Ferromagnetic semiconductor GaMnAs

Transcription

1 Ferromagnetic semiconductor GaMnAs The newly-developing spintronics technology requires materials that allow control of both the charge and the spin degrees of freedom of the charge carriers. Ferromagnetic semiconductors (SC) are considered suitable due to simultaneous presence of magnetic order and of semiconducting properties. GaMnAs is one of the most intensively studied ferromagnetic SC. In this paper we will review recent research and accomplishments regarding two technologically important properties magnetic anisotropy and interlayer coupling -- of GaMnAs-based multilayer structures, with an eye on their potential role in practical devices. Sanghoon Lee 1*, J.-H. Chung 1, Xinyu Liu 2, Jacek K. Furdyna 2 and Brian J. Kirby 3 1 Department of Physics, Korea University, Seoul, , Korea 2 Department of Physics, University of Notre Dame, Notre Dame, IN 46556, USA 3 Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA * slee3@korea.ac.kr Today s electronic devices involve two basic properties of the electron: its charge and its spin. Historically, different classes of materials have been needed to exploit these two properties. For example, semiconductors allow control over an electrical current via the electron charge (as in transistors, diodes, etc.), while magnetic metals are used to exploit the electron spin (as in magnetic hard drives, sensors, etc.). A key objective of spin-electronics (or spintronics ) research is to develop multifunctional, practical devices that allow precise and simultaneous control of both the charge and the spin properties of charge carriers. The development of a wide range of semiconductor alloys doped with magnetic ions referred to as diluted magnetic semiconductors (DMSs) represents a major step in this direction, opening up the prospect of utilizing the charge and the spin of the electron within the same material in order to create new device functionalities 1-4. A significant advance in our understanding of the interaction between charges and spins has been made by research on II-VI-based DMSs, i.e., materials achieved by alloying a II VI semiconductor host (such as HgTe, CdTe, and ZnSe) with magnetic ions, e.g., Mn 2+, whose valence is identical to that of the group II cations 5, 6. The random distribution of magnetic ions over the cation sublattices in DMS materials leads to important novel magnetic effects, e.g., large Zeeman splittings of electronic levels, giant Faraday rotation, magnetic-field-induced metalinsulator transition, and formation of bound magnetic polarons. These enhanced spin-dependent properties occurring in II-VI DMS systems were widely studied in the literature and are already summarized in many excellent review articles Although these materials have paved the way for our understanding of the interplay between semiconductor physics and magnetism, so far they have failed to find realistic applications in actual devices, because most of their potentially useful spin properties are manifested at very low temperatures. 14 APRIL 2009 VOLUME 12 NUMBER 4 ISSN: Elsevier Ltd 2009

2 Ferromagnetic semiconductor GaMnAs REVIEW A giant stride has been achieved in the last decade by introducing III-V-based DMSs in which Mn 2+ ions replace group-iii cations. Unlike the case of II-Mn-VI DMSs, Mn 2+ ions in III-V-based DMSs not only provide magnetic moments, but also act as acceptors and are therefore a source of free holes. One of the important effects of these extra charge carriers is that they mediate interactions between magnetic moments localized on the Mn ions, thus leading to ferromagnetic order in these III-V-based DMSs, with relatively high Curie temperatures The first successful growth of III-V-based DMS involved InMnAs films grown on GaAs substrates 18, 19 by lowtemperature molecular beam epitaxy (LT-MBE). Since then many other III-V-based DMSs - GaMnAs 20, 21, GaMnSb 22, 23, and InMnSb 24 have been fabricated. The successful growth of GaMnAs is of special interest, because of the prevalence of GaAs-based devices in industry already in place, from high speed transistors to light emitting diodes (LEDs) and laser diodes (LDs). Bringing a spin component into the picture is therefore of obvious interest, since this holds out the promise of integrating the existing GaAs-based electronic components with those involving GaMnAs, and in this way