APPLIED PHYSICS LETTERS 93, 262505 共2008兲

Effects of donor doping on Ga1−xMnxAs Y. J. Cho,1,a兲 K. M. Yu,2 X. Liu,1 W. Walukiewicz,2 and J. K. Furdyna1 1

Department of Physics, University of Notre Dame, Notre Dame, Indiana 46556, USA Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA

2

共Received 27 October 2008; accepted 10 December 2008; published online 31 December 2008兲 We investigate the effect compensating Mn acceptors in Ga1−xMnxAs films by doping with Si donors. For Ga1−xMnxAs with low Mn content 共e.g., x ⬍ 0.033兲, doping by Si decreases the Curie temperature TC compared to undoped Ga1−xMnxAs. At high values of x, however 共e.g., x ⬎ 0.10兲, Si doping is found to increase TC. We ascribe this to an increase in the hole mobility in high x samples due to changes in the relative occupancy of the hole impurity band associated with compensation by the Si donors. © 2008 American Institute of Physics. 关DOI: 10.1063/1.3063046兴 Since the successful growth of ferromagnetic 共III,Mn兲V semiconductors by low temperature molecular beam epitaxy 共LT-MBE兲,1 numerous studies on these materials were performed, with an eye on using spin properties of electrons for increased device functionality. It is now widely accepted that the Curie temperature TC of Ga1−xMnxAs is proportional to concentrations of substitutional Mn 共MnGa兲 and of the holes. However, the LT-MBE mode used for growing Ga1−xMnxAs inherently introduces Mn interstitial 共MnI兲 共Ref. 2兲 and As antisite 共AsGa兲 共Ref. 3兲 defects, whose main role is to act as double donors, making the sample less conductive. Thus they compensate the holes generated by the MnGa. Additionally, MnI tends to form antiferromagnetic 共AF兲 pairs with MnGa.5 Both these effects tend to weaken the ferromagnetism of Ga1−xMnxAs and to lower its TC.4,5 Currently postgrowth annealing is a widely used “remedy” for removing MnI.6 This increases the hole concentration p and 共by breaking the AF MnI – MnGa pairs兲 the population of active MnGa spins, thus increasing TC. While much effort has been devoted to improving TC of Ga1−xMnxAs either by annealing or by codoping with shallow acceptors 共e.g., Be兲,7 less attention was given to doping Ga1−xMnxAs by donors.8 In this paper we explore the effect of doping by Si on the properties of Ga1−xMnxAs with a wide range of Mn concentration 共x = 0.02– 0.16兲. All samples in this study were grown by LT-MBE 共see Ref. 2兲 at a substrate temperature Ts between 180 and 270 ° C, using the highest Ts that gave a good streaky 共1 ⫻ 2兲 reflection high-energy electron diffraction pattern for each Mn concentration. A GaAs buffer layer was deposited at this temperature, followed by a Si-doped Ga1−xMnxAs layer. An As-rich flux ratio 共As2 / Ga⬃ 15兲 was used in all LT-MBE growths. The Mn concentration was determined from Mn flux during the growth, and was confirmed by particleinduced x-ray emission 共PIXE兲. Si concentration in Si-doped samples was determined by secondary ion mass spectroscopy to be ⬃1 ⫻ 1020 / cm3. The value of TC for all samples was obtained from the temperature dependence of the magnetization measured in a field of 100 Oe using a superconducting quantum interference device. Temperature dependence of the resistivity was also measured to establish the electrical properties of the samples. Lattice location of Mn in the Ga1−xMnxAs layers a兲

Electronic mail: [email protected].

