Spintronics device concepts S.J. Pearton, D.P. Norton, R. Frazier, S.Y. Han, C.R. Abernathy and J.M. Zavada Abstract: Spin-dependent phenomena in semiconductors may lead to devices with new or enhanced functionality, such as polarised solid-state light sources (spin light-emitting diodes), novel microprocessors and sensitive biological and chemical sensors. The realisation of robust semiconductor spin-device technology requires the ability to control the injection, transport and detection of polarised carriers, and to manipulate their density by a field gating. The absence of Si- based or room-temperature dilute magnetic semiconductors has subdued the initial excitement over semiconductor spintronics, but recent reports demonstrate that progress is far from dormant. The authors give examples of a number of different spin-device concepts for polarised light emission, spin field-effect transistors) and nanowire sensors. It is important to re-examine some of the earlier concepts for spintronics devices, such as the spin field-effect transistor, to account for the presence of the strong magnetic field which has deleterious effects. In some of these cases, the spin device appears to have no advantage relative to the conventional charge-control electronic analogue. There have been demonstrations of device-type operation in structures based on GaMnAs and InMnAs at low temperatures. The most promising materials for room-temperature polarised light emission are thought to be GaN and ZnO, but results to date on realising such devices have been disappointing. The short spin-relaxation time observed in GaN/InGaN heterostructures probably results from the Rashba effect. Possible solutions involve either cubic phase nitrides or the use of additional stressor layers to create a larger spin-splitting, to get polarised light emission from these structures, or to look at alternative semiconductors and fresh device approaches. 1 Introduction The use of electron spin, in addition to electron charge, has been suggested to hold promise for a new class of devices with new or enhanced functionality [1–7]. Magnetism (and therefore electron spin) in metals has been the basis of information storage since the discovery of the giant magneto-resistance (GMR) effect [8] , in which the resistance of a thin-film ferromagnetic/nonmagnetic layer sandwich is strongly magnetic-field-dependent. The GMR effect is now used in most computer hard drives. The discovery of GMR in metallic multilayer structures has led to much more sensitive position sensors (used in automobile braking systems), perimeter defense systems, magnetometers, non- volatile memory chips and read heads used in computer hard drives and disk storage systems. The ability to produce ferromagnetism in semiconductors above room temperature has been suggested to lead to devices such as light-emitting and laser diodes with polarised output (which would enable much more information to be encoded in lightwave communication systems), transistors with novel designs [9], magnetic devices with gain, integrated logic and memory chips (leading to computers that would turn on immediately like a television set, without having to boot-up the operating system that is currently stored in magnetic memory in the hard drive), and powerful remote sensor systems that incorporate magnetic detection functions with on-chip signal processing and off-chip optical communication. Efforts are underway to achieve these goals through hybrid approaches that integrate the metallic magnetic elements on top of conventional Si circuits [10], or by injecting spin- polarised electrons from metals into semiconductors (Figs. 1a and b) [10] . However, the initial enthusiasm on many of these device concepts has been tempered by the lack of progress on realising operational devices and a re- examination of some of the proposed structures suggests they may have little or no advantage relative to conven- tional approaches. The most direct method of realising spintronic devices would be to induce ferromagnetism in a semiconductor at practical operating temperatures, by introducing appropriate magnetic dopants such as Mn at levels of a few per cent, producing a dilute magnetic semiconductor (DMS) (see Fig. 1c) [10]. DMSs are alloys where a stoichiometric fraction of the constituent atoms has been replaced by transition metal atoms. Such alloys are semiconducting, but can possess well-defined magnetic properties (e.g. paramagnetic, antiferromagnetic, ferromag- netic) that conventional semiconductors do not have. Thus, they can potentially serve as a means to inject spin to and control spin properties in adjacent nonmagnetic semicon- ducting layers. Besides GaMnAs, there are numerous other technologi- cally relevant materials, such as GaMnN, AlCrN and ZnCoO, that have exhibited ferromagnetism. In addition to their higher Curie temperatures, the nitride and oxide systems have several additional advantages over these narrower bandgap systems. The wider bandgap will allow device operation at elevated temperatures (4550 1C) and the ability to emit short wavelength light (in the blue/green S.J. Pearton, D.P. Norton, R. Frazier, S.Y. Han and C.R. Abernathy are with the Department of Materials Science and Engineering, University of Florida, Gainesville, FL 32611, USA J.M. Zavada is with the Electronics Division, Army Research Office, Research Triangle Park, NC 27709, USA E-mail: [email protected]fl.edu r IEE, 2005 IEE Proceedings online no. 20045129 doi:10.1049/ip-cds:20045129 Paper first received 13th September 2004 and in revised form 4th March 2005 312 IEE Proc.-Circuits Devices Syst., Vol. 152, No. 4, August 2005
Transcript
Spintronics device concepts
S.J. Pearton, D.P. Norton, R. Frazier, S.Y. Han, C.R. Abernathy and J.M. Zavada
Abstract: Spin-dependent phenomena in semiconductors may lead to devices with new orenhanced functionality, such as polarised solid-state light sources (spin light-emitting diodes), novelmicroprocessors and sensitive biological and chemical sensors. The realisation of robustsemiconductor spin-device technology requires the ability to control the injection, transport anddetection of polarised carriers, and to manipulate their density by a field gating. The absence of Si-based or room-temperature dilute magnetic semiconductors has subdued the initial excitement oversemiconductor spintronics, but recent reports demonstrate that progress is far from dormant. Theauthors give examples of a number of different spin-device concepts for polarised light emission,spin field-effect transistors) and nanowire sensors. It is important to re-examine some of the earlierconcepts for spintronics devices, such as the spin field-effect transistor, to account for the presenceof the strong magnetic field which has deleterious effects. In some of these cases, the spin deviceappears to have no advantage relative to the conventional charge-control electronic analogue.There have been demonstrations of device-type operation in structures based on GaMnAs andInMnAs at low temperatures. The most promising materials for room-temperature polarised lightemission are thought to be GaN and ZnO, but results to date on realising such devices have beendisappointing. The short spin-relaxation time observed in GaN/InGaN heterostructures probablyresults from the Rashba effect. Possible solutions involve either cubic phase nitrides or the use ofadditional stressor layers to create a larger spin-splitting, to get polarised light emission from thesestructures, or to look at alternative semiconductors and fresh device approaches.
1 Introduction
The use of electron spin, in addition to electron charge, hasbeen suggested to hold promise for a new class of deviceswith new or enhanced functionality [1–7]. Magnetism (andtherefore electron spin) in metals has been the basis ofinformation storage since the discovery of the giantmagneto-resistance (GMR) effect [8], in which the resistanceof a thin-film ferromagnetic/nonmagnetic layer sandwich isstrongly magnetic-field-dependent. The GMR effect is nowused in most computer hard drives. The discovery of GMRin metallic multilayer structures has led to much moresensitive position sensors (used in automobile brakingsystems), perimeter defense systems, magnetometers, non-volatile memory chips and read heads used in computerhard drives and disk storage systems. The ability to produceferromagnetism in semiconductors above room temperaturehas been suggested to lead to devices such as light-emittingand laser diodes with polarised output (which would enablemuch more information to be encoded in lightwavecommunication systems), transistors with novel designs[9], magnetic devices with gain, integrated logic and memorychips (leading to computers that would turn on immediatelylike a television set, without having to boot-up the operating
system that is currently stored in magnetic memory in thehard drive), and powerful remote sensor systems thatincorporate magnetic detection functions with on-chipsignal processing and off-chip optical communication.Efforts are underway to achieve these goals through hybridapproaches that integrate the metallic magnetic elements ontop of conventional Si circuits [10], or by injecting spin-polarised electrons from metals into semiconductors(Figs. 1a and b) [10]. However, the initial enthusiasm onmany of these device concepts has been tempered by thelack of progress on realising operational devices and a re-examination of some of the proposed structures suggeststhey may have little or no advantage relative to conven-tional approaches. The most direct method of realisingspintronic devices would be to induce ferromagnetism in asemiconductor at practical operating temperatures, byintroducing appropriate magnetic dopants such as Mn atlevels of a few per cent, producing a dilute magneticsemiconductor (DMS) (see Fig. 1c) [10]. DMSs are alloyswhere a stoichiometric fraction of the constituent atoms hasbeen replaced by transition metal atoms. Such alloys aresemiconducting, but can possess well-defined magneticproperties (e.g. paramagnetic, antiferromagnetic, ferromag-netic) that conventional semiconductors do not have. Thus,they can potentially serve as a means to inject spin to andcontrol spin properties in adjacent nonmagnetic semicon-ducting layers.
