero external field for both doped samples and undoped one. The coercivities for the samples are 50–130 Oe at RT and 500–800 Oe at 60 K, respectively. It is said that when the particle size larger than 10 nm, the magnetic property of pure CuO nanoparticle is similar with bulk CuO, which shows antiferromagnetic feature.12 The magnetization of Fe-doped samples is much larger than undoped CuO sample one, that we can ignore the contribution from pure CuO nanoparticle for the doped samples. This large change in magnetization should come from the doping of Fe. Comparing the M-H curves at 60 and 300 K in Fig. 6, it is clear that the magnetization is partially related to the measuring temperature of the samples. Both the saturated magnetization and coercivity of the samples at 60 K are larger than that at 300 K. Figure 7 shows the temperature dependence of magnetization for Cu1−xFexO 共x = 0, 0.05, 0.1, 0.15, and 0.2兲 measured from 55 to 400 K under a magnetic field of 200 Oe under field-cooling 共FC兲 of 2000 Oe. Apparently, the Fe-doped samples reveal ferromagnetic characteristic in the measured temperature range, as discussed in the following text. We have carried out detailed hysteresis loop measurements for the Cu0.85Fe0.15O sample as a function of temperature from 5 to 400 K 共shown in Fig. 8兲. The temperature dependence of remanence M r, coercivity Hc and saturated magnetization M s are shown in Fig. 9. The coercivity Hc remains nonzero 共⬃130 Oe兲 even at 300 K, revealing that the sample is still ferromagnetic at 300 K. A linear extrapo-

FIG. 7. 共Color online兲 Temperature dependence of magnetization for the Cu1−xFexO 共x = 0.00, 0.05, 0.1, 0.15, and 0.2兲 samples taken at applied magnetic field of 200 Oe.

lation of the data to higher T for M r → 0 yields Tc ⬎ 400 K, which indicates that the Curie temperature of Cu0.85Fe0.15O sample is over 400 K. The temperature dependence of magnetic susceptibility ␹ = M / H 共H = 200 Oe兲 under zero-FC 共ZFC兲/FC condition, and the difference susceptibility 共␹FC − ␹ZFC兲 are shown in Fig. 10. A couple of features in Fig. 10 should be noted. First, we know that ZFC/FC measured by a magnetometer employing SQUID 共superconducting quantum interference device兲 technology has been recognized as a reliable technique that can indirectly detect any ultrasmall magnetic nanoclusters in the matrix. If the ultrasmall magnetic nanoclusters existed then they would have normally displayed the superparamagnetic behavior with a low blocking temperature TB.22 But in this figure TB is not observed in the whole temperature range of 5–300 K, suggesting that there should be no tiny ferromagnetic nanoclusters. Therefore, we can conclude that the RT ferromagnetic of sample comes from the doped Fe ions. Second, the ZFC data is similar to that of bulk CuO,12,23 the Neel temperature TN can be determined by locating the peak in ⳵␹ / ⳵T versus T yields TN ⬇ 240 K, in close agreement with bulk CuO of TN ⬇ 230 K. Third, no steep rise of 共␹FC − ␹ZFC兲 in the whole temperature range is observed, which is in agreement with the result of CuO nanoparticles reported in Refs. 24 and 25, also indicates that the Curie temperature of Cu0.85Fe0.15O is over 400 K. Fourth, comparing the FC and ZFC data, we can find that the ZFC data only shows the antiferromagnetic property of CuO 共the magnified curve of ZFC is shown in the inset of Fig. 10兲 but the FC data present the ferromangnetic property of the sample, hence, we speculate that the Fe-doped samples are

FIG. 8. 共Color online兲 Magnetic hysteresis loops of the Cu0.85Fe0.15O sample at T = 15, 30, 50, 55, 60, 80, 100, 150, 200, 250, 300, 350, and 400 K. The inset is the magnetization of the ferromagnetic part of the sample as a function of applied magnetic field.

113908-5

J. Appl. Phys. 107, 113908 共2010兲

Li et al.