0003-6951/2008/93共26兲/262505/3/$23.00

was determined by Rutherford backscattering spectrometry and PIXE in channeling orientation.2 The hole concentrations were determined by electrochemical capacitancevoltage 共ECV兲 measurements, which avoid any complications inherent in Hall measurements that arise from the anomalous Hall effect.9 First, we discuss the results obtained in Ga1−xMnxAs with low Mn concentration. Temperature dependences of magnetization and resistivity of Ga0.097Mn0.033As and Ga0.097Mn0.033As: Si samples are shown in Figs. 1共a兲 and 1共b兲. Parts of both samples were annealed at 270 ° C for 1 h. In both as-grown and annealed cases the Ga0.967Mn0.033As: Si film shows a lower TC and a higher resistivity than the undoped Ga0.967Mn0.033As. Since Ga1−xMnxAs is strongly p-type, the Fermi level is close to the top of the valence band, and the formation energy of donor defects is low. The Si atoms are therefore expected to form donors by occupying Ga sites,10 thus compensating the holes and reducing the TC of the film. The ECV measurements give the values of p in undoped and Si-doped films as 3 ⫻ 1020 and 2 ⫻ 1020 / cm3, respectively, clearly showing that Si donors 共SiGa兲 compensate the Mn-generated holes in these samples.11 We now turn to Ga1−xMnxAs with higher values of x. Figure 2共a兲 shows the temperature dependence of the magnetization of the as-grown undoped and Si-doped Ga0.90Mn0.10As films. At this concentration, the TC of the two samples is almost identical. Figures 2共b兲 and 2共c兲 show the dependence of TC on Mn concentration x of the as-grown Ga1−xMnxAs and Ga1−xMnxAs: Si, respectively. For relatively low Mn concentrations, i.e., x ⱕ 0.10, the TC of Ga1−xMnxAs without Si is clearly higher than that of Ga1−xMnxAs: Si. However, for x ⱖ 0.10 the TC of undoped

FIG. 1. Temperature dependence of 共a兲 magnetization and 共b兲 resistivity of the as-grown and annealed Ga0.097Mn0.033As and Ga0.097Mn0.033As: Si. 93, 262505-1

© 2008 American Institute of Physics

Downloaded 06 Jan 2009 to 117.32.153.167. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

262505-2

Cho et al.

FIG. 2. 共a兲 Temperature dependence of magnetization of the as-grown undoped and Si-doped Ga0.90Mn0.10As. 关共b兲 and 共c兲兴 Curie temperature vs x of the as-grown undoped and Si-doped Ga1−xMnxAs.

Ga1−xMnxAs begins to decrease with x, while for Ga1−xMnxAs: Si TC continues to increase monotonically with x at least up to x ⬇ 0.16. Figure 3共b兲 shows the temperature dependence of the resistivity of the as-grown Ga0.84Mn0.16As and Ga0.84Mn0.16As: Si films. The undoped and Si-doped films 共both ⬃40 nm thick兲 were grown at 190 and 180 ° C, respectively. The undoped Ga0.84Mn0.16As shows a strongly insulating behavior, making it difficult to determine TC from the resistivity data. In contrast to the data for Ga1−xMnxAs with low Mn content shown in Fig. 1, Ga0.84Mn0.16As: Si now shows a lower resistivity 关Fig. 3共b兲兴 and a higher TC 关Fig. 3共a兲兴 than its undoped counterpart. To investigate the effects of annealing at high Mn concentrations in Si-doped and undoped Ga1−xMnxAs, both samples were annealed at 190 ° C for 2 h. Figure 3共a兲 shows that the TC of Ga0.84Mn0.16As: Si is again higher than that in undoped Ga0.84Mn0.16As. A comparison of the hysteresis loops of the as-grown

FIG. 3. 共a兲 Temperature dependence of magnetization of undoped and Si-doped Ga0.84Mn0.16As before and after annealing. 共b兲 Temperature dependence of resistivity. 共c兲 Hysteresis loop at 5 K of the as-grown Ga0.84Mn0.16As and Ga0.84Mn0.16As: Si. Magnetic field is applied along the in-plane ⬍100⬎ direction.