introducing the spin parameter into the device context in new ways. For example, attempts have already been made to integrate GaMnAs with GaAs-based nonmagnetic semiconductor systems to form spin-injecting structures 25 and spin dependent resonance tunneling devices 26. Furthermore, the magnetic properties of III-V-based DMSs can be controlled by changing carrier concentration via various external means, such as light illumination 27, 28, application of an electrical gate voltage 29, 30, or external carrier doping 31, 32. Such demonstrations of diverse manipulation techniques provide an indication that spintronic device applications based on the III-V ferromagnetic semiconductors are indeed a strong possibility. The most practical spintronic application of GaMnAs is probably a spin memory device, in which information is stored via the direction of magnetization. Since the magnetic anisotropy is one of the major physical quantities that determine the direction of magnetization in a ferromagnet, detailed understanding of the anisotropy is a critical first step for exploring ways to manipulate the magnetic properties in the GaMnAs-based structures The next step toward a functional device based on this material is to achieve a GaMnAs multilayer structure in which the spin configuration in GaMnAs layers is switchable between parallel and anti-parallel configurations. Here the interlayer exchange coupling (IEC) between the GaMnAs layers either antiferromagnetic (AFM) or ferromagnetic (FM) is expected to play a crucial role in controlling the spin configuration within the structures. This paper reviews the current understanding of both the magnetic anisotropy and the IEC in GaMnAs-based structures. Magnetic anisotropy of GaMnAs films The magnetic anisotropy of GaMnAs has been found to depend on many parameters, including temperature, strain and carrier density This understanding has stimulated intensive studies of GaMnAs via various experimental techniques, such as superconductor quantum interference device (SQUID) magnetometry 39-41, ferromagnetic resonance (FMR) 42, 43, magneto-transport 44-46, and magneto-optical Kerr effect (MOKE) 47, 48. These studies revealed a rather complex picture of magnetic anisotropy in GaMnAs, arising from cubic and uniaxial contributions. The behavior of magnetic anisotropy of a thin ferromagnetic film can be conveniently expressed via the magnetostatic free energy F. For ferromagnetic thin films with the zincblende structure (such as that of GaMnAs) the complete expression of F can be written as42, 49, 50 F = -MH[cosθ cosθ Η + sinθ sinθ Η cos(ϕ ϕ Η )] + 2πM 2 cos 2 θ (1) M[ H 2 cos 2 θ H 2 sin 2 θ sin 2 ϕ 1 2 H 4 cos 4 θ 1 4 H 4 sin 4 θ cos 2 2ϕ] where H is the applied magnetic field and M is the magnetization. The first term in Eq. (1) describes the Zeeman energy; the second term refers to the demagnetization energy (or shape anisotropy energy); and the last term gives the energy due to the uniaxial and cubic anisotropy. Here H 2 and H 4 are the perpendicular uniaxial and cubic anisotropy fields, respectively; H 2 and H 4 are the in-plane uniaxial and cubic anisotropy fields, respectively; the angles θ and θ Η are defined with respect to [001]; and the angles ϕ and ϕ Η are defined with respect to [110] in the (001) plane, as depicted in Fig. 1a. In the absence of an external magnetic field, the magnetization direction of the GaMnAs film is totally governed by magnetic anisotropy (i.e., the second and third terms in Eq. (1)), and thus a precise determination of anisotropy fields is important for correct understanding of the magnetization. Since the direction of the magnetization follows the position of the magnetic free energy minimum in the Stoner-Wohlfarth model 51, angular dependent measurements provide an opportunity for the experimental determination of anisotropy fields in a ferromagnetic film. The in-plane (H 2 and H 4 ) and out-of-plane (H 2 and H 4 ) components of anisotropy fields can be obtained with experimental configuration shown in Figs. 1b and 1c, respectively. Figure 2 shows a typical temperature dependence of magnetic anisotropy fields obtained for a Ga 1-x Mn x As film with x = grown on (001) GaAs substrate 52. Although the cubic anisotropy fields H 4 and H 4 have large values at low temperatures, they decrease rapidly as the temperature increases. The uniaxial anisotropy fields H 2 and H 2 are temperature independent, and thus become increasingly important at high temperature. The temperature dependence of the relative strengths of the cubic and uniaxial anisotropy fields causes the magnetic easy axes to change as a function of temperature. This change can be seen more clearly from the 3D free energy density diagrams shown in Fig. 3, which are constructed using the anisotropy fields given in Fig. 2. In this specific case the energy minima appear in the (001) plane, indicating that the APRIL 2009 VOLUME 12 NUMBER 4 15