FIG. 9. 共Color online兲 Temperature dependence of Hc, M r, and M s for the Cu0.85Fe0.15O sample.

mictomagnetic system which contains the ferromagnetic domains originated from the ferromagnetic coupling between the doping Fe ions and the antiferromagnetic CuO matrix. At the ZFC condition, the ferromagnetic domains randomly distributed in CuO matrix will be pinned by CuO matrix, and moreover, the applied magnetic field is much less than the coercivity, then the sample should only show the antiferromagnetic properties of CuO. On the contrary, when cooling under magnetic field HFC = 2000 Oe 共HFC Ⰷ Hc兲, the ferromagnetic domains play the prominent role. Last, we would like to discuss the faint inflexion around 90 K in Figs. 9 and 10, which probably mainly corresponds to the double super¯兴 exchange interaction between Fe–O–Cu–O–Fe along 关101 direction.26 Considering that our samples contain a large number of cation vacancies, the observed RT ferromagnetism might comes from the Fe–O–䊐 ferromagnetic coupling 共here “䊐” represents a vacancy兲, in which the coupling strength is stronger than that of the superexchange between Fe–O–Cu– O–Fe. As for low temperature ferromagnetism, it is supposed that the magnetization totally comes from Fe–O–Cu–O–Fe coupling, neglecting the contribution from the vacancies. In our previous work on Mn-doped CuO thin films, a chain model has been proposed, from which the saturation magnetization as a function of doping concentration can be calculated.26 To verify this proposed model, we compare the calculated results from the model and the experimental data of Cu1−xFexO 共x = 0.02, 0.05, 0.1, 0.15, and 0.2兲 at 60 K, as shown in Fig. 11. Here we take the magnetic moment of one Fe ion as 1 ␮B. The broadly similar trend of two plots implies that the chain model might also be suitable for nano-

FIG. 11. 共Color online兲 Experimental 共triangle signs “䉱”兲 and simulated 共circle signs 䊊兲 saturation magnetization M S as a function of Fe doping concentration 共neglecting the contribution from the vacancies兲 in the Cu1−xFexO samples.

crystalline Cu1−xFexO powders. That is to say, nearest Fe ions are antiferromagnetic coupled as Fe–O–Fe, meanwhile the ferromagnetic order is originated from the superexchange between the second-nearest Fe ions, for instance, Fe–O–Cu– ¯ 兴 direction. Moreover, it should be found O–Fe, along in 关101 that the experimental data does not fit the simulation very well. We believe that the mechanism of ferromagnetism for the Cu1−xFexO system at low temperature is more diverse and the double superexchange couping may be just the one reason. In summary, Fe-doped CuO NCs exhibit ferromagnetic properties at both low temperature and RT, and the magnetization decreases when the temperature rises up. The possible origin of the ferromagnetism might be the coupling between second-nearest Fe ions Fe–O–Cu–O–Fe for low temperature and Fe–O–䊐 共䊐 represents a vacancy兲 for RT. IV. CONCLUSION

DMS Cu1−xFexO 共x = 0, 0.02, 0.05, 0.1, 0.15, 0.2兲 of monoclinic phase are fabricated by combustion synthesis method. Structural and XPS analyses confirm that Fe3+ ions, along with cation vacancies, displace for Cu2+ ions in CuO lattice. RT ferromagnetic properties of the Fe-doped CuO are observed and the magnetization increases with decreasing temperature. The doped samples reveal mictomagnetic properties: ferromagnetic and antiferromagnetic. We speculated that the superexchange interaction of second-nearest Fe ions ¯ 兴 chain and Fe–O–Fe coupling may be coupling in the 关101 one of the possible origins of ferromagnetic and antiferromagnetic. The more accurate interpretation needs further theoretical study. ACKNOWLEDGMENTS

This work was supported by NSF of China 共Grant Nos. 50472092, 50672008, and 50971023兲. 1

FIG. 10. 共Color online兲 Temperature dependence of magnetic susceptibility ␹ZFC, ␹FC, and the difference susceptibility 共␹FC − ␹ZFC兲 of Cu0.85Fe0.15O sample. The inset shows the magnified curve for the ZFC data.