Appl. Phys. Lett. 93, 262505 共2008兲

Ga0.84Mn0.16As and Ga0.84Mn0.16As: Si at 5 K measured with magnetic field applied along the in-plane ⬍100⬎ direction is shown in Fig. 3共c兲. A much more abrupt magnetization reversal is observed for the Si-doped sample. Since our PIXE measurements show that the effective Mn concentrations xeff 共i.e., the net value xGa − xI, representing the number of active Mn spins that interact with the holes兲 in Ga0.84Mn0.16As and Ga0.84Mn0.16As: Si are 0.094 and 0.075, respectively, the weaker ferromagnetism of the undoped Ga0.84Mn0.16As cannot be explained by MnI. It has been reported that the density of AsGa increases rapidly under As-rich growth conditions with decreasing Ts,12,13 and that doping with Si serves to remove these defects in GaAs grown by LT-MBE.12 It is therefore informative to consider the effect of Si on the formation of AsGa in Ga1−xMnxAs. Based on Ts used for growing Ga1−xMnxAs with low x 共x ⬃ 0.03兲, it is estimated that in that case the AsGa concentration is typically below 1019 / cm3.13 This amounts to less than 1.5% of the total Mn in the sample, and is therefore not expected to affect its properties. However, Ga1−xMnxAs samples with x ⬇ 0.16 are grown at Ts ⬇ 190 ° C, which gives an expected AsGa concentration of ⬃1.5⫻ 1020 / cm3. Since AsGa are double donors, this reduces p by about 3 ⫻ 1020 / cm3. If we assume that the Si dopants remove a fraction of AsGa from Ga0.84Mn0.16As, such elimination of AsGa by Si doping 共⬃1 ⫻ 1020 / cm3兲 should increase p by ⬃2 ⫻ 1020 / cm3. Contrary to these expectations, however, ECV measurements show that p in as-grown undoped Ga0.84Mn0.16As 共p = 6 ⫻ 1020 / cm3兲 is larger than in Si-doped Ga0.84Mn0.16As: Si 共p = 4 ⫻ 1020 / cm3兲. Thus even at this high Mn concentration Si doping continues to reduce p in Ga0.84Mn0.16As, rather than increase it by preventing AsGa defects from forming during the growth. Despite the relative success of the Zener model in describing the ferromagnetism of 共III,Mn兲V semiconductors, there is an increasing number of theoretical14 and experimental15 papers that suggest the presence of a detached Mn-derived impurity band 共IB兲 at all values of x in Ga1−xMnxAs. Recent studies based on the concept of band anticrossing between the extended states of the GaAs valence band and the strongly localized Mn states further support this picture.16,17 The valence band anticrossing interpretation is substantiated by the optically observed shift of the GaAs band gap upon addition of Mn.16 It also explains the remarkably low hole mobility in Ga1−xMnxAs,16 and it correctly accounts for the metal-to-insulator transition 共MIT兲 recently observed in Ga1−xMnxAs1−y共N , P兲y alloys.17 The properties of Si-doped Ga1−xMnxAs reported in this paper can also be understood in terms of the Mn-derived IB model. It was recognized that at low Mn concentrations 共x ⬍ 0.05兲 the MIT occurs within an IB when the IB width exceeds lifetime broadening. The scattering of carriers within the band is then altered so as to enable extended transport.15 Thus the observation of the strikingly insulating behavior shown in Fig. 3共b兲 for undoped Ga1−xMnxAs with a high Mn concentration 共x ⬎ 0.1兲 is believed to arise from the fact that the IB in this case is relatively empty. By providing additional electron to the IB, Si doping changes its relative occupancy so as to make extended transport more likely. Note that the hole mobility obtained from resistivity and ECV measurements for the undoped Ga0.84Mn0.16As is 0.5 cm2 / V s, and for Ga0.84Mn0.16As: Si is 2.3 cm2 / V s,

Downloaded 06 Jan 2009 to 117.32.153.167. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

262505-3

Appl. Phys. Lett. 93, 262505 共2008兲

Cho et al.

showing directly the more extended nature of hole transport in the Si-doped material. The much lower mobility in undoped Ga0.84Mn0.16As also accounts for its lower TC since the number of Mn spins coupled by itinerant holes is reduced as the mean free path of the holes decreases. As seen from the results shown in Fig. 3共a兲, the arguments just presented also apply after annealing. Annealing increases the number of active MnGa acceptors by removing MnI,6 and is thus expected to broaden IB. While the increases in TC both in the undoped and Si-doped Ga0.84Mn0.16As are clearly related to this increase in the concentration of active MnGa, we also infer from these results that the larger value of TC observed for the Si-doped Ga0.84Mn0.16As is—again—a consequence of the lower occupancy of IB 共and thus a greater delocalization of the IB holes兲 produced by compensation, as discussed above. In summary, we have investigated the effect of doping of Ga1−xMnxAs films by Si donors, which compensate Mn acceptors in this alloy. For Ga1−xMnxAs with low Mn content 共e.g., x ⬍ 0.033兲 Si doping decreases TC compared to undoped Ga1−xMnxAs simply by decreasing p via compensation. At higher Mn concentrations, however 共e.g., for x ⬎ 0.10兲, Si doping has the desirable effect of increasing TC. As an example, for x = 0.16 the undoped Ga1−xMnxAs has a lower TC than Si-doped Ga1−xMnxAs with the same value of x, even though the Si-doped material has fewer holes. This unexpected result, observed in both as-grown and annealed Ga0.84Mn0.16As, arises from the reduction 共by Si compensation兲 in the relative occupancy of the IB, thus allowing the IB holes to move more freely. This latter effect is of key importance since it is the ability of the holes to move 共their delocalization兲 that determines the degree of ferromagnetic coupling between Mn spins in Ga1−xMnxAs. The work at Notre Dame was supported by the National Science Foundation Grant No. DMR06-03752. Work at LBNL was supported by the Director, Office of Science,