3 REVIEW Ferromagnetic semiconductor GaMnAs (a) (b) (c) Fig. 1 Coordinate system used to describe crystal directions and experimental configurations. The orientation of the applied magnetic field H is described by θ H and ϕ H. The resulting equilibrium orientation of the magnetization M is given by θ and ϕ. (b) and (c) Two experimental geometries used in angle-dependent measurements for obtaining anisotropy fields in GaMnAs layers. Fig. 2 Summary plot of anisotropy fields obtained from Ga 1-x Mn x As film with x = grown on a (001) GaAs substrate. While the uniaxial anisotropy fields are relatively insensitive to temperature, the cubic anisotropy fields decrease rapidly with increasing temperature. magnetic easy axes lie in the film plane. The two distinct magnetic easy axes near <100> at 3K gradually change their directions toward [11-0] as the temperature increases. This trend eventually makes the two easy axes merge into one along the [11-0] direction when the uniaxial anisotropy begins to dominate over the cubic anisotropy at a sufficiently high temperature. This temperature-dependent change of magnetic easy axes within the sample plane is typical of GaMnAs films under in-plane strain, characteristic of GaMnAs films grown on (001) GaAs substrates. The magnetic anisotropy of a GaMnAs film changes significantly under different strain conditions. For example, the Ga 0.97 Mn 0.03 As film grown on a Ga 0.8 In 0.2 As buffer layer (which has a larger lattice parameter than Ga 0.97 Mn 0.03 As) is under tensile strain 53. Unlike compressively strained GaMnAs film, the free energy of a GaMnAs layer under tensile strain is dominated by the out-of-plane components of the anisotropy, as shown in Fig. 4. The deepest energy minimum is seen to occur along the [001] direction, indicating that in this strain condition the film has an out-of-plane magnetic easy axis 43, 54. The transition of the magnetic easy axis from in-plane to out-of-plane can also be achieved by varying the temperature, as was observed by FMR 44. The magnetic anisotropy of GaMnAs is also expected to be strongly affected by the carrier density, due to the fact that the ferromagnetism in this material is carrier-mediated. The effect of carrier density on magnetic anisotropy has been investigated for a number of GaMnAs specimens with different carrier concentrations that were controlled either by the growth process or by post-growth thermal treatment in various gas environments (e.g., nitrogen or hydrogen) 40, 55. Direct doping during the growth (e.g., co-doping by Be) produces inconsistent results due to inadvertently introduced fluctuations of strain and/or chemical composition, all of which can affect magnetic anisotropy The effect of carrier density on the magnetic anisotropy of GaMnAs was later addressed more systematically in Ga 1-y Al y As/Ga 1-x Mn x As/Ga 1-y Al y As heterostructures (x = 0.062, y = 0.24), in which the top Ga 1-y Al y As barrier was modulation-doped with Be. The modulation doping technique allows one to control carrier density in the GaMnAs layer (i.e., to increase the hole density in the Be-doped structure relative to the un-doped structure), while keeping all other parameters constant 60. Figure 5 shows 2D free energy diagrams for the (001) plane constructed from magnetic anisotropy fields obtained for the undoped (p=(1.32 ± 0.01) cm 3 ) and the Be-doped (p=(7.72 ± 0.01) cm 3 ) Ga 1-y Al y As/Ga 1-x Mn x As/Ga 1-y Al y As structures, respectively. Both systems show significant deviations 16 APRIL 2009 VOLUME 12 NUMBER 4