Besides GaMnAs, there are numerous other technologi-cally relevant materials, such as GaMnN, AlCrN andZnCoO, that have exhibited ferromagnetism. In addition totheir higher Curie temperatures, the nitride and oxidesystems have several additional advantages over thesenarrower bandgap systems. The wider bandgap will allowdevice operation at elevated temperatures (4550 1C) andthe ability to emit short wavelength light (in the blue/green
S.J. Pearton, D.P. Norton, R. Frazier, S.Y. Han and C.R. Abernathy are withthe Department of Materials Science and Engineering, University of Florida,Gainesville, FL 32611, USA
J.M. Zavada is with the Electronics Division, Army Research Office, ResearchTriangle Park, NC 27709, USA
and UV regions). Advantageously, the nitrides possess ahigh tolerance to growth defects such as dislocations. All ofthese can be grown on Si substrates with good crystalquality, and have shown magnetic hysteresis above roomtemperature [11]. However, in most cases, there has beenless than clear evidence that this is due to carrierpolarisation rather than the inclusion of undesirableferromagnetic second phases. There is increasing interestin ZnO-based DMSs, where second phases generally cannotexplain the observed ferromagnetism. Of course, even insuch a system, there are wildly diverging results in theliterature, ranging from very low moments per transitionmetal ion [12] to values of over 2 Bohr magneton per Co[13].
The group of DiTusa et al. [14] demonstrated thatFe1–yCoySi, an n-type narrow-gap magnetic semiconductor,exhibits an unusually large anomalous Hall effect thatappears to derive from intrinsic band structure effects ratherthan impurity scattering as in conventional ferromagnets.This discovery of a Si-based magnetic semiconductorsuggests routes for the realisation of spin field-effecttransistors compatible with existing microprocessor circui-try. For example, CoSi2 is a commonly used low-resistancemetallic phase that is used in thin-film form to make ohmiccontacts for Si transistors. We can envision the use of theFeCoSi magnetic semiconductor layers as injectors andcollectors of spin-polarised current in Si transistors.Although FeCoSi has a Curie temperature of only 53K,the clear indication is that Kondo insulators in this familywith higher transition temperatures may form the basis of atrue Si-based spintronics technology. Toyosaki et al. [15]have observed the anomalous Hall effect (AHE) in n-typeTi1�xCoxO2�d (where d represents an oxygen deficiency),although the same authors found that Co metal was presentover half the oxygen pressure range used to generate thissame set of samples [16]. The AHE is often taken asevidence that the carrier population is polarised and that thematerial exhibiting it is a true DMS, although granularmagnetic systems such as Ni and Co in SiO2 have alsoshown the effect, so it is a necessary but not sufficientcondition for true DMS behaviour. Ti1�xCoxO2�d is aroom temperature transparent dilute magnetic semiconduc-tor, which is useful for transparent coatings on solarcells, for example. The results reinforce the findingthat electron-doped materials can exhibit high Curietemperatures, a feature often overlooked in theoreticaltreatments of ferromagnetism in semiconductors. Then-type semiconductors generally have larger carrier mobi-lities than their p-type counterparts, producing advantagesin terms of device speed. Ti1�xCoxO2�d may also presenttechnological payoffs in the form of thin-film spin field-
effect transistors, which can be used to drive displays forportable computers.
Previous articles have discussed some spintronic deviceconcepts, such as spin junction diodes and solar cells,optical isolators and electrically controlled ferromagnets[9, 17–51]. The realisation of light-emitting diodes with adegree of polarised output has been used to measure spininjection efficiency in heterostructures [17–19, 23–25]. Whilethe advertised advantages of spin-based devices includenonvolatility, higher integration densities and higher switch-ing speeds, there are many factors still to consider as towhether any of these can be realised. These factors includewhether the signal sizes due to spin effects are large enoughat room temperature to justify the extra development workneeded to make spintronic devices, and whether theexpected added functionality possible will materialise. Inaddition, many of the initial claims for spintronic devices donot stand up under close examination, as we will discuss inrelation to the so-called ‘spin-FET’. The push forsemiconductor spintronics is motivated by the materialscompatibility with traditional semiconductor electronicsand by the desire to produce true three-terminal spintronicdevices with potential applications such as nonvolatileprogrammable logic, spin-based optoelectronics and quan-tum computation. A persisting bottleneck for semiconduc-tor-based spintronics has been spin injection from theferromagnet into the semiconductor, which is a critical stepin the implementation of any spin logic devices. Despiteintensive efforts, efficient electrical spin injection at ambientconditions has proven challenging to achieve reproducibly.Traditionally, the spintronics community has tackled thisproblem from two fronts: materials synthesis and interfaceengineering. The work on materials has focused onsynthesis of magnetic semiconductors and other novelmaterials with high spin polarisation, while that oninterfaces has covered fabrication of tunnelling andSchottky barriers to facilitate different spin transport andphysical and chemical modification of the interface topreserve the high-bulk spin polarisation at the junction.
2 Candidate device concepts
A number of initial device configurations for GaMnAs [39–54] have already been demonstrated at low temperatures.This list includes exchange-biased samples [39], spin-dependent resonant tunnelling diodes [40], magnetic tunneljunctions [41, 42] and spin-polarised light-emitting diodes(spin LEDs) [43–46]. The presence of tunable wave-functionoverlap between magnetic ions and carriers confined toquantum structures has also been reported [46]. Morecontroversial are reports of very large magneto-transport
a
magnetic memory array
silicon waferwith circuitry
b
silicon
silicon
magnetic thin films
c
magnetic ions
host semiconductor
Fig. 1 Schematic of possible integration approaches for obtaining spintronic-functional materials [8]a Hybrid systemsb Hybrid devicesc Magnetic semiconductors
effects, the so-called ‘giant planar Hall effect’ [47]. There arealso numerous theoretical analyses of new bipolar deviceconfigurations combining n-doped semiconductors withGaMnAs [48–51].
Most existing device configurations that exploit the spindegree of freedom are extensions of ordinary electronicdevices, with enhanced but not qualitatively new function-ality. Attempts have been made to explore the potential offerromagnetic semiconductors for new functions such asreconfigurable logic, using a unipolar spin transistor, inwhich magnetic domain walls define regions whereincarriers of opposite spin play a role analogous to that ofelectrons and holes in conventional transistors [50],electrically tunable ferromagnetism has been explicitlydemonstrated in In1�xMnxAs-based field-effect devices [38].
2.1 Spin FETThe most common example of a spintronic device structureis the spin field-effect transistor (spin FET) [9]. A schematicdiagram of the device is shown in Fig. 2a [55]. In the spinFET, the drain and source of a conventional transistor aremade ferromagnetic, either by having an ohmic contactmetal scheme that is magnetic or using injection from amagnetic semiconductor into the channel. If the twoferromagnets are aligned, a spin-polarised current willbehave like a normal FET current. If the ferromagnets areanti-aligned the transistor will be shut off. The selection ofthe spin current is achieved via electric-field modulation.Electrons are injected with a definite spin orientation fromthe source, which is then controllably precessed in thechannel with a gate-controlled Rashba spin-orbit interac-tion, and finally sensed at the drain. At the drain end, theelectron’s transmission probability depends on the relative
alignment of its spin with the drain’s (fixed) magnetisation.By controlling the angle of spin precession in the channelwith a gate voltage, the relative spin alignment at the drainend can be controlled, and, hence, the source-to-draincurrent. This realises the basic ‘transistor’ action [55, 56].Through the application of a gate voltage, we can move thetwo spin-split sub-bands relative to the Fermi energy, whichcan effectively pinch off both channels (no current), pinch-off one channel (polarised current), or let both channelsthrough (unpolarised current). Compared to other spinfilters, this device offers selectivity and easy integration intoan integrated circuit. The interaction of an applied gatevoltage with polarised carriers can alter their spin alignmentand the spin FET is expected to be able to turn off at lowervoltages than a conventional charge-controlled transistor,leading to potential applications in very low-power micro-processors. Control of the spin alignment could also beattained dynamically, allowing for microprocessors thatreconfigure their hardware in real time.
However, recent re-examination of the spin-FET concepthas revealed serious shortcomings [55–58]. Cahay andBandyopadhyay [57, 58] modelled phase-coherent spintransport in the weakly disordered quasi-one-dimensionalchannel of a gate-controlled electron-spin interferometer.When the effects of an axial magnetic field in the channel ofthe interferometer (caused by the ferromagnetic contacts), aRashba spin-orbit interaction, and elastic (nonmagnetic)impurity scattering are all considered, it was shown that, inthe presence of an axial magnetic field, nonmagneticimpurities can cause spin relaxation in a manner similarto the Elliott–Yafet mechanism. The amplitudes and phasesof the conductance oscillations of the interferometer and thedegree of spin-conductance polarisation were found to bevery sensitive to the height of the interface barrier at thecontact, as well as the strength, locations and nature(attractive or repulsive) of just a few elastic nonmagneticimpurities in the channel. This can seriously hinder practicalapplications of spin interferometers. In the classic spinanalogue of the electro-optic modulator, the modulation ofthe spin-polarised current is carried out by tuning theRashba spin-orbit interaction with a gate voltage. Bandyo-padhyay and Cahay [55] recently proposed an analogousmodulator, where the modulation is carried out by tuningthe Dresselhaus spin-orbit interaction instead, using a splitgate (Figs. 2b and c). Additionally, the magnetisation of thesource and drain contacts in this new approach is transverseto the channel, whereas, in the classical device, it is along thechannel. In the new approach, there is no magnetic field inthe channel unlike in the case of the Datta–Das modulator.This can considerably enhance modulator performance, asthe deleterious effects of magnetic field have now been well-documented [57, 58].
Figure 3 shows schematics of spin FET structures basedon either GaN or Si(bottom). In the former case, thechannel may be either a DMS or simply the pure GaN,depending on the actual geometry. In the latter case, it maybe possible to use the room-temperature ferromagnetGaMnP [59], that is closely lattice matched to Si, as thespin injection source and drain.
In the context of spintronic field-effect devices, Bandyo-padhyay and Cahay [56] have recently re-examined some ofthe original concepts and have shown that it is generallyuntrue in the belief that spin-based devices will be faster andconsume less power than their electronic counterparts. Theydid an analysis of the switching voltages in a one-dimensional spin FET relative to a Si-based MOSFETand found that the former is actually not a lower-powerdevice. Their main conclusion was that, unless materials
source drain
gate
a
source drain
z
x
y
barrier
barrier
b
Schottkycontact
(split gate)ohmic
contact(drain)
c
Fig. 2 Schematic views of classical spin FET (a), new approachsuggested by Bandyopadhyay and Cahay (b) and top view of splitgates (c) [55]
with extremely strong spin-orbit interaction can be devel-oped, the spintronic devices will not measure up to theirelectronic analogues [56]. Several other manifestations ofthe spin FET have recently been proposed [60–62], but, ineach case, doubts have been raised whether these presentversions show any advantage over the original Datta–Dasdesign. Bandyopadhyay and Cahay [56] concluded recentlythat it is unlikely that currently proposed spin FETs willplay a significant role in digital, analogue or mixed signalcircuits but may be more suited to applications in memory,where high gain, high frequency etc. are not necessary. Theyalso concluded that spintronic devices may also have betternoise margin, because spin does not easily couple to strayelectric fields (unless the host material has very strong spin-orbit interactions). Thus, they concluded that it is alsopossible that spintronics may be able to outpace electronicsin nonconventional applications such as single spin logic[63], spin neurons [64] and using spin in a quantum dot toencode qubits [65–67].
2.2 Transparent ferromagnetA proposed photomagnet structure is shown in Fig. 4[68, 69]. The concept is based on the theory that ZnO:MnCris a half-metallic ferromagnet upon hole doping, while
ZnO:FeMn is a half-metallic ferromagnet upon electrondoping. For photons of appropriate energy, electrons andholes created in the GaAs substrate near the interface withthe ZnO:MnCr or ZnO:FeMn can be drawn into thosematerials by biasing, causing them to become half-metallicferromagnets. The presence of these ordered states can bedetected using the magneto-optical effect from anotherprobe beam of photons with energy lower than the ZnObandgap. The device is readily grown on conducting GaAsto provide good ohmic contact to the substrate. Thepotential use of this light- or field-controlled ferromagnethas been described in detail previously [68, 69] and ZnO hassome attributes such as ease of growth at low temperatureson low-cost substrates such as glass.
2.3 Spin LEDsRecently, efficient spin injection has been successfullydemonstrated in all-semiconductor tunnel diode structuresby using a spin-polarised DMS as the injector in one case[17–19], and by using a paramagnetic semiconductor underhigh magnetic field as a spin filter in the other [18]. In such ascheme, spin-polarised holes and unpolarised electrons areinjected from either side and recombine in a quantum well.The polarisation of the injected holes can be obtained bycomparing the intensity of the right- and left-circularlypolarised light in the electroluminescence spectra.
Among such devices the simplest seems to be the conceptof a light emitting diode (LED) with one of the contactlayers made ferromagnetic by incorporation of transitionmetal impurities, a so-called spin LED [17–19, 28, 29,31, 32]. A recent manifestation from Kioseoglou et al. [19]employed an n-type DMS, namely CdCr2Se4, to inject spinpolarised electrons into an AlGaAs/GaAs light-emittingdiode structure. An injected spin polarisation of 6% wasobtained at 5K. There is much to be learned about spininjection from such systems, but clearly it would bedesirable to try similar experiments in systems designedfor room-temperature operation. An obvious candidate isGaN, because of the robust light emission from thismaterial, even when grown on mismatched substrates suchas sapphire. An example of a GaN-based spin LED isshown in Fig. 5. Such a device should allow modulation ofthe polarisation of light emitted by the spin LED byapplication of an external magnetic field. Numerousexperiments and theoretical discussions have led to ageneral belief that efficient spin injection from a DMS intoa conventional semiconductor is promising because of thecomparable conductivity, good Fermi wave vector match-ing and clean interface.
The most straightforward approach to achieve a wide-bandgap spin LED would be to implant Mn into the topcontact p-GaN layer of the standard GaN/InGaN LED orgrow GaMnN as the top injection layer. The electrical andluminescent properties of such devices show that they doproduce electroluminescence, but due to the difficulty inannealing out all radiation defects the series resistance andthe turn-on voltage of such spin LEDs are higher than forordinary LEDs, and the electroluminescence intensity EL islower.
The spin injection efficiency was evaluated using theexcitation photon energy of about 5 eV, i.e. well above thebandgap of the top (Ga,Mn)N layer. Under theseconditions the preferential light absorption within the top(Ga,Mn)N layer should ensure that the dominant portionof the carriers and excitons participating in the radiativerecombination in the InGaN QW is supplied by themagnetic barrier [70]. The results of the performedmeasurements are summarised in Fig. 5b. The optical (spin)
(Ga, Mn)N
u-GaN
substrate
u-(Ga, Al)N
gate: Ti /Al/Pt /Au
source drain
a
Si
SiGe
Sip-GaMnPp-GaMnP
gatesource drain
b
Fig. 3 Schematic of (Ga,Mn)N based (a) and SiGe/Si-based (b)spin FET
V
hν < Eg(ZnO)
Eg(GaAs) < hν < Eg(ZnO)
ZnO
photo-induced ferromagnet
detection photons
GaAs substrate
holeelectron
or ZnO: Mn−xFex
ZnO: Cr1−xMnx
−+
Fig. 4 Schematic of ZnO-based transparent photomagnet (after[66, 67] and also H.Katayama-Yoshida, private communication,and presented at Spring MRS, San Francisco, April 2002)
polarisation of the QW PL was only detected in an appliedmagnetic field and was generally very weak (o2%). Theobserved PL polarisation reflects combined effects of thespin injection from the magnetic (Ga,Mn)M layer andintrinsic polarisation of the InGaN QW. The latter hasseparately been studied by tuning the excitation energybelow the bandgap of GaN (also presumably GaMnN),
i.e. by resonant optical excitation of the InGaN QW with aphoton energy of 2.9 eV. It gives up to 5–10% at 2K withan applied magnetic field of up to 5T, which is probably dueto population distribution between spin sublevels at a lowtemperature.
To shed light on the possible origin for the weak spinpolarisation of the InGaN QW, optical orientation experi-ments were performed where a chosen spin orientation ofexcitons/carriers in the InGaN QW was generated by therelevant circularly polarised excitation light resonantlypumping the InGaN QW [70]. No spin polarisation wasobserved at 0T, showing that the generated spin orientationwas completely lost during the energy relaxation process tothe ground state of the excitons, giving rise to the PL. Thisis in sharp contrast to the cases in II-VIs and GaAs, wherespin polarisation can usually be observed in such opticalorientation experiments. With an applied magnetic field, thespin polarisation of the InGaN QW is independent of thepolarisation of the excitation light which again proves thatthe spin relaxation in the InGaN QW is extremely efficient[70]. The spin loss in the structures is shown to be largelydue to fast spin relaxation (of the order of 20pS) within theInGaN MQW, which itself destroys any spin polarisationgenerated by optical spin orientation or electrical spininjection. In the wurtzite InGaN/GaN system, biaxial strainat the interfaces gives rise to large piezoelectric fieldsdirected along the growth axis. This built-in piezo fieldbreaks the reflection symmetry of confining potentialleading to the presence of a large Rashba term in theconduction band Hamiltonian, which is responsible for theshort spin-relaxation times. Recent theory suggests the shortspin-relaxation time in GaN/InGaN heterostructures resultsfrom the Rashba effect [71]. The Rashba splitting arisesfrom the inversion asymmetry of the confining potential. Insymmetrical InGaAs/GaAs/InGaAs quantum wells, theconfining potential has reflection symmetry and the Rashbaspin-orbit interaction is therefore weak. At first glance, wemight expect the same to hold for symmetrical InGaN/GaNquantum wells. The InGaN/GaN system, however, has awurzite crystal structure and the biaxial strain at theinterfaces gives rise to large piezoelectric fields directed
sapphire substrate
optical injection from the DMSlayer
p-GaN:Mg 2 µm
UID GaN 2 µm
n-GaMnN 120nm
Ni/Au ohmicring
Ti/Al/Pt/Au ohmicring
n-GaN:Si 20nm
InGaN 3 nmn-GaN:Si 10nm
UID GaN 10nm
5
x
a
GaMnN
exc PL
GaN
−2
0
2
420 440 460 480 500 520 540
wavelength, nm
T = 1.9K, exc = 244 nm
0T
3T PL
pola
risat
ion,
%
PL
inte
nsity
, arb
itrar
y un
its
b
Fig. 5 Schematic of GaN-based spin-led (a) and results of PLpolarisation measurements with optical injection of polarised carriersinto the GaMnN layer (b)
(Zn, Co)O
(Zn, Mn)O
ZnO-based spin FET
gate metalsource drain
insulator(gate oxide)
(Zn, Co)O
ohmic ring
ohmic ring
n-ZnMgO p-ZnMgO
ZnO
glass substrate
n+ - ZnO
n-ZnMgO
ZnO
p-ZnMgO
p-ZnOh�_
a
b
Fig. 6 Schematic of ZnMgCoO/ZnO/ZnMgCoO spin-led (a)and spin FET (b)
along the growth axis. This built-in piezo field breaks thereflection symmetry of confining potential leading to thepresence of a large Rashba term in the conduction bandHamiltonian. In highly strained structures this confinement-induced spin-orbit interaction presents the main source ofspin mixing and spin splitting being responsible for a strongspin relaxation via Elliott–Yafet and/or Dyakonov–Perelmechanisms. Using a multiband k.p approach, we havecalculated the built-in piezo-electric field, derived expres-sions for the Rashba coefficient, and have determined thespin-relaxation times.
One possibility for reducing the effects of fast spin-relaxation times in wurtzite hosts is to use the cubicsymmetry. ZnO is much more readily grown in the cubicphase than GaN and shows promise as a room temperatureDMS [13, 72, 73]. Figure 6 shows schematics of spin-FETand spin-LED structures in ZnO. The FET structure wassuggested by Sato and Katayama-Yoshida [68, 69] as ananalogue of the spin FET and takes advantage of the factthat (Zn,Mn)O can be grown as an antiferromagnetic spin-glass insulator, while hole- and Mn-codoped ZnO can be ahalf-metallic ferromagnet. In the structure of Fig. 6b, theorysuggests the application of negative gate bias brings holesinto the (Zn,Mn)O and may convert it to the half-metallicferromagnetic state, using ferromagnetic (Zn,Co)O as thesource and drain contact material, it should be possible tohave a highly spin-polarised electron flow in the (Zn,Mn)Ochannel. The device can be fabricated by growing thesource/drain materials on top of the (Zn,Mn)O and thenetching away in the gate region for selective growth of thegate oxide. In the LED structure, Mg doping is used toincrease the ZnO bandgap to create a heterostructure forcarrier confinement.
2.4 Selective Kerr rotatorAnother potential DMS-based optical device is the selectiveKerr rotator, as shown in Fig. 7. We show the use of GaNpurely as an example of a potential room-temperaturemanifestation. In this structure, the Kerr rotation iscontrolled by an electric field applied between two p–njunctions (the DMS being the p-material). At zero bias thesurface p-region is ferromagnetic, allowing incident light tobe reflected with large Kerr rotation angle. Whensufficiently biased, the surface depletion of holes wouldremove the ferromagnetism and any Kerr rotation.
3 Determination of spin polarisation
Amajor hindrance for the practical implementation of theseconcepts is that they require efficient spin-polarised carrierinjection and transport. Conventional ferromagnetic metals
are often incompatible with existing semiconductor technol-ogy. Moreover, the spin injection efficiency is often very lowdue to resistivity differences and to the formation ofSchottky barriers. A key materials aspect to overcomingthis problem is the use of DMSs.
3.1 Exchange constantThe degree of spin polarisation in DMSs is the mostrelevant parameter to any spintronics device applications.Of primary concern in many spin-based materials is themagnetic coupling strength between the mobile or localisedcharge carriers and the localised (Mn) ion moments.Complete spin polarisation is possible only if the conductionor valence-band Zeeman splitting in the ferromagnetic stateis larger than the Fermi energy, and the size of the splittingis directly proportional to the exchange parameter l.Measurement of l is therefore vital for device considera-tions of the grown materials.
Magneto-optical methods measure the magnetisation-induced conduction-band shifts to extract s,d-f exchange[2, 3, 74–79]. More recently, exciton splitting was used tomeasure the s,p-d exchange in III–V semiconductors and itwas found that the DMSs have one of the largest exchangeconstants [1–4, 6, 7]. Both methods need to be applied to thewide bandgap DMS to extract the exchange constant.Besides the optical methods, several electrical measurementsare effective for obtaining exchange constants. One of them,which works well with insulating materials, is the internalfield (Fowler–Nordheim) emission, which was successfullyapplied to Eu-chalcogenides [2, 3, 75] and (Cd,Mn)Te [2, 3,76]. In this method, a three-layer device composed of metal/insulating magnetic semiconductor/metal is constructed.The insulating layer is thick enough to prevent low-biastunnelling, but thin enough to provide a tunnelling currentwhen a large bias causes enough distortion to the barrier.The nonlinear I–V characteristics permit a quantitativeevaluation of the band splitting below the magnetic orderingtemperature or in the presence of a magnetic field. Thisyields a measure of the exchange parameter directlyassociated with the conduction band, in contrast to theoptical techniques which in the DMS measure a combi-nation of conduction- and valence-band splittings. Thisexchange constant can also be obtained frommeasurementsof Schottky-barrier capacitance [77] in lightly dopedmaterial, and tunnelling data [78] in heavily doped material.In the former, the band splitting due to the ferromagneticordering decreases the effective thickness of the barrier, thuscausing a change in the capacitance; and in the latter thesplitting changes the barrier height and the tunnellingprobability, causing a sharp decrease in the tunnelconductance. The electrical measurements have the added
Ti /Al /Pt /Au Ti /Al /Pt /Au
polarised light
n-GaN
Ti /Al /Pt /Au Ti/Al/Pt /Au
polarised light
ferromagnetic p-(Ga, Mn)N
depletion
ferromagnetic p-(Ga, Mn)N
depletion
V = 0 V > VG
n-GaN n-GaN n-GaN
Fig. 7 Principle of operation of selective Kerr rotator
advantage of being applicable to both direct gap andindirect gap materials.
In recent work, spin-dependent tunnelling between twoferromagnetic films separated by an insulating (I) filmshows junction magnetoresistance (JMR) of better than20% [10, 26, 35, 36]. This suggests that tunnelling may be amore effective way of achieving spin injection than diffusivetransport. A large magnetoresistance can also be obtained ifthe tunnel barrier is also ferromagnetic. A prototype device,shown in Fig. 8a, comprises a nonmagnetic electrode, aferromagnetic insulating tunnel barrier (the polariser), and aferromagnetic counterelectrode (the analyser). Below Tc ofthe ferromagnetic insulator, the tunnel barrier is spin split,giving a highly polarised tunnel current as indicated in theschematic diagram. If the moments of the ferromagneticcounterelectrode analyser are parallel (antiparallel) to themoments of the spin-polarised tunnel current, then theresistance is low (high). Magnetoresistance exceeding 100%has been obtained in an Al–EuS–Gd device [76] ,where EuS,the ferromagnetic insulator, has a Tc ¼16.8K and Gd, theanalyser, a Tc near room temperature. One key objective inthe future will be to find materials that enable operation ofthese novel spin filter devices at room temperature.Promising barrier materials might be transition-metal-doped ZnO, insulating ferrites (CoFe2O4), or insulatingDMS materials with localised carriers to provide thenecessary magnetic interactions [2, 3]. For satisfactoryoperation the localisation length xL must be less than thesample size (barrier thickness) but greater than the lengthcharacterising the magnetic interactions. An example ofsuch a device based on ZnO is shown in Fig. 8b.
The successful implementation of these measurementsdepends critically on the total Zeeman splitting, which isoften written as [2, 3, 75, 78, 79]: DE ¼ lxhSzi, where l maycontain a combination of the exchange integrals for thedonor and acceptor states, x is the Mn concentration, and
hSzi is the average z-component of the Mn spin. l has beenfound to be independent of x in (Ga,Mn)As, so themeasurable effects (splitting) directly depend on the amountof Mn incorporated homogeneously into the semiconduc-tor. Although Mn is known to be a p-type dopant in thesematerials, the large intrinsic n-background means that(Ga,Mn)N may remain insulating at large Mn concentra-tions, so these methods may be particularly suitable for thissystem.
3.2 Superconducting tunnellingAmethod that has been widely used to directly measure thespin polarisation of ferromagnetic metals is electrontunnelling with a superconducting counter electrode. Thistechnique, pioneered by Tedrow and Mersevey [80], isparticularly useful for the determination of spin polarisationat the interface. The experiment requires that a tunneljunction be fabricated, which can be accomplished viaepitaxial growth of an insulating barrier or surfacetreatment of the DMS. The superconductor has to be alight metal with minimal spin-orbit interaction so that thespin-up and spin-down density of states can be split with theapplication of a magnetic field. Al is a common choice andBe can also be used.
3.3 Spin-dependent ‘Hall effect’The technique of superconducting tunnelling requires theoften difficult task of creating a thin uniform tunnel barrierand is limited to temperatures below B1K. Anotherscheme to directly measure the spin polarisation is aconceptual analogue of the Stern–Gerlach experiment [2, 3]:a patterned Hall bar of the DMS is placed in a uniformmagnetic field gradient such that a force F ¼ rðm � BÞ isexerted on the conduction electrons with spin magneticmoment of 1mB. If the electron spins are polarised parallelto the field gradient, which may be accomplished throughcrystalline anisotropy or the application of a backgroundmagnetic field, then majority and minority spin populationswill separate and build up charge on the sample edges. Thiswill induce an electric field which will oppose further chargebuild up, exactly as is encountered in the ordinary Halleffect. A rough estimate with field gradient of 1T/cm gives1mV as an order-of-magnitude signal for a fully polarisedmaterial, a voltage that is easily measurable. A field gradientof this magnitude can be obtained at the centre of the topsurface of a SmCo ring-shaped magnet. This techniquerequires minimal processing and should be applicable toany ferromagnetic material over a wide temperature range.Its development should facilitate easy determination of thespin polarisation of the DMS.
3.4 Spin injectionNumerous experiments and theoretical discussions have ledto a general belief that efficient spin injection from a DMSinto a conventional semiconductor is promising because ofthe comparable conductivity, good Fermi wave vectormatching and clean interface. It has been more than adecade since the first semiconductor spintronics device, spinFET, was proposed [9]. There are three critical elements forthis and any other all-electronic spin logic device: injectionof polarised spins from a ferromagnet into the semiconduct-ing channel, coherent manipulation of spins in thesemiconducting channel, and spin detection at the drain.So far, the experimental efforts to implement the proposeddevice schemes have largely followed the traditionalmicroelectronics approach, except that the source and drainelectrodes are replaced by ferromagnetic metals or semi-conductors. However, the much more stringent requirement
M FI (↑) FM (↑↓)
J↑
a
i
Al metalAl contact
Al contact
ZnO substrate (n-type)
ZnO buffer layer
ZnO (Mn, Sn)
MgO (insulator)
ZnO (Mn, Sn)
V
b
Fig. 8 Spin tunnel device in which the insulating tunnel barrier isferromagnetic (a) and possible embodiment in the ZnO materialssystem (b)
of the interface quality for spin injection has thus farprevented the realisation of a spin FET.
The creation of spin-polarised currents and the control ofspin-relaxation processes in nanostructures are another ofthe challenging problems of spin-based electronics. Theseprocesses were investigated in detail for relatively lowcurrents [81–83], when the electric field is relatively smalland does not affect the energy and spin relaxation.However, electric fields are high in short-channel FETs.As strong electric fields drastically change the electrondistribution function, the spin-relaxation processes arestrongly affected. In the Dyakonov–Perel spin-relaxation
mechanism, the characteristic spin-relaxation rate ðtsÞ�1rapidly increases with the electron energy. Hence, theelectron heating by a strong electric field should lead to astrong field dependence of ts.
4 DMS-based sensors
There is a growing market for magnetoresitive DNA chips.These involve the use of MR sensors to detect biomolecularrecognition processes between an immobilised probe and amagnetically labelled target [84]. The integration of DNAand other biomolecules array and detection methods basedon a portable, inexpensive and fully electronic approach isbeing actively pursued [83]. Magnetic labels are often usedto provide a means for cell separation, drug delivery orcontrast enhancement in magnetic resonance (MR) ima-ging. Integrated MR sensors detect the fringe field of thelabel that binds to the hybridised target. For MR biochipapplications, a small magnetising field is applied tomagnetise the nonremanent particles. The magnetic fieldcan be created by either on-chip current lines or an externalelectromagnet. If an AC field is used, lock-in amplifiers areused for optimal signal-to-noise during particle detection.One concept based on DMS thin films is illustrated inFig. 9. The transition metal-doped DMS would serve as thesource S to inject spin-polarised electrons into a nanoscalewire in which the spin polarisation would be maintained,Fig. 9a. The electrons would then be collected at the drainD. This device is not a conventional field-effect transistor asthere is no gate. A specific receptor R, either for chemical orbiological sensing, would be attached to the nanowire,Fig. 9b. Many chemical or biological agents possessesmagnetic ions. Binding of such an agent A with thereceptor, Fig. 9c, would lead to changes in the localmagnetic field of the nanowire and would alter the polarisedcurrent. A single device could be replicated thousands oftimes with an assortment of receptors to form a microarray.
The aligned receptor array could consist of geneticallymanipulable and combinatorially selective molecular recog-nition units that would be responsive to an agent-receptorbinding event involving a specific reaction of eachsubelement. This would confer absolute specificity for eachindividual reaction based on the particular spin-coupledproperties of that subelement resulting in a unique patternfor all subelement responses taken together. The DMS-based array could also be integrated with Si microelec-tronics for real-time processing of the signals.
DMS materials offer rich choices of organic andbiomolecules for functionalisation which form strong bondson oxide surfaces. Using photo or electron-beam lithogra-phy, multiple electrodes can be patterned on a singlenanobelt, which forms a nanoscale FET on a Si/SiO2
substrate. Owing to the single-crystal quality of thenanobelts, the FETs with B0.05mm channel length shouldbe in the ballistic regime and high FET performance is
expected. Coupled with the large surface-to-volume ratio ofthe nanobelt geometry, these devices are expected to bechemical and biological sensors with extraordinary sensi-tivity. Nanoscale FETs based on Si nanowires have shownpromise as effective biosensors, but oxide nanobelt FETsmay have several advantages over those devices includingdemonstrated bulk electrical sensitivity to molecular bind-ing, rich surface functionalisation choices and structuralflexibility in assembling device arrays [85]. ZnO isparticularly well-suited to this purpose, because of the easewith which long (several mm) nanowires and nanobelts canbe formed, and due to its biocompatibility [85]. We haverecently shown the ability to reproducibly contact individualnanorods for fabrication of MOSFETs and chemicalsensors [86, 87]. An example is shown in Fig. 10.
5 DMS nanostructures
Reduced dimensionality is being explored not only inconventional semiconductors but also in DMS materials aswell. Generally, confined crystals have shown improvementsin the magnetic properties over their thin film or quantumwell counterparts. GaMnN nanowires with diameters in therange of 25–75nm have been reported which show room-temperature ferromagnetism and slightly higher coercivefields than for thin films. Similarly, ZnO:Mn-basedquantum dots have been fabricated which show evidenceof high Tc ferromagnetism [88].
While significant progress has been made in the synthesisand characterisation of thin-film DMS materials, far less isknown about the role of size effects in the transportbehaviour in heterostructures based on these materials,particularly for the wide-bandgap materials. The ability toaffect the band splitting through the control of the size in a
A
S D
a
R
b
c
Fig. 9 Schematic representation of detection of biological mole-cules using a nanowire spin filter (S. von Molnar and J.M. Zavada,private communication)a Spin-polarised electrons passing in nanoscale wire from sourceregion S to detector region Db Bioreceptors R attached to nanoscale wirec Bonding of a single receptor Ri with a bioagent A
quantum dot structure, for example, allows for theexploration of g-factor engineering, similar to the type ofbandgap engineering now routinely used in III–V photonicdevices. Dots can be tailored to give large g factors and thushave a high degree of spin polarisation. In addition, thepresence of piezoelectric effects at strained interfaces in theGaMnN/AlGaN or ZnMgO/ZnO interface can also have alarge effect in these types of nanostructures.
6 Summary and conclusions
There is a strong need for a practical device demonstrationshowing spin functionality at room temperature in a nitride-or oxide-based structure, such as spin LED or tunnellingmagnetoresistance devices [89–91]. The control of spininjection and manipulation of spin transport by externalmeans such as voltage from a gate contact or magneticfields from adjacent current lines or ferromagnetic contactsis at the heart of whether spintronics can be exploited indevice structures and these areas are still in their infancy.There has been tremendous progress in developingGaMnAs, TiO2, SnO2 and other materials, in addition tothe materials we have focused on here, and steadilyimproving Curie temperatures suggest that the control ofsynthesis is much better than even a year ago. The challengeis now to translate this to device embodiments. Inparticular, there is also a need to examine the deviceoperation from a theory viewpoint first, to ensure that
spintronics does in fact have an advantage for that devicefunction.
7 Acknowledgments
The authors are grateful to their colleagues S. von Molnar,P. Xiong, F. Ren, G.T. Thaler, M.E. Overberg, Y.W. Heo,M.P. Ivill, K. Ip, A.F. Hebard, J. Kelly, N. Theodoro-poulou, R. Rairigh, T. Steiner, Y.D. Park, I. Buyanova,W. Chen, C.J. Stanton, R.G. Wilson and J. Kim for theircontinued collaboration on wide-bandgap spintronics. Thework was partially supported by NSF-DMR 0400416 andby the US Army Research Office under grants, numbersARO DAAD 19-01-1-0710 and DAAD 19-02-1-0420 andby AFOSR F49620-03-1-0370. This work was partiallysupported by the Post-doctoral Fellowship Program ofKorea Science & Engineering Foundation (KOSEF).
8 References
1 Wolf, S.A., Awschalom, D.D., Buhrman, R.A., Daughton, J.M., vonMolnar, S., Roukes, M.L., Chtchelkanova, A.Y., and Treger, D.M.:‘Spintronics: a spin-based electronics vision for the future’, Science,2001, 294, p. 1488
2 von Molnar, S.: ‘Spin electronics: from concentrated to dilutedmagnetic semiconductors and beyond’, J. Supercond., 2003, 16,pp. 1–5
3 von Molnar, S., and Read, D.: ‘New materials for semiconductorspintronics’, Proc. IEEE, 2003, 91, p. 715
4 Awschalom, D.D., Loss, D. and Samarth, N. (Eds.).: ‘Semiconductorspintronics and quantum computation’ (Springer, Berlin, 2002)
5 Ball, P.: ‘Meet the spin doctors’, Nature, 2000, 404, p. 9186 Awschalom, D.D., and Kikkawa, J.M.: ‘Electron spin and optical
coherence in semiconductors’, Phys. Today, 1999, 52, p. 337 Awschalom, D.D., Flatt!e, M.E., and Samarth, N.: ‘Spintronics’, Sci.
Am., 2002, 286, pp. 66–738 Baibich, M. et al.: ‘Giant magnetoresistance of (001)Fe/(001)Cr
magnetic superlattices’, Phys. Rev. Lett., 1988, 61, p. 24729 Datta, S., and Das, B.: ‘Electrical analogue of the electrooptic
modulator’, Appl. Phys. Lett., 1990, 56, p. 66510 Jansen, R.: ‘The spin-valve transistor: a review and outlook’, J. Phys.
D, 2003, 36, pp. R289–R30811 Johnson, J.W., Piner, E.L., Vescan, A., Therrien, R., Rajagopal, P.,
Roberts, J.C., Brown, J.D., Singhal, S., and Linthicum, K.J.: ‘12W/mm AlGaN/GaN heterostructure field effect transistors on siliconsubstrates’, IEEE Electron. Dev. Lett., 2004, 25, p. 459
12 Tuan, C. et al.: ‘Epitaxial growth and properties of cobalt-doped ZnOon a-Al2O3 single-crystal substrates’, Phys. Rev. B, 2004, 70, p. 054424
13 Coey, J.D.M. et al.: ‘Donor impurity band exchange in diluteferromagnetic oxides’, Nature Mater., 2005, 4, p. 173
14 Manyala, N., Sidis, Y., DiTusa, J.F., Aeppli, G., Young, D.P., andFisk, Z.: ‘Large anomalous Hall effect in a Si-based magneticsemiconductor’, Nature Mater., 2004, 3, p. 255
15 Toyosaki, H. et al.: ‘Anomalous Hall effect governed by electrondoping in a room temperature transparent ferromagnetic semicon-ductor’, Nature Mater., 2004, 3, p. 221
16 Murakami, X. et al.: ‘Cobalt valence states and origins offerromagnetism in Co doped TiO2 rutile thin films’, J. Appl. Phys.,2004, 95, p. 5330
17 Ohno, Y., Young, D.K., Beschoten, B., Matsukura, F., Ohno, H., andAwschalom, D.D.: ‘Electrical spin injection in a ferromagneticsemiconductor heterostructure’, Nature, 1999, 402, p. 790
18 Feiderling, R., Keim, M., Reuscher, G., Ossau, W., Schmidt, G.,Waag, A., and Molenkamp, L.W.: ‘Injection and detection of a spin-polarized current in a light-emitting diode’, Nature, 1999, 402, p. 787
19 Kioseoglou, X. et al.: ‘Electrical spin injection from an n-typeferromagnetic semiconductor into a III–V device heterostructure’,Nature Mater., 2004, 3, p. 799
20 Oiwa, A., Slupinski, T., and Munekata, H.: ‘Control of magnetizationreversal process by light illumination in a ferromagnetic heterostruc-ture’, Appl. Phys. Lett., 2001, 78, p. 518
21 Schmidt, G., Ferrand, D., Molenkamp, L.W., Filip, A.T., and vanWees, B.J.: ‘Fundamental obstacle for electrical spin injection from aferromagnetic metal into a diffusive semiconductor’, Phys. Rev. B,Condens. Matter, 2000, 62, p. R4790
22 Yu, Z.G., and Flatt!e, M.E.: ‘Spin diffusion and injection insemiconductor structures: Electric field effects’, Phys. Rev. B, Condens.Matter Mater. Phys., 2002, 66, p. 235302
23 Hanbicki, A.T., Jonker, B.T., Itskos, G., Kioseoglou, G., and Petrou,A.: ‘Efficient electrical spin injection from a magnetic metal/tunnelbarrier contact into a semiconductor’, Appl. Phys. Lett., 2002, 80,p. 1240
24 Motsnyi, V.F., Safarov, V.I., De Boeck, J., Das, J., Van Roy, W.,Goovaerts, E., and Borghs, G.: ‘Efficient electrical spin injection froma magnetic metal/tunnel barrier contact into a semiconductor’, Appl.Phys. Lett., 2002, 81, p. 265
nanosensor for biodetection
electrodeelectrodenanorod
a
b
MAIC SEI 15.0kV X3,000 1µm WD 15.3mm
Fig. 10 SEM micrograph of single ZnO(Co) nanorod contactedby ohmic pads (a) and schematic of nanowire chemical sensor (b)
25 Van Dorpe, P., Motsnyi, V.F., Nijboer, M., Goovaerts, E., Safarov,V.I., Das, J., Van Roy, W., Borghs, G., and De Boeck, J.: ‘Highlyefficient spin injection in an MIS LED’, Jpn. J. Appl. Phys. 2, Lett.,2003, 42, p. L502
26 Van Dorpe, P., Van Roy, W., Motsny, V.F., Borghs, G., andDeBoeck, J.: ‘Efficient electrical spin injection in GaAs: A comparisonbetween AlOx and Schottky injectors’, J. Vac. Sci. Technol. A, Vac.Surf. Films, 2004, 22, p. 1862
27 Tanaka, M., Higo, Y., and Sugahara, S.: ‘Spin-polarized tunnelingin fully epitaxial semiconductor-based magnetic tunnel junctions’,J. Supercond., 2003, 16, p. 241
28 Jonker, B.T., Park, Y.D., Bennet, B.R., Cheong, H.D., Kioseoglou,G., and Petrou, A.: ‘Robust electrical spin injection into asemiconductor heterostructure’, Phys. Rev. B, Condens. Matter,2000, 62, p. 8180
29 Park, Y.D., Jonker, B.T., Bennet, B.R., Itzkos, G., Furis, M.,Kioseoglou, G., and Petrou, A.: ‘Robust spin injection across an air-exposed, processed interface’, Appl. Phys. Lett., 2000, 77, p. 3989
30 Park, Y.D., Hanbicki, A.T., Erwin, S.C., Hellberg, C.S., Sullivan, M.,Mattson, J.E., Ambrose, T.F., Wilson, A., Spanos, G., and Jonker,B.T.: ‘A group IV ferromagnetic semiconductor, MnGe’, Science,2002, 295, p. 651
31 Jonker, B.T., Hanbicki, A.T., Park, Y.D., Itskos, G., Furis, M.,Kioseoglou, G., and Petrou, A.: ‘Quantifying electrical spin injection:component-resolved electroluminescence from spin-polarized light-emitting diodes’, Appl. Phys. Lett., 2001, 79, p. 3098
32 Jansen, R., Anil Kumar, P.S., van’t Erve, O.M.J., Vlutters, R., deHaan, P., and Lodder, J.C.: ‘Thermal spin-wave scattering in hot-electron magnetotransport across a spin valve’, Phys. Rev. Lett., 2000,85, p. 3277
33 Anil Kumar, P.S., Jansen, R., van’t Erve, O.M.J., Vlutters, R., deHaan, P., and Lodder, J.C.: ‘Low-field magnetocurrent above 200%in a spin-valve transistor at room temperature’, J. Magn. Magn.Mater., 2000, 214, p. L1
34 Manago, T., and Akinaga, H.: ‘Spin-polarized light-emitting diodeusing metal/insulator/semiconductor structures’, Appl. Phys. Lett.,2002, 81, p. 694
35 Zhu, H.J., Ramsteiner, M., Kostial, H., Wassermeier, M., Sch.onherr,H.-P., and Ploog, K.H.: ‘Room temperature spin injection from Feinto GaAs’, Phys. Rev. Lett., 2001, 87, p. 016601
36 Motsnyi, V.F., De Boeck, J., Das, J., Van Roy, W., Borghs, G.,Goovaerts, E., and Safarov, V.I.: ‘Electrical spin injection in aferromagnet/tunnel barrier/semiconductor heterostructure’, Appl.Phys. Lett., 2002, 81, p. 265
37 Hanbicki, A.T., van’t Erve, O.M.J., Magno, R., Kioseoglou, G., Li,C.H., Jonker, B.T., Itskos, G., Mallory, R., Yasar, M., and Petrou,A.: ‘Analysis of the transport process providing spin injection throughan Fe/AlGaAs Schottky barrier’, Appl. Phys. Lett., 2003, 82, p. 4092
38 Ohno, H. et al.: ‘Electrical field control of ferromagnetism’, Nature,2000, 408, pp. 944–946
39 Eid, K.F. et al.: ‘Exchange biasing of the ferromagnetic semiconductorGa1�xMnxAs’, Appl. Phys. Lett., 2004, 85, p. 1556
40 Ohno, H. et al.: ‘Spontaneous splitting of ferromagnetic GaMnAsvalence band observed by resonant tunneling spectroscopy’, Appl.Phys. Lett., 1998, 73, p. 363
41 Tanaka, M., and Higo, Y.: ‘Large tunneling magnetoresistanceGaMnAs/AlAs/GaMnAs ferromagnetic semiconductor tunnel junc-tions’, Phys. Rev. Lett., 2001, 87, p. 026602
42 Chun, S.H. et al.: ‘Spin polarized tunneling in hybrid metal-semiconductor magnetic tunnel junctions’, Phys. Rev. B, Condens.Matter Mater. Phys., 2002, 66, p. 100408
43 Mattana, R. et al.: ‘Electrical detection of spin accumulation in ap-type GaAs quantum well’, Phys. Rev. Lett., 2003, 90, p. 166601
44 Ohno, Y. et al.: ‘Electrical spin injection in a ferromagneticsemiconductor heterostructure’, Nature, 1999, 402, p. 790
45 Johnston-Halperin, E. et al.: ‘Spin-polarized Zener tunneling inGaMnAs’, Phys. Rev. B, Condens. Matter Mater. Phys., 2002, 65,p. 041306
46 Myers, R.C., Gossard, A.C., and Awschalom, D.D.: ‘Tunable spinpolarization in III–V quantum wells with a ferromagnetic barrier’,Phys. Rev. B, Condens. Matter Mater. Phys., 2004, 69, p. 161305
47 Tang, H.X., Kawakami, R.K., Awschalom, D.D., and Roukes, M.L.:‘Giant planar Hall effect in epitaxial (Ga,Mn)As devices’, Phys. Rev.Lett., 2003, 90, p. 107201
48 Zutic, I., Fabian, J., and Das Sarma, S.: ‘Spin-polarized transport ininhomogeneous magnetic semiconductors: theory of magnetic/non-magnetic p-n junctions’, Phys. Rev. Lett., 2003, 88, p. 066603
49 Samarth, N. et al.: ‘Hybrid ferromagnetic/semiconductor hetero-structures for spintronics’, Solid-State Commun., 2003, 127, pp.173–179
50 Flatte, M.E., and Vignale, G.: ‘Unipolar spin diodes and transistors’,Appl. Phys. Lett., 2001, 78, pp. 1273–1275
51 Flatte, M.E., Yu, Z.G., Johnston-Halperin, E., and Awschalom,D.D.: ‘Theory of semiconductor magnetic bipolar transistors’, Appl.Phys. Lett., 2003, 82, pp. 4740–4742
52 Edmonds, K.W. et al.: ‘High-Curie-temperature Ga1�xMnxAsobtained by resistance-monitored annealing’, Appl. Phys. Lett., 2002,81, p. 4991
53 Ku, K.C.S.J., Potashnik, R.F., Wang, S.H., Chun, P., Schiffer, N.,Samarth, M.J., Seong, A., Mascarenhas, E., Johnston-Halperin, R.C.,Myers, A.C., Gossard, and Awschalom, D.D.: ‘Highly enhancedCurie temperature in low-temperature annealed [Ga,Mn]As epilayers’,Appl. Phys. Lett., 2002, 82, p. 2302
54 Gregg, J.F., Petej, I., Jouguelet, E., and Dennis, C.: ‘SpinelectronicsFa review’, J. Phys. D, Appl. Phys., 2002, 35, p. R121
55 Bandyopadhyay, S., and Cahay, M.: ‘Alternate spintronicanalogue of the electro-optic modulator’, Appl. Phys. Lett., 2004, 85,p. 1814
56 Bandyopadhyay, S., and Cahay, M.: ‘Reexamination of somespintronic field-effect device concepts’, Appl. Phys. Lett., 2004, 85,p. 1433
57 Cahay, M., and Bandyopadhyay, S.: ‘Conductance modulation ofspin interferometers’, Phys. Rev. B, Condens. Matter Mater. Phys.,2003, 68, p. 115316
58 Cahay, M., and Bandyopadhyay, S.: ‘Phase-coherent quantummechanical spin transport in a weakly disordered quasi-one-dimen-sional channel’, Phys. Rev. B, Condens. Matter Mater. Phys., 2004, 69,p. 045301
59 Theodoropoulou, N., Hebard, A.F., Overberg, M., Abernathy, C.R.,Pearton, S.J., Chu, S.N.G., and Wilson, R.G.: ‘Unconventionalcarrier-mediated ferromagnetism above room temperature in ion-implanted (Ga,Mn)P: C’, Phys. Rev. Lett., 2002, 89, p. 107203
60 Egues, J.C., Burkard, G., and Loss, D.: ‘Dutta-Das transistor withenhanced spin control’, Appl. Phys. Lett., 2003, 82, p. 2658
61 Schliemann, J.C., Egues, J.C., and Loss, D.: ‘Non-ballistic field effecttransistor’, Phys. Rev. Lett., 2003, 90, p. 146801
62 Cartoixa, X., Ting, Z.Y., and Chang, Y.C.: ‘A resonant spin lifetimetransistor’, Appl. Phys. Lett., 2003, 83, p. 1462
63 Bandyopadhyay, S.B., Das, S.B., and Miller, A.E.: ‘Supercomputingwith spin-polarized single electrons in a quantum coupled architec-ture’, Nanotechnology, 1994, 5, p. 113
64 Yu, N.J., Shibata, N., and Amemiya, Y.: ‘Boltzmann machine neurondevice using quantum-coupled single electrons’, Appl. Phys. Lett.,1998, 72, p. 3214
65 Bandyopadhyay, S., and Roychowdhury, V.P.: ‘Switching in areversible spin logic gate’, Superlattices Microstruct., 2002, 22, p. 411
66 Bandyopadhyay, S.: ‘Self-assembled nanoelectronic quantum compu-ter based on the Rashba effect in quantum dots’, Phys. Rev. B,Condens. Matter., 2000, 61, p. 13813
67 Bandyopadhyay, S.: ‘Computing with spins: from classical toquantum computing’, Superlattice Microstruct., 2005, 37, p. 77
68 Sato, K., and Katayama-Yoshida, H.: ‘First principles materialsdesign for semiconductor spintronics’, Semicond. Sci. Technol., 2002,17, p. 367
69 Sato, K., Katayama-Yoshida, H.: ‘Material design for transparentferromagnets with ZnO-based magnetic semiconductors’, Mater. Res.Soc. Symp. Proc., 2001, 666, p. F4.6.1; Jpn. J. Appl. Phys., 2001, 39,p. L555
70 Buyanova, I., Izadifard, M., Chen, W., Kim, J., Ren, F., Thaler, G.T.,Abernathy, C.R., Pearton, S.J., Pan, C., Chen, G., Chyi, J., andZavada, J.M.: ‘On the origin of spin loss in GaMnN/GaN LEDs’,Appl. Phys. Lett., 2004, 84, p. 2599
71 Kyrychenko, F.V., Stanton, C.J., Abernathy, C.R., Pearton, S.J., Ren,F., Thaler, G., Frazier, R., Buyanova, I., and Chen, W.M.: ‘Electronspin-relaxation in strained InGaN/GaN quantum wells’. 27th Int.Conf. Phys. Semicond., Flagstaff, AZ, USA, July 2004
72 Wakano, T., Fujimura, N., Morinaga, Y., Abe, N., Ashida, A., andIto, T.: ‘Magnetic and magneto-transport properties of ZnO:Ni films’,Physica E, 2001, 10, p. 260
73 Jalbout, A.F., Chen, H., and Whittenburg, S.L.: ‘Monte Carlosimulation on the indirect exchange interactions of Co-doped ZnOfilm’, Appl. Phys. Lett., 2002, 81, p. 2217
74 Szczytko, J., Mac, W., Stachow, Twardowski, A., Becla, P., andTworzydlo, J.: ‘The s,p-d exchange interaction in GaAs heavily dopedwith Mn’, Solid State Commun., 1996, 99, p. 927
75 Esaki, L., Stiles, P.J., and von Moln!ar, S.: ‘Magnetointernal fieldemission in junctions of magnetic insulators’, Phys. Rev. Lett., 1967,19, p. 852
76 Siegrist, T., von Moln!ar, S., and Holtzberg, F.: ‘Magnetic fielddependence of Fowler–Nordheim tunneling in Cd1�xMnxTe’, Appl.Phys. Lett., 1985, 47, p. 1087
77 Couturier, G., and von Moln!ar, S.: ‘Schottky barrier with thedegenerate dilute magnetic semiconductor Cd1�xMnxSe’, J. Appl.Phys., 1987, 62, p. 3253
78 Thompson, W.A., Holtzberg, F., McGuire, T.R., Petrich, G.: inGraham, D.C., Rhyne, J. (Eds.).: ‘Magnetism andmagnetic materials’.AIP Conf. Proc., New York, USA, 1971, Vol. 5, p. 827
79 Furdyna, J.K., and Kossut, J. (Eds.).: ‘Diluted magnetic semiconduc-tors’, Semicond. Semimet., 1988, 25, p. 296
80 Tedrow, P.M., and Meservey, R.: ‘Spin polarization of electronstunneling from films of Fe, Co, Ni, and Gd’, Phys. Rev. B., SolidState, 1973, 7, p. 318
81 Moodera, J.S., Nowak, J., and van de Veerdonk, R.J.M.: ‘Interfacemagnetism and spin wave scattering in ferromagnetic-insulator-ferromagnetic tunnel junctions’, Phys. Rev. Lett., 1998, 80, pp.2941–2944
82 LeClaire, P., Ha, J.K., Swagten, H.J.M., Kohlhepp, J.T., van de Vin,C.H., and de Jonge, W.J.M.: ‘Large magnetoresistance using hybridspin filter devices’, Appl. Phys. Lett., 2002, 80, p. 625
83 Parker, J.S., Watts, S.M., Ivanov, P.G., and Xiong, P.: ‘Spinpolarization of CrO2 at and across an artificial barrier’, Phys. Rev.Lett., 2002, 88, p. 196601
84 Freitas, P.P., Ferreira, H.A., Graham, D.L., Clarke, L.A., Amaral,M.D., Martins, V., Fonseca, L., and Cabral, J.S.: ‘MagnetoresitiveDNA chips’, in Johnson, M. (Eds.).: ‘Magnetoelectronics’ (Elsevier,Oxford, 2004), Chap. 7
