Spintronics PDF
Spintronics PDF
Spintronics PDF
2357
Reviews
Spintronics and spintronics materials
V. A. Ivanov,a T. G. Aminov,a V. M. Novotortsev,a and V. T. Kalinnikovb S. Kurnakov Institute of General and Inorganic Chemistry, Russian Academy of Sciences, 31 Leninsky prosp., 119991 Moscow, Russian Federation. Fax: +7 (095) 954 1279. E mail: vmnov@igic.ras.ru, aminov@igic.ras.ru, ivanov@ostrov.net bI. V. Tananaev Institute of Chemistry and Technology of Rare Elements and Mineral Raw Materials, Kola Research Center of the Russian Academy of Sciences, 14 ul. Fersmana, 184209 Apatity, Murmansk Region, Russian Federation. Fax: +7 (815 55) 7 9414. E mail: office@chemy.kolasc.net.ru
The review concerns the fundamentals of spintronics (spin transport electronics). The material covers spin spin interactions and spin relaxation in semiconductors as well as spin and spin injection related effects in the condensed matter. Examples of promising spintronic de vices are given, requirements for spintronic materials are formulated, methods of synthesis of spintronic materials are described, and the physicochemical properties of some materials are characterized. Organic spintronic materials are briefly outlined and the state of the art in the field of research on inhomogeneous magnetic semiconducting materials possessing high tem perature ferromagnetism is described. The emphasis is placed on the chemical bonding and electronic structure of magnetic impurities in semiconductors, consideration of the nature of ferromagnetism, and on the character of exchange interactions between localized spins in novel spintronic materials. Key words: spintronics, spin transport electronics, ferromagnetic semiconductors, dilute magnetic semiconductors, semimagnetic semiconductors, magnetic materials.
aN.
Introduction Spintronics is a new, quickly developing field of sci ence and technology, which deals with relationhips re sponsible for specific features of the spin interactions in metals, semiconductors doped with transition or rare earth elements, and heterostructures that ensure unique properties of these materials.*
The main avenues of the development of spintronics are (i) fabrication of magnetic nanostructures including novel materials, thin films, heterostructures, and multi functional materials; (ii) research on magnetism and spin control of magnetic nanostructures, theory of ferromag netic exchange in dilute magnetic semiconductors (DMS), tunneling effects and spin injection, ans spin transport and detection of magnetism; (iii) magnetoelectronics and
* The term spintronics was proposed in 1998 in a joint press release of Bell Labs and Yale University (USA), which defined the problem of design of devices for information storage by manipulating atoms of matter using electron spin encoded bits. Researchers at the Defence Advanced Research Project Agency (DARPA, USA) defined spintronics as spin transport electronics. Some other definitions of spintronics are as follows: this is (i) a science for which mutually consistent behavior of the electron charge and electron spin is of crucial importance;1 (ii) "electron spin based electronics, in which information is transmitted by employing the electron spin rather than electron charge, which creates prerequisites for design of a new generation of devices that combine conventional microelectronics and spin dependent effects";2 (iii) a "science of manipulation of electric current in semiconductors and heterostructures by changing electron and nuclear spin orientationss in magnetic and electric fields";3 and (iv) a "new branch of microelectronics where the electron spin and electron charge act as an active element for information storage and transmission, for integrated circuits and functional chips, multifunctional magneto opto electronic devices. Published in Russian in Izvestiya Akademii Nauk. Seriya Khimicheskaya, No. 11, pp. 22552303, November, 2004. 1066 5285/04/5311 2357 2004 Springer Science+Business Media, Inc.
2358
Ivanov et al.
devices employing the giant magnetoresistance (GMR) effect,4,5 tunneling devices, semiconductor heterostruc tures for spin injection, spin transport and detection, and pulsed ferromagnetism; (iv) magneto optical properties of dc magnetic semiconductor heterostructures and time resolved experiments, optical spin injection and detec tion, optically induced ferromagnetism, ultrafast mag neto optic switches; and quantum information transmis sion; (v) pattern recognition; imaging and metrology in cluding magnetic pattern recognition and anomalous Hall effect; and (vi) instrument engineering and applied studies. Semiconductor spintronics is a science dealing with the coexistence of the charge and spin degrees of freedom in doped semiconductors and nanostructures, the nature of ferromagnetism in and methods of synthesis of these materials, and fabrication of devices employing the spin characteristics in semiconductors. Magnetic semiconductors were obtained after the dis covery of materials based on type IIIV matrices (mainly (Ga,Mn)As) because they not only retain the semicon ducting properties but acquire the ferromagnetic proper ties, being lightly doped with magnetic impurities. From this viewpoint the known magnetic semiconductors (chal cogenides of rare earth metals, magnetic chalcospinels) would be well to call semiconducting magnets because these compounds (in the case of with stoichiometric com positions) are ferromagnets with intrinsic magnetic sub lattices. Synthesis of the type (III,Mn)V DMS, namely, (In,Mn)As alloys68 and (In,Mn)As/(Ga,Al)Sb hetero structures9 was basic to the breakthrough in technology of preparation of novel ferromagnetic semiconductors with high Curie temperatures (TC). At present, to distinguish between the DMS* which have mostly diamagnetic matrices and conventional mag netic semiconductors, the latter are called concentrated magnetic semiconductors (CMS). Doping them with other elements cannot cause significant changes in those in trinsic physicochemical characteristics of CMS that differ from the properties of the known semiconductors. At the same time, distinctions between the crystal lattices of CMS and the known semiconductors imposes severe restric tions on the possibility for CMS to be used in semicon ductor devices. It is assumed that spintronic devices will employ nonequilibrium spin density created in semiconductors, manipulation of spin orientations by external fields, and detection of the spin state thus formed.10 In rela tion to relaxation processes, degradation mechanisms are of importance for spintronics (and microelectronics).
* In this review the terms "dilute magnetic semiconductors" and "high temperature ferromagnetic semiconductors" cover novel magnetic semiconductors; the discovery of these materials gave rise to a new branch of science and technology, a new branch of electronics, that is, spintronics.
However, spin currents induce a much lower degrada tion of integrated circuits compared to electric currents capable of inducing rearrangement of the atomic struc ture of integrated circuits at high current densities. There fore, spintronic devices can act as counterparts or in some cases be used instead of conventional electronic devices. If traditional electronics employs the charge degrees of freedom of substances, spintronics deals with the spin (magnetic) characteristics of materials. In contrast to elec tronic devices, spintronic devices employ the dependence of transfer of the current carriers on the electron spin orientation. A specific feature of spintronics is the joint action of the electron charge and electron spin; the smaller the object, the more pronounced the effect. A magnetic field induced electron spin flip requires a much lower energy and occurs faster than the displacement of elec tron charges by an electric field. Therefore, one can ex pect that manipulation of spin states will permit in the future fabrication of ultrasmall logic elements and ultrafast computer chips of high information capacity and low power consumption. The subject of nanospintronics11,12 is research on the behavior of spins in mesoscopic systems and on the indirect spin interactions in nanoparticles that has been poorly studied as yet. Engineering spintronics employs the spin degrees of freedom of the metals used in computer engineering (hard disks and MRAM chips for personal computers). At present, fabrication of semicon ductor structures doped with magnetic impurities is placed on the agenda. The electronic structure of such materials is determined by the chemical bonds while the magnetic structure is governed by strong interactions between elec trons of the transition metal or rare earth metal (REM) impurity atoms. Yet another aspect of computer engineering can also be related to spintronics. The case in point is that opera tion of computers in the binary number system requires two fast switchable states corresponding to "0" and "1" that must be stable in time and resistant to thermal fluc tuations. These functions could be operated by an elec tron in a two level system (e.g., a diatomic molecule) by hopping between atoms. At present, hundreds and thou sands of electrons per operation are "consumed" in elec tronic devices. Quantization of electron spin (with con ventional "spin up" and "spin down" orientations) indi cates that the spin is a natural logic cell with two states corresponding to "1" and "0". In this case, a logic opera tion can be encoded using a single electron. Therefore, the use of electron spin is much more beneficial com pared to the electron charge. Intensive basic research into the physical and chemi cal processes in solids with allowance for the spin degrees of freedom was carried out by researchers working in do mestic institutes before introducing the term "spintronics" (see, e.g., Refs. 1331). Considerable advances have been
2359
made in molecular design (including proteins and spin systems), design of molecule based ferromagnets;32 and investigations of the chemical, electrical, and magnetic properties of nanoclusters, nanotubes, and nanowires.33 It was established that the spin dynamics is quickly re sponsive to the molecular dynamics of an elementary chemical act.26 Spin chemistry emerged, which studies nuclear spin related spintronic effects27 and has some fea tures in common with nuclear spintronics that is in the early stage of development. The available reviews mainly concern new effects and applications of spintronic devices,34,35 ferromagnetic al loys which have found some practical applications, and some novel materials with the spin degrees of free dom.3647 Some avenues of progress in spintronics have been elucidated.4855 Coexistence of the spin and charge, the ferromagnetic and semiconducting properties in the same material makes magnetic semiconductors of prime importance for spintronics applications; however, there is a lack of reviews covering the field. Here, we try to bridge the gap. At first, we briefly outline the fundamentals of spin tronics and consider spin spin interactions and spin re laxation in semiconductors as well as spin related and spin injection related effects in condensed matter. We present examples of promising spintronic devices; formu late requirements for spintronic materials, and character ize the physicochemical properties of certain spintronic materials and relevant methods of synthesis. Organic spintronic materials are also touched upon and research in the field of inhomogeneous magnetic semiconduct ing materials exhibiting high temperature ferromag netism is discussed. Particular attention is paid to prob lems of the chemical bonding and electronic structure of magnetic impurities in semiconductors, the nature of ferromagnetism, and the character of exchange interac tions between localized spins in the novel spintronic ma terials. 1. Band structure and spin interactions in spintronic materials 1.1. Band structure and impurity states The nature of the charge carriers and spin carriers is closely related to the filled energy bands of semiconduc tors, namely, the valence band (this band is generally filled) and the conduction band (usually, this band is empty in p type semiconductors) separated from the va lence band by an energy gap (1.42 eV in GaAs, 2.26 eV in GaP, and 3.2 eV in GaN at 300 K). The electronic struc ture of semiconductors is governed by the chemical bonds. In semiconductors with the diamond, zinc blende or wrtzite crystal structures these are the hybrid sp bonds involving valence electrons, which are directed toward
the vertices of a tetrahedron, being stronger than the s and p bonds.56 For two atoms in the unit cell, the non degen erate bonding (antibonding) combination of s orbitals and the three fold degenerate bonding (antibonding) combi nation of p orbitals give rise to the valence band (conduc tion band). Among the bonding and antibonding states of the sp3 orbitals in an AnBVIIIn type semiconductor (B is a pnictogen, chalcogen, halogen), the bonding p orbitals of the anion sublattice are mainly near the top of a wide valence band, while the antibonding s orbitals of the cat ion sublattice are near the bottom of the conduction band. In semiconductors, the spin orbit interaction (see be low) removes the six fold energy degeneration of the bonding combinations of the p orbitals in the valence band, shifting doubly degenerate states with the total an gular momentum j = 1/2 down from the heavy hole states with j = 3/2 by hundreds of meV (0.35 eV for GaAs). The four fold degenerate energy states with j = 3/2 form the heavy hole band with the projections of the angular mo mentum jz = (3/2) and the light hole band with the projections jz = (1/2). The heavy hole band is of par ticular importance for the establishment of indirect ex change interaction between magnetic impurities in DMS and related isoelectronic systems. The valence band is characterized by an enhanced electron density in the interatomic space, whereas in the conduction band the electron density is localized on atoms. The width of the energy gap between these bands is in fact determined by the bond dissociation energy or by the lowest energy required for an electron to go from the bound state to the conduction band. There are two types of charge carriers in semiconductors, called con duction electrons (electrons in the conduction band) and "holes" (missing electrons in the valence band). Holes also have spins, because all spins in the filled valence band are compensated and removal of an electron means the appearance of a spin (vice versa, removal of an elec tron causes the appearance of a positively charged hole). At low temperatures, the conductivity of semiconductors is mainly determined by the electrons and holes supplied by impurity atoms, defects, dislocations, etc., having their impurity levels in the energy gap. The donor or acceptor character of an impurity is determined by its higher or lower ionization potential compared to the ionization po tential of the replaced atom. In type IIIV compounds, substitutional donor impurities for the anion sublattice are atoms of Group VI and VII elements and nitrogen atoms which have rather deep intrinsic s levels. A semi conductor with donor impurities is called an n type semi conductor (here, electrons are current carriers), while a semiconductor with acceptor impurities is called a p type semiconductor (positively charged holes are current car riers). In a non degenerate impurity semiconductor, ei ther donor electrons or acceptor holes are spatially local ized near corresponding impurities. The radius of a cur
2360
Ivanov et al.
rent carrier (or the radius of wave function localization) is given by the expression
2 r0 = rB(m/m*) = (h )/(m*e2),
where m* is the effective mass of the current carrier, is the dielectric permeability of the material, rB is the ground state radius of hydrogen atom (so called Bohr radius), m is the mass of electron, and h = h/(2) (h is the Planck constant). Typical semiconductors are characterized by = 1020, m* = 101m, and r0 = (100200)rB; the last mentioned parameter is much larger than the lattice con stant a rB. This means that electrons in semiconductors (in contrast to, e.g., transition metal oxides) cause weak lattice distortions and the impurity potentials can be con sidered as point potentials because the range of their ac tion is much shorter than the radius of wave function localization. However, already at a low impurity concen tration, x a3/r03 per lattice site, the wave functions of current carriers are overlapped, thus forming an impurity band and the non degenerate impurity semiconductor becomes a degenerate one.57 Impurity states in the electron energy spectrum of a semiconductor can be divided into shallow (hydrogen like) and deep states depending on whose chemical prop erties (of the semiconductor matrix or the dopant, re spectively) influence them to the greatest extent. Posi tions of shallow impurity levels are determined by the effective mass of the charge carrier and the dielectric con stant of the semiconductor, i.e., by the chemical nature of the matrix, and the properties of such impurities manifest themselves only as a weak distortion of the central cell (in addition to the charge). Shallow donor (acceptor) impu rity levels lie near the bottom (top) of the conduction band (valence band) and can be correctly described by the effective mass theory. In type IIIV compounds, Group IIII metals that are chemically similar to the replaced cations do not create deep levels in the forbid den energy band. On the contrary, positions of deep im purity levels are governed by the chemical nature of the impurities and specific features of their non filled elec tron shells. In type IIIV semiconductors, transition metal impurities in the cation sublattice almost without exception create deep levels. In the theory of deep isoelectronic impurity levels in semiconductors the forbidden band is much narrower than the allowed bands58 and the spherically symmetrical im purity potential gradually decreases at distances of the order of the lattice constant. In this case, creation is pos sible of virtual heavy hole states in the conduction band and of multiply charged levels with the formation of nega tively charged impurity centers. The theory58 of deep im purity levels established the principles of classification and the properties of the states localized on such levels. In creating deep levels an important role is played by the states of the non filled impurity electron shells.59,60
Deep impurities cannot be described using the ap proaches suitable for the description of shallow impurities (KohnLattinger and KosterSlater models), according to which shallow levels are created as a result of potential scattering of the band carriers by the substitutional poten tial. Different symmetries of the valence band and the conduction band in the semiconductor predetermines their different contributions to the impurity states, which is ignored by the KosterSlater model that implies equal involvement of carriers from the allowed bands in scatter ing by the impurity with a real symmetry. Deep levels are due to resonant scattering of the band electrons by the d levels of the non filled shells.61,62 For transition metal and REM impurities in semiconductors, an important role is also played by the inter electron interaction. 1.2. Spin interaction in semiconductors Spin interaction in semiconductors can be divided into several types. Dipole dipole interaction or spin spin coupling is a di rect interaction between the spin magnetic moments of electrons, which is proportional to the product of the magnetic moments and inversely proportional to the cube of the distance between them. Being proportional to (v/c)2 (v is the speed of electron and c is the speed of light), the dipole dipole coupling is too weak to play a significant role in semiconductors. The energy of the dipole dipole interaction between two electrons occupying adjacent lat tice sites is about 1K. The energy of the spin spin cou pling is comparable in order of magnitude with the low energy of the interaction between the spin magnetic mo ment of an electron and the anisotropy field. However, it is the weak spin spin coupling that is of great importance for the useful properties of some magnetic semiconduc tors. In DMS the direct spin spin coupling plays so insig nificant role that the magnetization of DMS can easily be changed by applying experimentally attainable external magnetic fields. Relatively weak magnetic fields affect the characteristics of DMS to a greater extent compared to the parameters of normal magnetic semiconductors. Spin orbit interaction is the interaction between the spin magnetic moment of an electron and the mag netic moment of the orbital motion of the electron. The spin orbit interaction is due to the fact that an electron in an electric field of strength traveling at speed v E = grad(r) = (r /r)(/r) (e.g., in the field of an atomic nucleus with potential (r)), "sees" the magnetic field with the magnetic induction
(c is the speed of light) proportional to the angular mo mentum of the electron l = r mv . In this magnetic field the electron with the magnetic moment = g B s ac
2361
to the spin orbit interaction, the spin of a moving elec tron or hole can be influenced by both static and alternat ing external electric fields. In semiconductors, the spin orbit coupling is several orders of magnitude stronger than in atoms or metals; therefore, it can manifest itself even at thermal velocities of electrons. The spin orbit coupling becomes stronger with an increase in the charge of atomic nucleus; its energy is 0.04 and 0.29 eV in Si and Ge semiconductors, respectively. Often, this interaction is responsible for spin relaxation and provides an interplay between the transport of current carriers and spin related phenomena. The spin orbit interaction relates the spin coordinates of an electron to its spatial coordinates and produces an effective intracrystalline magnetic field (mag netic anisotropy field), because the orbital motion of a particle is related to the crystallographic directions. A particular case of the spin orbit interaction for two dimensional systems with asymmetrical potential was re ported.20 Such a potential is induced in the channel of a silicon field effect transistor. In this case, the shape of the potential can be manipulated by applying a voltage to the transistors base, thus varying the strength of the spin orbit coupling. Strain produced in the crystal lattice of supported DMS layers also affects the spin orbit cou pling. Depending on the type of the strain produced, that is, compression or tension, the DMS magnetization (fer romagnetic easy magnetization axis) lies either in the plane of the ferromagnetic layer or is parallel to the growth direction of this layer, i.e., the magnetic anisotropy field in DMS also depends on the lattice distortion. Exchange interaction is a part of inter electron Cou lomb interaction, which depends on permutation of two electrons following the Pauli exclusion prin ciple: Js 1s 2. Such a "quantum force", which depends on the mutual spin orientation, originates from different orbital motions of two electrons that tend to form elec tronic configurations either with distant electrons (prob ability for both electrons to be in close proximity to each other is low) or with close located electrons (here, this probability is high). In this case, the total spin of the electron pair, S = s 1 + s 2, equals either 1 (spin triplet) or 0 (spin singlet). The exchange energy (integral) J equals a halved energy difference between these two electronic configurations. The intensity of the exchange interaction is somewhat weaker than the halved strength of electro static Coulomb interaction. If the electron shell of an atom contains several electrons (d and f shells of transi tion metal atoms and rare earth elements), the Coulomb repulsion between them is minimum in the case of aligned spins, i.e., the exchange integral is positive, JH > 0 (intra atomic Hund exchange). In this case the electron con figurations tend to have the maximum possible total spin
(the first Hund rule). For the same reason in an isolated atom the spins of the s and d electrons of the atomic shell are also aligned. The appearance of a magnetic impurity makes the situation in the non magnetic metal more com plicated, namely, the spin orientation of the s conduction electron can be parallel or antiparallel to the spin orienta tion of the d impurity electron. The result depends on the intra atomic repulsion between the d electrons, which is characterized by the AndersonHubbard parameter U, and on position of the impurity level in the energy band of the metal; creation of uncompensated magnetic moment is facilitated at a low density of conduction electrons and low degree of bond covalency.63 For covalent bonding to occur, the energetically favorable situation must involve orbital motion of electrons with the maximum electron density in the region between positively charged nuclei (J < 0), so covalent bonding produces a spin singlet. As the bond covalency in M3+AnM3+ and M4+AnM4+ (An is an anion) increases in the order of anions F, O2, S2, Se2, the exchange integral J also increases. Mutual spin orientation and, eventually, ferromagnetism are gov erned by the exchange interaction of the atoms of the metal impurity Mn+, which, however, is not responsible for the total spin orientation with respect to the crystallo graphic axes. Degeneration of the total magnetization of the crystal relative to crystallographic axes is partially re moved by the spin orbit interaction. Combined with the Pauli exclusion principle, the kinetic energy of electrons tends to destroy the spin order; the temperature at which the material no longer possesses ferromagnetic properties is said to be the Curie temperature. In semiconductors, direct exchange interaction between delocalized electrons or holes plays an insignificant role owing to the low den sity of charge carriers and abrupt (exponential) weaken ing of the exchange interaction with distance. The exclu sions are a group of CMS with rather low Curie tempera tures and the semiconductorferromagnetic interfaces that are of importance for spin injection phenomena. In some cases, an exchange interaction between localized d or f electrons and conduction electrons (sd interaction characterized by the exchange parameter Jsd) or valence holes (pd interaction characterized by the exchange pa rameter Jpd) is phenomenologically introduced for mag netic semiconductors.64 Owing to peculiarities of the elec tron density distribution over the semiconductor energy bands, electrons move along magnetic cations, whereas holes move along non magnetic anions. Therefore, ferro magnetic exchange between the localized electrons and conduction electrons must be stronger than their antifer romagnetic exchange interaction with holes in the va lence band: Jsd > |Jpd|. Without dwelling on specific fea tures of the structure of chemical bonds between impuri ties and the semiconductor matrix, note that, phenom enologically, these exchange interactions in magnetic
2362
Ivanov et al.
semiconductors are treated as contact interactions cen tered on magnetic atoms. Indirect spin interactions. In the crystal, direct spin interaction between distant atomic cores (electronic con figurations), e.g., magnetic impurity ions in semiconduc tor is negligible. The spin interaction between such ions occurs via disturbance of electron states in the atoms shared by these ions. If itinerant electrons are localized on, e.g., non mag netic anions (F, O2, S2, Se2, etc.), being involved in the chemical bonding between identical magnetic ions M (M3+AnM3+ and M4+AnM4+ ions in perovskites LaMO3 and CaMO3 or Cr3+FCr3+ ions in CrF3), the orbitals of electrons localized on the magnetic ions are overlapped. In this case, an electron can be trans ferred from one ion to another ion with the kinetic energy t. In the case of semiconductors doped with magnetic impurities the hopping integral t appears due to the direct overlap of the impurity orbitals (wave functions). Having arrived at the filled orbital of another ion, an electron with antiparallel spin orientation acquires an additional energy of electrostatic repulsion, U >> t, and the energy of electron pair is governed by competition between the ki netic energy of electron transfer and the repulsion energy between electrons in the same orbital:
E = 2 t + 2U
( is the overlap of the orbitals of localized electrons). Minimization gives the energy necessary for spin disor dering, E = (t2/U). In particular, two localized elec trons are stabilized with antiparallel spin orientations and the antiferromagnetic exchange integral is
J = (4t2/U ) < 0.
This indirect exchange interaction is called (kinetic) superexchange, or Anderson superexchange. If the atom of a main group element is replaced by an atom of other main group element with a differ ent valence (e.g. La3+ A2+ in (La,A2+)MnO3), the magnetic manganese ions become mixed valence ions. Electrons of the Mn atoms are now delocalized and can move between distant magnetic ions through the anion, e.g ., as follows: Mn3+ O 2 Mn 4+ and Mn4+ O2 Mn3+.65 Such a "hopping" electron with a given spin orientation avoids tunneling toward a site occupied by an electron with antiparallel spin orienta tion, due to strong repulsion U, which is important in the case of antiferromagnetic superexchange. Then, compe tition between a low kinetic energy of electron tunneling between cations (t) and a strong Hund intra atomic ferro magnetic exchange between localized and delocalized electrons (JH > 0) stabilizes aligned cation spins. Indeed, an electron can execute free motion between ions, thus reducing the total energy of the system, only if the spins of
all ions are aligned. This type of ferromagnetic ordering of cations with TC t is due to two transitions involving the non magnetic anion and is therefore called double ex change.65 In fact, the double exchange interaction ex changes different valence states (Mn3+ Mn4+ in 2+ (La,A )MnO3); therefore, it occurs in mixed valence compounds. The mechanism of double exchange was re peatedly refined.66,67 Sometimes, this type of magnetic exchange between two different valence ions is called Zener (ferromagnetic) exchange, though it was proposed earlier by Frhlich and Nabarro for nuclear ferromag netism.68 Long before Zener, mutual biasing of the local ized and delocalized (itinerant) electrons was pointed out by Vonsovsky6971 who considered indirect exchange of establishment of ferromagnetism of localized electrons involving itinerant electrons in the case where all non magnetic ions in the crystal lattice are replaced by atoms of transition metals or rare earth elements (sd model) and, similarly to Zener,65 calculated changes in the con ductivity on transition to the ferromagnetic phase. The phenomenological sd model was also used by Zener.72 The Zener exchange65 is a particular case for double ex change studied earlier by Vonsovsky,70,71 namely, the limiting case of a very strong Hund intra atomic ex change JH >> t. Because of this, it would be more cor rect to call the double exchange mechanism7072 the VonsovskyZener mechanism. This mechanism was ex tended to ferromagnetic semiconductors with the magnetic sublattice and to antiferromagnetic semiconductors.7375 The superexchange and double exchange mechanisms were used to describe a number of magnetic semiconduc tors with intrinsic magnetic ion lattices (i.e, concentrated magnetic conductors).64 Superexchange responsible for antiferromagnetism oc curs between magnetic ions with different spin orienta tions and equal charges and removes the energy degen eracy of the two electron virtual state with high energy U. Superexchange induces antiferromagnetism in semicon ductor transition metal compounds with relatively low Neel temperatures TN t2/U. Double exchange respon sible for ferromagnetism occurs between differently charged ions involving delocalized electrons and removes the energy degeneracy of real states due to transition via a virtual anion state. Therefore, the double exchange mechanism is inapplicable to the description of ferromag netism in semiconductors doped with magnetic impuri ties with integer valences. In spite of the fundamental difference between mixed valence CMS and integer va lence DMS the latter are often studied using the double exchange mechanism (see., e.g., Refs 7679). Some authors postulate the existence of two exchange contact interactions between the impurity d electrons and itinerant s and p electrons. Because of the density distri bution of the delocalized carriers over the conduction band and the valence band this exchange must be ferro
2363
magnetic in character in the conduction band (Jsd > 0) and antiferromagnetic in character in the valence band (Jpd < 0). The fitting character of these interactions led to discrepancies between the estimates of the exchange in teraction Jpd in the best studied DMS, (Ga,Mn)As. They ranged from 4.5 (see Ref. 80), 3.3 (see Ref. 81), and 2.5 eV (see Ref. 82) to much smaller values, 1.2 (see Ref. 83), 1.0 (see Ref. 84), and 0.6 eV (see Ref. 86) (negative sign corresponds to antiferromagnetic exchange interaction of the d and p electron spins). The scatter in the data depends on which methods were used to calcu late Jpd from experimental data on (Ga,Mn)As in differ ent phenomenological models or which "first principle" numerical methods were employed ("first principles" based on density functional were analyzed in Ref. 86). The dif ference between estimates of the magnitude and sign of the parameter of ferromagnetic exchange interaction be tween localized d electrons and delocalized s conduction electrons (Jsd 101 eV) is much smaller. An unjustified use of the double exchange mechanism led77 to prediction of high TC only in p type DMS, which contradicts facts, namely, high temperature ferromag netism in some n type magnetic semiconductors, e.g. (Ga,Mn)N, and high temperature magnetic semiconduc tors similar to (CdGe,Mn)P2. Recent cyclotron resonance experiments* with DMS in magnetic fields of up to 500 T, carried out taking p type In1xMnxAs films (0 < x < 2.5) grown by molecular beam epitaxy on GaAs substrates as examples, indicated the absence of double exchange in DMS. When considering magnetic impurities in semicon ductors, it is of great importance to allow for the tunnel ing energy V of the impurity d electron to the band of itinerant carriers and back to another impurity atom. The parameter V is related to the degree of hybridization of the localized and delocalized electrons, being in fact a mea sure of covalency of the bond between them, i.e., the hybridization parameter is the one electron energy. The magnitude of the hybridization parameter of the d impu rity electrons and itinerant p holes (Vpd > 1 eV) in DMS and HTFS is comparable with the width of the heavy hole band. It is important that tunneling occurs with conserva tion of the spin projection. If we introduce the density of states of itinerant electrons () = N/, where N is the number of electron states in the energy interval , the quantity ()V gives the proportion of the conduction electrons in the energy interval (, + ), which are polarized by the electrons with the same spin projection localized on the impurity. Double tunneling from the impurity to the band and from the band to another impu
* Y. H. Matsuda, G. A. Khodaparas, M. A. Zudov, J. Kono, Y. Sun, F. V. Kyrychenko, G. D. Sanders, C. J. Stanton, N. Miura, S. Ikeda, Y. Hashimoto, S. Katsumoto, H. Munekata, http://arXiv.org/cond mat/0404635.
rity (as follows: d1 p d2) in DMS and HTFS governs coupling between distant impurities with the transfer en ergy proportional to V2 (dependence on the distance is omitted), which is of importance for occurrence of kine matic exchange interaction. Hyperfine coupling between electron and nuclear spins is, similarly to the spin orbit coupling, relativistic in na ture. In the case of semiconductors it provides an addi tional, as compared to metals, possibility of manipulation of electron spin by not only external magnetic field but also the field of atomic nuclei present in the semiconduc tor lattice. Overhauser87,88 described interaction between the electron and nuclear spins in condensed matter start ing from contact interaction between them in isolated atoms (Fermi hyperfine interaction). It can be written in the form
,
where (r ) is the Dirac delta function, summation is performed over positions r i of all atomic nuclei with the spins Ii, g is the gyromagnetic ratio for electron, and A is the hyperfine coupling (HFC) constant. This interaction is equivalent to the existence of a long lived strong mag netic field B n (Overhauser field) exerted on the electron spin S by the spin polarized nuclei distributed over some domain in the material. Semiconductors are characterized by rather long nuclear spin relaxation times (102103 s at helium temperatures), which much exceed the electron spin relaxation time (~107 s). This means that nuclear depolarization is a slower process compared to electronic depolarization. The Overhauser field can be very strong, its magnitude depends on the type of atoms (via the HFC constant A) and on the degree of nuclear spin polariza tion.8993 For GaAs semiconductor, the HFC constant is negative and the Overhauser field B n tends to polarize electron spins in the same way as nuclear spins, reaching a value of 5.3 T at 100% nuclear polarization. In connection with importance of the hyperfine cou pling between the electron and nuclear spins, nuclear spintronics can be considered as a separate branch.* The nuclear spin can be used instead of electron spin. In prin ciple, it is possible to fabricate a device in which informa tion is transmitted to nuclei by light via electrons. The information capacity of such a nuclear memory will be many orders of magnitude higher than and faster than conventional semiconductor based memory. Electronics, photonics, and magnetism can give rise to spin photonics in order to design devices operating at teraHertz fre quencies.