4 Ferromagnetic semiconductor GaMnAs REVIEW Fig. 3 3D plot of free energy density for a Ga 1-x Mn x As film with x = grown on a (001) GaAs substrate. Energy minima occur in the (001) plane near the <100> directions. The directions of the energy minima rotate toward [11-0] with increasing temperature due to the increased importance of uniaxial anisotropy along the [11-0] direction as the cubic terms decrease. from the <100> four-fold symmetry dictated by the cubic anisotropy, implying a strong influence of uniaxal anisotropy. Note, however, that in the un-doped structures the directions of the energy minima (and thus the directions of the easy axes) rotate toward the [1-10], and in Be-doped structures to the [110] directions, indicating that the uniaxial anisotropy axis has changed from [1-10] to [110] due to Be doping. Such complex behavior of magnetic anisotropy, which affects the orientation of magnetization and complicates the domain structure 61, 62, makes the investigation of IEC difficult in GaMnAs multilayers. Fig. 4 3D plot of free energy density for a Ga 0.97 Mn 0.03 As film grown on a Ga 0.8 In 0.2 As buffer layer. The deepest energy minima appear along the [001] direction, indicating that in this case the easy axis of magnetization is normal to the plane of the film. Interlayer exchange coupling in GaMnAsbased multilayers We now consider a current passing across an interface between two magnetic layers. The nature of charge carrier scattering strongly depends on the type of spin alignment, either parallel or anti-parallel, between two adjacent ferromagnetic layers. Thus the magnitude of Fig. 5 2D plot of free energy densities for two Ga 1-y Al y As/Ga 1-x Mn x As/Ga 1-y Al y As structures (x = 0.062, y = 0.24), without and with Be-doping in the top Ga 1-y Al y As barrier. The directions of the energy minima are different in the two structures due to a change of the uniaxal anisotropy from the [1-10] direction in the un-doped sample to [110] in the Be-doped structure. APRIL 2009 VOLUME 12 NUMBER 4 17

5 REVIEW Ferromagnetic semiconductor GaMnAs electric current passing across the multilayers can be controlled by manipulating the relative orientations of the magnetization vectors in these layers, reaching a maximum for parallel and a minimum for antiparallel configurations. Significant changes of electrical resistance the so-called giant magnetoresistance (GMR) have been observed in metallic magnetic multilayers when such spin configurations were controlled by external magnetic field 63. Key ingredients for observing a GMR effect are a field-dependent relative alignment of two magnetic layers separated by non-magnetic layer, a mobility difference between majority and minority carriers in the ferromagnetic layers, and sufficiently weak spin-relaxation throughout the structure. In the case of the conventional ( current-in-plane ) geometry, the spinrelaxation length needs to be large compared to the transport meanfree path. In this paper we concentrate on the IEC, which determines antiferromagnetic (AFM) or ferromagnetic (FM) spin alignment between magnetic layers. Spontaneous (AFM) IEC has been observed in various metallic and semiconductor multilayer structures. The discovery of GMR effect in metallic multilayers led to a host of new spin based electronic devices, including magnetic random access memory (MRAM) a practical spin memory device. For the realization of GMR-like spin memory devices in the ferromagnetic semiconductor GaMnAs one needs to obtain a detailed understanding of the nature of IEC between GaMnAs layers in multilayer structures. Theoretical studies based on the k p kinetic exchange and tight binding models predicted oscillatory behavior between AFM and FM IEC for carrier-mediated DMS-based superlattices The strength and period of the oscillation were predicted to depend on both the thickness of the nonmagnetic spacers and on the carrier concentration in the layers. Experimentally, the IEC properties of various configurations of GaMnAs-based multilayer structures including trilayers and superlattices were investigated by various characterization techniques While several groups have observed FM IEC in GaMnAs structures, spontaneous AFM IEC has only very recently been realized in GaMnAs, achieved by introducing additional charge carriers into the GaAs spacers that separate the GaMnAs layers 80. This observation of AFM IEC in GaMnAs-based multilayers demonstrated the high potential of III-V-based ferromagnetic semiconductors for spin memory applications. Observation of FM IEC in GaMnAs multilayer structures The simplest and most informative structure for studying IEC is the trilayer structure, in which two magnetic