S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. Von Molnar, M. L. Roukes, A. Y. Chtchelkanova, and D. M. Treger, Science 294, 1488 共2001兲. 2 N. Tansu, A. Quandt, M. Kanskar, W. Mulhearn, and L. J. Mawst, Appl. Phys. Lett. 83, 18 共2003兲. 3 H. Zhu, F. Zhao, L. Q. Pan, Y. P. Zhang, C. F. Fan, Y. Zhang, and J. Q. Xiao, J. Appl. Phys. 101, 09H111 共2007兲.

113908-6 4

J. Appl. Phys. 107, 113908 共2010兲

Li et al.

Y. P. Zhang, L. Q. Pan, Y. S. Gu, F. Zhao, H. M. Qiu, J. H. Yin, H. Zhu, and J. Q. Xiao, J. Appl. Phys. 105, 086103 共2009兲. 5 R. A. Borzi, S. J. Stewart, G. Punte, R. C. Mercader, G. A. Curutchet, R. D. Zysler, and M. Tovar, J. Appl. Phys. 87, 4870 共2000兲. 6 X. G. Zheng, Y. Kodama, K. Saito, E. Tanaka, Y. Tomokiyo, H. Yamada, and C. N. Xu, Phys. Rev. B 69, 094510 共2004兲. 7 C. T. Meneses, J. G. S. Duque, L. G. Vivas, and M. Knobel, J. Non-Cryst. Solids 354, 4830 共2008兲. 8 A. Barcs, L. Bottyan, B. Molnar, D. L. Nagy, N. S. Ovanesyan, and H. Spiering, Hyperfine Interact. 55, 1187 共1990兲. 9 A. Filippetti and V. Fiorentini, Phys. Rev. B 74, 220401 共2006兲. 10 T. Kimura, Y. Sekio, H. Nakamura, T. Siegrist, and A. P. Ramirez, Nature Mater. 7, 291 共2008兲. 11 G. N. Rao, Y. D. Yao, and J. W. Chen, IEEE Trans. Magn. 41, 3409 共2005兲. 12 A. Punnoose, H. Magnone, M. S. Seehra, and J. Bonevich, Phys. Rev. B 64, 174420 共2001兲. 13 T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science 287, 1019 共2000兲. 14 P. Ravindranathan, S. Komarneni, and R. Roy, J. Mater. Sci. Lett. 12, 369 共1993兲.

15

D. A. Fumo, M. R. Morelli, and A. M. Segadães, Mater. Res. Bull. 31, 1243 共1996兲. 16 B. H. Toby, J. Appl. Crystallogr. 34, 210 共2001兲. 17 L. D. Zhang and J. M. Mu, Nanomaterials and Nanostructure, 1st ed. 共Science Publishing House, Beijing, 2001兲. 18 S. Åsbrink and L.-J. Norrby, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 26, 8 共1970兲. 19 D. X. Wu and Q. M. Zhang, Phys. Rev. B 73, 235206 共2006兲. 20 J. Ghijsen, L. H. Tjeng, J. Vanelp, H. Eskes, J. Westerink, G. A. Sawatzky, and M. T. Czyzyk, Phys. Rev. B 38, 11322 共1988兲. 21 T. Fujii, F. M. F. de Groot, G. A. Sawatzky, F. C. Voogt, T. Hibma, and K. Okada, Phys. Rev. B 59, 3195 共1999兲. 22 Y. Xin, J. Lu, P. A. Stampe, and R. J. Kennedy, Appl. Phys. Lett. 88, 112512 共2006兲. 23 M. S. Seehra, Z. Feng, and R. Gopalakrishnan, J. Phys. C 21, L1051 共1988兲. 24 X. G. Zheng, T. Mori, K. Nishiyama, W. Higemoto, and C. N. Xu, Solid State Commun. 132, 493 共2004兲. 25 A. Punnoose and M. S. Seehra, J. Appl. Phys. 91, 7766 共2002兲. 26 F. Zhao, H. M. Qiu, L. Q. Pan, H. Zhu, Y. P. Zhang, Z. G. Guo, J. H. Yin, X. D. Zhao, and J. Q. Xiao, J. Phys.: Condens. Matter 20, 425208 共2008兲.