* I. D. Vagner, Nuclear Spintronics: Quantum Hall and Nano Systems, http://arXiv.org/cond mat/0403087.
2364
Ivanov et al.
1.3. Spin relaxation mechanisms The most important problem in studies of spin phe nomena in the condensed matter is electron spin relax ation (in other words, spin decoherence) generally inter preted as a result of the action of fluctuating magnetic fields.10 Usually, these are the effective fields originating from the spin orbit and exchange interactions. The fluc tuating magnetic fields are characterized by their ampli tudes (more exactly, root mean square amplitudes) and correlation times, c (during the time interval c the field can be considered constant). In some cases, it is conve nient to use the mean frequency, , of spin precession in the fluctuating field instead of the field amplitude. Then spin relaxation can be imagined as follows. An electron spin precesses about a certain random direction of the effective magnetic field with the characteristic frequency over the characteristic time c. Then, both the direction and magnitude of the magnetic field are changed and the spin now precesses about a new direction. Finally, after a number of random changes the electron "forgets" the ini tial direction of spin orientation. Real fate of the electron spin in the semiconductor is governed by the dimension less parameter c, i.e., the characteristic angle of rotation of the precessing spin and the correlation time. Usually, the angle of precession is small (c << 1) and the spin orientation experiences a slow angular diffu sion. During the time interval the electron will be in volved in /c independent random changes (the angle of precession squared in each step is (c)2) and the total angle of precession squared equals (c)2t/c. The spin relaxation time, s, is defined as the time taken for the angle of precession squared to reach a value of the order of a unity, s 1/(2c). Otherwise, at c >> 1, the electron spin will execute many rotations about the magnetic field direction during the correlation time. The time averaged (over the time interval of the order of 1/) spin projection perpendicular to the magnetic field direction completely disappears, whereas the spin projection on the magnetic field direc tion is retained. After the time c, the magnetic field di rection will be changed and the initial spin polarization will vanish. Thus, in this case one has s c, i.e., the spin relaxation time is of the same order of magnitude as the spin correlation time. Different mechanisms are responsible for creation of the fluctuating magnetic fields governing the spin relax ation. The DyakonovPerel mechanism9496 relating the effective magnetic field to the momentum dependent spin splitting of the conduction band dominates in the case of mobile carriers in the semiconductors based on type IIIV and IIVI compounds and two dimensional electrons in quantum wells. This and other mechanisms involving the motion of electrons are inapplicable to electrons local ized on donor sites or in quantum dots and the spin relax
ation times are very long. The problem of spin depolariza tion is of crucial importance for realization of spin based quantum computers. The hole spin relaxation times are much shorter than the electron spin relaxation times and spin exchange be tween electrons and holes enhances decoherence of both types of carriers. Therefore, creation of spin coherence requires the absence of holes, as is the case of n type semiconductors. In the absence of holes spin decoherence is mainly due to the spin orbit coupling that is relativistic in nature. Namely, an electron moving in the electric field of a crystal "senses" a part of the field as magnetic field; as a consequence, the electron spin begins precession about the field direction. Initially excited spin polarized elec trons are characterized by a scatter of velocities and there fore each of them precesses in its own fashion. Two elec trons with initially equal spin projections can change them after a short time. As the degree of spin disordering in creases, the total spin polarization on the average de creases and the spin coherence vanishes. Experiments with novel magnetic semiconductors showed that the spin co herence times can be much longer than the observation times, thus providing conditions for efficient operation of spintronic devices in the future. 1.4. Basic spin effects Pioneering experiments97 on manipulation of electron spin orientation by external electromagnetic fields were concerned with observation of depolarization of mercury vapor luminescence by applying a transverse magnetic field (here, terrestrial magnetic field). The mechanism of the effect observed98 involves precession of atomic elec tron spins about the direction of the magnetic field. On exposure to continuous irradiation the average spin pro jection on the direction of observation, which determines the degree of circular polarization of luminescence, de creases owing to precession. As a result, the degree of polarization decreases as the transverse magnetic field increases. Measurements of the dependence of lumines cence depolarization on the magnetic field allows both the spin relaxation time and the recombination time to be determined. Later, detailed studies of optical pumping of atoms were carried out.99,100 At present, depolarization of atomic luminescence by transverse magnetic field has been efficiently used in experiments on spin orientation in semi conductors. There is a number of effects following from interrela tion between spins and charge transfer.10 One of them (Mott effect) is related to the fact that spin orbit coupling produces asymmetry of electron scattering by a charged center relative to the plane passing through the momen tum vector and the direction of spin orientation. This effect induces an additional electric current with the den
2365
sity j = E S rotS , where S is the spin density vector, E is the electric field strength of, e.g. light wave, and the coefficients and are proportional to the spin orbit coupling constant. The first term describes the so called anomalous Hall effect and the second term de scribes the additional electric current that is due to the spin orbit coupling and nonuniform spin density distri bution,101,102 which was confirmed experimentally.103 The reverse effect is also possible, namely, electric current can induce spin orientation near the surface of the sample.104 In gyrotropic semiconductors (here, the crystal lattice has no center of inversion), electric current can also be in duced by a uniform nonequilibrium spin density distribu tion; this was predicted theoretically105,106 and then con firmed experimentally.107 Optical polarization of electron spins in semiconduct ing materials (e.g., silicon108) is in principle similar to optical pumping in atomic physics and differs from the latter in that in semiconductors we deal with spin polar ized conduction electrons rather than bound electrons in the atom. The polarization of nonequilibrium spin polarized electrons can be transferred with ease to atomic nuclei in the crystal lattice. Long lived nuclear spin polarization creates a very strong effective magnetic field, which in turn affects the electron spins (rather than the orbital motion of electrons). Owing to hyperfine coupling, spin polarized electrons and the lattice of oriented nuclear spins form a tightly bound system. Using circular polar ization of luminescence,109,110 in this system one can ob serve such nonlinear phenomena as slow undamped os cillations and hysteresis.111 Interband optical pumping experiments108 permitted attainment of 32% nuclear po larization, which provided112 an internal magnetic field of 1.7 T. An improved technique113 involves nuclear spin polarization by creating nonequilibrium spins in the elec tron subsystem, which transfers its spin polarization to the nuclear subsystem during relaxation to the thermody namic equilibrium state. This technique is much more efficient than spin polar ization of nuclei by external magnetic fields H. In the latter case the degree of nuclear spin polarization IH/kBTI is expressed through world constants, viz., the Boltzmann constant kB and nuclear magneton I. Though the nuclear magneton is small compared to the Bohr magneton, nuclear spin polarization is possible even in weak fields due to low temperature TI of the nuclear spin system. This effect and a strong feedback between the electron and nuclear spin systems in the semiconductor114 can also be employed in spintronics. However, low initial tempera tures (liquid helium) and the lack of isotopically pure materials preclude practical implementation of these physical effects. Note also a report on the effect of nuclear spins on the low temperature magnetotransport.115
In the ferromagnetic tunneling contact, or the mag netic tunneling transition, the height of the potential bar rier is different for electrons with different spin orienta tions, thus providing different tunneling probabilities. Therefore, the contact barrier will pass electrons with preferable spin orientation. Spin injection from a ferro magnetic tip into GaAs via electron tunneling was first demonstrated116 before the term "spintronics" was pro posed. Magnetic tunneling transitions are of paramount importance for fabrication of highly sensitive field trans ducers and memory devices. Recently, experiments on spin injection into DMS have been carried out. Spin dependent resistance in ferromagnetic metals was also studied117,118 long before the emergence of spin tronics. It was shown that at sufficiently low temperatures the electrons whose magnetic moments are parallel and antiparallel to the magnetization direction of a ferromag netic are not mixed in scattering processes. Then, the total conductivity can be represented by the sum of inde pendent and unequal contributions of two different ori entations of electron spins. This concept was general ized119121 to explain different magnetoresistance effects. In the novel DMS the temperature dependence of resistivity, (T ), is nonmonotonic at the magnetic impu rity concentration corresponding to the highest TC. Be low TC, one has d/dT > 0 (cf. d/dT < 0 above TC). The resistivity of magnetic materials also depends on an exter nal magnetic field. At a specified temperature, the resis tivity can be reduced by applying a magnetic field, the greatest resistivity drop being near TC. This phenomenon is often observed in the DMS and magnetic metals. It is due to the spin dependent scattering of charge carriers by spin fluctuations owing to exchange interaction. This scat tering is most pronounced near TC where the magnetic correlation length is comparable with the Fermi carrier wavelength.122 As the magnetic field increases, additional ordering of localized spins occurs, thus suppressing the spin dependent scattering and being responsible for nega tive magnetoresistance. Ferromagnetic materials (Sn,Mn)Te, 123 124 (GeTe)1x(MnTe)x and (Pb,Sn,Mn)Te 125 have not received wide acceptance because of the Curie tempera tures and of the properties of type IVVI matrices that are of limited use in electronics. 1.5. Spin injection Operation of "spin memory" devices, "spin transistor", and "spin quantum computer" requires spin injection polarized electrons, rather long spin relaxation times in electron transport processes, and detection of the spin state of an electron. One must create quantum state of electrons with specified spin orientation, retain the state during the operation of the device, and then read the
2366
Ivanov et al.
electron state at the output. Seemingly, a natural solution to the problem of spin injection into semiconductors was to use ferromagnetic injectors made of Fe, Co, of Ni. Spin polarization of current in the ferromagnetic con taining free carriers is due to the difference between the density of states of electrons with spin orientations "up" and "down" and to subsequent difference between the conductivities of the systems of electrons with antiparallel spin orientations. This spin injection technique was suc cessfully employed in the ferromagnetic metal/supercon ductor system. Spin injection from a ferromagnetic into normal metal was first proposed theoretically126 and ex perimentally observed nearly a decade later.127 However, attempts to realize this approach for the ferromagnetic metal/semiconductor system failed. First experiments on spin injection of electrons from ferro magnetic Ni into GaAs failed owing to low quality of heterostructures though researchers did observe a sub stantial change in the coercive force on exposure of the structure to a weak luminous flux with an intensity of only 5 mW cm2.128 In the case of electron injection into a semiconductor through a ferromagnetic metal/semicon ductor contact, electrons in the semiconductor have nonequilibrium spins that contain information on the elec tron spins in the ferromagnetic; i.e., spins of electrons in the semiconductor can act as, e.g., detectrors of the state of the magnetic film. In turn, the orientation of electron spins in semiconductors can be detected both optically and electrically. The magnetic properties of the ferromag nets placed in contact with semiconductors in hetero structures can also be controlled.* To date, the highest efficiency of spin injection from ferromagnetic metal into a semiconductor (up to 30%) was achieved only in ex periments with a scanning tunneling microscope.129131 A reason for low efficiency of spin injection through the metal/semiconductor interface (~1%)132 is a large con ductivity difference between these materials.133 The efficiency of spin injection can be enhanced using the Schottky barriers (electrostatic barriers formed at the metal/semiconductor interface due to defect formation) which can act as tunneling barriers, thus weakening the effect of the difference between the electrochemical po tentials of the ferromagnetic metal and semiconductor on the transport of spin polarized electrons through the in terface. This allowed an efficiency of 2% in the spin injec tion from a Fe contact into a GaAs/(In,Ga)As light emit ting diode to be achieved at room temperature.134 Efficient solution to the problem requires that the in jector be a ferromagnetic semiconductor or a semicon ductor placed in an external magnetic field. Then, the spin polarization of electrons can be as high as nearly
* V. L. Korenev, VI Ross. konf. po fizike poluprovodnikov [VI All Russia Conference on Physics of Semiconductors] (St. Petersburg, October 2731, 2003), Sankt Petersburg, 2003.
100% owing to the exchange interaction between the con duction electrons and the magnetic impurity.135 First, it was proposed to use manganese doped type IIVI semiconductors whose conductivity is com parable with the conductivity of the non magnetic semi conductor containing injected spin polarized electrons. The spin polarization efficiency was 50% for the (Zn,Mn)Se/GaAs contact, 136 more than 50% for (Cd,Mn)Te/CdTe,137 and 86% for spin injection from a paramagnetic semiconductor BexMnyZn1xySe into a quantum well light emitting diode AlGaAs/GaAs.138 Re combination of spin polarized charge carriers causes emis sion of circularly polarized light; by measuring the degree of polarization of luminescence it is possible to estimate the efficiency of spin injection. Severe problems posed when using type IIVI magnetic semiconductors for in jection are due to the necessity of operation at low tem peratures (<10 K) because these materials are para magnetics and an increase in temperature causes a dra matic decrease in their magnetization at a specified exter nal magnetic field.* Clearly, efficient spin injection reqiures ferromagnetic semiconductors that could inject, transfer, and easily ori ent the spin polarized carriers in the semiconductor heterostructures. Therefore, the most promising candi dates for spin injectors are high TC ferromagnetic semi conductors that are technologically compatible with nor mal semiconductors. Among possible candidates are the DMS based on type IIIV matrices, the Geissler al loys X2YZ (X and Y are transition elements and Z are Group IIIV elements), and semimetallic and half me tallic ferromagnetic oxides. After the synthesis of novel DMS, experiments on electron injection from a ferromagnetic injector layer into normal semiconductors were carried out.139,140 Spin polarized holes were injected from a ferromagnetic (Ga,Mn)As electrode into a non magnetic GaAs elec trode.139 In spite of difficulties originated from the quality of the material and the presence of defects served as elec tron scattering centers, researchers did measure the spin polarization of the material. The efficiency of spin injec tion from a p type ferromagnetic (Ga,Mn)As (TC = 110 K) into a photoluminescent diode with non magnetic quan tum well InGaAs/GaAs was 1%. 139 Highly efficient (8210%) spin injection from (Ga,Mn)As (TC = 120 K) into an (Al,Ga)As light emitting diode at 4.6 K was dem onstrated.141 At present, search has been carried out for novel ferromagnetic semiconductors with higher TC, which could serve as room temperature spin injectors at weak or zero external magnetic fields. In spite of the problems faced by the crystal growth and doping tech niques, high TC ferromagnetism in semiconductors per
* A. A. Toropov, http://link.edu.ioffe.ru/winter/2001/main/ toporov.
2367
mits optimistical expectations of fabrication of spin de vices operating at room temperature. (The problem of spin transport in microelectronic devices has been the subject of a review.*) Most spin injection experiments involve passage of electric current from a ferromagnetic into a semiconduc tor through a tunneling barrier. High degree of spin polar ization can be achieved using a magnetic tunnel transis tor.142,143 At present, spins are injected into semiconduc tors by passing spin polarized electric current from the magnetic semiconductor under the action of voltage ap plied. This is the principle of the operation of a spin filter. However, the process is hampered by interference at the ferromagnetic/semiconductor interface. If we change the sign of the voltage, the ferromagnetic will pass only those electrons whose spins are aligned with the magnetization vector of the ferromagnetic semiconductor rather than all non polarized electrons flowing from the normal semi conductor. Electrons with antiparallel spin orientations will not pass through the interface and thus will accumu late in the semiconductor. This pure electronic technique was implemented in the experiments on spin injection using a GaAs layer 500 nm thick placed in contact with a ferromagnetic semiconductor MnAs (TC 600 K) layer 25 nm thick at T = 7.5 K.144 The highest efficiency of accumulation of spin polarized electrons was achieved at a voltage of 1.5 V. This is several times higher than the efficiency of spin injection from ferromagnetic semicon ductor into normal semiconductor. The degree of spin polarization was determined from magneto optical mea surements. Thus, spin injection into semiconductors was realized using electric current from normal semiconduc tor to ferromagnetic semiconductor.** Spin injection of current carriers is closely related to two practically important problems, namely, how long are the distances appropriate for the spin excitation trans fer and at which rates can the spin states be switched on and off. It was proposed to create a nonequilibrium spin system in semiconductors by passing electric current through a ferromagnetic contact.145 If an electron leaved from the ferromagnetic by tunneling, it must retain the spin orientation in the semiconductor at a distance of the order of the spin diffusion length , where D is the diffusion coefficient and s is the spin relaxation time. At the ferromagnetic/semiconductor interface, potential bar riers to electrons with the spins oriented parallel and anti parallel to the magnetization of the ferromagnetic be come different owing to the exchange interaction. For this reason, any deviation from the spin equilibrium (due to electric current through the interface or to exposure to
* A. S. Borukhovich, N. A. Viglin, and V. V. Osipov, Elec tronic journal " Issledovano v Rossii" [Studied in Russia ], http://zhurnal.ape.relarn.ru/articles/2001/. ** A. V. Ivanov, Scientific.ru, 11.04.04, http://www.scientific.ru.
light) will cause injection of spins into the semiconduc tor. In this case the nonequilibrium spin orientation will be retained at distances of the order of the spin diffusion length . Macroscopic spin transfer was first demonstrated tak ing n type GaAs semiconductor as an example.146 A laser pulse excited coherent precession of electron spins and then electrons were transferred along the crystal over dis tances longer than 100 m nearly without loss of spin polarization by applying an electric field. These distances are much longer than the characteristic sizes necessary for modern microelectronics. Coherent precession of elec tron spins was excited by a 150 fs laser pulse, i.e., spins could be manipulated thousands of times during the spin relaxation time. Since modern electronics widely employs heterostructures, the results obtained are also of impor tance in relation to spintronics. n Type semiconductors GaAs also exhibit retention of optically induced spin po larization of electrons with long (longer than 100 ns) spin relaxation times.147 After relaxation of electron spins on nuclear spins the relaxation time of GaAs was extended to 300 ns.148 Electric voltage induced spin injection from GaAs semiconductor layer into a ZnSe heterostructure layer through a pn junction at 5 K was also reported.149 (Un til now, it was assumed that electrons passed through the pn junction always lose their spin orientations due to scattering.) Ordered groups of electron spins that were transferred by the applied electric field also sustain their spin polarizations for a rather long time at room tempera ture. The total transferred spin retains its characteristics as the spin reservoir both in the first semiconductor and in the adjacent semiconductor layer. Applying an external electric field makes possible the reversal of spin current. Because spins were manipulated by electric rather than magnetic field, the effect detected points to considerable promises for fabrication of multifunctional electronic de vices (e.g., spin transistors) in which logic functions are combined with memory functions. Such devices could be simultaneously manipulated by both electric and mag netic field, thus completely employing the electron charge and electron spin degrees of freedom. 2. Spintronic facilities and devices An attractive feature of many devices employing the spin dependent properties of materials is their similarity to classical valves, diodes, transistors, etc., used in inte grated circuits. The operation of the novel devices can be understood with ease based on analogy between the elec tron spin flip flop for the materials with different polar ization of electron spins and a conventional pn junc tion for p type and n type semiconductors. It is also pos sible to design combined ferromagnetic semiconductor
2368
Ivanov et al.
devices using the advantages of microcircuitry. Consider spintronic related effects and spintronic devices in more detail. Time resolved optical experiments. Time resolved op tical experiments150 with picosecond temporal resolution revealed a marked spin precession as well as the electron and hole spin relaxation. The method provides a unique possibility to observe fine details of various spin related processes in semiconductors. Electric field controlled ferromagnetism. A character istic feature of novel magnetic semiconductors based on type IIIV matrices and isoelectronic materials is the dependence of TC on the concentration of charge carri ers. The density of charge carriers can be changed by doping, by applying an electric field, or by irradiation with light. A ferromagnetic DMS (In,Mn)As changes its magnetic properties on applying an electric field (see Refs 150153). As a positive potential was applied to the grounded layer of (In,Mn)As (TC = 30 K in the absence of external fields), the electric field inside the layer caused a decrease in the hole density and disappearance of ferro magnetism in the layer. Vice versa, at negative potentials applied or at zero potentials, ferromagnetism restored. Electric field induced changes in the magnetization of small domains of samples in the vicinity of electrodes were detected from changes in the Hall resistance. The Hall effect in magnetic materials is the sum of the normal Hall effect induced by the action of the Lorentz force on the charge moving in the magnetic field and the anoma lous Hall effect due to the magnetization induced asym metric scattering of current carriers
RHall = (R0/d)H + (RM/d)M.
Here, R0 is the normal Hall coefficient, H is the magnetic field strength, RM is the anomalous Hall coefficient, and M is the magnetization perpendicular to the layer of thick ness d. In magnetic materials, the anomalous Hall effect is much stronger than the normal Hall effect, which al lows the sample magnetization to be determined by mea suring the transport properties. At present, these labora tory experiments are of academic interest for elucidation of the nature of ferromagnetism because the electric volt age applied was 125 V. One could decrease the thickness of the active layer, but at d = 5 nm ferromagnetism in (Ga,Mn)As disappeared. Ferromagnetism in Ge1xMnx can be manipulated at much lower voltages (~1 V) and higher temperatures (~50 K).154 These experiments can appear to be basic to design of a new type of recording devices. Non magnetic spin manipulation in strained semicon ductors. Direct measurements of coherent electron spin precession in zero magnetic field as the electrons drifted in response to an applied electric field were reported.155 Ultrafast optical spectroscopy was employed to study the
electron spin dynamics in the strained GaAs and InGaAs epitaxial layers. A spin splitting arising from strain in the semiconductor films was observed. The method permits electrical control over electron spins with allowance for strain produced in the crystal lattice. The strain induced shift of the spin resonance frequency can be as high as nearly 30 MHz. Spin polarization of electrons by electric field. A possi bility of creation of spin polarized electron beams, which can find practical application in spintronics, was re ported.156 The authors considered the spin orbit interac tion in the presence of current flowing in the plane of oriented two dimensional electron gas with the Hamilto nian20 (orientation is due to, e.g., asymmetry of the quan tum well). Evaluation of the effect for a typical case of GaAs/AlGaAs heterostructure with an electron concen tration of 5 10 11 cm 2 and an electron mobility of 104 cm2 (V s)1 and an electric field applied of 10 V cm1 showed that a total of 108 electrons per cm2 become ori ented. Though this relative polarization is very small, the effect observed allows the degree of polarization to be controlled by applying an electric field. DattaDas spin transistor. Almost all studies concern ing spintronic devices mention the so called DattaDas transistor,157 that is, a ballistic (without scattering) field effect transistor as a future spintronic device. Unlike a conventional field effect transistor, here the emitter is a source of spin polarized electrons and the collector is a spin filter; depending on the base voltage, the current can be switched on or off. At a nonzero voltage, the emitter injects electrons with oriented spins that precess during the time of flight of an electron to the collector. Spin precession is due to the spin orbit coupling and the volt age applied, which for the moving electrons is transformed into the effective magnetic field (BychkovRashba ef fect20,158). Since the magnetic collector detects electrons with a particular spin orientation, the electric current is an oscillating function of the voltage applied to the cir cuit. The device proposed was called an electronic analog of electro optic modulator. Various improvements of the spin transistor, related to the mutual orientation of the magnetization vectors of the emitter and collector can also be imagined. However, conventional spin independent electron scattering in the transistor causes mixing of the spin states due to the spin orbit coupling. Thus, successful operation of the spin tran sistor is only possible in the ballistic regime; however, in this case the device loses its advantages over conventional ballistic transistor. The spin transistor was "improved"159,160 with allow ance for the so called Dresselhaus correction to the spin orbit coupling for three dimensional systems, whose crys tal lattices are asymmetrical with respect to inversion of the spatial coordinates. In this case, coherent carrier trans port between coupled channels creates an additional spin
2369
precession and at the same Rashba and Dresselhaus con stants in the spin orbit coupling scattering becomes spin independent, so the spin transistor can now operate in the non ballistic regime. Though fabrication of spin transistors faces some prob lems such as stray magnetic fields and low efficiency of spin injection, the results of recent experiments show that these difficulties can be avoided using novel DMS. Spin dependent light emitting diode. A spin diode and a bipolar transistor have been reported.161 The materi als used were a semimagnetic n type semiconductor BeMnZnSe with 100% spin polarization at 30 in a magnetic field of 2 T 162 and a dilute magnetic semicon ductor p Ga1xMnxAs with TC > 100 K and 98% spin polarization at room temperature and a Zeeman splitting of 100 meV. These devices seem to be promising for mag netic sensors, non volatile memory devices, and re programmable logic devices. The aim of current research is to achieve technological integration of optical and mag netic characteristics. Opto electronics employs a light emitting diode in which the electron hole recombination causes spontaneous emission of light (silicon light emit ting diode163). Magnetoresistive transducers that employ electron spin polarization have also been reported. Spin valve based on organic semiconductor. Compared to traditional materials such as Si, organic semiconduc tors are easier to produce and their resistivity can also be changed by doping. The first organic semiconductor spin valve164 is based on an organic semiconductor made from Al and 8 hydroxyquinoline (100 nm thick) sand wiched in between ferromagnetic layers (Co and LaSrMg based alloy). This organic material is processable because it is used in some light emitting di odes and TV displays. As the orientation of the electrode magnetizations changed from antiparallel to parallel, the electric current increased by 40%, thus exhibiting a GMR effect. This organic spin valve operates at temperatures between 262 C and 40 C and can be arranged on a chip with a surface area of 1/3 square inch. Spin valve luminescent transistor165 modulates the lu minescence intensity in the near IR region depending on the magnetic field applied. High sensitivity to magnetic field (200% optical signal modulation on a change in the magnetic field by tens of Gauss) and a relatively small size (900900 m2) of this transistor offer prospects for re mote detection and imaging of magnetic fields and for fabrication of magnetic memory devices with optical data output channel. Similarly to a conventional transistor, the new device comprises a nano assembled collector, an emitter, and a base. The collector is a pin junction grown by molecular beam epitaxy on GaAs substrates. The p type impurity content in the AlGaAs layer is an order of magnitude higher than the n type impurity con centration; therefore, this is a hole rich heterostructure. By increasing the n type impurity concentration it is pos
sible to extend the lifetime of the incoming electrons and thus to enhance the efficiency of electron hole optical recombination at the center of the transition (GaAs layer 10 nm thick) and, hence, the intensity luminescence. The transistors base comprises two NiFe and Co magnetic layers 5 nm thick obtained by high temperature vacuum deposition and forms a Schottky contact with the collec tor. The emitter (Al layer 6 nm thick) was deposited on an insulating Al2O3 layer 100 nm thick, which separates the emitter and the base. As a voltage is applied, electrons move from the emitter to the collector and recombine with the excess holes, thus emitting photons that are de tected. The problem is of what happens to the electron current as it passes through two ferromagnetic layers. If the NiFe and Co layers are magnetized in parallel, the spin polarized electrons moving from the emitter pass through the base to the collector without scattering. Anti parallel magnetization of the NiFe and Co layers causes a decrease in the electron current and a 200% decrease in the luminescence intensity (experimentally observed). This device combines a spin valve transistor and an opti cal output. Spin photocurrents. Numerous results of studies on the spin photocurrents (also known as circular photogalvanic effect) were reported in a review.* Detection of spin polarized electric current in semi conductor layers irradiated with circularly polarized light was reported.166 The effect originates from a specific prop erty of the materials with the quantum wells formed by the "sandwich' of different semiconductors. Namely, elec trons with antiparallel spin orientations acquire differ ently directed velocities owing to asymmetry of the crystal lattice. Then circularly polarized laser radiation directed perpendicular to the quantum well plane results in a spin polarized electron beam. By changing the light polariza tion it is possible to reverse the spin current. This effect can be used for fabrication of circularly polarized light driven spin logic switches. Theoretically, the spin current can be generated by two beam laser irradiation of the electron system in the semiconductor.167 Using interference of the laser beams, one can not only produce electron spin polarization and spatial orientation of electron spins but also generate the electron spin current without electric voltage. All optical magnetic resonance. Nearly five decades experimental NMR techniques have employed radio frequency electromagnetic fields for orientation of the magnetic moments of atomic nuclei. With allowance for the electron spin degrees of freedom the NMR technique can be improved as follows. One laser excites the electron spins whose magnetic moments cause orientation of the spins of the nearest nuclei owing to the hyperfine cou
* S. Ganichev and W. Prettl, http://arXiv.org/cond mat/ 0304268.