layers are separated by a nonmagnetic spacer. The best known ferromagnetic semiconductor trilayer structures are Ga 1-x Mn x As/Ga 1-z Al z As/Ga 1-y Mn y As layer combinations grown on a GaAs substrate. In these structures, the two GaMnAs layers differ in either Mn concentration, hole concentration, or layer thickness, chosen to give the two GaMnAs layers different magnetic properties. Whether the two magnetic layers are coupled or independent can be inferred from magnetization and/or magneto-transport measurements. Since the IEC in the ferromagnetic semiconductor trilayer structure is expected to be sensitive to the parameters of the nonmagnetic spacer (such as potential height and the layer thickness), the investigation of IEC in trilayer structures has typically been performed with series of samples, in which one of the spacer parameters is systematically varied. The first experimental investigation of IEC between GaMnAs layers was performed on Ga 1-x Mn x As/Ga 1-z Al z As/Ga 1-y Mn y As trilayer structures by magneto-transport and magnetization measurements 74. In that study the otherwise independent characteristics of the Hall resistance and of the hysteresis of the two GaMnAs layers were gradually observed to develop into a coupled behavior as the potential barrier height of the Ga 1-z Al z As layer was lowered by changing its alloy composition, or by decreasing its thickness. The systematic dependence of magnetic properties of the assembly on the nonmagnetic spacer parameters implied the presence of IEC between the two GaMnAs layers. The study of IEC between GaMnAs layers was extended using diverse multilayer structures as well as measurement techniques 76-79, 81, 82. Although the above experimental studies have demonstrated the existence of the IEC between the GaMnAs layers in multi-layer structures, all previous observations revealed only FM IEC, regardless of the parameters of the non-magnetic spacer. The absence of AFM IEC in the GaMnAs-based multilayers was rather surprising from the theoretical viewpoint, which predicted both FM and AFM IEC, depending on the spacer properties. However, the experimental techniques used in the above studies SQUID magnetometry, FMR, and magneto-transport all have limitations in unambiguously detecting AFM IEC, since they are sensitive to the collective behavior of all the layers in a multilayer structure as a whole, and are not able to directly access the spin configuration in individual GaMnAs layers. Clearly, to fully understand IEC in a multilayer structure, one requires an experimental technique that can directly probe the spin alignment in individual magnetic layers. Direct measurement of spin alignment in magnetic layers In the early days of GMR, light scattering from spin waves provided important evidence of AFM IEC in metallic ferromagnetic trilayers 63. Observation of spin waves via light scattering, however, is limited to simple structures with high magnetization densities. In the case of GaMnAs the magnetization is usually very low: it is indeed a diluted ferromagnet, as Mn 2+ ions typically constitute less than 10% of the total chemical composition, and not all Mn ions contribute to the ferromagnetic exchange. Thus a probe that is directly sensitive to even a weak magnetization is crucial in this case. Polarized neutron reflectometry provides such a probe 83, 84. This technique is sensitive 18 APRIL 2009 VOLUME 12 NUMBER 4

6 Ferromagnetic semiconductor GaMnAs REVIEW to the depth profiles of the nuclear density and magnetization in thin films and multilayers 66 even when the magnetization is small, as in the case of GaMnAs 69, 79, 80, 85, 86, 87. Polarized neutron reflectivity has already been successfully used by Kepa et al. 79 to observe FM IEC in GaMnAs-based multilayers. Using 50-period multilayers of Ga 0.94 Mn 0.06 As/GaAs, Kepa and co-workers observed magnetic contributions to the reciprocal space Bragg peak corresponding to the multilayer periodicity 79. Since it is very difficult to obtain a field that is ideally zero at the sample during measurement, simple observation of parallel alignment does not provide definitive evidence of spontaneous FM IEC. The authors 79 observed that some samples developed single domains with net magnetization along the direction opposite to the external magnetic field, and therefore concluded that the FM IEC was intrinsic. Later, antiparallel alignments between ferromagnetic GaMnAs layers were also observed using polarized neutron reflectivity. Using Ga 0.95 Mn 0.05 As/GaAs/Ga 0.95 Mn 0.05 As trilayers, where modulation doping was applied on one side of the structure, Kirby and coworkers showed that magnetization of ferromagnetic layers can be reversed individually by using the difference in coercivity 87. Although this did not involve spontaneous coupling, the experiment of Kirby et al. 87 demonstrated that polarized neutron reflectivity can be used to directly observe AFM alignments in GaMnAs-based multilayers. Very recently, definitive evidence of true AFM IEC was finally reported by Chung et al. 80 using 10-period multilayers of Ga 0.97 Mn 0.03 As/GaAs, in which the carrier concentration was increased directly in the non-magnetic GaAs spacers by Be doping 80. The sample was deposited on a GaAs (001) substrate by molecular beam epitaxy, and the Be concentration in the spacers is estimated as 1.2 x cm -3. Here we make a more detailed description of the polarized neutron reflectivity method, which is conceptually depicted in Figure 6. Polarized neutron beams (arrows) were used to probe all of the layers in the multilayer stack, revealing a 6.95 nm Mn-doped FM layer thickness d FM, a 3.47 nm Be-doped p-type spacer layer thickness d S, and the magnetization orientations of each of the layers under a range of field and temperature conditions. The neutron spins flip when they are scattered from layers whose magnetization orientations are perpendicular to them. This scattering process is called spin flip and is typically very weak for diluted magnets. On the other hand, the neutron spins are maintained when they are scattered from layers whose magnetization vectors are parallel or antiparallel to them. These two scattering processes are called non-spin flip (NSF). While the NSF intensities also include non-magnetic scattering, magnetic components can be extracted because phase shifts occur depending on the relative orientations between the neutron spins and the magnetization vectors. Therefore, net magnetization of each layer is revealed as splittings between two NSF reflectivities with opposite neutron polarizations, (++) or (--), respectively. Figure 7 shows a summary of the field-dependent NSF polarized neutron reflectivity Fig. 6 Schematic representation of a polarized neutron reflectometry measurement used to detect the AFM IEC reported in 80. Polarized neutron beams (arrows) probe each of the layers in the multilayer stack, allowing for determination of the individual layer thicknesses and magnetizations. data at 30K. A characteristic spin-split AFM Bragg peak was clearly observed at a wave vector transfer corresponding to twice the structural periodicity of the multilayer, Q 2π/2(d FM + ds) 0.03 Å -1 (see Fig. 7(a)). This AFM Bragg peak was suppressed when the applied field was increased beyond the coercive field of the GaMnAs layers, and spin-dependent changes were observed at the FM Bragg peak position Q 2π/2(d FM + ds) 0.06 Å -1, revealing a switch to FM alignment of the GaMnAs layers (see Fig. 7(b)). The AFM Bragg peak was not recovered when the field was lowered to below the coercive field at 6K, due to the strong cubic anisotropy field in the GaMnAs layers. At 30K, in contrast, the AFM Bragg peak was repeatedly recovered even after cycling the field several times, demonstrating that the observed AFM IEC is spontaneous and robust (see Fig. 7). In contrast, only FM IEC was observed when the spacers were not doped with Be, indicating that the Be doping of the non-magnetic spacer layers was responsible for the observed AFM IEC. While this observation substantially brightened the prospect of achieving all-semiconductor-based spintronic devices, we note that so far only multilayers with spacers of approximately 12 monolayers (3.5 nm thick) have been investigated with polarized neutron reflectivity. Thus a large parameter space still remains unexplored, providing considerable hope for optimization. CONCLUDING REMARKS GaMnAs thin film structures exhibit very intricate and interesting spin properties associated with their magnetic anisotropy, including two magnetic easy axes (i.e., four easy orientations of magnetization) that originate from strong cubic anisotropy. This feature automatically offers the opportunity for increasing storage capacity in magnetic recording devices. For example, a four-level magnetic memory concept APRIL 2009 VOLUME 12 NUMBER 4 19

7 REVIEW Ferromagnetic semiconductor GaMnAs Fig. 7 AFM (left) and FM (right) splittings in polarized neutron reflectivity (multiplied by Q 4 ) measured in a Ga 0.97 Mn 0.03 As/GaAs:Be/Ga 0.97 Mn 0.03 As multilayer at 30K. The data were collected serially in the order from (a) to (g). The solid lines show calculated reflectivity corresponding to multilayer models shown in the middle column. had already been demonstrated by Lim et al. 88 with GaMnAs grown on a vicinal GaAs surface, where four distinct Hall resistance states were realized due to the combined effects of the planar and the anomalous Hall effects. Later, Lee et al. 89 further extended this concept by using stable muti-domain formations arising from the presence of the four magnetization directions in a GaMnAs layer. The magnetic anisotropy of GaMnAs responsible for such rich magnetization behavior depends in a sensitive manner on the material parameters such as strain and carrier density, which then provide a handle for manipulating the magnetization. However, the nature of magnetic anisotropy in GaMnAs layers e.g., the origin of in-plane uniaxial anisotropy in this material is still not fully understood, and remains to be uncovered for achieving precise control of magnetization orientation in GaMnAs device structures. From the point of view of memory device applications, IEC in the GaMnAs multilayer structures must also be thoroughly understood. Although the theoretically predicted AFM IEC was indeed recently observed, the agreement between theory and experiment is only qualitative at the present time. So far the AFM coupling was observed in only a single sample, and the presence of IEC oscillation in GaMnAs multilayer structures is yet to be revealed. Furthermore, the thickness of GaMnAs layers which produce AFM IEC in a multilayer geometry is much thicker than what was considered in theoretical calculations. The dependence of IEC on the structure parameters must therefore be further investigated quantitatively, both in theory and in the laboratory. The possibility of electrical control of magnetic anisotropy of a GaMnAs layer 90 along with controlling IEC in GaMnAs multilayers by doping, as discussed in this paper, suggests that such multilayers can potentially provide significant device advantages over metallic ferromagnetic multilayers. The remaining obstacle for the implementation of GaMnAs-based devices is the fact that ferromagnetic properties of GaMnAs films are now only observed below room temperature. Increasing the ferromagnetic transition temperature T C in GaMnAs is therefore the primary challenge that needs to be addressed. Recent work suggests, however, that there is hope in this regard, since values of T C as high as 170K have already been demonstrated in bare GaMnAs films with heavy Mn doping 91, and T C of nearly 250K has been achieved in delta- and modulationdoped GaMnAs-based heterostructures 92. There is also some evidence that the proximity of an Fe layer can induce room temperature ferromagnetism in a thin GaMnAs layer 93. If room temperature ferromagnetism can be achieved in GaMnAs, there is real potential to exploit the novel magnetic anisotropy properties, together with the demonstrated FM and AFM coupling in GaMnAs-based multilayers described in this paper, to realize entirely new devices with unprecedented properties in the area of magnetic memory storage and manipulation. 20 APRIL 2009 VOLUME 12 NUMBER 4

8 Ferromagnetic semiconductor GaMnAs REVIEW ACKNOWLEDGEMENTS This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korean government (MEST) (No. R ); by the Seoul R&DB Program; by Korea University Grant; by KOSEF through the Nuclear R&D Programs (M N ); and by NSF Grant No. DMR REFERENCES 1. Ohno, H., Science (1998) 281, Haury, A., et al., Phys. Rev. Lett. (1997) 79, Prinz, G. A., Science (1990) 250, Wolf, S. A., et al., Science (2001) 294, Furdyna, J. K., and Kossut, J., (ed) Semiconductors and Semimetals, New York: Academic, (1988), vol Furdyna, J. K., J. Appl. Phys. (1982) 53, Galazka, R. R., and Kossut, J., Lecture Notes in Physics, Springer, Berlin, (1980), Vol. 132, p Mycielski, J., in Recent Developments in Condensed Matter Physics, Devreese, J. T., (ed.), Plenum, New York (1981), p Brandt, N. B., and Moshchalkov, V. V., Adv. Phys. (1984) 33, Goede, O., and Heimbrodt, W., Phys. Status Solidi B (1988) 146, Furdyna, J. K., J. Appl. Phys. (1988) 64, R Gaj, J. A., Semiconductors and Semimetals, in Diluted Magnetic Semiconductors, Furdyna, J. K., and Kossut, J., (eds.), Academic, Boston, (1988) Vol. 25, chap Dietl, T., Diluted Magnetic Semiconductors, in Handbook of Semiconductors, ed. Mahajan, S., Vol. 3B, North Holland, Amsterdam, (1994). 14. Furdyna, J. K., et al., Optical Properties of Semiconductor Nanostructures (NATO Science Series vol 81) Sadowski, M. L., Potemski, V., and Grynberg, M., (eds.), Dordrecht, Kluwer, (2000) p Ohno, H., J. Magn. Magn. Mater. (1999) 200, Furdyna, J. K., et al., J. Korean Phys. Soc. (2003) 42, S579 Topical Review R MacDonald, A. H., et al., Nat. Mater. (2005) 4, Munekata, H., et al., Phys. Rev. Lett. (1989) 63, Ohno, H., et al., Phys. Rev. Lett. (1992) 68, De Boeck, J., et al., Appl. Phys.Lett. (1996) 68, Ohno, H., et al., Appl. Phys. Lett. (1996) 69, Matsukura, F., et al., J. Appl. Phys. (2000) 87, Lim, W. L., et al., Physica E (2004) 20, Wojtowicz, T., et al., Appl. Phys. Lett. (2003) 82, Ohno, Y., et al., Nature (1999) 402, Ohno, H., et al., Appl. Phys. Lett. (1998) 73, Koshihara, S., et al., Phys. Rev. Lett. (1997) 78, Oiwa, A., et al., Phys. Rev. Lett. (2002) 88, Ohno, H., et al., Nature (2000) 408, Chiba, D., et al., Science (2003) 301, Yuldashev, Sh. U., et al., Appl. Phys. Lett. (2003) 82, Lee, S., et al., J. Appl. Phys. (2003) 93, Tang, H. X., et al., Phys. Rev. Lett. (2003) 90, Hamaya, K., et al., J. Appl. Phys. (2003) 94, Dietl, T., et al., Science (2000) 287, Sawicki, M., et al., Phys. Rev. B (2004) 70, Abolfath, M., et al., Phys. Rev. B (2001) 63, Dietl, T., et al., Phys. Rev. B (2001) 63, Sawicki, M., et al., J. Superconductivity (2003) 16, Sawicki, M., et al., Phys. Rev. B (2005) 71, Wang, K. Y., et al., Phys. Rev. Lett. (2005) 95, Liu, X., et al., Phys. Rev. B (2005) 71, Sasaki, Y., et al., J. Appl. Phys. (2002) 91, Liu, X., et al., J. Appl. Phys. (2005) 98, Shin, D. Y., et al., Phys. Rev. B (2007) 76, Nazmul, A. M., et al., Phys. Rev. B (2008) 77, Lang, R., et al., Phys. Rev. B (2005) 72, Titova, L. V., et al., Phys. Rev. B (2005) 72, Moore, G. P., et al., J. Appl. Phys. (2003) 94, Farle, M., Rep. Prog. Phys. (1998) 61, Stoner, E. C., and Wohlfarth, E. P., Philos. Trans. R. Soc. London, Ser. A (1948) 240, Son, H., et al., J. Appl. Phys. (2008) 103, 07F Liu, X., et al., Phys. Rev. B (2003) 67, Shen, A., et al., J. Cryst. Growth (1997) 175, Thevenard, L., et al., Phys. Rev. B (2007) 75, Yu, K. M., et al., Appl. Phys. Lett. (2002) 81, Limmer, W., et al., Phys. Rev. B (2005) 71, Ku, K. C., et al., Appl. Phys. Lett. (2003) 82, Yu, K. M., et al., Phys. Rev. B (2002) 65, Wojtowicz, T., et al., Appl. Phys. Lett. (2003) 83, Shin, D. Y., et al., Phys. Rev. Lett. (2007) 98, Kim, J., et al., Phys. Rev. B (2008) 78, Grünberg, P., et al., Phys. Rev. Lett. (1986) 57, Parkin, S. S., et al., Phys. Rev. Lett. (1990) 64, Parkin, S. S., et al., Phys. Rev. Lett. (1991) 67, Borchers, J. A., et al., Phys. Rev. Lett. (1999) 82, Bloemen, P. J., et al., Phys. Rev. B (1994) 50, Rhyne, J. J., et al., J. Magn. Magn. Mater. (1998) , Kepa, H., et al., Europhys. Lett. (2001) 56, Kepa, H., et al., Phys. Rev. B (2003) 68, Jungwirth, T., et al., Phys. Rev. B (1999) 59, Sankowski, P., and Kacman, P., Phys. Rev. B (1999) 71, (R). 73. Giddings, A. D., et al., Phys. Stat. Sol. (c) (2006) 12, Akiba, N., et al., Appl. Phys. Lett. (1998) 73, Chiba, D., et al., Appl.Phys. Lett. (2000) 77, Yuldashev, Sh. U., et al., Jpn. J. Appl. Phys. (2004) 43, Ge, Z., et al., Appl. Phys. Lett. (2007) 91, Chung, S. J., et al., J. Appl. Phys. (2004) 95, Kepa, H., et al., Phys. Rev. B (2001) 64, (R). 80. Chung, J. H., et al., Phys. Rev. Lett. (2008) 101, Chiba, D., et al., Phys. Rev. Lett. (2004) 93, Dziatkowskia, K., et al., Acta Phys. Pol. A (2007) 112, Felcher, G. P., Phys. Rev. B (1981) 24, 1595(R). 84. Majkrzak, C. F., Physica B (1996) 221, Kirby, B. J., et al., Phys. Rev. B (2006) 74, Kirby, B. J., et al. Phys. Rev. B (2007) 76, Kirby, B. J., et al., J. Appl. Phys. (2008) 103, 07D Lim, W. L., et al., Phys. Rev. B (2006) 74, Lee, S., et al., Appl. Phys. Lett. (2007) 90, Chiba, D., et al., Nature (2008) 455, Ohya, S., et al., Appl. Phys. Lett. (2005) 87, Nazmul, A. M., et al., Phys. Rev. Lett. (2005) 95, Maccherozzi, F., et al., Phys. Rev. Lett. (2008) 101, APRIL 2009 VOLUME 12 NUMBER 4 21