Office of Basic Energy Sciences, Division of Materials Sciences and Engineering, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. 1

H. Munekata, H. Ohno, S. Von Molnar, A. Segmüller, L. L. Chang, and L. Esaki, Phys. Rev. Lett. 63, 1849 共1989兲. 2 K. M. Yu, W. Walukiewicz, T. Wojtowicz, I. Kuryliszyn, X. Liu, Y. Sasaki, and J. K. Furdyna, Phys. Rev. B 65, 201303共R兲 共2002兲. 3 S. Sanvito and N. A. Hill, Appl. Phys. Lett. 78, 3493 共2001兲; X. Liu, A. Prasad, J. Nishio, E. R. Weber, Z. Liliental-Weber, and W. Walukiewicz, ibid. 67, 279 共1995兲. 4 R. C. Myers, B. L. Sheu, A. W. Jackson, A. C. Gossard, P. Schiffer, N. Samarth, and D. D. Awschalom, Phys. Rev. B 74, 155203 共2006兲. 5 J. Blinowski and P. Kacman, Phys. Rev. B 67, 121204共R兲 共2003兲. 6 S. J. Potashnik, K. C. Ku, S. H. Chun, J. J. Berry, N. Samarth, and P. Schiffer, Appl. Phys. Lett. 79, 1495 共2001兲; K. W. Edmonds, P. Boguslawski, K. Y. Wang, R. P. Campion, S. N. Novikov, N. R. S. Farley, B. L. Gallagher, C. T. Foxon, M. Sawicki, T. Dietl, M. B. Nardelli, and J. Bernholc, Phys. Rev. Lett. 92, 037201 共2004兲. 7 K. M. Yu, W. Walukiewicz, T. Wojtowicz, W. L. Lim, X. Liu, U. Bindley, M. Dobrowolska, and J. K. Furdyna, Phys. Rev. B 68, 041308共R兲 共2003兲. 8 M. A. Scarpulla, K. M. Yu, W. Walukiewicz, and O. D. Dubon, AIP Conf. Proc. 772, 1367 共2005兲. 9 K. M. Yu, W. Walukiewicz, T. Wojtowicz, W. L. Lim, X. Liu, Y. Sasaki, M. Dobrowolska, and J. K. Furdyna, Appl. Phys. Lett. 81, 844 共2002兲. 10 W. Walukiewicz, Physica B 302, 123 共2001兲. 11 The ECV measurements were done at room temperature. However, because the hole concentration is very large, it is reasonable to assume that the hole concentration remains constant in the whole temperature range used in our experiments. 12 M. Missous and S. O’Hagan, J. Appl. Phys. 75, 3396 共1994兲. 13 A. Wolos, M. Kaminska, M. Palczewska, A. Twardowski, X. Liu, T. Wojtowicz, and J. K. Furdyna, J. Appl. Phys. 96, 530 共2004兲. 14 Y. Zhang and S. Das Sarma, Phys. Rev. B 72, 125303 共2005兲; S. Sanvito, P. Ordejón, and N. A. Hill, ibid. 63, 165206 共2001兲. 15 K. S. Burch, D. B. Shrekenhamer, E. J. Singley, J. Stephens, B. L. Sheu, R. K. Kawakami, P. Schiffer, N. Samarth, D. D. Awschalom, and D. N. Basov, Phys. Rev. Lett. 97, 087208 共2006兲; L. P. Rokhinson and Y. Lyanda-Geller, Phys. Rev. B 76, 161201 共2007兲. 16 K. Alberi, K. M. Yu, P. R. Stone, O. D. Dubon, W. Walukiewicz, X. Liu, and J. K. Furdyna, Phys. Rev. B 78, 075201 共2008兲. 17 P. R. Stone, K. Alberi, S. K. Z. Tardif, J. W. Beeman, K. M. Yu, W. Walukiewicz, and O. D. Dubon, Phys. Rev. Lett. 101, 087203 共2008兲.

Downloaded 06 Jan 2009 to 117.32.153.167. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp