Ferromagnetism in Cobalt-Implanted ZnO

D.P. Norton⁽¹⁾, J.D. Budai⁽²⁾, L.A. Boatner⁽²⁾, J.S. Lee⁽³⁾, Z.G. Khim⁽³⁾, Y.D. Park⁽³⁾, M.E.Overberg⁽¹⁾, S.J. Pearton⁽¹⁾ and R.G. Wilson⁽⁴⁾

⁽¹⁾ Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA

⁽²⁾ Oak Ridge National Laboratories, Oak Ridge, TN 37831, USA

⁽³⁾ School of Physics, Seoul National University, Seoul 151-747, Korea

⁽⁴⁾ Consultant, Stevenson Ranch, CA 91381, USA

ABSTRACT

The magnetic and structural properties of cobalt-implanted ZnO single crystals are reported. High quality, (110)-oriented single crystal Sn-doped ZnO substrates were implanted with Co so as to yield transition metal concentrations of 3-5 at.% in the near-surface (~2000Å) region. The implantation was performed at ~350°C to promote dynamic annealing of ion-induced damage and prevent amorphization. After implantation, the samples were subject to a 5 min rapid thermal annealing at 700°C. Four-circle x-ray diffraction indicates the presence of (110)-oriented hexagonal phase Co in the ZnO matrix. From the 2-q full-width-half-maximum (FWHM) of the Co (110) peak, the nanocrystal size is estimated to be ~ 3.5 nm, which is below the superparamagnetic limit at room temperature. In-plane x-ray diffraction shows that the nanocrystals are epitaxial with respect to the ZnO host matrix. Magnetization measurements indicate ferromagnetic behavior at low temperatures. Hysteresis is observed in the M vs. H behavior at 5 K. Coercive fields were £ 100 Oe at this measurement temperature. Temperature-dependent magnetization measurements showed evidence for ordering temperatures of > 300 K, although hysteresis in the M vs. H behavior was not observed at room temperature. The magnetic properties are consistent with the presence of Co nanocrystals, but do not preclude the possibility that a component of the magnetism is due to Co substitution on the Zn site in the ZnO matrix.

ZnO is an attractive semiconductor for use in UV light emitters, gas sensors, and transparent electronics. It is available in bulk crystal form, is readily cleavable, and has a large exciton bonding energy (60 meV) relative to its III-N counterparts (25 meV).^(1-9) The commonly observed residual conductivity in ZnO is n-type, which originates from either Zn interstitials⁽³⁾ or hydrogen⁽⁴⁾. Heavily doped n-type ZnO can be realized via excess Zn or with Al, Ga or In doping. P-type doping has proven difficult to achieve.^(5-7) In recent years, there has emerged significant interest in achieving magnetic functionality in semiconducting materials, including ZnO. Transition metal doped ZnO has been investigated as a promising dilute magnetic semiconductor for implementing spintronic device concepts. Dietl et al.⁽¹⁰⁾ first predicted a Curie temperature of ³ 300K for Mn-doped p-type ZnO, while electron-doping of Fe, Co or Ni-doped ZnO was predicted to stabilize high Curie temperature ferromagnetism.^(11,12) Carrier-induced ferromagnetism has been addressed theoretically by others for the case of hole doping of ZnO(Mn)^(13,14), while methods for improving p-type doping have also been suggested.⁽¹⁵⁾ Numerous reports of the magnetic properties of transition metal-doped ZnO have appeared recently.^(1,16-21) In particular, Ueda et al.⁽²⁰⁾ reported Curie temperatures above 300 K for Co-doped ZnO A key issue in this and other transition metal doped semiconductors is to delineate the origin of magnetism. For Co-doped ZnO, there is some evidence that the observed ferromagnetism may be due to Co precipitates instead of carrier-mediated exchange in the ZnO matrix.⁽²²⁾ If the ferromagnetism reflects formation of a dilute magnetic semiconductor, then this provides an excellent material system for investigating various semiconductor-based spintronic device concepts. If the magnetic properties originate from Co nano-precipitates, there may be opportunities to exploit these for magnetic functionality, particularly if the orientations of the hard and easy magnetic axes can be tailored.^(23-28)

In this letter, we report on the magnetic and structural properties of bulk ZnO crystals implanted with Co. The ZnO single-crystals were doped with Sn during vapor transport growth, yielding an electron concentration of ~10¹⁸ cm^-3. These crystals were subsequently implanted with 250 keV Co⁺ ions at doses of 3 or 5x10¹⁶ cm^-2 into the (110) growth face. The samples were held at ~350°C during the implantation step to avoid amorphization of the ZnO lattice. The projected range of the implanted ions is ~2000 Å, producing an average transition metal concentration of ~ 3 or 5 at.%. The samples were subsequently annealed at 700°C for 5 min. under flowing N₂ gas to repair implant damage. The magnetic properties were examined using a Quantum Design SQUID magnetometer. Screening for the formation of secondary phases was performed using four-circle x-ray diffraction (XRD).

X-ray diffraction measurements of the implanted samples yielded evidence for hcp Co nanocrystal precipitates that are epitaxial with respect to the ZnO crystal lattice. Figure 1 shows the x-ray diffraction 2θ-scan parallel to the surface normal. The ZnO (110) peak at 2θ = 56.6˚ is evident in the scan. A broad diffraction peak at 2θ = 75.5˚, corresponding to a d-spacing of 1.259 Å, is consistent with either hexagonal Co(110) or cubic Co(220). Note that cobalt can exhibit either the hexagonal ABABAB… stacking or the fcc cubic structure (stable above 450˚C) with ABCABC… stacking. For this case, standard θ-2θ x-ray scans do not provide sufficient information to distinguish between the hexagonal and cubic structure. Note also that the broad shoulder at 2θ ≈ 20˚ likely reflects substrate damage due to the implant. A FWHM in 2q of 2.9˚ for the peak at 75.5˚ corresponds to an x-ray coherence distance (estimate of the nanocrystal size) of 36 Å. This is smaller than the superparamagnetic critical diameter for Co at room temperature, which is relevant to the magnetization measurements to be discussed later.

In order to further investigate the origin of the 2θ = 75.5˚ peak, four-circle x-ray diffraction was performed. Assuming that the Co peak is hexagonal Co(110), an in-plane scan at χ = 49.8˚ would correspond to the hexagonal Co (011). A φ-scan through the hexagonal Co (011), shown in Fig. 2a), reveals two pairs of peaks, one set at φ = 44.6˚ and –135˚, corresponding to a “hex on hex” alignment of the Co with ZnO, and a second set at φ = -48˚ and 131.5˚, which is close to a 90˚ rotation of the former. In addition, an off-axis diffraction scan parallel to the L axis, but shifted away from the origin, was also performed as shown in Fig. 2b). Based on the crystal symmetry of the two Co phases, this scan should show peaks at integral L positions for hexagonal stacking and at integer/3 positions for cubic stacking as was demonstrated elsewhere.⁽²⁹⁾ Figure 2b) shows an L-scan along the Co(01L). The hex Co(01-1) and hex Co(011) peaks are visible, although weak and broad. The hexagonal (001) peak appears too weak to resolve. Most important, there is no evidence for the integer/3 peaks corresponding to the cubic phase.

The x-ray diffraction results are significant for two reasons. First, it demonstrates the utility of x-ray diffraction in detecting and identifying precipitate phases in transition metal doped semiconductors. Often, it is assumed that x-ray diffraction will be ineffective at searching for secondary phases with doping levels of only a few percent. However, if the secondary phases that form are epitaxial, as in the present case, the x-ray diffraction intensity becomes sufficient not only for phase detection in a 2θ scan, but also for in-plane characterization of phase orientation. Second, in the present case, the Co nanocrystals are epitaxial, thus yielding a unique orientation of both the hard and easy magnetic axes. For the development of nanomagnetic arrays for high density memory, controlling the orientation of the magnetic axes is critical to maximizing the density of addressable bits. Assuming that the magnetic anisotropy is the same as in bulk material, the easy axis of the Co nanocrystals is in-plane for the present case. More generally, this approach provides a means by which the magnetic anisotropy of the magnetic nanocrystals embedded in a semiconductor matrix can be tailored via selection of the host crystal lattice orientation.

In addition to the x-ray diffraction study, the magnetization behavior of the implanted samples was also investigated. Figure 3 shows magnetization versus field behavior at 5K (top) and 300K (bottom) from samples implanted with the 5 at.% dose of Co. Hysteresis is clearly observed at 5 K, with a coercivity < 300 Oe. At room temperature, no hysteresis is observed in the magnetization versus field data within the resolution of the measurements. Figure 4 shows the temperature dependence of the magnetization from a 3 at.% Co-implanted sample. Unimplanted control samples showed paramagnetic behavior, indicating that the introduction of the Co is the cause of the observed ferromagnetism. Based on the observed magnetization behavior, it is reasonable to assign the magnetic behavior of the implanted samples to the Co nanocrystals detected in x-ray diffraction. Macroscopic Co precipitates would be ferromagnetic with a bulk Curie temperature of 1382 K. However, at ~3.6 nm in diameter, the Co nanocrystals should exhibit superparamagnetic behavior. For hexagonal Co spheres at room temperature, the critical diameter is approximately 5 nm.^(30,31) Based on the x-ray diffraction data, the average size of the Co nanocrystals is below the superparamagnetic limit at room temperature, which is consistent with the magnetic properties observed. Low temperature ferromagnetism is observed, while the samples do not exhibit magnetic hysteretic behavior at room temperature.

While the presence of Co nanocrystals can account for magnetism in these implanted samples, it does not preclude the possibility that magnetism originating from Co substitution on the Zn sites also contributes. Total energy calculations suggest that Co doping of n-type ZnO would produce ferromagnetism.⁽¹¹⁾ Note that the magnetization does not show the expected Brillouin-like dependence on temperature, but is consistent with the disorder model of Bhatt and co-workers.^(21-25) Their model takes into account the effects of positional disorder of the magnetic impurities, in which the carriers are allowed to hop only between the transition metal dopants. The interaction between the carriers and the magnetic ions is an antiferromagnetic Heisenberg exchange type. The shape of the M-T plot is a function of the wide distribution of exchange couplings because some transition metal atoms do not order until lower temperatures. There are numerous models in the literature, none of which reproduce all of the experimental data, including the carrier-induced Zener ferromagnetism⁽¹⁰⁾ and bound magnetic polarons.⁽³⁵⁾ The Dietl et al.⁽¹⁰⁾ near-field model considers the ferromagnetism to be mediated by delocalized or weakly localized holes in the p-type materials.⁽¹⁰⁾ The magnetic transition metal ion provides a localized spin. The spin-spin coupling is assumed to be a long-range interaction, allowing use of a mean-field approximation. The Curie temperature for a given material, transition metal concentration, and hole density is then determined by a competition between the ferromagnetic and antiferromagnetic interactions. The model takes into account the anisotropy of the carrier-mediated exchange interaction related with the spin-orbit coupling in the host material. The T_C is predicted to be proportional to the density of transition metal ions and carrier density.

In the absence of carriers, the magnetization M_O(H) is dependent on the Brillouin function B_S according to⁽¹⁰⁾

where g is the degeneracy factor, m_B is the Bohr magneton, S is the localized spin state, N_O is the concentration of cation sites, N_OX_eff is the effective spin concentration, k_B is Boltzmann’s constant and the antiferromagnetic temperature T_AF describes the sum of the exchange interactions to the Curie-Weiss temperature. In the presence of carriers, the magnetization is represented as⁽¹⁰⁾

where F_C[M]is the hole contribution to the free-energy functional F, which is dependent on the magnetization of the localized spin. From this relation, T_C can be expressed⁽¹⁰⁾

where b is the p-d exchange integral, A_F is the Fermi liquid parameter and P_S is the total density of states. As mentioned earlier, neither this model nor any of the other existing models can be considered definitive at this point. One clear difference of our data with the mean field model is that it assumes that holes are the mediators of the interactions between Mn ions, whereas our samples are Co-doped and n-type. While our results are consistent with the theoretical prediction of Sato and Katayama-Yoshida⁽¹¹⁾ from total energy calculations that indicated Co doping would induce ferromagnetism in n-type ZnO, the presence of Co nanocrystals prevents a unique determination as to whether this mechanism is relevant.

In summary, the magnetic and structural properties were obtained from bulk n-type ZnO crystals directly implanted with Co at concentrations of 3-5 at.%. The origin of the ferromagnetism appears related to the formation of Co nanocrystals, although contributions from carrier-induced magnetism or the percolation of magnetic polarons can not be excluded. The M-T curves are non-classical and are consistent with recent disorder models for dilute magnetic semiconductors.

Acknowledgments

The work at UF is partially supported by NSF (DMR-0101856 and DMR-0101438) and by ARO, while the work at SNU is partially supported by KOSEF and Samsung Electronics Endowment through CSCMR.

References

1. A.Tiwari, C. Jin, A. Kuit, D. Kumar, J.F. Muth and J. Narayan, Solid-State Comm. 121, 371 (2002).

2. Z.K. Tang, P. Yu, G.K.L. Wong, M. Kawasaki, A. Ohtomo, H.H. Koinuma and Y. Segawa, Solid-State Comm. 103, 459 (1997).

3. D.C. Look, J.W. Hemsky and J.R. Sizelove, Phys. Rev. Lett. 82, 2552 (1999).

4. C.G.Van de Walle,Phys.Rev.Lett.85 1012(2000).

5. K. Minegishi, Y. Koiwai, Y. Kikuchi, K. Yano, M. Kasuga and A. Shimizu, Jap. J. Appl. Phys. 36, L1453 (1997).

6. T. Yamamoto and H. Katayama-Yoshida, Jap. J. Appl. Phys. 38, L166 (1999).

7. M. Joseph, H. Tabata and T. Kawai, Jap. J. Appl. Phys. 38, L1205 (1999).

8. T. Aoki, Y. Hatanaka and D.C. Look, Appl. Phys. Lett. 76, 3257 (2000).

9. Y.R. Ryu, S. Zhu, D.C. Look, J.M. Wrobel, H.M. Jeong and H.W. White, J. Crystal Growth 216, 330 (2000).

10. T. Dietl, H. Ohno, F. Matsukura, J. Cibert and D. Ferrand, Science 287, 1019 (2000).

11. K. Sato and H. Katayama-Yoshida, Jap. J. Appl. Phys. 40, L334 (2001).

12. K. Sato and H. Katayama-Yoshida, Jap. J. Appl. Phys. 39, L555 (2000).

13. K. Sato and H. Katayama-Yoshida, Physica E10, 251 (2001).

14. K. Sato and H. Katayama-Yoshida, Mat. Res. Soc. Symp. Proc. 666, F4.6.1 (2001).

15. T. Yamamoto and H. Katayama-Yoshida, Jap. J. Appl. Phys. 38, L166 (1999).

16. T. Fukumura, Z. Jin, M. Kawasaki, T. Shono, T. Hasegawa, S. Koshikara, S. Koshihara and H. Koinuma, Appl. Phys. Lett. 78, 958 (2001).

17. T. Wakano, N. Fujimura, Y. Morinaga, N. Abe, A. Ashida and T. Ito, Physica E10, 260 (2001).

18. Z. Jin, K. Hasegawa, T. Fukumura, Y.Z. Yoo, T. Hasegawa, H. Koinuma and M. Kawasaki, Physica E10, 256 (2001).

19. S.W. Jung, S.-J. An, G.-C. Yi, C.U. Jung, S.-I. Lee and S. Cho, Appl. Phys. Lett. 80, 4561 (2002).

20. K. Ueda, H. Tabota and T. Kamai, Appl. Phys. Lettt. 79, 988 (2001).

21. H.-J. Lee, S.-Y. Jeong, C. R. Cho, and C. H. Park, Appl. Phys. Lett 81, 4020 (2002).

22. J. H. Kim, H. Kim, D. Kim, Y. E. Ihm, and W. K. Choo, J. Appl. Phys. 92, 6066 (2002).

23. M. Klimenkov, J. von Borany, W. Matz, D. Eckert, M. Wolf, and K.-H. Müller, Appl. Phys. A 74, 571 (2002).

24. M. Sato, S. Takada, S. Kohiki, T. Babasaki, H. Deguchi, M. Oku, and M. Mitome, Appl. Phys. Lett. 77, 1194 (2000).

25. S. I. Woods, J. R. Kirtley, S. Sun, and R. H. Koch, Phys. Rev. Lett. 87, 137205-1 (2001).

26. C. Petit, J. Legrand, V. Russier, and M. P. Pileni, J. Appl. Phys. 91, 1502 (2002).

27. O. Fruchart, P.-O. Jubert, C. Meyer, M. Klaua, J. Barthel, and J. Kirschner, J. Magnetism and Mag. Materials 239, 224 (2002).

28. M. Spasove, U. Wiedwald, R. Ramchal, M. Farle, M. Hilgendorff, and M. Giersig, J. Magnetism and Mag. Materials 240, 40 (2002).

29. J. D. Budai, C. W. White, S. P. Withrow, M. F. Chisholm, J. Zhu and R. A. Zuhr, Nature 390, 384 (1997).

30. C. P. Bean, J. Appl. Phys. 26, 381 (1955).

31. D. J. Clerk, Magnetism: Principles and Applications (Wiley, Chichester 1995).

32. M. Berciu and R.N. Bhatt, Physica B 312/313, 815 (2002).

33. M. Berciu and R.D. Bhatt, Phys. Rev. Lett. 87, 107203 (2001).

34. D.E. Angelescu and R.H. Bhatt, Phys. Rev. Lett. 65, 075211 (2002).

35. A.C. Durst, R.N. Bhatt and P.A. Wolff, Phys. Rev. B 65, 235205 (2002).

36. R.N. Bhatt, M. Berciu, M.D. Kennett and X. Wan, J. Superconductivity: Incorporating Novel Magnetism 15, 71 (2002).

Figure Captions

Figure 1. X-ray diffraction θ-2θ scan for cobalt-implanted ZnO crystal showing the Co (110) peak.

Figure 2. X-ray diffraction (a) φ-scan and (b) L-scan indicating that the Co nanocrystals are in-plane aligned and possess the hexagonal structure

Figure 3. Magnetization loops at 5K (top) and 300K (bottom) for field applied parallel to the plane of a ZnO sample implanted with 5 at.% Co.

Figure 4. Temperature dependence of magnetization at a field of 1000 G for a ZnO sample implanted with 3 at.% Co.