2370
Ivanov et al.
pling. During the process, the nuclei are controlled by the second laser beam. This method was first demonstrated taking GaAs semiconductor as an example.113 This basi cally new alternative to conventional NMR considerably enhances the resolution of the technique, because light can be focused much better than the radiofrequency elec tromagnetic fields. All optical NMR provides the possi bility of coherent manipulation of individual nuclear spins, being a step toward establishment of optical control over spin polarized nuclei in solids.168 Microwave radiation in spintronics. A magnetic multi layer nanoscale heterostructure can transform electric cur rent into high frequency magnetic waves, i.e., nano magnets can serve as sources of the spin polarized current driven microwave radiation.169 Experiments were carried out with a copper nanorod with elliptic cross section (prin cipal axes 130 and 70 nm long) in which a layer of "thick" (40 nm) Co ferromagnetic film with the magnetization directed perpendicular to the axis of the cylinder and a layer of thin Co magnetic film (3 nm thick) were sepa rated by a short distance. A weak dc current (~1 mA) was passed along the nanorod axis. After passage through the thick magnetic layer, electrons were found to be spin polarized and the spin polarization was transferred to the thin Co layer. As a result, the magnetization of the thin layer began precession and emission in the microwave range. A microwave detector integrated with the circuit detected this emission whose frequency varied over a wide range up to ~18 GHz depending on the electric current strength and the external magnetic field applied. Earlier, the spin precession dynamics in the microwave region was studied by exposing the material to microwave radia tion. The reverse effect, namely, microwave emission gen erated by the nanomagnet magnetization precession was studied.169 The results obtained permitted evaluation of the intensity of oscillations at different frequencies as func tion of the electric current and magnetic field and com parison of the estimates with the results of calculations. Technological applications require that the threshold cur rent density (107 A cm2) be reduced to 105106 A cm2 while the power of the output signal be increased, al though being forty times higher than room temperature noise. This effect can be employed in spintronic nano devices for generation of microwave emission and in vari ous oscillators. On spin currents. In contrast to the electron charge, the electron spin can be transferred with a low energy loss. For instance, the spin relaxation time in semicon ductor GaAs is 100 ps (three orders of magnitude longer than the hole lifetime). Therefore, a topical problem in spintronics is to study the laws of spin motion. It was proposed170 that an electric field applied to a semicon ductor will produce a flow of electron spins due to the spin orbit coupling and topological effects. The direction of the spin current is determined following a rule accord
ing to which the electric field strength, spin orientation, and spin current density vectors must be mutually per pendicular. Seemingly, the voltage applied to the semi conductor to control the dissipation free motion of elec tron spins (spintronics) can be used instead of using volt age for electron transfer (electronics). However, strictly speaking, the electron spin is not conserved in the presence of spin orbit coupling. There fore, one must critically look at calculations of the spin currents. Switching on the electric field disturbs symme try of the system relative to inversion of the spatial coor dinates. Therefore, in the case of time reversal both the direction of current and the spin orientation change, i.e., the spin current is conserved. The conductivity and elec tric field strength appeared in the Ohm law remain un changed on time reversal, which eventually causes un avoidable release of the Joule heat. On the contrary, the spin conductivity relating the spin current density to the electric field strength appears to be170 dissipation free as in, e.g., the superconductor. However, the spin currents derived cannot serve as transport currents, i.e., they can not be used for spin tansport and spin injection into mate rials. Indeed, persistent spin currents in the semiconduc tors, whose crystal lattices have no center of inversion (e.g., GaAs and GaP with zinc blende structures), must also exist in the thermodynamic equilibrium state without external electric fields. Thus, although being dissipation free, the spin currents proposed170 appear to be pure back ground currents. Rigorous calculations showed that the background spin currents do not contribute to spin trans fer both in the two dimensional143,171 and in one dimen sional172 cases.* Giant planar Hall effect. In the normal Hall effect the current flowing through a planar conductor is deviated by the magnetic field applied perpendicularly to both the direction of the current and the plane of the conductor. As a result, the resistivity of the sample along some direc tions decreases. This type of the magnetoresistance anisot ropy is observed in magnetic metals, being rather weak. The anisotropy effect observed in the experiments173 with the dilute magnetic semiconductor (Ga,Mn)As at 45 was found to be four orders of magnitude stronger than in ferromagnetic metals. This effect called giant planar Hall effect can favor the development of improved versions of magnetic resonance spectroscopy, fabrication of magnetic transducers and, probably, elements of solid state quan tum computers based on DMS. Supergiant magnetoresistance effect. In ballistic mag netoresistance the size of a sensor can be reduced to the size of a cluster of ferromagnetic atoms connected by two wires. Here, the term "ballistic" means that the size of the sensor is less than the scattering length of electrons; be cause of this, the electron trajectories are straight lines.
* A. V. Ivanov, Scientific.ru, 16.04.04, http://www.scientific.ru.
2371
Hence, electron scattering is due to the magnetic effects rather than collisions with the atoms of the sensor or with impurities in the transducer. This significantly enhances the sensitivity of information reading process. Spin po larized electrons moving through the transducer are scat tered stronger or weaker (which implies a larger or smaller resistance) depending on the magnetization of the atoms of the electrode material. The ballistic magnetoresistance of nickel nanocontacts was studied experimentally.174 The sensor is so small (of the order of a few nanometers) that the spin of an electron passing through the contact does not "sense" the spin of the second electrode. But if the latter differs from the spin of the electrode material, the effect of giant magnetoresistance is produces. An enor mously high change in the magnetoresistance, up to 3150% (room temperature effects reach 100% for GMR and up to 1300% for colossal magnetoresistance) was detected. This is the strongest room temperature spin dependent magnetoresistance effect ever observed in spintronic de vices. Since the size of the transducer is small, the infor mation storage density can be increased to terabytes per square inch. Tunneling magnetoresistance. A low temperature study175 of the tunneling magnetoresistance of a three layer heterostructure (Ga,Mn)As/GaAs/(Ga,Mn)As at 0.39 K revealed a 290% effect at nearly zero base voltage. The spin polarization was determined from the combined Hanle effect98 and magneto optical Kerr effect mea surements. Magnetoresistance measurements of a mag netic tunnel junction Ga 1x Mn xAs ( x = 0.04, 50 nm)/AlAs/Ga1xMnxAs (x = 0.033, 50 nm)176 with an AlAs layer less than 1.6 nm thick revealed a spin polariza tion of 75% in Ga1xMnxAs at 8 . This spin polarization of the DMS is comparable with that of the manganite LaSrMnO, being greater than those of the classical ferromagnets Fe and Co (<50%), Ni (~30%) and lower than that of the metallic ferromagnetic CrO2 (~80%) only. Spin precession and magnetization of nanoscale objects. Information storage in memory devices is based on rever sal of the magnetization vector of a certain domain of a magnetic compound involving local magnetic fields. An increase in the recording density in a memory device faces the problem of induction of a controllable magnetic field in a small spatial domain of the compound. The use of magnetic fields produced by local currents requires unrealistically heavy electric currents for the magnetiza tion reversal of the small volume elements of the com pound. In this connection, a promising way is to employ spin driven manipulation of the micromagnetization of domains. A number of studies were concerned with the possibility of the magnetization reversal of a small do main of a magnetic substance by spin polarized electron beam. The magnetic field of such an electron beam is too weak to cause the magnetization reversal, and the dipole
interaction between the electron spins and the magnetic moments of atoms is also weak. Here, utilization of the exchange interaction between the beam electrons and electrons of the atoms of the material can appear to be efficient. Experimentally,177 spin polarized photoemission elec tron beams from GaAs were generated using circularly polarized light and then passed through the suspended Fe, Co, and Ni magnetic films a few nanometers thick, whose magnetizations were perpendicular to the direction of the spin polarization of the beams. A spin filter detected co herent precession of the magnetic moments of electrons in the beam. The reverse effect of the spin polarized elec tron beam on precession of the magnetization of a small domain in the ferromagnetic films became pronounced with decreasing the size of the domain. Both the direct and reverse effects can be used for information recording and reading. A switching frequency of tens of gigaHertz seems to be attainable. Quantum dot spin filter. Spin orientation can be achieved by using the spin orbit interaction between spins and the crystal lattice of the material, whih allows the design of the spin device to be significantly simplified. For instance, in a spin filter178 operating without using external magnetic fields, non polarized electrons are passed through a T shaped quasi one dimensional chan nel. Partial spin selection of electrons near the channel intersection is due to the spin orbit interaction. The pro cess can be controlled using an external electrode. When the incident electron energy is in resonance with the quan tum dot states at the intersection, the efficiency of spin selection for the trasmitted electrons approaches 100%. Interest in the spin effects in quantum dots is due to the expected long spin relaxation times. Indeed, here the relaxation mechanism is related either to phonons or to hyperfine interactions between electron and nuclear spins. Transport of electrons through quantum dots must be spin dependent owing to the Pauli exclusion principle. If the quantum dot contacts the spin polarized reservoir, one can manipulate the spins of the incident electrons or the spins of the electrons passing out, which changes the electric current flowing through the quantum dot. Hence, the spin state of the quantum dot can be encoded by using an external electron added. Vice versa, the spin state of the quantum dot can be probed or changed by adding an electron or by applying an external magnetic field. Such a unique single charge or single spin control can play an important role in the future solid state quantum comput ers in which the quantum system with two discernible states, capable of storing a bit of information, comprises specially created spin states.179 It is also of interest to employ electron spins in quantum dots for fabrication of functional elements for quantum computers.180 Carbon nanotubes and dilute magnetic semiconductors. Carbon nanotubes are mechanically stable and can resist
2372
Ivanov et al.
the elevated temperatures and electric currents of density up to 109 A cm2 (cf. ~105 A cm2 for normal metals). The first field effect transistor based on a nanotube with metallic contacts181 was of low quality owing to the chemi cal incompatibility of the materials. Electric contacts of the carbon nanotube with the semiconductor material were found to be much more promising and and techno logically integrable. For instance, a single walled carbon nanotube was attached to the GaAs/AlAs and (Ga,Mn)As heterostructures grown by molecular beam epitaxy on GaAs substrates. 182 In fact, a nanotriode was fabri cated, in which a dilute magnetic semiconductor, p type Ga0.95Mn0.05As only 2050 nm thick, acts as an emitter and a collector, the n type GaAs semiconductor layer 20 nm thick being a base. The emittercollector distance was of the order of 1 m. The technique permits fabrica tion of (i) field emission pnp transistors using car bon nanotubes in which the emitter, collector, and base are chemically compatible semiconductors and (ii) nano lasers with semiconductor cavities made of doped carbon nanotube. Great prospects are offered for fabrication of hybrid devices comprised of nanotubes and semiconduc tors in traditional semiconductor chips, which is conve nient for studies of spin polarized transport from DMS to nanotubes. Atomic nuclei and spintronics. The field produced by the magnetic moments of atomic nuclei (Overhauser field) governs the effective Zeeman spin splitting of electronic states and eventually the spin polarization of a nanoscale object. A new mechanism was proposed183 of manipula tion of electron spins in nanostructures by using a spa tially nonuniform distribution of the atoms of isotopes with nonzero nuclear spins. Polarization of the nuclear spins in a small domain of a semiconductor material cre ates a local potential attracting electrons with a particular spin orientation and repulsing electrons with antiparallel spin orientation. As a result, the polarized nuclear spins produce an effective magnetic field, which can create bound electron, whose energies are up to a few meV in semiconductor nanoscale objects of size greater than 200 . Such spin dependent potentials can form a new type of spin nanostructures, namely, spin quantum dots, spin quantum wells, etc., in real time using coherent polarization or depolarization of nuclear spins. Thus it makes possible spin engineering of nanoscale objects with polarized nuclear spins. Here, the magnetic field of atomic nuclei strongly influences the electron spins, resulting in magnetization of the nanoscale objects, although the Overhauser field does not manifest itself due to small nuclear magnetic moments. Manipulation of nuclear spins. Compared to magnetic fields, electric fields are much easier to induce and con trol. Therefore, electric field controlled manipulation of nuclear spins is desired. The first step in this direction was reported in Ref. 184. First, a laser pulse polarizes electron
spins in an AlxGa1xAs quantum well. A disk of polarized electrons of radius 2000 nm and 20 nm thick, formed in the quantum well can be transferred with nanometric ac curacy by varying the voltage applied. The atomic nuclei confined within the domain filled with the polarized elec trons will also be polarized, being aligned with the elec tron spins. As a result, position of a small domain com prised of ten polarized nuclei can be exactly located and changed by applying an electric voltage. Such thin layers of polarized nuclei can appear to be the main element of storage/memory devices based on nuclear spins. Quantum computer and nuclear spins. The idea of quan tum computing185 has been the subject of keen discussion since 1986.186 It was proposed185,186 to employ operations with the states of a quantum system, each of them being a superposition of states in contrast to the classical system. However, a boom of projects of quantum computers be gan after a lecture and publications by P. Shor187 who proposed an algorithm for factoring large n digit numbers into their prime components. The amount of operations needed to solve the problem using modern computers grows as an exponential function of the size of the num ber. It is complexity (nonsolvability) of this problem for modern computers that is employed in encrypting infor mation in cryptographic systems. A quantum computer needs n3 operations to solve the problem, with the speedup coefficient that can be very large at large n. At the same time there are algorithms executed on traditional com puters, which do not accelerate on quantum computers. Shors algorithm seems to provide an example of the phe nomenon where the complexity class of a task changes basically depending on which physical princilpes underlie computations. When using a traditional computer, the task belongs to the same class, whereas the quantum na ture of qubits (a quantum bit, or qubit, is the smallest amount of information in quantum computing) in the quantum computer significantly accelerates the process of computation. Having intrinsic parallellism in large scale computations, the quantum computer can decode almost any message. The use of quantum logic cells for information recording, storage, and writing will provide in the future fabrication of quantum memory devices ca pable of self correcting errors and suitable for eternal in formation storage. To date, a number of projects of quantum computers was proposed,188 based on different matrix materials in cluding liquids (helium) and solids. To realize a solid state quantum computer, one would employ the experi ence of microelectronic technology. Then, quantum com puters could be similar to microchips. The goal of most studies in this field is to integrate modern microelectron ics technologies with quantum computing ideas with spintronics underlying research efforts. It is of great im portance that materials for quantum computers based on ferromagnetic semiconductors dont need have high TC.
2373
In solid state quantum computers information can also be encoded by nuclear spins because they interact with the environment weaker than do the electron spins and therefore retain their state for a longer time. In addtion, small magnetic moment of a nucleus can be more readily reoriented by external magnetic fields compared to the magnetic moment of an electron. If one treats the nuclear spins I of a donor impurity in the semiconductor as qubits, the atomic nuclei of the semiconductor must have I = 0 to prevent possible spin spin interaction. This excludes all type IIIV semiconductors from being candidates for the matrix material because none chemical element has stable isotopes with I = 0 in this case. To this end, silicon seems to be the most appropriate element because methods of preparation of isotopically pure 28Si (I = 0) have been best developed compared to other technologies and con siderable experience in fabrication of nanoscale objects has been accumulated. The only shallow donor in Si is the 31P isotope (spin I = 1/2). At a low 31P concentration and T = 1.5 , the electron spin relaxation time in the Si31P system is of the order of 103 s while the 31P spin relaxation time exceeds ten hours. Seemingly, at temperatures about 1 mK the 31P spin relaxation time limited by the interac tion with phonons only will be of the order of 1018 s, which makes this system unbeatable for quantum com puting.189 Low temperatures favor computations involv ing nuclear spins because it excludes the ionization of donors. Then, the conduction electrons are accumulated near the donor nuclei, thus providing a sufficiently strong HFC between the electron and nuclear spins. On this basis it was proposed to use the 31P nuclear spins injected into a single crystal of isotopically pure silicon as qubits for the quantum computer.190 Qubits were manipulated by using static magnetic and electric fields and a radio frequency magnetic field. At T 0.1 , a silicon plate was placed in a constant magnetic field B0 2 T, in which nearly 100% of electron spins were polarized; the nuclear spins were ordered via the hyperfine exchange interaction with the electron spins. The results obtained were quite unpretending, namely, one or two qubit spin interac tions. The authors of Ref. 191 improved the ideas190 using electron spin resonance in epitaxial Si/Ge heterostructures with the electron energy bands dependent on the compo sition of the material. Possible alternatives. Though such DMS as (GaMn)As, (GaMn)P, and (GaMn)N have sufficiently high TC, they also have defect structures and are crystallochemically incompatible with silicon. At the same time silicon is a nonmagnetic substance and attempts at doping of silicon with magnetic impurities led to destruction of the crystal structure, the multiphase composition of the sample, and reduction of conductivity. Potentially, an alternative to DMS is provided by tran sition metal monosilicides. For instance, compounds Fe1xMnxSi and Fe1yCoySi based on the narrow gap in
sulator FeSi, in which iron atoms are replaced by Mn or Co atoms in the whole range of concentrations, were reported more than two decades ago.192 Recently, de tailed studies of the magnetotransport properties of these substances have begun in relation to spintronics demands.193 At x < 0.8, the alloy Fe1xMnxSi remains a p type paramagnetic down to 1.7 . Compound Fe1yCoySi is characterized by elelctronic conductivity, being at y < 0.3 a helicoid magnetic with a long period structure and a highest transition temperature of 53 . Metallic magnetic Fe1yCoySi with TC = 53 and the best studied DMS (Ga,Mn)As with TC = 110 are strongly different in current carrier density (1.51020 for (Ga,Mn)As and 4.410211.31022 cm3 for Fe1yCoySi) and in effective masses of current carriers (0.5me and 0.08me respectively for the heavy and light holes in (GaMn)As and 30me for Fe 1yCo ySi). Nevertheless, Fe1yCoySi and (GaMn)As have similar magnetizations and the magnetic field affects the electric properties of Fe1yCoySi and (GaMn)As in a similar fashion. The Co enriched material Fe1yCoySi retains the crystal struc ture of the FeSi matrix, thus providing yet another route to preparation of spintronic materials. 3. Magnetic spintronic materials 3.1. Requirements for spintronic materials Being an interdisciplinary field of science and tech nology, spintronics has been the subject of intensive stud ies carried out in a continuously increasing number of research laboratories. At present, computer hard disks, reading heads, and MRAM chips are produced based on ferromagnetic metallic alloys belonging to the first gen eration of spintronic materials. More than 90% of hard disks contain platinum. It is the properties of the CoPt alloy that make possible production of hard disks with information capacity greater than 20 Gb. If the thermal stability of the properties is important, computer disks are produced from magnetic materials based on the SmCo alloy. The major drawback of this type of hard disks is high cost of samarium and cobalt, which hampers wide use of these alloys. GMR reading heads are produced using films of FeNi magnetic alloys with specified com positions, characterized by the saturation induction which is two to four times greater than that of the widely used ferromagnetic alloy permalloy. The major drawback of these spintronic materials is that they are metals. Most ferromagnetic materials are metals. First of all, these are elemental ferromagnets (nine elements includ ing three 3d metals, namely, Fe, Co, and Ni and six 4f metals, that is, Gd, Dy, Tb, Ho, Er, and Tm) and a large number of metallic alloys and compounds. Most non metallic magnetic materials are antiferromagnets or ferrimagnets. A ferrimagnet (antiferromagnet) is a mate rial with antiparallel spin orientations of the magnetic
2374
Ivanov et al.
ions in different magnetic sublattices and different (equal) magnetic moments of the sublattices, which provides a nonzero (zero) total magnetization in the magnetically ordered state. Ferrimagnets (e.g., magnetite Fe3O4) have long been known, whereas actually experimental studies of antiferromagnetism began in 1938 taking MnO as an example.194 Usually, ferrimagnets have the crystal struc ture of spinel AB2O4 in which spins of the magnetic ions in the tetrahedral (A) and octahedral (B) cation sites have antiparallel orientations. Non metallic properties of magnetite FeFe22+/3+O4 and, hence, retention of the band gap both below and above the Vervey charge phase transition temperature (TV 120 K), and a ferrimag neticmetal phase transition at 865 K are due to strong exchange and Coulomb interactions between electrons in the non filled atomic shells of the Fe ions.195 Interest in magnetite as a spintronic material increased after experi ments196 that revealed a giant negative magnetoresis tance effect in nanocontacts at room temperature. Mag netic hysteresis permits the use of magnetite for produc tion of memory elements and switches in spintronic de vices. In this connection mention may be made of the synthesis of the first dilute ferrimagnetic semiconductor [ZnGa2O4]1 x[Fe 3O4]x * based on the semiconductor spinel ZnGa2O4 with a band gap of 4.1 eV 197 and TC 200 at x = 0.15. In spite of practical use of the first generation spin tronic materials mentioned above, they possess no useful semiconducting properties employed in microelectron ics. They can hardly be doped and cannot be used for signal enhancement in semiconductor devices. The mag netic properties of these materials are preset by the mag netic sublattice characterized by ordered arrangement of the magnetic ions. At present, the key problem in spintronics is the lack of novel ferromagnetic materials possessing a necessary combination of the semiconductor, magnetic, and optical properties. As mentioned above, magnetic semiconduc tors (inlcuding novel materials) exhibit interesting effects that underlie a number of spintronic devices proposed. However, until now wide use of these devices is hampered by the lack of magnetic semiconductors suitable for re tention of the necessary operating properties at tempera tures somewhat exceeding room temperature. Efforts into the development of spintronics on the basis of traditional semiconductor matrices appear to be inefficient since most of these materials are ferromagnets at very low tempera tures or non magnetic at all. And since the operating temperatures of, e.g. the surface components in comput ers can be as high as 50 C, the production of cooling systems would make computers bigger and more ex pensive.
* A. S. Risbud and R. Seshadri, http://arXiv.org/cond mat/ 0404147.
The present day demands for the Curie temperatures, TC, of ferromagnetic spintronic materials are similar to the earlier requirements for the operating temperatures of materials for solid state electronics. For instance, the op erating temperatures of germanium transistors were at most 70 C. Because of the lack of methods for the surface passivation the operating characteristics of these transis tors were unstable, especially at elevated temperatures, and transistors per se were non durable. Only replace ment of germanium by a new semiconductor, silicon, after elaboration of methods for the synthesis of ultrapure Si (to 106 at.% of impurities) allowed the operating tem perature of transistors to be increased to 120150 C at stable characteristics. Elaboration of planar technology permitted a reliable protection of the surface of the silicon crystal against the action of external medium by a thin SiO2 film; then, chips were fabricated, which marked a revolution in microelectronics. There are at least three criteria that must be met by the most promising spintronic semiconducting materials198: 1) ferromagnetic properties of these materials contain ing both n and p type mobile current carriers must be retained at the operating temperatures of semiconductor devices (above room temperature); 2) among technologically important are relatively simple and inexpensive synthetic routes to spintronic ma terials and methods of integration of items with conven tional semiconductor circuits; 3) it is desired that ferromagnetic semiconductors not only possess the necessary magnetic properties but also retain the structure and physicochemical properties of the starting semiconductor matrices without deterioration of their useful characteristics. The use of the complex growth techniques and high level experimental methods for the synthesis of spintronic materials is expensive. Ferromagnetic semiconductors for spintronics applications are as a rule synthesized using expensive vacuum deposition techniques including mo lecular beam epitaxy, laser irradiation, and ion implanta tion. At the same time traditionally the most widely used method for the preparation of chemical compounds is relatively fast and cheap synthesis of samples with con trolled compositions and structure from high purity pow ders. Versatility of these methods makes them perfect for the synthesis of substances with specified properties and characteristics. Production of cheap spintronic materials requires a close collaboration of physicists, chemists, ma terials scientists, and engineers and integration of the prin ciples of spintronics with technology of novel ferromag netic semiconductors. 3.2. Concentrated magnetic semiconductors Design of a material exhibiting the properties of a ferromagnet and a semiconductor (coexistence of the spin and charge degrees of freedom) is an age long dream.
2375
However, the first ferromagnetic semiconductor CrBr3 with TC = 37 was synthesized as long as 1960.199 The second generation of spintronic materials is represented by ferromagnetic semiconductors. The family of concentrated magnetic semiconductors (CMS) comprises europium and lead chalcogenides, chro mium chalcogenide spinels (e.g., CdCr2Se4, CuCr2S4), and a number of transition metal oxides (manganites) and pnictides57 in which the magnetic ions have own mag netic sublattices. In EuO and europium chalcogenides EuX (X = S, Se, Te)200202 the internal magnetic field of the Eu2+ ions can produce a giant Zeeman splitting of the spin subbands (nearly 0.5 eV for EuS).200,201 In the mean field approxi mation, the energy of the total spin splitting between two subbands in the electron spectrum is given by the expres sion E = gBH + 2Js S , where the first term is the Zeeman spin splitting due to the external magnetic field H and the second term describes the magnetic interaction between the spin s of the conduction electron and the averaged magnetization S of magnetic ions, which is "sensed" by the electron owing to the magnetic exchange interaction J. The second term is much larger than the first one. It is the interaction described by the second term that is the reason for giant changes in the energies between the band electrons with antiparallel spin orienta tions. Because of this, magnetic insulators are efficient spin filters in, e.g., metal/magnetic insulator/metal hetero structures.203 However, europium chalcogenides are of limited use owing to low TC that are much lower than room temperature. The Curie temperatures of chalcospinels are some what higher, namely, 84.5 K for CdCr2S4, 106 K for HgCr2Se4, and 130 K for CdCr2Se4. The Curie tempera tures of semiconducting materials CuCr2Se4xBrx exceed room temperature but the materials are unstable in air. Of interest are also the families of multicomponent semicon ductor solid solutions based on ferromagnetic com pound CuCr2S4 ( TC = 367 K) and antiferromagnets Cu0.5M0.5Cr2S4 (M = In, Ga, Al) or ferrimagnets CoCr2S4 and MnCr2S4, which possess above room temperature ferromagnetism.57 But, having retained some properties of the ferromagnetic matrix CuCr2S4 with p type conduc tivity, they are still far from meeting purity requirements for electronic materials. A number of transition metal compounds with ferro magnetic or ferrimagnetic order, namely, magnetite Fe3O4 (see above),204 CrO2,205 Geissler alloys X2YZ (see Sec tion 1.5),206 and pnictides MnAs,207 MnSb,208 CrAs,209 and CrSb210 are also attractive for spintronics applica tions. These materials are called semimetals and half metals. Unfortunately, the properties of the Geissler alloys are unstable and vary over a wide range depending on the stoichiometry. Only Ni2MnGa has a Curie temperature
exceeding room temperature and exhibits a good hyster esis.211 Other compounds exhibit a non classical tem perature dependence of magnetization and their proper ties cannot be described using the available models. Fer romagnetism in Geissler alloys disappears on disordering of magnetic ions. Synthesis of thin films of these materials on semiconductor substrates is also a complicated task. The exclusion is Ni2MnSb with TC 730 K, which was studied in order to reveal tunneling effects.212 A number of complex oxides possess ferromagnetism with relatively low Curie temperatures. These are BiMnO3 (TC = 105 K),213 GeCuO3 (TC = 26 K),214 YTiO3 (TC = 29 K),215 and EuO (TC = 79 K).216 Recently,217 it was found that CuO doped with Mn (3.515 at.%) also ex hibits ferromagnetism with TC = 80 K. Transition metal and rare earth element oxides, fluorides, and phosphides are poorly compatible with the semiconductor technolo gies in physicochemical parameters. The properties of these and other semimetals are briefly outlined in Ref. 218. At present, only one oxide, CrO2, simultaneously pos sesses metallic properties and ferromagnetism. Though this compound is widely used in recording devices, until recently its electronic structure was poorly studied and band structure calculations predicted semimetalic prop erties.219 Chromium dioxide is a saturated ferromagnet (one spin subband is completely filled, whereas the other is empty), at least at low temperatures. Experimental pho toemission studies show that at room temperature 80% of spins in CrO2 remain polarized. Because of this CrO2 is of paramount interest as a material for spintronic devices.220 Chromium dioxide (TC = 400 K) crystallizes in the rutile structure and has an integer magnetic moment of 2 B per Cr ion. This material is obtained in the form of a single crystalline film epitaxially grown on rutile substrates;221 textured CrO2can be grown on sapphire (Al2O3) sub strates.222 It is unknown whether is it possible to obtain CrO2 in the form of a textured or epitaxial thin film on GaAs or Si. The material exhibits complex magneto transport properties.205,223 The dependence of the Hall resistance on the magnetic field differs from that charac teristic of the ferromagnetic metal in the temperature range from helium temperatures to 100 K and has both the hole and electron contributions. Consideration in the frame work of the two band model shows that CrO2 has highly mobile holes that move coherently with more abundant but less mobile electrons, i.e., the band structure of the material appears to be more complex than that predicted by calculations. However, all spin polarization related ki netic characteristics224 and the results obtained by tun neling spectroscopy225 indicate that at low temperatures all carriers in CrO2 are spin polarized. Thus, CrO2 can be a candidate for being a source of spins or a spin detector in spintronic devices. Manganites based on antiferromagnetic matrices AMnO3 (A is REE) doped with Ca, Sr, Ba, etc., are also
2376
Ivanov et al.
of interest as spintronic materials. Applied value of these materials is due to the possibility of fabrication of high quality thin films and to colossal negative magnetoresis tance of some heavily doped manganites at room tem perature. These materials possess specific properties. Namely, being heavily doped, a number of them are fer romagnets with TC 350 ; however, they undergo struc tural transitions depending on the degree of doping and on the magnitude of external magnetic field. But manga nites are poorly integrable with semiconductor micro electronic devices owing to differences in crystallo chemical parameters. Compared to perovskites AMnO 3, ferromagnetic semimetallic transition metal oxides having the double perovskite structure A2BBO6 (A is an alkali earth metal ion and B and B are transition metals) and possessing room temperature magnetoresistance effects can appear to be more attractive for spintronics. For instance, the Curie temperature of Sr2FeMoO6 is 420 K,226 whereas that of Sr2CrReO6 is 635 K.227 Sr2CrReO6 films fabri cated by magnetron sputtering on SrTiO3 substrates ex hibit a TC of 620 .228 A homogeneous polycrystalline material Sr2FeMoO6 (TC = 416 K) has a magnetoresis tance effect of 45% at 5 K (H = 9 T) and of 6.5% at room temperature (H = 0.3 T).229 Doping with Al permits an increase in the magnetoresistance of Sr2Fe1xAlxMoO6 (x = 0.15) to nearly 60% at 5 (H = 1 T) and improve ment of electron transport properties.230 Differences between the crystal structures of normal semiconductors GaAs and Si and the CMS or semimetals and half metals prevent the latter from being integrated into semiconductor heterostructures for spintronics ap plications. 3.3. Semimagnetic semiconductors In spite of the fact that CMS possess a number of interesting spin related properties, these materials do not meet requirements imposed by semiconductor engineer ing. Spintronics applications require thick films, layers, and quantum dots that retain the useful physicochemical properties of semiconductors. Therefore, it is important to synthesize and study such magnetic semiconductors that are compatible with tetrahedrally coordinated semi conductors and mechanically stable against structural dis tortions. Of considerable interest are magnetic semicon ductors based on type AIIBVI and AIVBVI (AII = Zn, Cd, Hg; AIV = Pb, Sn; BVI = S, Se, Te) matrices, namely, CdMnSe, CdMnTe, PbSnMnTe, etc., in which the tran sition metal ions Fe2+, Co2+, or Mn2+ randomly replace the A elements in the lattice sites. For instance, doping with manganese (compound Zn2+1xMn2+xSe) introduces into the matrix only randomly distributed spins S = 5/2, leaving the charge carrier density in the semiconductor unchanged. These semiconductors are sometimes called
semimagnetic semiconductors because their magnetiza tion can be readily changed by applying external mag netic fields but the materials retain the characteristics of the semiconductor matrix. Considerable attention was paid ro Cd1xMnxTe with a high solubility of Mn (up to 77%). Many properties of these materials are due to the exchange interaction between the d electrons responsible for the local magnetic moments and the conduction elec trons or holes in the valence band. Owing to predominant antiferromagnetic exchange interaction between magnetic ions most AIIBVI based semimagnetic semiconductors ap peared to be weak ferromagnets with TC < 10 K,231233 antiferromagnets, paramagnetics, or even spin glasses (see reviews135,234) depending on the concentration of mag netic ions. Spin glasses are materials characterized by relatively stable arrangement of randomly oriented spins, which is established at temperatures below the so called glass tran sition temperature. Metallic spin glasses were discovered in the experiments on light doping of noble metals with transition metals. Here, an indirect magnetic exchange interaction between spins of the localized transition metal ions occurs involving the conduction electrons (so called RudermanKittelKasuyaYosida, or RKKY interac tion235237) with the energy
where JRKKY is the constant of the RKKY spin spin inter action, J is the parameter of the exchange interaction between a shared charge carrier (conduction electron in the case of spin glasses) and a localized electron (in the vicinity of an ion the spin orientations of the conduction electrons are parallel to the spin orientation of the mag netic ion), kF is the Fermi wave vector related to the volume density of free carriers n (kF n1/3), R 1/x1/3 is the distance between two spins, S 1 and S 2, of the mag netic atoms with the concentration x, and F = (h kF)2/(2m) is the Fermi energy.238 At high current carrier density and low concentration of magnetic atoms, one has kFR >> 1 and the RKKY interaction becomes oscillating in charac ter and weakens with the spin spin separation
,
i.e., in the spin glass state the interaction between spins of different pairs of the transition element atoms can be both ferromagnetic and antiferromagnetic. The oscillating char acter of indirect RKKY exchange interaction provides a correct explanation for magnetism of rare earth metals. And vice versa, at high concentrations of spin carriers (magnetic impurities) and low charge carrier densities
2377
one gets kFR << 1, the indirect magnetic RKKY exchange always being ferromagnetic in character, i.e., no spin glass state is formed and one deals with the VonsovskyZener double exchange. In this case the concentration de pendence of the RKKY interaction has the form JRKKY ~ n . Though, in principle, memory elements can be based on spin glasses, the glass transition tem peratures of these materials are low. The RKKY inter action manifests itself provided that the exchange pa rameter is smaller than the Fermi energy (J << F) or the band gap, which holds for metals and spin glasses but not for novel magnetic semiconductors (here, J >> F). Then, the RKKY interaction is inefficient near the critical tem peratures of magnetic disordering because in these re gions damping of free electrons simulated by de Broglie planar waves is significant. Despite the fact that the RKKY interaction was theoretically proposed nearly five decades ago, direct experimental confirmation of the existence of this interaction was obtained only recently. The RKKY spin polarization of conduction electrons was first de tected239 in the GMR measurements of heterostructures containing REE (Dy, Gd, Nd) layers. Spin glasses also exhibit a giant Zeeman splitting and can be used for fabri cation of ferromagneticnormal semiconductor contacts and other hybrid systems possessing ferromagnetic and semiconducting properties.240242 Strong exchange interactions in the doped type IIVI paramagnetic materials also favor high spin polarization of carriers in external magnetic fields. Indeed, though a strong exchange interaction between the Mn 3d states and the semiconductor states causes no ferromagnetic ordering of the Mn moments, an external magnetic field polarizes the manganese spins; the intrinsic magnetic field of the Mn spins causes a giant spin splitting. For instance, the spin splitting in (BeZn,Mn)Se is 400 mRy,138,243 be ing five orders of magnitude larger than the characteristic Zeeman energy of elementary magnetic moment in 1T magnetic field. A combination of strong exchange in teraction between the Mn 3d states and the carriers in the semiconductor matrix with very low temperatures (or even absence) of magnetic phase transitions makes semi magnetic semiconductors based on type IIVI and IVVI matrices very interesting objects suitable for studies of exchange interactions. This class of materials is also ap propriate for fabrication of spin injection devices. Doped materials based on type IVVI matrices also exhibit low TC ferromagnetism, but these materials and their heterostructures including widely used semiconductors can hardly be synthesized. Passage to type IIVI binary compounds (hosts for semimagnetic semiconductors) is illustrated in Scheme 1 (second line): atoms of Group IVa elements are replaced by their second neighbors on the left and on the right in the Mendeleev periodic system. Synthesis of spintronic
materials based on the isoelectronic type IIIIVI2 ma trix (lowest line), e.g., CuGaTe2 with well studied phase diagram is also of interest.
Scheme 1
Although semimagnetic semiconductors based on type IIVI matrices can also serve as spin filters or for spin transport,138 their use as sources of polarized spins in spintronic semiconductor devices is inefficient. Seemingly, an epitaxially growth semimagnetic semi conductor (Zn,Mn)Te confirms this conclusion because the highest C is reached at a p type carrier density of 1.531019 cm3, being only 2.4 .244 But the discovery of ferromagnetism in the DMS (Zn,Cr)Te 245 shows some promise of the future of these materials in spintronics. Detailed magnetization and magneto optical response measurements of (Zn,Cr)Te at a Cr concentration, , of 0.2 revealed a Curie temperature of 30010 K.246 In terest in the magnetic semiconductors based on type IIVI matrices also renewed after the discovery of high tem perature ferromagnetism in transition metal doped ZnO. But most of these materials are inhomogeneous so far (see Section 3.7). Note that doping of the starting Group IV material with transition metals can also give rise to ferromagnetism. For instance, oriented MnxGe1x(001) films on Ge and GaAs(001) substrates were grown at 70 C.154 These tem peratures of the substrates rule out the formation of Mn11Ge8 clusters with high TC similar to room tempera ture; these clusters are formed at substrate temperatures of 300350 C. MnxGe1x materials were characterized by semiconductor p type conductivity with a carrier den sity of about 10191020 cm3 at the Mn concentration, x, lying in the range 0.006 < x < 0.035 and exhibited ferromagnetic properties with 25 < C < 116 K de pending on the impurity content. Ferromagnetism in Co0.12Mn0.03Ge0.85 epitaxial film (TC 270 K) is also of interest.247 3.4. Dilute magnetic semiconductors Type IIIV semiconductors with the zinc blende or wrtzite lattices are in great request on the market of lasers operating in the IR and visible regions, reading devices (GaAs), and magnetic sensors (InAs). Broad for bidden bands in the energy spectrum make these materi
2378
Ivanov et al.
als transparent in corresponding optical ranges. Excellent electric and optical characteristics of GaAs and InP pro vide a high switching rate and low power consumption. Passage from the semiconductors based on Group IVa elements to type IIIV binary compounds (matrices for DMS) and to isoelectronic type IIIVV2 ternary com pounds (high temperature ferromagnetic semiconductors) is shown in Scheme 2.
Scheme 2
TC/K 140 1 2 3 4 5 6 7 8
100
60
20 2 4 6 Mn content (%)
Fig. 1. Curie temperature (TC) plotted vs. Mn content (at.%) in (Ga,Mn)As. Experimental data were taken from the following references: Curve Reference 1 * 2 81 3 260 4 261 5 262 6 257 7 263 8 264
Following the CMS, semimetals, half metals, and semimagnetic semiconductors, the DMS synthesized in the 1990s serve as the basis for next generation spintronic materials. In these compounds certain atoms of Group III elements of the semiconductor matrix (InAs, InP, GaAs, GaP or GaN) are randomly replaced by transition metal atoms with non filled 3d electron shells. It is important that DMS retain the crystal structure of the semiconduc tor matrix, being isoelectronic to silicon (see Scheme 2). A distinctive feature of this class of magnetic semicon ductors is that their Curie temperatures and other mag netic properties depend on the dopant concentration or current carrier density. Search for and studies of the ferromagnetic compounds based on type IIIV matrices were initiated by the dis covery of ferromagnetism in (In,Mn)As (TC = 7.5 ).8 More recently, a TC of 350 was reached in InAs:Mn quantum dot layers 248 and the Curie temperature of (In,Mn)As thin films was raised to 35 9,249252 and then to 60 K (x = 0.063).253 InP with ion implanted Mn+ is characterized by TC 90 .254 One can overcome the thermodynamic limit of man ganese solubility in the type IIIV semiconductor ma trix using low temperature molecular beam epitaxy (T < 523 K; the phase diagram is presented in Ref. 140). Usually, molecular beam epitaxy imvolves Ga, Mn, In, Al, and As targets. (Ga,Mn)As epitaxial films are grown on GaAs(001) substrates at growth rates of 0.60.8 m h1. Prior to epitaxy, the buffer (Ga,Mn)As, GaAs or (Al,Ga)As layers are built up. To relieve strain in the material synthesized, in some cases a thick (In,Ga)As layer (~1 m) is used, with the lattice constant greater than that of (Ga,Mn)As. Usually, the Mn content in Ga1xMnxAs films is determined from the lattice con stant using a calibration curve. If a monophase material (Ga,Mn)As is grown at T 473523 K, the Mn impuri
ties replace atoms in the Ga sublattice. This was estab lished by EXAFS 255 and is consistent with the linear dependence of the lattice constant on the Mn content (Vegard law): a = 0.566(1 x) + 0.598x (nm).256 The lattice constant depends on the synthesis conditions (As pressure, annealing temperature) because another rela tion, a = 0.5654(1 x) + 0.5901x (nm), was obtained in another study.257 If the substrate temperature or the Mn concentration are high, yet another phase, metallic ferro magnetic MnAs, is formed on the surface of (Ga,Mn)As samples. A (Ga,Mn)As sample with zinc blende struc ture, which was synthesized in 1998, exhibited a record (at that time) TC = 110 43,81,258 at a hole density of 1020 cm3 and a resistivity of the order of 10 mOhm cm, which is four orders of magnitude higher than the copper resistivity at 300 (2.24 Ohm cm); i.e., (Ga,Mn)As is a "bad" metal. Thus, the DMS conductivity is similar in order of magnitude to the conductivity of non magnetic semiconductors. This is a great advantage of the DMS and HTFS over metals when used for spin injection appli cations. At first, (Ga,Mn)As exhibited mostly nonmonotonic dependences of TC (see Fig. 1) and of the density of free holes on the dopant concentration. The magnetization of samples was also not in proportion to the Mn content. Annealed samples grown by low temperature molecu lar beam epitaxy exhibited a more monotonic depen dence TC(x). Interestingly, the Curie temperature of com pound (In0.5Ga0.5)0.93Mn0.07As (~110 K)259 is the same as that of Ga0.945Mn0.055As.81 At present, Ga1xMnxAs materials with x 910 at.% and heterostructures of corresponding composi tion255257,265283,** are the best studied DMS. Manga
* B. Gallagher, http://ep2ds14.fzu.cz/ms/level2.nott.php. ** D. Chiba, Y. Sato, T. Kita, F. Matsukura, and H. Ohno, http://arXiv.org/cond mat/0403500.
2379
nese atoms not only replace Ga atoms during the forma tion of GaMnAs alloy but can also occupy interstitial sites in poorly annealed samples,284,285 though GaAs samples grown at high temperatures have a stoichiometric compo sition. However, GaMnAs is obtained at relatively low temperatures. In this case a fraction of As atoms can occupy Ga sites, thus forming so called antisites AsGa which, in addition to the manganese interstitial defects, act as donors compensating the holes supplied by the Mn ions in gallium positions and thus reduce TC. Besides, the magnetic moments of the interstitial Mn ions tend to form antiferromagnetic bonds with substitutional impuri ties, thus reducing the average magnetic moment per Mn atom in (Ga,Mn)As. The Mn interstitial defects in (Ga,Mn)As films can be removed by coating the material with As layer followed by annealing at 473 K for 2 h.* Being a substitutional impurity with respect to Ga, man ganese is an acceptor and accepts an electron from the valence band of the GaAs matrix in order to saturate its own bonds, thus supplying a delocalized hole into the matrix. Deuteration causes saturation of all manganese bonds and abruptly decreases the density of delocalized holes, leaving the Mn content unchanged. This approach can be used to suppress ferromagnetism in ferromagnetic Ga0.963Mn0.037As film 320 nm thick with TC = 70 .286 Ferromagnetic (In,Ga,Mn)As films were grown on InP substrates (001) by molecular beam epitaxy.287,288. Using low temperature molecular beam epitaxy, a manganese concentration of 69 at.% can be attained in Ga1xMnxAs film; a specific multilayer technique (digital alloys)252 allow a Mn content of up to 14 at.% to be achieved. At higher Mn concentrations, the second phase is formed with the crystal structure different from the zinc blende structure. Multiphase composition and reduction of lattice symmetry to hexagonal (NiAs type) or orthor hombic (MnP type) dramatically deteriorates the opto electronic properties of the material. DMS are also synthesized from organometallic com pounds using vapor phase epitaxy. (Ga,Mn)As films thus grown contain up to 4 at.% of Mn but also in clude nanoclusters formed at high growth temperatures (673873 K) that are necessary for decomposition of the compounds mentioned.289,290 The Curie temperatures of dilute magnetic semicon ductor (Ga,Mn)As reported in different studies are plot ted vs. manganese content in Fig. 1. At this writing a "three layer" system (Ga,Mn)As/GaAs/(Ga,Mn)As ex hibited TC = 160 K 291 while Mn doped GaAs/Be doped p type AlGaAs heterostructures (the so called doping, or digital alloy) were reported to have TC = 172 K.292
* M. Adell, V. Stanciu, J. Kanski, L. Ilver, J. Sadowski, J. Z. Domagala, P. Svedlindh, F. Terki, and C. Hernandez, http://arXiv.org/cond mat/0406584.
A (Ga,Fe)As alloy with n type conductivity and Fe content up to 10 at.% was obtained293 by molecular beam epitaxy on GaAs substrate in the temperature range from 533 to 623 K. The lattice constant of the doped material decreased with increasing the Fe concentration because the FeAs bond is shorter than the GaAs bond. At lower temperatures, gallium atoms are replaced by iron atoms in the zinc blende lattice sites and the (Ga,Fe)As films become paramagnetic. At higher temperatures, Fe clusters and FeAs complexes possessing ferromag netic properties294 are formed. Molecular beam epitaxy was also used for the synthe sis of (Ga,Cr)As with Cr concentration up to 10 at.%. A number of samples exhibited no ferromagnetism; at the same time a TC of nearly 45 K was reported. Similarly to (Ga,Fe)As, these materials can be inhomogeneous. It should be noted that (In,Mn)As, (Ga,Mn)As, and (Ga,Mn)P samples with the highest TC have a high hole density (~1020 cm3) and their characteristics are typical of ferromagnetic materials. For instance, a domain struc ture (stripes of width 1.5 m) was observed at 5 K in ferromagnetic Ga 0.957Mn 0.043As film grown on Ga0.84In0.16As substrate.295 Rectangular magnetization hysteresis loops also favor normal ferromagnetism in DMS. As the temperature increases, the magnetization and coercive force of the materials decrease, the magneti zations of DMS and isoelectronic HTFS being not pro portional to the dopant (magnetic ion) concentration. Recently, above room temperature ferromagnetism in dilute magnetic semiconductor DMS (Ga,Mn)P with the zinc blende structure was reported. Ion implantation dop ing of p type GaP:C (carrier density ~1020 cm3) with manganese to ~6 at.% of Mn followed by annealing at 700 C allowed TC = 270 K to be achieved.296 Using molecular beam epitaxy, one can obtain a DMS (Ga,Mn)P with TC = 330 K.297,298 According to theory (see below), a TC(x) dependence should also be expected for (Ga,Mn)P type DMS and HTFS. Owing to weakened spin orbit coupling in the type IIIN semiconductors compared to type IIIAs semiconductors, doped nitrides are expected to have longer spin relaxation times and spin coherence lengths compared to arsenides. First attempts to detect ferromag netism in wrtzite type (Ga,Mn)N crystallites grown at high temperatures (~1200 C) have failed.299 Using dif ferent regimes of Mn diffusion at 5231073 K in epitaxial GaN films grown on sapphire substrates, (Ga,Mn)N samples with TC = 220370 K were prepared.300,301 The samples obtained by ion implantation were characterized by lower TC.302 (Ga,Mn)N samples with n type conduc tivity and Mn content of 69 at.% were also obtained by molecular beam epitaxy at 853993 K.303 Measurements of the temperature dependence of magnetization carried out up to 750 K and extrapolation of this dependence to
2380
Ivanov et al.
zero revealed TC = 940 K. However, the results ob tained in this study and in other studies (TC > 900 K in (Ga,Mn)N) require verification because no ferromagnetic manganese compounds with gallium or nitrogen, possess ing so high TC have been reported so far. At the same time, a ferromagnetic MnGa alloy is characterized by TC 750 K.304 n Type (Ga,Mn)N was reported to have TC = 320 K.305 Room temperature anomalous Hall effect mea surements confirmed that manganese doped gallium ni trides and phosphides are homogeneous ferromagnets. However, EXAFS data and a study306 point to p type conductivity of the HTFS (Ga,Mn)N and to the presence of some amount of manganese clusters in the samples, which increases with the Mn concentration. Studies of (Ga,Mn)N are important for establishment of the nature of ferromagnetism in DMS because (Ga,Mn)N is a high TC material with n type conductivity, whereas other high TC DMS, e.g., (Ga,Mn)As and (Ga,Mn)P have a p type conductivity. Gallium nitride containing implanted Fe 307 and Ni 308 exhibits ferromagnetism with TC 200 and 50 K, respec tively. (Ga,Fe)N films grown by molecular beam epitaxy are characterized by TC < 100 K.309 Of interest is a DMS (Ga,Cr)N with TC > 400 K, prepared310 by molecular beam epitaxy at 973 K on sapphire substrate. A (Ga,Cr)N sample was characterized311 by TC = 280 K, n type con ductivity with a carrier density of 91018 cm3, a mobility of 150 cm2 (V s)1, and a magnetic moment of 1.1 B per Cr ion. Characteristics of DMS also depend on the annealing conditions.263,271,312 Materials belonging to the (III,Mn)V family possess new spintronics related properties of doped semiconductors, such as injection of long lived spins, car rier induced and optically controlled ferromagnetism (see a review44), and electric current driven magnetization re versal of the heterostructures.313 Yet another material is aluminum nitride AlN, which is used in opto electronic devices that operate in the green, blue, and UV regions of the optical spectrum. Chromium doped AlN also exhibits interesting optical properties. In addition, doping of this material with chromium occurs at higher temperatures compared to the doping of other type IIIV matrices with manganese, which favors the syn thesis. Pioneering results314 included the synthesis of Al1xCrxN films with chromium content up to x = 0.357 and Curie temperatures exceeding 340 K. (Al,N)Cr films (TC > 600 K) obtained by the same molecular beam epit axy technique were studied.315 Here, the expected mag netic moment of the Cr3+ ion occupying an Al vacancy is 3 B. The measured magnetic moments were equal to 1.0 B per Cr atom at 300 K and 1.2 B per Cr atom at 10 K. From here it follows that at room temperature only 33% of Cr3+ ions are magnetoactive, while at low temperatures the proportion of such ions in the doped
AlN is 40%. The presence of magnetic phases CrO2 or Cr2O3 in the (Al,N)Cr films grown cannot be ruled out. Al1xCrxN thin films (0.02 < x < 0.1) on sapphire substrates were synthesized.* At 300 and 50 K, the mag netic moments of Cr3+ ions were respectively 0.62 and 0.71 B per Cr atom. At x = 0.027, ferromagnetism with TC > 900 K was observed, probably, TC approaching 1000 K. Indium nitride InN possesses useful spintronics re lated transport properties, e.g., a high carrier drift veloc ity. Epitaxial (In,Cr)N films were grown on a sapphire substrate at 623723 K using a buffer GaN layer to smooth out differences between the lattice constants of the sub strate and the film grown. X Ray analysis revealed a monophase single crystalline structure, while photolumi nescence measurements confirmed the existence of a band gap of width 0.78 eV, as in the starting InN matrix. SQUID magnetic measurements of the (In,Cr)N fields revealed a nearly room temperature Curie point of the material. Noteworthy is that Hall effect measurements in (In,Cr)N revealed n type conductivity with an electron concentra tion of ~1020 cm3.** Among type IIIV semiconductor matrices, InSb has the narrowest band gap (0.17 eV) and the highest electron mobility. This material is used for fabrication of sensitive Hall sensors; therefore, interest in the preparation of InSb based DMS is quite understandable. (In,Mn)Sb epi taxial films obtained316 by molecular beam epitaxy exhib ited ferromagnetism with TC = 20 K (samples with p type conductivity and a hole density of 1.11020 cm3), whereas the samples with n type conductivity and an electron den sity of 8.61018 cm3 were found to be paramagnetics. 3.5. High temperature ferromagnetic semiconductors At present, DMS based on type IIIV matrices have been studied in most detail. High degree of spin polariza tion and the possibilities of optical or electric control of the magnetic properties and of integration with semicon ductor circuits make the DMS quite attractive from tech nological viewpoint. However, fabrication of DMS using molecular beam epitaxy technique can produce anti structural, interstitial, and other uncontrollable defects that deteriorate the transport properties of materials. To date, defects in the (III,Mn)V systems are removed by prolonged controlled annealing. This allows the hole mobility to be enhanced to, e.g., 10 cm 2 (V s)1 in (Ga,Mn)As. Most of DMS compounds are ferromagnetic p type semiconductors with spin polarized holes, which leads to lower carrier mobilities and shorter spin relax
* D. Kumar, J. Antifakos, M. G. Blamire, and Z. H. Barber, http://arXiv.org/cond mat/0404436 (unpublished results). ** R. Rajaram, R F. C. Farrow, G. Solomon, J. S. Harris, and S. S. P. Parkin, Opening Event of the IBM Stanford Spintronic Science and Applications Center (April 26, 2004), 2004.
2381
ation times compared to those of the n type materials. Yet another drawback of the DMS materials is that some of them must be synthesized using Ga, which is incompat ible with the most widely used material for microelec tronics, silicon, and toxic arsenic. Nevertheless, spintronic DMS materials are expected to be of great promise be cause, theoretically, their Curie temperatures can lie above room temperature. Digital alloys ( doping) have higher TC compared to homogeneous DMS. Other semiconductors, namely, type IIIVV2 and type IIIIVI2 compounds crystallizing in the chalco pyrite structure,200 are also of interest. Chalcopyrites ZnGeP2 and CdGeAs2 possess unusual nonlinear optical properties and can be used for fabrication of optical modu lators and frequency converters. Compound ZnSnAs2 is also a candidate for optical applications in the far IR region. The lattice constants of wide gap chalcopyrites ZnGeN2 and ZnSiN2 are similar to those of GaN and SiC. Because of this, ferromagnets based on these hosts would favor fabrication of hybrids of magnetic sensors or switches with lasers operating in the blue/green/UV re gions of the optical spectrum, light emitting diodes, and microwave electronic devices based on GaN and SiC. A matter of no small consequence is that the width of the forbidden band in ZnGexSi1xN2 is a linear function of the germanium content and varies over a wide range, from 3.2 eV (x = 1) to 4.46 eV (x = 0). Yet another advantage of the type IIIVV2 and IIIIVI2 com pounds as hosts for novel ferromagnetic semiconductors is facilitation of p type doping with high carrier mobility in these ternary compounds. The synthesis of ferromagnetic semiconductors CdGeP2:Mn (averaged Mn/Cd ratio of 20%) with TC = 320 K 317 and ZnGeP2:Mn with TC = 350 K 318 marked the emergence of novel spintronic materials, namely, high temperature ferromagnetic semiconductors (HTFS). Similarly to semimagnetic or dilute semiconductors, in the HTFS the atoms of the semiconductor host are ran domly replaced by transition metal and, probably, REE impurities. Compounds AIIBIVCV2 are electronic analogs of compounds AIIIBV because the average valence of AII and BIV in HTFS equals the valence of the element AIII in DMS. Here, the density of itinerant holes (charge carrier density) is an order of magnitude lower than the Mn impurity content (density of localized spins). Analysis of the properties of HTFS based on type IIIVV2 and IIIIVI2 compounds shows that, compared to DMS, the Mn2+ impurity in HTFS can replace both the atoms of Group II or I elements and the atoms of Group IV or III elements, thus playing the role of accep tors and donors of mobile holes. This was confirmed by measuring the EPR spectra of (ZnGe,Mn)P2 319,320 that revealed two types of substituted manganese sites, MnZn and MnGe.
Recently, high temperature ferromagnets ZnGeP2:Mn (TC = 312 K),321 ZnSnAs2:Mn (TC = 329 K)322 and ZnSiGeN2:Mn (TC 300 K)323,324 were discovered. Stud ies of the transport properties of epitaxial HTFS with composition ZnSiGeN2:Mn (5 at.% Mn; TC 300 K), grown on sapphire substrate, revealed n type conductiv ity.324 Compared to DMS, HTFS are characterized by higher mobility of both p and n type carriers. Completely substituted chalcopyrites MnGeP2 and MnGeAs2 possess ferromagnetism with TC = 320 and 340 K, respectively,325 and are of academic interest because their semiconductor characteristics differ from those of the starting materials ZnGeAs2 and ZnGeP2. High temperature ferromagnetic semiconduc tors (CdGe,Mn)P 2 and (ZnGe,Mn)P 2 were ob tained317320,326 by Mn deposition on pre synthesized single crystalline substrates CdGeP2 and ZnGeP2, respec tively, at nearly 673 K followed by manganese diffusion into the interior of the samples. The HTFS thus obtained contained no MnP phase. A study327 of (ZnGe,Mn)P2 showed that the manganese diffusion profile reflects in homogeneity of the sample with a metallic surface layer covering (ZnGe,Mn)P2. Layer by layer etching revealed ferromagnetism in both phases (the surface phase and the internal phase). A 8 mm 8 mm single crystal and polycrystals of (ZnGe,Mn)P2 were synthesized starting from Zn, Ge, Mn3P2, and P in quartz ampules by stepwise heating to 1400 K followed by slow cooling. A single crystalline sample with a Mn concentration, x, of 0.03 and polycrys talline samples with x = 0.045, 0.056, and 0.2 (calculated for the composition Zn1xMnxGeP2) were obtained. The samples with x = 0.056 and 0.2 were characterized by TC = 312 K and exhibited antiferromagnetism at tem peratures below 47 K. HTFS samples of composition (ZnGe,Mn)P2 with a Mn content, x, of 0.08 and 0.15 were synthesized accord ing to Ref. 321 and then studied.328 Measurements of the 55Mn and 31P NMR spectra of the samples with x = 0.15 at temperatures above and below TC 300 K showed that more than 90% of Mn atoms are clustered to form an impurity phase MnP with cluster sizes of the order of 10 nm. At present, intensive theoretical and experimental re search198 on novel HTFS has been carried out. Single crystals and polycrystals of (CdGe,Mn)As2 were first ob tained using pre synthesized CdAs2, Mn powder, Ge, and As.329 Single crystals of (CdGe,Mn)As2 containing 0.5 wt.% Mn, were grown using a vertical modification of Bridgeman technique followed by heat treatment. Ther mal e.m.f. measurements at 300 K revealed a p type con ductivity of the materials containing less than 6 wt.% of Mn and X ray phase analysis confirmed that Cd and Ge atoms in the lattice sites are replaced by Mn atoms.
2382
Ivanov et al.
(CdGe,Mn)As2 samples containing 1, 3, and 6 wt.% of Mn appeared to be novel HTFS with TC 355 K.329 Considerable attention has also been paid to the prepa ration of HTFS based on carborund SiC, which is used in high temperature high power electronics. Carborund has more than seventy polymorphs, some of them being struc turally similar to wrtzite. The 6H polymorph has a broad forbidden energy band (3 eV), p type conductivity, good transport properties, and can be doped. By ion implanta tion of Ni, Fe, or Mn (up to ~5 at.%) followed by anneal ing of the 6H SiC polymorph of carborund one can obtain ferromagnets with TC = 50 (6H SiC:Ni), 250 (6H SiC:Mn), and 270 K (6H SiC:Fe)* with a hole den sity of ~1017 cm3 and a coercive force of 150 and 50 Oe (at 10 K) for the last mentioned two samples.330 3.6. Organic spintronic materials At present, spintronics employs "sandwiches" that are sputtered or grown on thick substrates. But at long dis tances travelled by electrons passing through one layer to another layer the electron spins lose their orientation. This significantly deteriorates the current switching prop erties of devices. A decrease in the layer thickness to a few atoms must favor solution to this problem. At present, it is possible to fabricate thin layers of organic conductors and even superconductors on the surface of a polymer matrix. This is an ancestor of conducting organic strips in elec tronic circuits. Organometallic compounds possess a va riety of properties and have definite chemical composi tions. One can influence the molecules in the conducting layer, which permits control of the electrical properties. For instance, a quasi two dimensional material based on bis ethylenedithiotetrathiofulvalene (BEDT TTF) mol ecules has the same chemical composition but can be a conductor, antiferromagnet, or a superconductor depend ing on the mutual arrangement of molecules in layers or on the layer thickness (see Ref. 30 and references cited therein). A new class of ferromagnetic spintronic materials, namely, organic ferromagnets were discovered at the end of the 20th century. Among them, crystals with structures comprised of alternating conducting layers based on the BEDT TTF molecules and insulating layers (oxalates of magnetic ions Fe3+, Cr3+) can be pointed out. This class of materials opens the possibility of electric current driven control by changing the spin orientation in the magnetic field. A possible practical application of such crystals are devices for molecular and spin electronics. Organic ferro magnets also comprise molecular ferromagnets based on organic paramagnetic molecules (stable radicals and polyradicals and high spin carbene molecules) or para magnetic complexes with magnetic ions encapsulated in a
* J. Kim, F. Ren, S. J. Pearton, C. R. Abernathy, M. E. Overberg, G. T. Thaler, and Y. D. Park, unpublished results.
ligand shell comprised of organic molecules. To fabricate spintronic materials with different structures and spin den sities, molecules and organometallic complexes contain ing ten to twenty and more electron spins per particle and a magnetic moment of up to several tens of Bohr magne tons are synthesized. Molecular ferromagnets have some drawbacks, namely, a low spin density and a large propor tion of chemical "garbage" and therefore weak exchange interaction and low TC (330 K). But molecular ferro magnets are of great interest as promising high tech ma terials; interest in them has even increased in connection with the development of spintronics. Recent information on the preparation of high temperature molecule based ferromagnets with TC = 315330 K stimulated this inter est. Unfortunately, unambiguous structural characteriza tion of these substances has not been reported so far.* Among organic spintronic materials, mention may also be made of recently synthesized magnetic plastic based on vanadium tetracyanoethanide.331 This material pos sesses magnetic properties at high temperatures (above boiling point of water) and can therefore be used in spintronic devices without using a cooling system. A current vs. voltage curve of a benzene 1,4 dithiol molecule connecting two gold contacts was studied.332 In the future such nanostructures fabricated from molecular wires based on organic molecules connecting transition metal nanocontacts will make it possible to realize spin dependent transport through organic molecules, thus opening the field of molecular spintronics.333 3.7. Inhomogeneous magnetic materials Monophase magnetic materials are best for efficient spin injection into semiconductors. However, inhomoge neous materials with high TC of the magnetic component can also be useful in spintronics, especially for magneto optical applications. An example is provided by the semiconducting mate rials TiO2:Co 334,335 with the anatase structure, which sometimes possess room temperature ferromagnetism. Magneto optical and magnetic measurements336 revealed a large hysteresis at room temperature. However, the scat ter of magnetization depending on the method of prepa ration of the sample indicates the presence of Co or an unknown compound of the system CoTiO in the ana tase matrix. The results of the temperature measurements of resistance337,338 indicate a multiphase composition of the material and the possibility for cobalt clusters to be present in the TiO2 matrix. At the same time, the Ti0.92Co0.08O2 films deposited on SrTiO3 substrates ex hibit room temperature ferromagnetism only in a limited range of carrier concentrations (2101851022 cm3),339
* A. L. Buchachenko and R. Z. Sagdeev, Perspektivnye Tekhnologii [Advanced Technologies], 10, Iss. 5, March 15, 2003 (in Russian).
2383
which excludes clustering of cobalt. Clearly, further physi cochemical studies of TiO2:Co are required. Iron doped TiO2 epitaxial films with the rutile struc ture possess ferromagnetism with nearly room TC,340 while TiO2:V thin films with the anatase structure, grown on a LaAlO3 substrate are ferromagnets with TC > 400 K, which was confirmed by anomalous Hall effect measurements.341 It should be noted that the first studies of (In,Mn)As also revealed ferromagnetic inclusions similar to the MnAs phase with TC 300 K; homogeneous DMS were ob tained later.6,7 It is important that interesting results were obtained in experiments with the multiphase and nonequilibrium tran sition metal pnictides. For instance, a material prepared by deposition of 3 nm MnSb islands on a GaAs matrix is characterized by an easy magnetization reversal and has a giant magnetoresistance in low fields (effect is as high as 880% in a magnetic field of 1 T and 320000% in a 2 T field at room temperature).342 The formation of nanostructures (including ferromagnetic ones) in MnAs films (TC = 310 K) deposited on GaAs(001) substrate was observed at 30 C.343 Unstable compounds CrAs 209,344 and CrSb 345 with the zinc blende structure were prepared by deposition on GaAs substrates. These materials possess room temperature ferromagnetism. Because of 100% spin polarization they can serve excellent sources of spins grown on technologically important semiconductor sub strates. Yet another example of inhomogeneous spintronic ma terials is provided by the materials based on ZnO, which is widely used in opto electronics and cell phones. Smaller in size and faster optical modulators, detectors, lasers, etc. based on zinc oxide can be produced with ease. Zinc oxide crystallizes in the wrtzite structure, possesses piezo electric properties, and has a broad forbidden band (3.3 eV) and a high limit of transition metal solubility. From this, interest in doping ZnO with transition metals to obtain ZnO based ferromagnetic semiconductors possessing pi ezoelectric properties is quite understandable. The doped material must not contain mobile carriers capable of screening electric fields induced in the material. The Mn, Fe, or Co doped nanocrystalline ZnO thin films were obtained by sol gel technique.346350 Recent studies of bulk and thin film ZnO:M (M = Co,351,352 Mn,353,354 and V 355) samples confirmed their room temperature ferromagnetism (e.g., TC = 290380 K for ZnO:Mn) in contrast to studies356,* where a spin glass behavior was assumed. At the same time, detailed measurements357 of the magnetic properties of ZnO:M (M = Cr, Mn, Fe, Co, Ni, Cu) epitaxial films revealed no
* G. Lawes, A. S. Risbud, A. P. Ramirez, and R. Seshadri, http://arXiv.org/cond mat/0403196.
magnetic ordering down to 3 K. However, ferromagnetism with TC 50 K was observed in a study of diffusion doping of four pod ZnO nanostructures with manganese at 873 and 1073 K.358 The nickel doped ZnO nanocrystals with TC > 350 K were synthesized from solution.359 The ZnO:Ni2+ nanocrystals with a characteristic size of the order of 6 nm exhibit paramagnetism but their aggrega tion gives rise to ferromagnetism, which is probably due to additional doping with nickel during the growth (Ni content in the material was about 1 at.%). The results of structural, magnetic, and optical measurements of Zn0.9Co0.1O epitaxial films obtained by laser deposition on a sapphire substrate were reported;360 the films exhibit ferromagnetism with TC = 300 K (the presence of nano clusters and inhomogeneities is excluded). The hysteresis loop was observed up to T = 350 K in the Zn0.91Co0.09O films kept in Zn vapors* and up to T = 300 K in (Zn,Co)O nanocrystals.361 However, in similar situations one must feel certain of homogeneity of the isoelectronically doped ZnO. In the case of ZnO one can not rule out the possi bility of formation of noninteracting antiferromagnetic clusters comprising impurity atoms with an increase in the dopant concentration; this was confirmed in the stud ies of monophase (Zn,Mn)O and (Zn,Co)O samples pos sessing no ferromagnetism above 2 K.** Oriented ZnO(110) films on sapphire substrates were doped*** with transition metals by laser deposition at 600 C. Room temperature ferromagnetism in the films was observed at Sc, Ti, V, Fe, Co, and Ni concentrations of about 5 at.%, whereas the films doped with Cr, Mn, and Cu possessed no ferromagnetism. Ferromagnetism was also observed in ZnO:Co 361 and CdSe:Co 362 quan tum dots. Additional doping of the transition metal containing ZnO bulk phases with copper increases the density of current carriers that provide with indirect exchange inter action between magnetic impurities, which can cause ferromagnetism. For instance, a TC of 550 K in a Zn0.94Fe0.05Cu0.01O sample was reported.363 Ferromagnetic nanocrystals CdSe:Mn,364 ZnS:Mn,365 ZnSe:Mn,366 InAs:Mn,367 and TiO2:Co **** can be prom ising as ferromagnetic coatings for spintronics appli cations. A novel ferromagnetic semiconductor material with TC > 400 K and a structure comprised of alternating thin layers (a few atoms thick) of GaSb, GaMn, and their
* D. A. Schwartz and D. R. Gamelin, http://arXiv.org/ cond mat/0404518. ** G. Lawes, A. S. Risbud, A. P. Ramirez, and R. Seshadri, http://arXiv.org/cond mat/0403196. *** M. Venkatesan, C. B. Fitzgerald, J. G. Lunney, and J. M. D. Coey, http://arXiv.org/cond mat/0406719. **** J. D. Bryan, S. M. Heald, S. A. Chambers, and D. R. Gamelin, http://arXiv.org/cond mat/0407451.
2384
Ivanov et al.
mixture was prepared by molecular epitaxy.368 The au thors believed that the material also retains the desired properties at higher temperatures, although a TC of 400 K is sufficient for successful use of the material in spintronic devices at room temperature. Doping of hosts with the rutile structure is also of interest. A transparent Sn0.95Co0.05O2 film grown by laser deposition on a sapphire substrate exhibited high temperature ferromagnetism with TC = 650 K, but a very high magnetic moment (7.50.5 B per Co atom) can be indicative of inhomogeneous sample.369 Early studies of (Ga,Mn)As 370 pointed to a low Mn solubility in the ma trix and the formation of magnetic nanoclusters that com plicate the determination of TC. Therefore, criticism must be expressed when analyzing the results obtained in stud ies concerned with novel ferromagnetic semiconductors possessing nearly room temperature TC (see, e.g., a re view44). For instance, ferromagnetism in lanthanum doped compound Ca1xLaxB6 (0.01 < x < 0.5) with TC 900 K was reported,371373 but the authors of study* found that in this case ferromagnetism is due to the presence of Fe and Ni impurities contaminated the material during synthesis. DMS based on silicon hosts would also be of applied value. Unfortunately, (Si,Mn) samples obtained by depo sition technique374 or molecular bean epitaxy375 have a two phase composition; the ferromagnetic phase (tenta tively, SiMn) has TC 30 K and the antiferromagnetic phase Si3Mn5 has the Neel temperature TN 95 K.375 In this connection of considerable interest is a re cent magnetic and structural study376 of single crystal line Mn0.05Si0.95 films obtained by heat treatment of amorphous films deposited on the Si(001) surface. SQUID Measurements revealed ferromagnetism with TC > 400 K; however, in this case multiphase composi tion of the material also cannot be ruled out.376 Investigations of light induced ferromag netism251,377,378,** or electric field controlled ferromag netism151,152,154 allow one to distinguish between true fer romagnetism in magnetic semiconductors, which origi nates from indirect exchange interaction between impu rity spins involving delocalized carriers and ferromag netism induced by magnetic nanoclusters. A photo induced phase transition to ferromagnetic state in (Ga,Mn)As films ( x = 0.011, hole density less than 1020 cm3, TC = 30 K) was realized at 60 K,377 the magnetization being equal to 15% of the saturation mag netization.
* C. Bennett, J. van Lierop, E. M. Berkeley, J. F. Mansfield, C. Henderson, M. C. Aronson, D. P. Young, A. Bianchi, Z. Fisk, F. Balakirev, and A. Lacerda, http://arXiv.org/cond mat/ 0306709. ** G. A. Khodaparast, J. Wang, J. Kono, A. Oiwa, and H. Munekata, http://arXiv.org/cond mat/0305017.
4. Ferromagnetism in spintronic materials 4.1. Models of ferromagnetism in dilute magnetic semiconductors DMS and HTFS exhibit a number of new properties of doped semiconductors, which are of importance for spintronics applications. These are a high degree of spin polarization, an injection of long lived spins, an electri cally and optically controllable ferromagnetism, and a possibility for DMS and HTFS of being embedded in the known semiconductor circuits. Although ferromagnetism in the best studied DMS (Ga,Mn)As was first observed as early as 1996 with TC = 60K, the microscopic nature of ferromagnetism in the DMS based on type IIIV hosts and in isoelectronic HTFS based on type IIIVV2 hosts, etc., is not well theoretically understood as yet. It is unclear how can magnetism in DMS and HTFS occur provided rather long distances between magnetic impurities and even greater separations between current carriers. The well known models for exchange interac tions (direct or double one) cannot give explanations. A characteristic feature of DMS and HTFS is the depen dence of their magnetic, electric, and optical properties on the dopant concentration (x) or on current carrier density, but an empirical rule TC = (2000x10) K 81 is invalid for Ga1xMnxAs. Ferromagnetism in DMS and HTFS belongs to the most interesting problems in basic research. In contrast to cryoelectronics that employs the BardeenCooperSchriefferBogolubov microscopic theory or the GinzbirgLandau phenomenological ap proach for calculations, no substantiated spintronics ori ented concepts of the mechanism of magnetic interaction between impurity spins and of the origin of high TC in DMS and HTFS have been proposed so far. To make the DMS and HTFS real rather than hypothetical spintronic materials, one should answer the question as to the nature of ferromagnetism in and characteristics of the semicon ductor matrix and the magnetic dopants responsible for high TC. Most theories of ferromagnetism in such DMS as (Ga,Mn)As, (Ga,Mn)P or (Ga,Mn)N are based on (semi)phenomenological models. There are three ap proaches, which can be followed to study magnetic states in DMS and HTFS. The first approach postulates the existence of local magnetic moments, a ferromagnetic exchange interaction of these moments with valence elec trons (exchange parameter Jpd < 0), and an antiferromag netic interaction of these moments with the conduction electrons of the semiconductor matrix. This approach is used for the description of ferromagnetism in such DMS as (Ga,Mn)As, (Ga,Mn)P, and (Ga,Mn)N (see a re view379 and Refs 79, 380382). Some of these theories take into account the role of shallow acceptors383 or deep
2385
resonance levels.384,385 Note that the expression77,78 (see also Ref. 79) for TC containing the phenomenological fitting parameter Jpd of the exchange interaction between localized and delocalized electrons, which was used in numerous studies of magnetic semiconductors (especially, DMS) coincides, correct to the change Jpd Jsd, with the AbrikosovGorkov equation derived in the mean field approximation for the TC of alloys of magnetic impurities with non magnetic metals386:
,
where a is the lattice constant, S is the spin of the im purity whose concentration in the material is x, and h = 0.5g2B2() is the magnetic susceptibility of holes in the valence band of the semiconductor or of the conduc tion electrons of the metal (Pauli susceptibility). The sec ond term in this relation is proportional to m*kF p1/3 (kF is the Fermi wave vector and p is the density of holes with the effective mass m*); as applied to DMS, this means that TC xp1/3, which contradicts experimental data. It should be emphasized that the AbrikosovGorkov rela tion is applicable to the systems in which the concentra tion, x, of spin carriers (magnetic impurities) is much lower than the density, n, of charge carriers (e.g., to dilute alloys CuMn). But in the case of DMS and HTFS we deal with the reverse situation, namely, x >> n (hole den sity), which means that the distance between delocalized holes much exceeds the average distance between local ized impurities and, hence, direct Heisenberg type impu rity exchange or indirect RKKY magnetic exchange is impossible. It is the condition x >> n that is responsible for the fact that in the external magnetic field, which polarizes localized magnetic moments, the itinerant car riers "see" a large number of Mn2+ ions (strong effective magnetic field of these ions causes giant Zeeman splitting in DMS and semimetals and half metals). At sufficiently high concentration of charge carriers, a reverse effect is possible, namely, magnetization of localized magnetic moments by the charge carriers. As can be seen, the TC value is independent of the sign of the parameter Jpd. But if for metals the exchange inter action between the conduction electrons and impurity can be treated as a contact interaction (this is more cor rect for the f rather than d impurity orbitals), this approxi mation is at least rough in the case of semiconductors. The case in point is that, unlike metals, replacement of atoms of the semiconductor matrix by a non isoelectronic transition metal atom causes a strong disturbance and the localized atomic state is "pumped" from the heavy hole band. Owing to the character of the pd chemical bond the impurity d orbitals are mixed with p orbitals and the degree of delocalization of d orbitals strongly depends on the position of the d level in the semiconductor band and
on the concentration of magnetic impurities. These fea tures of the electronic structure cannot be described using a single parameter of contact exchange interaction Jpd, though intra atomic interactions of d electrons, Hund exchange JH, and the AndersonHubbard repulsion U undoubtedly remain among important factors. Semiphenomenological models of DMS are also re lated to some versions of the density functional theory (see, e.g., a review* and references cited therein). The Local Density Approximation (LDA) calculations of DMS ignore strong intra impurity electron correlations that sup press fluctuations in the number of d electrons. As a re sult, these calculations give unrealistically high d state density near the top of the valence band, underestimate the role of the 4p component of holes in the valence band, and overestimate the strength of spd hybridiza tion.387,388 Other versions of the density functional ap proach (e.g. LDA+U 389), which include the Coulomb repulsion between impurity 3d electrons (Ander sonHubbard parameter U) allow a more realistic elec tron energy spectrum and the 4p character of holes in the valence band to be obtained. However, first principle (ab initio) density functional calculations cannot86 intrinsically to correctly allow for high energy states induced in DMS and HTFS by mag netic impurities. Although application of this method led to some advances in the theory of simple metals, as ap plied to semiconductors, the approach has a fundamental drawback caused by energy variation over the electron density. This procedure cannot be correctly performed for DMS and HTFS because the minimum change in the electron density requires excitation of an electron hole pair with a finite energy equal to the width of the band gap in semiconductor, which here is not infinitesimal, as re quired in the variational method. First principle calculations are employed in the sec ond approach to studies of the magnetic properties of DMS and HTFS. It is based on a generalized cluster method; namely, a low magnetic impurity concentration is simulated using a cubic supercell with the magnetic ion at the center and then the cell is involved in calculations of the density of spin polarized states in the entire homo geneously doped crystal.80,390396 Sometimes an exchange Hamiltonian is introduced phenomenologically, with the effective magnetic exchange constant playing the role of a fitting parameter in numerical calculations.80,391 The clus ter approach is based on the assumption of a local pertur bation produced by an impurity in the matrix and treats the impurity region in the semiconductor crystal as a de fect quasi molecule in which the impurity ion is chemi cally bound to the neighboring matrix atoms.397,398 The matrix elements of the potential for a substitutional impu
* J. Sinova, T. Jungwirth, and J. erne, http://arXiv.org/ cond mat/0402568.
2386
Ivanov et al.
rity are related to the "oxidative ability", i.e., they are assumed to be proportional to the energy difference be tween the impurity orbital and the corresponding orbital of the atom to be replaced. Most methods used in the framework of the cluster approach are based on the tight binding approximation in which the electron wave func tions are represented by linear combinations of AO or MOs and provide some idea of the energy spectrum of a semiconductor containing one impurity. But these are mainly one electron methods, which cannot correctly al low for the intra impurity electron electron interactions (electron correlations) inside the transition or rare earth metal atoms, that are necessary for the description of the magnetic properties. Yet another problem of electronic structure calculations in the framework of the cluster ap proach is concerned with the outermost ("surface") atoms of the cluster, which have dangling bonds; localized en ergy levels of the surface atoms in the electron energy spectrum should be differentiated from the energy levels of the bulk atoms. If the supercell is a cluster whose size exceeds the localization radius of the impurity state, in a periodic structure comprised of such clusters the local ized levels are broadened to create an impurity band. In spite of drawbacks of the cluster approach, numeri cal calculations of DMS and HTFS show that differences between the physical and chemical nature of transition metal impurities and cations in the semiconductor cause significant changes in the electron energy spectrum. Heavy transition metal impurities in the crystal give rise to quite essential states that are few in the allowed energy bands. The electron electron interactions and covalent effects governing the structure of chemical bonds play an impor tant role in the formation of the properties of impurities with non filled shells in the semiconductor hosts. The third strategy of studying the magnetic states in DMS and HTFS is presented in Sections 4.2 and 4.3. 4.2. Electronic structure of magnetic impurities in dilute magnetic semiconductors and high temperature ferromagnetic semiconductors A consistent theory of exchange interactions in DMS and HTFS, which explains the reasons for magnetic or dering must allow for the structure of chemical bonds between the impurity and the matrix as well as for the Coulomb and electron electron exchange interactions in the 3d shells (intra atomic correlations). Usually, the energy levels of simple isoelectronic sub stitutional impurities lie near the forbidden energy band, being comprised of the contributions of the valence elec trons of the semiconductor matrix. At the same time, taking Pb1xSnxTe as an example, it was proposed399 that this rule can be violated for charged impurities of Group III elements in the type IVVI hosts. This violation appeared to be logic for DMS and HTFS. Experimental data sub
stantiating the existence of shallow impurity (Cr, Mn, and Fe) levels in narrow gap type IIIV semiconductors (InAs, InSb) were first reported in a monograph.189 According to the third approach, in studying the mag netic states and ferromagnetism in DMS and HTFS vari ous interactions involving d electrons are derived taking into account the chemical bonds and electronic structure of type IIIV, IIIVV2, etc. semiconductors doped with transition metals. Such a theory permits the descrip tion of the electronic structure of these impurities, con tains all ingredients for an accurate derivation of indirect magnetic exchange between impurities, and allows the properties of DMS and HTFS to be predicted. Peculiari ties of the electronic structure of a semiconductor with one d impurity were studied back in the mid 1970s 61,62 (see also monographs400402). A tetrahedral crystal field characteristic of semicon ductors splits five fold degenerate (with respect to the orbital momentum projection) impurity d levels into three fold degenerate t2 and doubly degenerate e sublev els, the latter lying lower on the energy scale. In other words, the localized d electrons of transition metal impu rities possess t2 and e symmetry of the semiconductor point symmetry group Td, so the magnetic impurity ion replacing atoms of Group II(s2) or III(s2p) elements can have the dn(ert2nr) or dn1(ert2nr12) configuration. Co valent bonds involving the t2 orbitals are nearly type and strong, whereas the e orbitals form weak bonds involving p orbitals. The electronic configurations of the 4fn6s2 REE impurities in the semiconductor hosts are similar to those of transition metal impurities. However, these apparent d and f configurations are violated in the case of the impurities with half filled electron shells that retain its high spin state in accord with the Hund rule. Namely, in the case of cationic replacement manganese ions in the GaAs and GaP hosts retain the 3d5 state while europium ions in InP retain the 4f7 state. In both cases these impurities give rise to acceptor levels in the forbid den energy band. Having determined positions of the ac ceptor levels relative to the top of the valence band from the optical absorption edge and the position of the bottom of the band gap from the photo ionization energy of the semiconductor, one can determine the "absolute" posi tions of the bottom and top of the forbidden band and of the impurity levels relative to vacuum.401,402 Thus it was established that the true reference point for the energies of the 3d levels in type IIIV (for impurities from V to Cu) and IIVI (for impurities from Sc to Zn) semi conductors is the photo ionization threshold of free 3d atom. In other words, the energy levels of magnetic impurities in semiconductors are not related to the top or bottom of the band gap or to other characteristics of the electronic structure of different hosts; rather, they are calculated relative to the absolute reference point, namely, the vacuum level. This means that as the forbidden band
2387
of the semiconductor matrix is broadened (i.e., with a decrease in the bond covalency in the matrix), the energy levels of magnetic impurities calculated with respect to the top of the valence band increase. Numerous experi mental data indicate a significant redistribution of the electron and spin density in the region between the impu rity and the neighboring atoms, the spin density perturba tion near the impurity site extending up to the fifth coor dination sphere; this was confirmed by analyzing double electron nuclear resonance spectra.400 Typical of transition metal impurities in the semicon ductor hosts of DMS and HTFS is tunneling of impurity d electrons to heavy hole p states and back, described by a very important hybridization parameter V that exceeds 1 eV (parameter of hybridization with light holes is small). Therefore, changes in the electron d states are due to the effect of the electronic states of the crystal environment. Unlike other impurities in semiconductors, transition metal ions can create impurity levels not only in the al lowed energy band but also in the forbidden energy band. The bonding configurations of the p orbitals of the va lence band with d orbitals, where the d component domi nates, are classified as the crystal field resonances (CFR). The antibonding configurations with predominant con tribution of the p component of the valence band are called the dangling bond hybrids (DBH). Manganese ions Mn2+(d5) and Mn3+(d4) involved in the exchange inter action in DMS and HTFS have populated electron con figurations d4(e2t2) and d5(e2t3) in accord with the Hund rule. The d6(e3t3) charge state lies much higher than these configurations on the energy scale due to strong intra atomic Coulomb repulsion (AndersonHubbard param eter U ). Empty CFR levels (d6/d5) are pinned deep in the conduction band, while the antibonding e orbitals should be excluded from consideration, which will be done in analyzing the Hamiltonian (see below). Empty CFR lev els can be responsible for antiferromagnetic ordering of the impurity ions, which is rather weak in DMS and HTFS at optimum impurity concentrations. The qualitative picture of the electronic structure of 3d impurities presented above is confirmed by the results of rigorous calculations using the following Hamiltonian for magnetic impurities in the semiconduc tor matrix:
. (1)
hybridization parameter, V, with impurity d electrons. is the creation (annihilation) operator of Here, the hole with the momentum p, the spin orientation , and the energy . The next term in Hamiltonian (1),
,
describes the subsystem of impurity d electrons in the semiconductor matrix, where Ed is the initial non renormalized level of the magnetic impurity in the semi is the number operator for the conductor, and d electrons (with the spin orientation = ,) of the impurity that randomly occupies the ith lattice site. The AndersonHubbard parameter U includes the Coulomb repulsion between d electrons in the same orbital. If nec essary, one can allow for the degeneracy of d orbitals and include in the Hamiltonian the additional intra atomic correlations, namely, the inter orbital Coulomb and ex change interactions. The last term in Hamiltonian (1) describes the reso nant (r) and potential (p) scattering of heavy holes by impurities:
,
(2)
Here, and are respectively the pd hybridization parameter and the scattering poten tial, both of them being only slightly dependent on the hole momenta, and H.c. stands for "Hermitian conju gation". The parameter W is the pseudopotential differ ence between the atom of the semiconductor host and the substituent magnetic impurity atom. Hamiltonian (1) proposed for DMS and HTFS includes magnetic impu rities randomly distributed over the ith lattice sites (their coordinates are denoted as Rj in expression (2) of the Hamiltonian); the impurity concentration is specified by particular problems. Impurity states in condensed matter can be best de scribed using Greens functions. The method is based on separation of the Hamiltonian into the unperturbed part and perturbation
.
includes only the heavy holes characterized by higher density of states compared to light holes and the greatest
2388
Ivanov et al.
/eV
E1 R1
Conduction band
E2 R2
V 2/W
i i
where G 0 = ( function:
( is the unity matrix). From the knowledge of Greens function G() one can calculate the energy spectrum and any characteristics of an impurity crystal. The advantages of Greens function technique are also that the perturba tion can not necessary be small, defects in the crystal can be taken into account, and systems from a perfect crystal to an impurity can be studied. The method was also used in studies of the effect of short range impurity potentials (point impurities or defects in crystal) on the quantum states and vibrational spectra,403 magnetically ordered crystals,404 and electronic properties of semicon ductors.405,406 General properties of disordered systems were considered in a monograph.407 Greens function method was used to study the influence of an isolated magnetic impurity on the spectrum of semiconductor.61,62 Theory presented here also employs Greens function formalism and treats the first two terms of Hamilto nian (1), H0 = Hh + Hd, which describe noninteracting subsystems of heavy holes and impurity electrons, as the "major" terms. The interaction between these subsystems occurs by scattering of current carriers in the semicon ductor by d electrons with Hamiltonian (2) treated as perturbation:
.
Fig. 2. Calculated scheme of energy levels in GaAs:Mn. The d5/d4 CFR levels (E1, E2) lie below the bottom of the heavy hole band, being 3.47 eV lower than the top of the valence band. The energy of the initial impurity d level is Ed = 2.6 eV. As the Mn content increases, the impurity band merges with the va lence band. Shown is splitting of the CFR levels of two Mn impurities into bonding and antibonding energy levels.
The results of electronic structure calculations of manganese impurities in two DMS, (Ga,Mn)As and (Ga,Mn)N, are shown in Figs 2 and 3, respectively. In GaAs:Mn with one impurity (i = 1 in Hamilto nian (1)), which corresponds to a very low impurity con centration in the real system, the appearance of the ac ceptor level is due to potential scattering of heavy holes from the band of width w = 2.9 eV 417 by manganese ions with the potential W = 1.02 eV. The energy of the accep
Conduction band
Ed
R1 U
R2
Computational details have been reported in Refs 408416. All characteristic parameters of semicon ductor hosts (widths of allowed and forbidden bands and masses of heavy holes) and magnetic impurities (energies of initial impurity levels Ed, Coulomb repulsion between d electrons U, pd hybridization parameter V, and poten tials W ) were taken from the literature and then used in calculations carried out with a realistic semi elliptic den sity of states for itinerant carriers:
,
V 2P11() 0
R1
R2
where w is the width of the heavy hole band (usually, w 23 eV in DMS and HTFS) and is the heavy hole energy counted from the top of the valence band ( 102 meV, i.e., << w).
Fig. 3. Calculated scheme of energy levels in GaN:Mn. The d5/d4 CFR levels (denoted as R1,2) of each Mn impurity lie deep in the forbidden energy band. Energies of the initial impurity d level (Ed = 1.2 eV) and of the bottom of the heavy hole band (h = 2 eV) were calculated relative to the top of the valence band. Shown is splitting of the CFR levels of two Mn impurities into bonding and antibonding energy levels.
2389
tor level (Eacc) counted from the top of the valence band is given by,
Eacc i = W[1 w/(4W )]2, (3)
being equal to 85 meV for GaAs:Mn. Position of the CFR d5/d4 level is determined from the self consistent solution to the equation
ECFR Ed Ei Ed = V2P11(Ei), (4)
is the real part of Greens function of the holes in the )1 in the ma valence band, which has the form ( trix representation, and integration in P11() is performed over the interval from the bottom of the heavy hole band to the top of the overlapping impurity band. A graphical solution to Eq. (4) for GaAs:Mn is shown in Fig. 2, a. Resonant scattering with the hybridization parameter V = 1.22 eV renormalizes position of the initial impurity d level Ed = 2.6 eV in GaAs:Mn into position of the CFR level with Ei = 3.47 eV below the bottom of the heavy hole band. Coulomb repulsion between d electrons with en ergy U 4.5 eV shifts the empty d6/d5 CFR to the con duction band of (Ga,Mn)As. Resonant scattering is in sufficiently strong to form a DBH at w = 2.9 eV and the acceptor level in GaAs:Mn is due to potential scattering in accord with Eq. (3). The results of electronic structure calculations of the system GaP:Mn are qualitatively similar to those shown in Fig. 2 for GaAs:Mn; they only differ in that the accep tor level originates from the DBH. In GaP:Mn, for the energy of the initial impurity d level one has Ed = 1.3 eV; w = 2.6 eV, and the resonant scattering occurs with hybri dization parameter V = 1.31 eV. Higher position of the impurity d level in GaP:Mn compared to GaAs:Mn is due to broadening of the forbidden band (2.26 eV in GaP vs. 1.42 eV in GaAs), being also responsible for the CFR level with energy ECFR = 3.28 eV and a DBH with EDBH i = 400 meV with respect to the top of the valence band, which are induced by resonant scattering of holes by the impurity. The crystal field resonances and dangling hybrid bonds are two solutions to Eqn. (4), which can also be represented graphically (similarly to the left side in Fig. 2). But in contrast to GaAs:Mn, in the case of GaP:Mn the straight line EEd intersects both peaks of the real part of Greens function P11(E) and give rise to the DBH and CFR.
In GaAs:Mn, the calculated position of the CFR level (ECFR = Ei = 3.47 eV) nearly coincides with the mea sured value (3.4 eV)416b,418 and lies in the valence band below the bottom of the heavy hole band (2.9 eV). The situation with the CFR in GaP:Mn is also similar. There fore, the lowest d5/d4 CFR levels in (Ga,Mn)V (V = As, P) are always filled and, hence, Mn2+ ions retain the fifth electron in the 3d shell and the neutral impurity state A0(3d5p ), where p denotes a loosely bound hole occupying the acceptor level in the forbidden energy bands of GaAs and GaP and A is the spectroscopic notation of the impurity state. Our assumptions of the chemical regu larities of the electronic structure of impurities in DMS and the results of calculations are confirmed by the detec tion of the 3d5p complexes in GaAs:Mn by EPR 419 and IR spectroscopy.420 Acceptor levels originate from resonant scattering (GaP:Mn) or potential scattering (GaAs:Mn) and lie 400 meV (GaP:Mn) or 100 meV (GaAs:Mn) above the top of the valence band. The last mentioned value is much larger than a value of 30 meV obtained from calcu lations using the theory of effective mass.* pd Hybridization (parameter V differs from zero only for the same spin states of a d electron and a p hole) is only possible if the localized spin and the spin of the itinerant hole have the same orientation during tunnel ing. Strong intra atomic Coulomb interaction between d electrons (parameter U ) and the Hund coupling pro vide half filled 3d5 shells of the manganese impurity ion in the lowest CFR levels. In the case of two impurity atoms ( i = 1, 2 in Hamiltonian (1)) tunneling of d electrons to empty states near the top of the valence band and back with the energy ~V 2 (dependence from the distance between impurities is omitted for simplicity) causes splitting of the acceptor and CFR levels into bond ing and antibonding sublevels (see Fig. 2, lowest CFR). As the impurity concentration increases, the distance be tween impurities is shortened while the number of accep tor levels and the splitting value increases. As a result, the acceptor levels are broadened to give an impurity band, which is overlapped with the valence band in (Ga,Mn)As and in type (IIIV,Mn)V2 HTFS. In Ga1xMnxAs, the overlap begins at x < 1 at.%, which was confirmed experi mentally.260,421 An IR spectroscopy study421 of com pounds (Ga,Mn)As with TC = 70 K revealed energy levels of antisites AsGa at 500 meV, which lie above the acceptor levels (~100 meV). The overlap of the impurity band and the valence band in (Ga,Mn)As with an increase in the Mn content is schematically shown in Fig. 2. In (Ga,Mn)P, the DBH levels lie deep in the forbidden energy band, so the impurity band remains isolated from the valence band with an increase in the Mn content.
2390
Ivanov et al.
(d5/d4)
Resonant nature of the manganese CFR t2 level near the middle of the band gap in (Ga,Mn)N was confirmed experimentally422 and by numerical calcula tions.391,390 The electronic structure of GaN:Mn (see Fig. 3) was calculated using the following parameters of the GaN matrix and the Mn impurity: position of initial impurity d level Ed = +1.2 eV, the heavy hole band width w = 2 eV, and hybridization parameter V = 1.2 eV. Posi tion of the only impurity CFR level d5/d4 was determined by graphical solution to the Eq. (4), which is presented on the left in Fig. 3. The CFR level is above the top of the valence band (ECFR = 1.7 eV), lying nearly in the middle of the forbidden band (for GaN with the zinc blende structure the width of the forbidden band is 3.233.25 eV at 300 K and 3.3 eV at 0 K). The manganese CFR t2 levels (d5/d4) appear deep in the valence bands of the GaAs and GaP hosts (see Fig. 2) but deep in the forbidden band of GaN (see Fig. 3) due to the narrower band gaps, higher position of the valence bands in GaAs and GaP compared to GaN, and referenc ing to vacuum of the Mn 3d level. In GaN:Mn, the CFR t2 level is empty is the Ga3+ ion is replaced by Mn3+(d4) impurity and becomes magnetically active only upon n doping. The configuration Mn3+(d4) corresponds to a neutral state; each Mn impurity ion introduced in GaN gives an empty CFR level near the middle of the band gap, this level playing the role of a deep acceptor. Therefore, in contrast to (Ga,Mn)As and (Ga,Mn)P, there is no need of binding a hole to retain electrical neutrality of (Ga,Mn)N. This behavior is also characteristic of the early transition metal ions (Sc, Ti, V) in all compounds of the IIIV type. Replacing Ga3+, the Mn2+ impurity retains its high spin state with a half filled d shell 3d5 due to strong intra atomic Hund exchange. In other words, the manganese impurities in the (Ga,Mn)V systems create hole attract ing potentials, acting as acceptors in (Ga,Mn)As and (Ga,Mn)P, and increase the magnetic moments of the p type (III,Mn)V compounds. On the contrary, a Mn impurity in the wide gap n type semiconductor GaN do nates one d electron to the valence band and thus under goes transition to the Mn3+(d4) state. The impurity band (not shown in Fig. 3) of n type (Ga,Mn)N appears near this state inside the band gap. It is important for the electronic structure of a semi conductor containing a magnetic impurity that the abso lute position of the CFR level depends only slightly on the semiconductor matrix, because the CFR energy cal culated relative to vacuum is compensated by the com peting bonds with the electronic states in the valence band and in the conduction band. On the contrary, the DBH energies are closely related to the density of states in the heavy hole band, their positions being dependent on the strength of pd hybridization.408416 It should be noted that qualitative schemes of the energy levels of 3d impurities calculated in an ex
tended IIIV supercell390 are similar to the diagrams shown in Figs. 2 and 3 obtained408414,416b using Greens function formalism. According to calculations,390 doping with manganese cations leads to the appearance of holes in the shallow acceptor levels of GaAs and GaSb and in the deep CFR level of GaN. Positions of the CFR and DBH levels of the 3d impurity in type IIIV compounds obtained using Greens function technique408416 and from numerical calculations390 (see also Refs 80 and 391) are similar because first principle calculations in the supercell begin with calculations for a single impurity in a finite cluster. However, further use of the "supercell" ap proach to describe the magnetic state by buildup of a periodic cluster structure does not go beyond the Stoner phenomenological model of itinerant ferromagnetism. Therefore, it is quite reasonable that in similar studies the DMS and HTFS are analyzed and discussed in terms of the exchange splitting and the spin subbands correspond ing to spin orientations "up" and "down" (see., e.g., Refs 80, 390396). Below we present the third strategy of research into the nature of magnetism in DMS and HTFS. Here, the exchange interaction and the magnetic properties are ana lyzed with allowance for the electron states of transition metal impurities in different semiconductors. The elec tron energy spectrum is formed by the chemical bonds between the 3d orbitals of the impurity ions and the p orbitals electrons in the valence band, whereas the mag netic state is governed by the Coulomb and exchange interaction inside the atomic 3d shells (intra atomic cor relations electron). 4.3. Nature of ferromagnetism in dilute magnetic semiconductors and high temperature ferromagnetic semiconductors Consideration of the chemical bonds formed by mag netic impurities in semiconductors helps to clarify the nature of exchange interaction between electrons local ized on distant magnetic impurities in the semiconductor. Owing to peculiarities of the electronic structure of tran sition metal impurities it is impossible to directly extrapo late the ferromagnetic properties of p type (Ga,Mn)As with relatively narrow band gap to high temperature DMS, e.g., n type (Ga,Mn)N with a broad forbidden band or p type (Ga,Mn)P. However, this often occurs in phe nomenological approaches (see, e.g., Ref. 77). Two Mn impurities in p (Ga,Mn)As and p (Ga,Mn)P have relatively shallow acceptor levels (see Fig. 2, lev els R1 and R2) that are absent in p (Ga,Mn)N and a deep CFR level in the allowed band. In the n type (Ga,Mn)N, the localized states are deep CFR levels in the forbidden band (see Fig. 3, levels R1 and R2). In these two different cases indirect exchange interactions between electrons occupying the CFR levels R1 and R2 of two impurities are basically different. The nature of ferromagnetic exchange
2391
in the p type and n type DMS and in the isoelectronic HTFS should be considered separately.408416 Spin selective electron transfer with the energy V 2 between impurities is governed by pd hybridization and strong intra atomic electron correlations (Coulomb in teractions and Hund exchange) (see Section 4.2). In other words, transfer is only possible if the electron spins Si=1,2 of neighboring impurities (e.g., Mn2+) with the wave func tions 1 and 2 have parallel orientations. In this case, tunneling removes the degeneracy of the initial impurity states 1,2, their energy levels become split (E+ E = 2V 2), and now one deals with a bonding and antibonding combinations of the wave functions of two impurities, namely, + = (1 + 2)/ and = (1 2)/ (spin triplet S = 1); the energy difference between these states is no more than the exchange energy E E+ = 2Jex . Indeed, the Hamiltonian of two impurities has the form
E E+ = | | +| |+ = 21| |2, (5)
pendence of tunneling on the distance between impurities is left out of consideration). Rigorous total energy calculations takes into account all changes in the energy spectrum and significant rear rangement of the impurity band and the heavy hole band with allowance for both resonant and potential scattering. The total energy of the system is given by the spur of the Hamiltonian, E = Tr , and the change in the total en ergy due to scattering of current carriers by impurities is given by
,
i.e., the exchange interaction Jex of two isolated electrons formally coincides with the energy V 2 = 1| |2 of electron tunneling between impurities due to pd hybri dization. If the impurity spins have antiparallel orienta tions, no tunneling occurs and the impurity levels do not split. But splitting of the levels of two localized electrons does not provide an indirect magnetic exchange. In the semiconductor with the heavy hole band of width w the proportion of the holes involved in magnetic exchange between identical impurities with the CFR energy levels ECFR is [(w/ECFR)V ]2. As a result, an indirect exchange interaction occurs between electrons localized on impuri ties, which can be written as follows taking into ac count Eq. (5)
Jex V2[(w/ECFR)V ]2 = 3V 4(w/ECFR)2, (6)
where G() = G() G0() is the difference between the total and unperturbed Greens functions (for computa tional details, see Refs 408416). To calculate the pair exchange interaction, one should know the electron en ergy spectrum of the semiconductor with two magnetic impurities (cf. Section 4.2). Owing to indirect spin de pendent overlap of the impurity orbitals of two magnetic impurities in the semiconductor the change in the energy of the system is the ferromagnetic energy gain relative to the spin disordered state, i.e., the exchange energy
, (7)
where
and
are the real and imaginary parts of heavy hole Greens function:
P12() + (i/2)12()
We called it kinematic exchange interaction because it is due to tunneling, i.e., electron transfer between impuri ties via the empty band and impurity states. Kinematic exchange is accompanied by a decrease in the kinetic energy of itinerant heavy holes owing to their hybridiza tion with impurity d electrons with the same spin orien tation. Otherwise, in accord with the Hund rule an elec tron hopping to the impurity atom with a different spin orientation requires an additional energy. If the magnetic moments of impurities are not aligned, the Hund rule will block transfer of an electron with the spin orientation parallel to the angular momentum direction of the first impurity to the second impurity with the "wrong" angular momentum direction. Splitting of the initial impurity lev els, not suppressed by the Hund rule, and the kinematic exchange occur only if not all available bands and impu rity states are filled, that is () 0 in Eq. (6). (Here, for qualitative description of kinematic exchange (6) the de
and integration in Jex is performed from the Fermi en ergy to the top of the valence band, t. Here, k = (2m)1/2/h is the wave vector of the hole with the mass m moving with the energy and R is the average distance between impurities whose concentration is x = a3/(4R3), where a is the lattice constant. At t = 0, the energy of exchange interaction (7) can be written in explicit form as
(8)
where kF = is the Fermi wave vector, ECFR is the position of the CFR impurity level, kb = ,
2392
Ivanov et al.
and (F) is the density of hole states in the Fermi level F(x) (F is counted from the top of the valence band and depends on the dopant concentration). In Eq. (8) the coefficient on the left side before the brackets coincides with the qualitative estimate (6). Kinematic exchange interaction in p type DMS and in HTFS involves virtual transitions of d electrons to un occupied valence and impurity states. It self consistently allows for three important interrelated contributions to the magnetic energy, namely, dp tunneling (hybridiza tion parameter V ), CFR d resonances, and DBH dan gling p bonds and can be reduced to none of the conven tional exchange interactions. Unlike phenomenological models dealing with localized impurity spins and itinerant holes in the allowed band, the kinematic exchange takes into account a change in the hole density of states and therefore the magnetic response of holes owing to the resonant scattering and impurity band formation. From Eq. (8) it follows that only at unreally high concentrations of magnetic impurities (kbR << 1) kinematic exchange changes its sign similarly to RKKY indirect exchange (see Section 3.3). At low impuriy concencentrations (kbR >> 1), as is typical of DMS and HTFS, the kinematic exchange is proportional to 1/R, being much stronger than the RKKY exchange. The kinematic exchange differs from the VonsovskyZener double exchange mechanism65,7072 used 65 to explain ferromagnetism in manganites (La,Sr2+)MnO3 (mixed valence compounds with intrin sic Mn sublattice) by the exchange of different valence states (Mn3+ Mn4+) via oxygen anions. In p type DMS and in HTFS, transition metal impurity ions have the same integer valence and occupy the CFR states iden tically (cf. the levels R1 and R2 in Fig. 2); therefore, no Zener exchange interaction65 occurs in these substances. Nevertheless, density functional calculations388,395 of p type DMS (In,Mn)As and (Ga,Mn)As were performed with the Zener model in spite of the absence of mixed valence in the DMS and the experimental results men tioned above.* In the molecular field approximation the spin Hamil tonian of the kinematic exchange interaction (7), (8) in DMS and HTFS has the form
. (9)
their antiparallel orientations is Jex (7), (8). In a conven , see., e.g., Ref. 423) this energy tional spin model (J gain is 2J, where J is the direct Heisenberg exchange interaction. From Eq. (9) it follows that the Curie tem perature of the ferromagneticparamagnetic phase tran sition is given by
, (10)
Here, summation is performed over all positions of the magnetic impurities with the orbital moment Ji in the semiconductor lattice. The coefficient 1/2 accounts for the fact that, by implication of kinematic exchange, the energy gain for two spins with parallel orientations vs.
* Y. H. Matsuda, G. A. Khodaparas, M. A. Zudov, J. Kono, Y. Sun, F. V. Kyrychenko, G. D. Sanders, C. J. Stanton, N. Miura, S. Ikeda, Y. Hashimoto, S. Katsumoto, H. Munekata, http://arXiv.org/cond mat/0404635.
where J is the total spin of impurities (angular momen tum) and z is the coordination number of the magneti cally active impurities in a given DMS or HTFS. An important feature of magnetic impurities in semi conductors, which was considered in the previous Sec tion, is that the manganese ion retains their valence. When manganese replaces the higher valence atoms in DMS and HTFS, the Pauling principle of electrical neutrality is obeyed owing to the formation of the Mn2+ complexes with heavy holes. The total "spin" of the complex appears to be lower than S(Mn2+) = 5/2. For instance, in (Ga,Mn)As and (Ga,Mn)P the total momentum of the complex Mn(3d5p ) equals J = S j = 1, because it is formed by the orbital momentum j = 3/2 of a loosely bound hole p , involved in antiferromagnetic interaction with the spin of the Mn2+ ion (S = 5/2). The antiferro magnetic exchange constant of the complex Mn(3d5p ) was measured by Raman scattering.424 Coexistence of weakly and strongly localized states in (Ga,Mn)As was also confirmed by conductivity measurements.281 Ternary compounds, e.g., HTFS based on manga nese doped type IIIVV2 hosts exhibit a more im pressive picture of cation substitution compared to bi nary IIIV compounds. Namely, Mn2+ ions replace both the atoms of Group II elements with retention of the spin S = 5/2 and the atoms of Group IV elements. For the same reason (antiferromagnetic exchange of holes and the local spin) the MnGe ion in the HTFS (ZnGe,Mn)P2 forms a complex with the momentum J = 1/2 with two holes, which was confirmed by ESR.319 Replacement of the Ge4+ ion with the smaller radius by Mn2+ ion in the HTFS (CdGe,Mn)As2 was confirmed by an increase in the lattice constant329 and by estimates of the average angu lar momentum from magnetic measurements413,415,416a (J < 5/2 per formula unit of (CdGe,Mn)As2). The Fermi energy appeared in Eqns (7) and (8) de pends on the magnetic impurity concentration x and is related to the volume density, n(x), of charge carriers in the material:
(coefficient 8 is due to the presence of four pairs of atoms of Group IIIV or isoelectronic chemical elements in the unit cell of volume a3). Usually, the parameter n(x) is
2393
found from Hall effect measurements, by IR spectros copy,424 using an electrochemical cell, or other methods. To calculate the TC(x) of a p type DMS, one must substitute the total spin of impurities J and the coordina tion number z in Eq. (10). Calculations of the kinematic exchange (7) require substitution of the known character istics of the material and magnetic impurities, namely, the mass and concentration of heavy holes, the heavy hole band width, hybridization parameter, and the energy of the impurity level. If the role of the impurity band is insignificant (t = 0), one can use expression (8). Using the impurity content x and the corresponding hole con centration n in the samples of the material under study, an n(x) extrapolation can be constructed. Studies408416 present the results of TC(x) calculations for various mate rials, carried out using the n(x) dependences, that are in excellent agreement with experimental data. For such DMS as (Ga,Mn)As or (Ga,Mn)P one must substitute the spin J = 1 in Eq. (10). The results obtained for a particular case of calculations are shown in Fig. 4. For comparison, Fig. 4 also presents the results of our calculations using the phenomenological theory of DMS,78 which significantly differ from the experimental data due to ignoring the chemical bond of the Mn impu rity in GaAs:Mn. The plots shown in Fig. 4 were calcu lated using relation (10) with the hole densities measured for seven (Ga,Mn)As samples with TC 140 K.425 As can be seen from Fig. 4, the kinematic theory predicts a maxi mum TC of about 160 K for annealed samples. Later,
TC/K 160 1 2 3
TC/K 120 90 60 30 1 1 2
0.4
0.8
1.2
p/nm3
Fig. 5. Curie temperature plotted vs. hole concentration in Ga0.94Mn0.06As: experimental data425 (1) and results of TC(p) calculations using phenomenological theory78 (1). The smooth curve 1 obtained from TC(p) calculations for kinematic ex change mechanism.
120
80
40
Mn content (%)
Fig. 4. Curie temperature TC plotted vs. Ga1xMnxAs composi tion. The solid curves represent the results obtained from TC(x) calculations414 for kinematic exchange starting from the experi mental hole density. The filled squares, open circles, and open triangles respectively denote the experimental TC values of an nealed (1) and as grown (2) materials;425 and the results ob tained425 for the as grown samples (3) based on the phenom enological theory78 (see Website*). * http://ep2ds14.fzu.cz/ms/setparam.ausprg.php.
(Ga,Mn)As samples with TC = 150160 K were synthe sized.426428,* In contrast to phenomenological models dealing with localized spins and free holes near the top of the valence band, the kinematic exchange (7) allows for the change in the hole density states and thus their magnetic "response" owing to resonant scattering and impurity band forma tion. As a result, one gets realistic TC(x) dependences that are consistent with the experimental data (see Fig. 4 and the TC(x) dependences reported for other DMS samples in Refs 408412) rather than a linear plot of Curie tem perature vs. impurity concentration (TC xp1/3). Phase transitions in DMS and HTFS from the para magnetic to ferromagnetic state can be induced either by a decrease in the temperature at a constant density of current carriers or by an increase in the carrier density at constant temperature. The carrier density can be changed by doping, annealing, applying an electric field or by ex posing the material to light. The dependence of TC on the hole density p in (Ga,Mn)As containing 6 at.% of Mn calculated using Eqns (8) and (10) on the basis of the experimentally measured hole densities in the sample af ter four stages of annealing in different regimes is pre sented in Fig. 5. The TC(p) curve reaches a saturation point of 158 K at a hole density, p, of 3.0 nm3 (not shown in Fig. 5), i.e., the TC(p) dependence does not obey the relation TC xp1/3 used in phenomenological theory7779 which predicts an infinite increase in TC with increasing the hole concentration. The results of TC(Ta) calculations using the measured hole density429 for three annealing temperatures Ta are in
* K. W. Edmonds, P. Boguslawski, K. Y. Wang, R. P. Campion, N. R. S. Farley, B. L. Gallagher, C. T. Foxon, M. Sawicki, T. Dietl, M. B. Nardelli, and J. Bernholc, http://arXiv.org/ cond mat/0307140.
2394
Ivanov et al.
TC/K
2 100 1
75
280
300
320
340
Ta/ C
Fig. 6. Curie temperature plotted vs. annealing temperature of (Ga,Mn)As: results of calculations413 (1) and experimental data429 (2).
good agreement with the experimentally measured Curie temperatures (Fig. 6). In the case of p type (Ga,Mn)N the kinematic ex change theory should be slightly modified. Here, Mn re mains a neutral isoelectronic substitutional defect with the configuration Mn3+(d4), because the CFR level of Mn2+/3+ falls in the middle of a broad forbidden energy band of GaN (see Fig. 3). If the hole concentration is higher than the Mn impurity content, ferromagnetism in p type (Ga,Mn)N has a kinematic exchange mechanism, similarly to (Ga,Mn)As and p (Ga,Mn)P, but with the localized moments of the Mn3+(e2t2) magnetic impuri ties J = 2. If one deals with a doped DMS such as n type (Ga,Mn)N, a fraction of Mn3+ impurities traps donor electrons and becomes charged ions Mn2+(d5). As a re sult, the impurity band in (Ga,Mn)N, which appears near the CFR level, contains both Mn 2+ and Mn 3+ ions in contrast to (Ga,Mn)As, (Ga,Mn)P, and p type (Ga,Mn)N. Here, a problem is posed of the magnetic exchange interaction in n type (Ga,Mn)N with the im purity band in the deep CFR level, which remains sepa rated from the allowed bands in the whole range of Mn concentrations. If other shallow acceptors do not create free holes near the top of the valence band, the Fermi level must lie in the impurity band and electron tunneling between the CFR levels must occur due to the heavy hole p orbitals mixed with the d orbitals owing to pd hybri dization. In this case the role of the p orbitals in (Ga,Mn)N is similar to that played by the oxygen p orbitals in the double exchange in (La,Sr)MnO3 65 and the magnetic interaction in n type (Ga,Mn)N is provided by the exchange of the valence states Mn2+ Mn3+. Here, the TC value of n type (Ga,Mn)N equals the half width of the impurity band, averaged over positions of the impurity levels;413,416b that is, the highest TC is achieved if the impurity band is half filled and position of the Fermi
level coincides with the CFR level. The Curie tempera ture calculated with the hybridization parameter V = 1.2 eV and the "spin" J = 2 varies from 100 K at x = 0.01 to 1000 K at x = 0.05, being in agreement with the experimental data (TC = 940 K 303,430433 in (Ga,Mn)N epitaxial films). If the Fermi level lies near the bottom of the conduction band, TC is exponentially low with respect to the impurity concentration.434 In the Hd term of the Hamiltonian (1), the three fold degeneracy of the d orbitals with t2 symmetry is signifi cant when calculating the broadening of the impurity band with increasing the impurity concentration. The case in point is that the band p states obey the Fermi statistics. But in the statistics of the atomic d states hybridized with the band p states the d orbital cannot be filled with two electrons with (i) parallel spin orientations due to the Pauli exclusion principle and (ii) antiparallel spin orien tations due to the Coulomb repulsion characterized by the AndersonHubbard parameter U. The description of the states of magnetic impurities is similar to the HeitlerLondon treatment of hydrogen molecule (1927). If the energy of the Mn2+(d5) impurity is ECFR and the energy of the Mn3+(d4) impurity is (chemical poten tial), their statistical sum is Z = exp[ECFR/(kBT )] + + 3exp[/(kBT )]. From here the occupation numbers of the impurity atomic shells differ from the Fermi distri bution:
,
where nF(ECFR ) = (exp{(ECFR )/(kBT )} + 1)1 is the Fermi distribution. As can be seen, if ECFR > 0 (chemical potential is lower than the CFR level of the fifth electron of the Mn2+ impurity), one has nF(ECFR ) = 0 and the occupation numbers are n(d5) = 0 and n(d4) = 1/3 at low temperatures. If > ECFR, one has n(d5) = 1 and n(d4) = 0. The quantum statistics of magnetic impurities in semiconductors is derived on the basis of the Fermi statistics and the atomic statistics in such a way that the impurity states obey the atomic statistics that allows each impurity level to be occupied by one electron only (even in the case of degenerate t2 states) and the impu rity band to be filled with integer numbers of elec trons.408412,414,416b, Density functional calculations of such HTFS as (Cd,Mn)GeP2 and (Zn,Mn)GeP2 using the Zener model predict ferromagnetism326 while the supercell calcula tions435,436 of the same compounds predict an antiferro magnetic ground state. The data,392,393 according to which both CFR and DBH favor the ferromagnetic state of (CdGe,Mn)P2, seem to be the best approximation to the
2395
theory of the kinematic exchange in HTFS. In p type (CdGe,Mn)As2, ferromagnetic ordering is favored by the (Cd,VCd,Mn)GeAs2 vacancies and nonstoichiometry of the type (CdGe,Mn)GeAs2, which increase the concen trations of the holes and empty states involved in kine matic exchange near the top of the valence band.329 In type IIIV DMS, the highest Curie temperatures corre late with the width of the forbidden band in spite of differ ent exchange mechanisms in the p and n type com pounds, Namely, TC increases in the order (Ga,Mn)As, (Ga,Mn)P, (Ga,Mn)N. Consideration of type IIIVV2 compounds gives the following order of a decrease in the stability of the ferromagnetic phase of the manganese doped hosts198:
GaN GaP GaAs CuGaS2
CuGaSe2 GaSb CuGaTe2.
In fact, the theory of kinematic exchange interaction in DMS and HTFS has no fitting parameters; all quanti ties are expressed through the available characteristics of materials. For instance, the energy spectrum and the cal culated curves TC(x) and T(p) of annealed samples shown in Figs 2 and 46 were obtained using the same known parameters of the (Ga,Mn)As system, viz., the heavy hole mass m = 0.51m0, the width of the heavy hole band w = 2.9 eV, pd hybridization parameter V = 1.22 eV, coordi nation number z = 4.0, and the energy of the impurity d level Ed = 2.6 eV. With these parameter values, the Curie temperatures were determined from the experimen tal hole concentrations. The calculated dependences of TC on the impurity concentration, hole density, and an nealing temperature (see also Refs 408416b) are in good agreement with experimental data. Greater hole density provides a higher conductivity, i.e., the DMS and HTFS exhibiting the highest TC must be good conductors. Formally, the kinematic exchange interaction between magnetic impurities is given by the integral (7), which allows for reconstruction of the whole electron spectrum. This differs the mechanism of kinematic exchange from the double and other types of exchange mechanism in volving a specific parameter of exchange interaction that is usually introduced phenomenologically. The effect of antisites (e.g., AsGa) or interstitial defects and possible clustering of impurities are included in this theory via the effective coordination number z and statistics. A simpli fied picture of the structure of the valence band (heavy hole band with a semi elliptic density of states) does not present an obstacle to the development of this theory, which can be improved with allowance for, e.g., the light hole band. The theory is also suitable for the description of ferro magnetism in other DMS, e.g., an elemental dilute semi conductor (Ge,Mn)154,247,437,438 or a wide gap ZnO ma
trix doped with Mn and Fe,346,349 where magnetic impu rities (Mn2+ and Fe2+) isoelectronically replace Zn ions, and IIIIVI2 based HTFS. Transition metal ions in Ge are usually interstitial impurities that are in some cases involved in complexation. The empty impurity CFR levels (d6/d5) lie deep in the conduction band due to strong intra atomic Coulomb repulsion (AndersonHubbard parameter U ) and can be a reason for antiferromagnetic ordering in the impurity band. In zinc blende type crystals, indirect exchange in volving empty electron states in the conduction band is weak because the hybridization Vds between the impurity d and s electron states near the bottom of the conduction band is low from symmetry considerations.401,402 In con nection with the empty states in the electronic spectrum of a material the BlombergenRowland mechanism of indirect exchange439 is worth noting, according to which interaction between the impurity spins in DMS involves virtual states in the allowed bands, created by electron transitions from the bands to the impurity acceptor levels. This mechanism was the subject of a few studies383,443,* (see also a discussion in Refs 440, 441) that postulated the existence of a contact dp or ds exchange interaction ig noring the chemical bonds of magnetic impurities in semi conductors. The properties of DMS were analyzed* using the BlombergenRowland mechanism and the Hamilto nian (1). However, the TC of n (Ga,Mn)N was found to be proportional to |Vdp|2|Vds|2, i.e., unrealistically low.* In the case of DMS and HTFS of particular theoretical in terest is to allow for the distribution functions of random local magnetic fields over the lattice sites chaotically oc cupied by magnetic impurities442,443 (at present, in the framework of the Ising model that can be solved exactly), which leads to unusual temperature dependences of the magnetization. Conclusion The review presents the state of the art in the field of spintronics, a rapidly developing field of science and tech nology, which employs the experience, concepts, and ideas from different fields of knowledge of the condensed matter. Among them are the band theory of solids; micro electronics; magnetism and various versions of magnetic resonance; superconductivity in nanosystems includ ing single electron electronics, Josephson tunneling, Abrikosov vortices, all types of transport mechanisms (Boltzmann, ballistic, hole, tunneling); specific methods such as optical spin orientation and photo EMF; materi als science;1 physical chemistry of materials and materials design; and strong electron interactions in condensed matter.
* V. Barzykin, http://arXiv.org/cond mat/0311114.
2396
Ivanov et al.
Among the variety of spintronics related aspects, in this review the emphasis was placed on the fundamentals of spintronics; formulation of criteria for spintronic ma terials; consideration of the first generation spintronic materials (ferromagnetic metal alloys); characterization of the physicochemical properties of the second genera tion spintronic materials, namely, concentrated magnetic semiconductors and semimetals and half metals, semi magnetic semiconductors, dilute magnetic semiconduc tors (DMS), and high temperature ferromagnetic semi conductors (HTFS). Research on inhomogeneous mag netic materials and high TC nanocrystals including or ganic materials suitable for spintronics applications was also outlined. (Semi)phenomenological and numerical cluster approaches used to explain the magnetic proper ties of novel spintronic materials (DMS and HTFS) were analyzed. The electronic structures of semiconductors doped with magnetic impurities were considered.400402 It is the structure of the chemical bonds formed by the magnetic impurity in the semiconductor matrix that is responsible for both the mechanism of exchange interac tion and the nature of ferromagnetism408416 that occurs in the DMS and HTFS with an increase in the transition metal content. It was not the authors goal to provide with exhaustive literature on the subject. Most of the studies selected from the rapidly growing body of information are concerned with the subject of this review, namely, spintronic materi als and magnetic properties of DMS and HTFS. (for con tinuously renewable literature on DMS and HTFS, please visit the Website*.) Without pretending to cover all aspects of spintronics, which has got a new impetus to development after the discovery of DMS and HTFS, in conclusion the authors of this review take the risk of making some prognoses. Currently, spintronic devices based on first generation materials are increasingly used, whereas the DMS and HTFS based devices are still in the stage of test models or demonstration facilities. Spintronics serves as a generator of new ideas and methods in materials science, solid state physics, and solid state chemistry. Research and devel opment in order to solve spintronics related problems involves design of setups; elaboration of methods of syn thesis of single crystals and nanocrystals, and hetero structures with alternating DMS and HTFS layers; syn thesis of novel materials; improvement of methods of liq uid phase epitaxy, chemical vapor deposition, and low temperature molecular beam epitaxy. In studying the properties of spintronic materials re searchers discover new types of magnetic interactions that cannot be described using the available concepts of mag netism. Probably, in the case of DMS and HTFS we deal with a novel type of indirect magnetic exchange, which
* http://unix12.fzu.cz/ms/allpubl.php.
does not occur in the transition metal oxides described by Zener theory of ferromagnetic insulators. Novel spintronic materials based on semiconductors seem to be quite prom ising because, according to theory, they can exhibit Curie temperatures much exceeding room temperature, high current carrier mobilities, and high degree of spin polar ization. At present, investigations in the field of spintronics are carried out at research institutes and in research laborato ries at many universities and largest corporations. Great investments in the field are due to the fact that spintronics is among the top priority research avenues in the nano technology programs. Since 2003, a research program on spin dependent effects in solids and spintronics has been developed in the Russian Academy of Sciences (Division of Physical Sciences). The main avenues of basic research carried out at the Russian Academy of Sciences (Section "Chemical Sciences and Materials Science")* include syn thesis and investigations of novel compounds, design of materials and nanomaterials including magnetic materi als, which also covers the field of spintronics. The aforesaid of course gives no assurances that spintronics, in its basic research aspects and as a novel technology, will always offer exclusive prospects and, simi larly to electronics, will appear to be an irreplaceable applied discipline in the nearest decades. However, irre spective of how long will be the lifetime of hopeful tech nology assessment (by the way, this also concerns nano technology), investigations in the field of spintronics and spintronic materials undoubtedly belong to the topical avenues of research on condensed matter. This is of con siderable importance for, e.g., the development of new principles of interaction between distant magnetic impu rities and creation of a new generation of HTFS and devices based on these materials. Probably, the most ex citing aspects of spintronics will be related to the effects, devices, and materials that cannot be imagined as yet. Addressees of this review, that is, physicists, chemists, and materials scientists must be ready for future changes. References
1. E. I. Rashba, J. Superconductivity: Incorporating Novel Mag netism, 2002, 15, 13. 2. S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. von Molnar, M. L. Roukes, A. Y. Chtchelkanova, and D. M. Treger, Science, 2001, 294, 1488. 3. Spin Electronics , Eds M. Ziese and M. J. Thornton, Springer, Berlin, 2001, 500 pp. 4. M. N. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, and F. Petroff, Phys. Rev. Lett., 1988, 61, 2472. 5. G. Binasch, P. Grnberg, F. Saurenbach, and W. Zinn, Phys. Rev., B, 1989, 39, 4828. * Poisk [Poisk], Iss. 35 (145), August 29, 2003 (in Russian).
2397
6. H. Munekata, H. Ohno, S. von Molnar, A. Segmller, L. L. Chang, and L. Esaki, Phys. Rev. Lett., 1989, 63, 1849. 7. H. Ohno, H. Munekata, S. von Molnar, and L. L. Chang, J. Appl. Phys., 1991, 69, 6103. 8. H. Ohno, H. Munekata, T. Penney, S. von Molnar, and L. L. Chang, Phys. Rev. Lett., 1992, 68, 2664. 9. H. Munekata, A. Zaslavsky, P. Fumagalli, and R. J. Gambino, Appl. Phys. Lett., 1993, 63, 2929. 10. M. I. Dyakonov, in Future Trends in Microelectronics: The Nano, the Giga, and the Ultra, Eds S. Luryi, J. Xu, and A. Zaslavsky, Wiley IEEE Press, New York, 2004, p. 424. 11. K. E. Drexler, Proc. Natl. Acad. Sci. USA, Chemistry Sec tion, 1981, 78, 5275. 12. K. E. Drexler, Nanosystems: Molecular Machinery, Manu facturing, and Computation, Wiley, New York, 1992, 556 pp. 13. L. P. Gorkov and G. M. Eliashberg, Zh. Eksp. Teor. Fiz., 1965, 48, 1407 [Sov. Phys. J. Exp. Theor. Phys., 1965, 21, 940 (Engl. Transl.)]. 14. E. A. Shapoval, Zh. Eksp. Teor. Fiz., 1964, 47, 1007 [Sov. Phys. J. Exp. Theor. Phys., 1964, 20, 675 (Engl. Transl.)]. 15. E. I. Rashba, Fizika Tverdogo Tela, 1960, 2, 1224 [Sov. Phys. Solid State, 1960, 2, 1109 (Engl. Transl.)]. 16. E. I. Rashba, Usp. Fiz. Nauk, 1964, 84, 557 [Sov. Phys. Usp., 1965, 7, 823 (Engl. Transl.)]. 17. R. L. Bell, Phys. Rev. Lett., 1962, 9, 52. 18. V. V. Kveder, Yu. A. Osipyan, and A. I. Shalynin, Pisma v Zhurn. Eksp. Teor. Fiziki, 1984, 40, 10 [Sov. Phys. JETP Lett., 1984, 40, 729 (Engl. Transl.)]. 19. V. V. Kveder, V. Ya. Kravchenko, T. R. Mchelidze, Yu. A. Osipyan, and D. E. Khmelnitskii, Pisma v Zhurn. Eksp. Teor. Fiziki, 1986, 43, 202 [Sov. Phys. JETP Lett., 1986, 43, 255 (Engl. Transl.)]. 20. Yu. A. Bychkov and E. I. Rashba, Pisma v Zhurn. Eksp. Teor. Fiziki, 1984, 39, 66 [Sov. Phys. JETP Lett., 1984, 39, 78 (Engl. Transl.)]. 21. V. K. Kalevich and V. L. Korenev, Pisma v Zhurn. Eksp. Teor. Fiziki, 1990, 52, 859 [Sov. Phys. JETP Lett., 1990, 52, 230 (Engl. Transl.)]. 22. A. V. Vedyayev, B. Dieny, and N. Ryzhanova, Europhys. Lett., 1992, 19, 329. 23. E. L. Frankevich and E. I. Balabanov, Pisma v Zhurn. Eksp. Teor. Fiziki [Sov. Phys. JETP Lett.], 1965, 1, 33 (in Russian). 24. E. L. Frankevich, V. I. Lesin, and A. I. Pristupa, Zhurn. Eksp. Teor. Fiziki, 1978, 75, 415 [Sov. Phys. J. Exp. Theor. Phys., 1978, 48, 208 (Engl. Transl.)]. 25. I. A. Sokolik and E. L. Frankevich, Usp. Fiz. Nauk, 1973, 111, 261 [Sov. Phys. Usp., 1973, 111 (Engl. Transl.)]. 26. A. L. Buchachenko, R. Z. Sagdeev, and K. M. Salikhov, Magnitnye i spinovye effecty v khimicheskikh reaktsiyakh [Magnetic Field Induced and Spin Effects in Chemical Reac tions], Nauka, Novosibirsk, 1978, 296 pp. (in Russian). 27. A. L. Buchachenko, Usp. Khim., 1999, 68, 99 [Russ. Chem. Rev., 1999, 68, 85 (Engl. Transl.)]. 28. V. A. Ivanov and K. Kanoda, Molec. Cryst. Liq. Cryst., 1996, 285, 211. 29. V. A. Ivanov and K. Kanoda, Physica C, 1996, 268, 205. 30. V. A. Ivanov, E. A. Ugolkova, and M. Ye. Zhuravlev, Zhurn. Eksp. Teor. Fiziki, 1998, 113, 715 [J. Exp. Theor. Phys., 1998, 86, 395 (Engl. Transl.)].
31. V. F. Gantmakher and M. V. Feigelman, Usp. Fiz. Nauk, 1998, 168, 113 [Physics Uspekhi, 1998, 41, 105 (Engl. Transl.)]. 32. V. I. Ovcharenko and R. Z. Sagdeev, Usp. Khim., 1999, 68, 381 [Russ. Chem. Rev., 1999, 68, 345 (Engl. Transl.)]. 33. A. L. Buchachenko, Usp. Khim., 2003, 72, 419 [Russ. Chem. Rev., 2003, 72, 375 (Engl. Transl.)]. 34. M. Oestreich, M. Brender, J. Hubner, D. H. W. W. Rhle, T. H. P. J. Klar, W. Heimbrodt, M. Lampalzer, K. Voltz, and W. Stolz, Semicond. Sci. Technol., 2002, 17, 285. uti 35. I. Z , J. Fabian, and S. Das Sarma, Rev. Mod. Phys., 2004, 76, 323. 36. J. P. Ansermet, J. Phys.: Condens. Matter, 1998,10, 6027. 37. J. Bass and W. P. Pratt, Jr., J. Magn. Magn. Mater., 1999, 200, 274. 38. J. M. Daughton, A. V. Pohm, R. T. Fayfield, and C. H. Smith, J. Phys., D, 1999, 32, 169. 39. M. A. M. Gijs and G. E. W. Bauer, Adv. Phys., 1997, 46, 285. 40. G. Prinz, Science, 1998, 282, 1660. 41. M. D. Stiles, in Ultrathin Magnetic Structures III , Eds B. Heinrich and J. A. C. Bland, Springer, New York, 2004. 42. P. M. Tedrow and R. Meservey, Phys. Rep., 1994, 238, 173. 43. H. Ohno, Science, 1998, 281, 951. 44. S. J. Pearton, C. R. Abernathy, M. E. Overberg, G. T. Thaler, D. P. Norton, N. Theodoropoulou, A. F. Hebard, Y. D. Park, F. Ren, J. Kim, and L. A. Boatner, J. Appl. Phys., 2003, 93, 1. 45. A. J. Epstein, MRS Bull., 2003, 28, 492. 46. A. M. Goldman, V. Vasko, P. Kraus, K. Nikolaev, and V. A. Larkin, J. Mag. Magn. Mater., 1999, 200, 69. 47. K. Tsukagoshi, B. W. Alphenaar, and H. Ago, Nature, 1999, 401, 572. 48. Magnetic Interactions and Spin Transport , Eds A. Chtchelkanova, S. A. Wolf, and Y. Idzerda, Kluwer Academic DordrechtPlenum, New York, 2003, 572 pp. 49. Magnetic Multilayers and Giant Magnetoresistance. Funda mentals and Industrial Applications, Ed. U. Hartman, Springer Verlag, BerlinHeidelbergNewYork, 2000, 320 pp. 50. E. Hirota, H. Sakakima, and K. Inomata, Giant Magneto Resistance Devices, Springer, Berlin, 2002, 177 pp. 51. P. M. Levy and I. Mertig, in Spin Dependent Transport in Magnetic Nanostructures, Eds S. Maekawa and T. Shinjo, Taylor and Francis, New York, 2002, 47 pp. 52. S. Maekawa, S. Takahashi, and H. Imamura, in Spin Dependent Transport in Magnetic Nanostructures , Eds S. Maekawa and T. Shinjo, Taylor and Francis, New York, 2002, p. 143. 53. S. S. P. Parkin, in Spin Dependent Transport in Magnetic Nanostructures, Eds S. Maekawa and T. Shinjo, Taylor and Francis, New York, 2002, p. 237. 54. T. Shinjo, in Spin Dependent Transport in Magnetic Nanostructures, Eds S. Maekawa and T. Shinjo, Taylor and Francis, New York, 2002, p. 1. 55. V. L. Ginzburg, Usp. Fiz. Nauk, 2002, 172, 213 [Physics Uspekhi, 2002, 45, 205 (Engl. Transl.)]. 56. Ya. K. Syrkin and M. E. Dyatkina, Khimicheskaya svyaz i stroenie molekul [The Chemical Bonding and Molecular Struc
2398
Ivanov et al.
ture], Goskhimizdat, MoscowLeningrad, 1946, 588 pp. (in Russian). 57. L. I. Koroleva, Magnitnye poluprovodniki [Magnetic Semi conductors], Izd vo MGU, Moscow, 2003, 312 pp. (in Russian). 58. L. V. Keldysh, Zhurn. Eksp. Teor. Fiziki, 1963, 45, 364 [Sov. Phys. J. Exp. Theor. Phys., 1963, 18, 253 (Engl. Transl.)]. 59. V. N. Ryabokon and K. K. Svidzinskii, Fizika tverdogo tela, 1969, 11, 585 [Sov. Phys. Solid State, 1969, 11, 473 (Engl. Transl.)]. 60. V. N. Ryabokon and K. K. Svidzinskii, Fizika i Tekhnika Poluprovodnikov, 1971, 5, 1865 [Sov. Phys. Semiconductors, 1971, 5 (Engl. Transl.)]. 61. V. N. Fleurov and K. A. Kikoin, J. Phys. C, 1976, 9, 1673. 62. F. D. M. Haldane and P. W. Anderson, Phys. Rev., B, 1976, 13, 2553. 63. P. W. Anderson, Usp. Fiz. Nauk, 1979, 127, 19 [Sov. Phys. Usp., 1979 (Engl. Transl.)]. 64. E. L. Nagaev, Fizika magnitnykh poluprovodnikov [Physics of Magnetic Semiconductors], Nauka, Moscow, 1979, 432 pp. (in Russian). 65. C. Zener, Phys. Rev., 1951, 82, 403. 66. P. W. Anderson and H. Hasegawa, Phys. Rev., 1955, 100, 675. 67. P. G. de Gennes, Phys. Rev., 1960, 118, 141. 68. F. Frhlich and F. R. N. Nabarro, Proc. Roy. Soc. (Lon don), A, 1940, 275, 382. 69. S. Schubin and S. Wonsowsky, Sow. Phys., 1935, 7, 292. 70. S. V. Vonsovsky, Zhurn. Eksp. Teor. Fiziki, 1946, 16, 981 (in Russian). 71. S. Vonsovsky, J. Phys., 1946, 10, 486. 72. C. Zener, Phys. Rev., 1951, 81, 440. 73. Yu. P. Irkhin and E. A. Turov, Fizika Metallov i Metallovedenie, 1957, 4, 9 [Phys. Metals Metallography, 1957, 4 (Engl. Transl.)]. 74. Yu. P. Irkhin, Fizika metallov i metallovedenie, 1959, 7, 3 [Phys. Metals Metallography, 1959, 7 (Engl. Transl.)]. 75. E. A. Turov and Yu. P. Irkhin, Fizika metallov i metallovedenie, 1960, 9, 488 [Phys. Metals Metallography, 1960, 9 (Engl. Transl.)]. 76. H. Katayama Yoshida and K. Sato, Physica B , 2003, 327, 337. 77. T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science, 1998, 287, 139. 78. T. Jungwirth, W. A. Atkinson, B. H. Lee, and A. H. MacDonald, Phys. Rev., B, 1999, 59, 9818. 79. T. Dietl, H. Ohno, and F. Matsukura, Phys. Rev., B, 2001, 63, 195205. 80. S. Sanvito, P. Ordejon, and N. A. Hill, Phys. Rev., B, 2001, 63, 165206. 81. F. Matsukura, H. Ohno, A. Shen, and Y. Sugawara, Phys. Rev., B, 1998, 57, 2037 82. J. Szczytko, W. Mac, A. Stachow, A. Twardowski, P. Becla, and J. Tworzydlo, Solid State Commun., 1996, 99, 927. 83. J. Okabayashi, A. Kimura, O. Rader, T. Mizokawa, A. Fujimori, T. Hayashi, and M. Tanaka, Phys. Rev., B, 1998, 58, 4211. 84. J. Szczytko, W. Bardyszewski, and A. Twardowski, Phys. Rev., B, 2001, 64, 075306.
85. H. Ohno, N. Akiba, F. Matsukura, A. Shen, K. Ohtani, and Y. Ohno, Appl. Phys. Lett., 1998, 73, 363. 86. Z. V. Popovi , V. A. Ivanov, M. J. Konstantinovi , A. Cantarero, J. Martnez Pastor, D. Olgun, M. I. Alonso, M. Garriga, O. P. Khuong, A. Vietkin, and V. V. Moshchalkov, Phys. Rev., B, 2001, 63, 165105. 87. A. Overhauser, Phys. Rev., 1953, 92, 411. 88. A. Overhauser, Phys. Rev., 1954, 94, 768. 89. C. P. Slichter, Principles of Magnetic Resonance, 3rd ed., Springer Verlag, Berlin, 1990, 655 pp. 90. M. I. Dyakonov and V. I. Perel, in Modern Problems in Condensed Matter Science, Vol. 8, Eds F. Meier and B. P. Zakharchenya, North Holland, Amsterdam, 1984, p. 11. 91. A. Berg, R. R. Gerhards, and K. von Klitzing, Phys. Rev. Lett., 1990, 64, 2563. 92. K. R. Wald, L. P. Kouwenhoven, and P. L. McEuen, Phys. Rev. Lett., 1994, 73, 1011. 93. S. E. Barrett, G. Dabbagh, L. N. Pfeiffer, K. W. West, and R. Tycko, Phys. Rev. Lett., 1995, 74, 5112. 94. M. I. Dyakonov and V. I. Perel, Zhurn. Eksp. Teor. Fiziki, 1971, 60, 1954 [Sov. Phys. J. Exp. Theor. Phys., 1971, 33, 1053 (Engl. Transl.)]. 95. M. I. Dyakonov and V. I. Perel, Fizika Tverdogo Tela, 1971, 13, 3581 [Sov. Phys. Sol. State, 1971, 13, 3023 (Engl. Transl.)]. 96. M. I. Dyakonov and V. Yu. Kachorovsky, Fizika i Tekhnika Poluprovodnikov, 1986, 20, 178 [Sov. Phys. Semiconductors, 1986, 20, 100 (Engl. Transl.)]. 97. R. W. Wood and A. Ellett, Phys. Rev., 1924, 24, 243. 98. W. Hanle, Z. Physik, 1924, 30, 93. 99. J. Brossel and A. Kastler, Compt. Rend. Hebd. Acad. Sci., 1949, 229, 1213. 100. A. Kastler, Science, 1967, 158, 214. 101. M. I. Dyakonov and V. I. Perel, Pisma v Zhurn. Eksp. Teor. Fiziki, 1971, 13, 657 [Sov. Phys. JETP Lett., 1971, 13, 467 (Engl. Transl.)]. 102. N. S. Averkiev and M. I. Dyakonov, Fizika i Tekhnika Poluprovodnikov, 1983, 17, 629 [Sov. Phys. Semiconductors, 1983, 17, 393 (Engl. Transl.)]. 103. A. A. Bakun, B. P. Zakharchenya, A. A. Rogachev, N. N. Tkachuk, and V. G. Fleisher, Pisma v Zhurn. Eksp. Teor. Fiziki, 1984, 40, 464 [Sov. Phys. JETP Lett., 1984, 40, 1293 (Engl. Transl.)]. 104. M. I. Dyakonov and V. I. Perel, Phys. Lett., A, 1971, 35, 459. 105. E. L. Ivchenko and G. E. Pikus, Pisma v Zhurn. Eksp. Teor. Fiziki, 1978, 27, 604 [Sov. Phys. JETP Lett., 1978, 27, 604 (Engl. Transl.)]. 106. V. I. Belinicher, Phys. Lett., A, 1978, 66, 213. 107. V. M. Asnin, A. A. Bakun, A. M. Danishevskii, E. L. Ivchenko, G. E. Pikus, and A. A. Rogachev, Solid State Commun., 1979, 30, 565. 108. G. Lampel, Phys. Rev. Lett., 1968, 20, 491. 109. M. I. Dyakonov and V. I. Perel, Zhurn. Eksp. Teor. Fiziki, 1973, 65, 362 [Sov. Phys. J. Exp. Theor. Phys., 1973, 38 (Engl. Transl.)]. 110. V. G. Fleisher and I. A. Merkulov, in Optical Orientation, Eds F. Maier and B. P. Zakharchenya, North Holland, Amsterdam, 1984, p. 173. 111. A. I. Ekimov and V. I. Safarov, Pisma v Zhurn. Eksp. Teor. Fiziki, 1972, 15, 453 [Sov. Phys. JETP Lett., 1972, 15, 179 (Engl. Transl.)].
2399
112. P. Paget, G. Lampel, B. Shapoval, and V. I. Safarov, Phys. Rev., B, 1977, 15, 5780. 113. J. M. Kikkawa and D. D. Awshalom, Science , 2000, 287, 473. 114. V. G. Fleisher, R. I. Dzhioev, and B. P. Zakharchenya, Pisma v Zhurn. Eksp. Teor. Fiziki, 1976, 23, 22 [Sov. Phys. JETP Lett., 1976, 23, 18 (Engl. Transl.)]. 115. S. Kronmller, W. Dietsche, J. Weis, K. von Klitzing, W. Wegscheider, and M. Bichler, Phys. Rev. Lett., 1998, 81, 2526. 116. S. F. Alvarado and P. Renaud, Phys. Rev. Lett., 1992, 68, 1387. 117. N. F. Mott, Proc. R. Soc. London, Ser. A, 1936, 153, 699. 118. N. F. Mott, Proc. R. Soc. London, Ser. A, 1936, 156, 368. 119. I. A. Campbell, A. Fert, and A. R. Pomeroy, Phil. Mag., 1967, 15, 977. 120. A. Fert and I. A. Campbell, Phys. Rev. Lett., 1968, 21, 1190. 121. T. Valet and A. Fert, Phys. Rev., B, 1993, 48, 7099. 122. S. von Molnar and T. Kasuya, Phys. Rev. Lett., 1968, 21, 1757. 123. M. Escorne, A. Ghazalli, and P. Leroux Hugon, Proc. 12th Intern. Conf. on the Physics of Semiconductors , Ed. M. H. Pilkuhn, Teubner, Stuttgart, 1974, p. 915. 124. R. W. Cochrane, M. Plischke, and J. O. Strm Olsen, Phys. Rev., B, 1974, 9, 3013. 125. T. Story, R. R. Galazka, R. B. Frankel, and P. A. Wolff, Phys. Rev. Lett., 1986, 56, 777. 126. A. G. Aronov, Zhurn. Eksp. Teor. Fiziki, 1976, 24, 37 [Sov. Phys. JETP, 1976, 24, 32 (Engl. Transl.)]. 127. M. Johnson and R. H. Silsbee, Phys. Rev. Lett., 1985, 55, 1790. 128. R. I. Dzhioev, B. P. Zakharchenya, and V. L. Korenev, Phys. Solid. State, 1995, 37, 1929. 129. V. P. LaBella, D. W. Bullock, Z. Ding, C. Emery, A. Venkatesan, W. F. Oliver, G. J. Salamo, P. M. Thibado, and M. Mortazavi, Science, 2001, 292, 1518. 130. W. F. Egelhoff, Jr., M. D. Stiles, D. P. Pappas, D. T. Pierce, J. M. Byers, B. Johnson, B. T. Jonker, S. F. Alvarado, J. F. Gregg, J. A. C. Bland, and R. A. Buhrman, Science, 2002, 296, 1195. 131. W. F. Egelhoff, Jr., M. D. Stiles, D. P. Pappas, D. T. Pierce, J. M. Byers, B. Johnson, B. T. Jonker, S. F. Alvarado, J. F. Gregg, J. A. C. Bland, and R. A. Buhrman, Science, 2002, 296, 5571. 132. P. R. Hammar, B. R. Bennet, M. J. Yang, and M. Johnson, Phys. Rev. Lett., 1999, 83, 203. 133. G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip, and B. J. van Wees, Phys. Rev., B, 2000, 62, 4790. 134. H. J. Zhu, M. Ramsteiner, H. Kostial, M. Wassermeier, H. P. Schonherr, and K. H. Ploog, Phys. Rev. Lett., 2001, 87, 016601. 135. Semiconductors and Semimetals, Eds J. K. Furdyna and J. Kossut, Vol. 25, Academic Press, New York, 1988, 496 pp. 136. B. T. Jonker, Y. D. Park, B. R. Bennett, H. D. Cheong, G. Kioseoglou, and A. Petrou, Phys. Rev., B, 2000, 62, 8180. 137. M. Oestreich, J. Hubner, D. Hagele, P. J. Klar, W. Heimbrodt, W. W. Ruhle, D. E. Ashenford, and B. Lunn, Appl. Phys. Lett., 1999, 74, 1251. 138. R. Fiederling, M. Keim, G. Reuscher, W. Ossau, G. Schmidt, A. Waag, and L. W. Molenkamp, Nature, 1999, 402, 787.
139. Y. Ohno, D. K. Young, B. Beschoten, F. Matsukura, H. Ohno, and D. D. Awschalom, Nature, 1999, 402, 790. 140. H. Ohno, J. Magn. Magn. Mater., 1999, 200, 110. 141. P. Van Dorpe, Z. Liu, W. Van Roy, V. F. Motsnyi, M. Sawicki, G. Borghs, and J. De Boeck, Appl. Phys. Lett., 2004, 84, 3495. 142. X. Jiang, R. Wang, S. van Dijken, R. Shelby, R. Macfarlane, G. S. Solomon, J. Harris, and S. S. P. Parkin, Phys. Rev. Lett., 2003, 90, 256603. 143. E. I. Rashba, Phys. Rev., B, 2003, 68, 241310(R). 144. J. Stephens, J. Berezovsky, J. P. McGuire, L. J. Sham, A. C. Gossard, and D. D. Awschalom, March Meeting of the APS (Montreal, 2226 March 2004) , Montreal, 2004. 145. A. G. Aronov and G. E. Pikus, Fizika i Tekhnika Poluprovodnikov, 1976, 10, 1177 [Sov. Phys. Semiconduc tors, 1976, 10, 698 (Engl. Transl.)]. 146. J. M. Kikkawa and D. D. Awschalom, Nature, 1999, 397, 139. 147. J. M. Kikkawa and D. D. Awschalom, Phys. Rev. Lett., 1998, 80, 4313. 148. R. I. Dzhioev, V. L. Korenev, I. A. Merkulov, B. P. Zakharchenya, D. Gammon, Al. L. Efros, and D. S. Katzer, Phys. Rev. Lett., 2002, 88, 256801. 149. I. Malajovich, J. J. Berry, N. Samarth, D. D. Awschalom, Nature, 2001, 411, 770. 150. J. A. Gupta, X. Peng, A. P. Alivisatos, and D. D. Awschalom, Phys. Rev., B, 1999, 59, 10421. 151. H. Ohno, D. Chiba, F. Matsukura, T. Omiya, E. Abe, T. Dietl, Y. Ohno, and K. Ohtani, Nature, 2000, 408, 944.WW 152. D. Chiba, M. Yamanouchi, F. Matsukura, and H. Ohno, Science, 2003, 301, 943. 153. D. D. Awshalom and R. K. Kawakami, Nature, 2000, 408, 923. 154. Y. D. Park, A. T. Hanbicki, S. C. Erwin, C. S. Hellberg, J. M. Sullivan, J. E. Mattson, T. F. Ambrose, A. Wilson, G. Spanos, and B. T. Jonker, Science, 2002, 295, 651. 155. Y. Kato, R. C. Myers, A. C. Gossard, and D. D. Awschalom, Nature, 2004, 427, 50. 156. L. I. Magarill, A. V. Chaplik, and M. V. Entin, Fizika i Tekhnika Poluprovodnikov, 2001, 35, 1128 [Semiconduc tors, 2001, 35, 1081 (Engl. Transl.)]. 157. S. Datta and B. Das, Appl. Phys. Lett., 1990, 56, 665. 158. Yu. A. Bychkov, E. I. Rashba, J. Phys. C, 1984, 17, 6039. 159. J. Schliemann, J. C. Egues, and D. Loss, Phys. Rev. Lett., 2003, 90, 146801. 160. J. C. Egues, G. Burkard, and D. Loss, Appl. Phys. Lett., 2003, 82, 2658. 161. M. E. Flatte and G. Vignale, Appl. Phys. Lett., 2001, 78, 1273. 162. B. K nig, U. Zehnder, D. R. Yakovlev, W. Ossau, T. Gerhard, M. Keim, A. Waag, and G. Landwehr, Phys. Rev., B, 1999, 60, 2653. 163. W. L. Ng, M. A. Lourenco, R. M. Gwilliam, S. Ledain, G. Shao, and K. P. Homewood, Nature, 2001, 410, 192. 164. Z. H. Xiong, Di Wu, Z. Valy Vardeny, and J. Shi, Nature, 2004, 427, 821. 165. J. Appelbaum, K. J. Russell, D. J. Monsma, V. Narayanamurti, C. M. Marcus, M. P. Hanson, and A. C. Gossard, Appl. Phys. Lett., 2003, 83, 4571.
2400
Ivanov et al.
166. S. D. Ganichev, E. L. Ivchenko, S. N. Danilov, J. Eroms, W. Wegscheider, D. Weiss, and W. Prettl, Phys. Rev. Lett., 2001, 86, 4358. 167. R. D. R. Bhat and J. E. Sipe, Phys. Rev. Lett., 2000, 85, 5432. 168. R. K. Kawakami, Y. Kato, M. Hanson, I. Malajovich, J. M. Stephens, E. Johnston Halperin, G. Salis, A. C. Gossard, and D. D. Awschalom, Science, 2001, 294, 131. 169. S. I. Kiselev, J. C. Sankey, I. N. Krivotorov, N. C. Emley, R. J. Schoelkopf, R. A. Buhrman, and D. C. Ralph, Na ture, 2003, 425, 380. 170. S. Murakami, N. Nagaosa, and S. Zhang, Science, 2003, 301, 1348. 171. E. I. Rashba, Phys. Rev., B, 2003, 68, 241315 R. 172. A. G. Malshukov, C. S. Tang, C. S. Chu, and K. A. Chao, Phys. Rev., B, 2003, 68, 233307. 173. H. X. Tang, R. K. Kawakami, D. D. Awschalom, and M. L. Roukes, Phys. Rev. Lett., 2003, 90, 107201. 174. H. D. Chopra and S. Z. Hua, Phys. Rev., B, 2002, 66, 020403 R. 175. D. Chiba, F. Matsukura, and H. Ohno, Physica E, 2004, 21, 966. 176. M. Tanaka and Y. Higo, Phys. Rev. Lett., 2001, 87, 026602. 177. W. Weber, S. Riesen, and H. C. Siegmann, Science, 2001, 291 1015. 178. A. Kiselev and K. W. Kim, Appl. Phys. Lett., 2001, 78, 775. 179. M. Ciorga, A. Wensauer, M. Pioro Ladriere, M. Korkusinski, J. Kyriakidis, A. S. Sachrajda, and P. Hawrylak, Phys. Rev. Lett., 2002, 88, 56804. 180. D. Loss and D. P.. DiVincenzo, Phys. Rev. A, 1998, 57, 120. 181. S. J. Tans, A. R. M. Verschueren, and C. Dekker, Nature, 1998, 393, 49. 182. A. A. Jensen, J. R. Hauptmann, J. Nygrd, J. Sadowski, and P. E. Lindelof, Nano Lett., 2004, 4, 349. 183. V. Fleurov, V. A. Ivanov, F. M. Peeters, and I. D. Vagner, Physica E, 2002, 14, 361. 184. M. Poggio, G. M. Steeves, R. C. Myers, Y. Kato, A. C. Gossard, and D. D. Awschalom, Phys. Rev. Lett., 2003, 91, 207602. 185. Yu. I. Manin, Vychislimoe i nevychislimoe [Computable and Non Computable], Sov. Radio, Moscow, 1980, 128 pp. (in Russian). 186. R. P. Feinman, Foundations Phys., 1986, 16, 507. 187. P. W. Shor, Proc. 35th Annual Symposium on Foundations of Computer Science (IEEE Press), 1994, p. 124. 188. K. A. Valiev, Vestnik RAN [Herold of the Russian Academy of Sciences],... , 2000, 70, 688 (in Russian). 189. E. M. Omel yanovskii and V. I. Fistul , Primesi perekhodnykh metallov v polyprovodnikakh [Transition Metal Impurities in Semiconductors], Metallurgiya, Moscow, 1983 (in Russian). 190. B. E. Kane, Nature, 1998, 393, 133. 191. R. Vrijen and D. Di Vincenzo, Phys. Rev., A, 2000, 62, 012306. 192. J. Beille, J. Voiron, and M. Roth, Solid State Commun., 1983, 47, 399. 193. N. Manyala, Y. Sidis, J. F. DiTusa, G. Aeppli, D. P. Young, and Z. Y. Fisk, Nature Materials, 2004, 3, 255. 194. S. Krupicka, Physics of Ferrites and Related Magnetic Ox ides, Academia, NCSAV, Prague, 1969 (in Czech).
195. V. A. Ivanov and R. O. Zaitsev, J. Magn. Magnetic Mater., 1989, 81, 331. 196. J. J. Versluijs, M. A. Bari, and J. M. D. Coey, Phys. Rev. Lett., 2001, 87, 026601 1. 197. S. K. Sampath and F. Cordaro, J. Am. Ceram. Soc., 1998, 81 649. 198. V. A. Ivanov, in Sovremennye problemy i dostizheniya obshchei i neorganicheskoi khimii [Advances and Modern Problems in General and Inorganic Chemistry], Ed. N. T. Kuznetsov, Nauka, Moscow, 2004, p. 150 (in Russian). 199. I. Tsubokawa, J. Phys. Soc. Jpn, 1960, 15, 1664. 200. S. Methfessel and D. C. Mattis, Handb. Phys., 1968, 18, 389. 201. P. Wachter, CRC Crit. Rev. Solid State Sci., 1972, 13, 189. 202. F. Holtzberg, S. von Molnar, and J. M. D. Coey, The Handbook on Semiconductors, 1980, 3, 815. 203. L. Esaki, P. J. Stiles, and S. von Molnar, Phys. Rev. Lett., 1967, 19, 852. 204. M. Penicaud, B. Siberchicot, C. B. Sommers, and J. Kbler, J. Magn. Magn. Mater., 1992, 103, 212. 205. S. M. Watts, S. Wirth, S. von Molnr, A. Barry, and J. M. D. Coey, Phys. Rev., B, 2000, 61, 9621. 206. J. M. D. Coey, in Spin Electronics, Eds M. Ziese and M. J. Thornton, Springer, Berlin, 2001, p. 277. 207. C. P. Beam and D. S. Rodbell, Phys. Rev., 1962, 126, 104. 208. H. Akinaga, M. Mizuguchi, K. Ono, and M. Oshima, J. Magn. Soc. Jpn, 2000, 22, 451. 209. H. Akinaga, T. Manago, and M. Shirai, Jpn J. Appl. Phys., Part 2 Letters, 2000, 39, L1118. 210. J. H. Zhao, F. Matsukura, K. Takamura, E. Abe, D. Chiba, and H. Ohno, Appl. Phys. Lett., 2001, 79, 2776. 211. J. W. Dong, L. C. Chen, and C. J. Palmstrom, Appl. Phys. Lett., 1999, 75, 1443. 212. C. T. Tanaka, J. Nowak, and J. S. Moodera, J. Appl. Phys., 1999, 86, 6239. 213. H. Chiba, T. Atou, and Y. Syono, J. Solid State Chem., 1997, 132, 139. 214. K. Kohn, K. Inoue, O. Horie, and S. I. Akimoto, J. Solid State Chem., 1976, 18, 27. 215. J. D. Garrett, J. E. Greedan, and D. A. MacLean, Mater. Res. Bull., 1981, 16, 145. 216. B. T. Matthias, R. M. Bozorth, and J. H. V. Vleck, Phys. Rev. Lett., 1961, 7, 160. 217. S. G. Yang, T. Li, B. X. Gu, Y. W. Du, H. Y. Sung, S. T. Hung, C. Y. Wong, and A. B. Pakhomov, Appl. Phys. Lett., 2003, 83, 3746. 218. C. M. Fang, G. A. de Wijs, and R. A. de Groot, J. Appl. Phys., 2002, 91, 8340. 219. K. Schwarz, J. Phys., 1986, 16, 211. 220. R. J. Soulen, Jr., J. M. Byers, M. S. Osofsky, B. Nadgorny, T. Ambrose, S. F. Cheng, P. R. Broussard, C. T. Tanaka, J. Nowak, J. S. Moodera, A. Barry, and J. M. D. Coey, Science, 1998, 282, 85. 221. B. L. Chamberland, CRC Crit. Rev. Solid State Sci., 1977, 7, 1. 222. S. Ishibashi, T. Namikawa, and M. Satou, Jpn J. Appl. Phys., 1978, 17, 249. 223. M. S. Laad, L. Craco, and E. Mller Hartmann, Phys. Rev., B, 2001, 64, 214421. 224. Y. Ji, G. J. Strijkers, F. Y. Yang, C. L. Chien, J. M. Byers, A. Anguelouch, G. Xiao, and A. Gupta, Phys. Rev. Lett., 2001, 86, 5585.
2401
225. J. S. Parker, S. M. Watts, P. G. Ivanov, and P. Xiong, Phys. Rev. Lett., 2002, 88, 196601. 226. K. I. Kbayashi, T. Kimura, H. Sawada, K. Terakura, and Y. Tokura, Nature, 1998, 395, 677. 227. C. Ritter, M. R. Ibarra, L. Morellon, J. Blasco, J. Garcia, and J. M. De Teresa, J. Phys.: Condens. Matter, 2000, 12, 8295. 228. H. Asano, N. Kozuka, A. Tsuzuki, and M. Matsui, Appl. Phys. Lett., 2004, 85, 263. 229. M. Retuerto, J. A. Alonso, M. J. Martinez Lope, J. L. Martinez, and M. Garcia Hernandez, Appl. Phys. Lett., 2004, 85, 266. 230. Y. Sui, X. J. Wang, Z. N. Qian, J. G. Cheng, Z. G. Liu, J. P. Miao, Y. Li, and W. H. Su, Appl. Phys. Lett., 2004, 85, 269. 231. A. Hauri, A. Wasiela, A. Arnoult, J. Cibert, S. Tatarenko, T. Dietl, and Y. M. dAubign, Phys. Rev. Lett., 1997, 79, 511. 232. S. H. Wei and S. B. Zhang, Phys. Rev., B, 2002, 66, 155211. 233. L. Hansen, D. Ferrand, G. Richter, M. Thierley, V. Hock, N. Schwarz, G. Reuscher, G. Scmidt, A. Waag, and L. W. Molenkamp, Appl. Phys. Lett., 2001, 79, 3125. 234. Y. Murayama, Mesoscopic Systems: Fundamentals and Ap plications, Wiley VCH, Berlin, 2001. 235. M. A. Ruderman and C. Kittel, Phys. Rev., 1954, 96, 99. 236. T. Kasuya, Prog. Theor. Phys., 1956, 16, 45. 237. K. Yosida, Phys. Rev., 1957, 106, 893. 238. E. I. Kondorskii, Zonnaya teoriya magnetizma (chast 2) [Band Theory of Magnetism (Chapter 2)], Izd vo MGU, Moscow, 1977, 95 pp. (in Russian). 239. A. T. Hindmarch and B. J. Hickey, Phys. Rev. Lett., 2003, 91, 116601. 240. S. M. Ryabchenko and Yu. G. Semenov, Spektroskopiya kristallov [Spectroscopy of Crystals], Fizmatlit, Leningrad, 1983, 432 pp. (in Russian). 241. I. I. Lyapilin and I. M. Tsidilkovskii, Usp. Fiz. Nauk, 1985, 146, 35 [ Sov. Phys. Usp., 1985, 28 , 349 (Engl. Transl.)]. 242. I. M. Tsidilkovskii, Usp. Fiz. Nauk, 1992, 162, 63 [Sov. Phys. Usp., 1992, 35, 85 (Engl. Transl.)]. 243. C. Gould, G. Schmidt, G. Richter, R. Fiederling, P. Grabs, and L. W. Molenkamp, Appl. Surf. Science, 2002, 190, 395.W 244. D. Ferrand, J. Cibert, A. Wasiela, C. Bourgognon, S. Tatarenko, G. Fishman, T. Andrearczyk, J. Jaroszynski, S. Kolesnik, T. Dietl, B. Barbara, and D. Dufeu, Phys. Rev., B, 2001, 63, 085201. 245. H. Saito, V. Zayets, S. Yamagata, Y. Suzuki, and K. Ando, J. Appl. Phys., 2002, 91, 8085. 246. H. Saito, V. Zayets, S. Yamagata, and K. Ando, Phys. Rev. Lett., 2003, 90, 207202. 247. F. Tsui, L. He, L. Ma, A. Tkachuk, Y. S. Chu, K. Nakajima, and T. Chikyow, Phys. Rev. Lett., 2003, 91, 177203. 248. M. Holub, S. Chakrabarti, S. Fathpour, P. Bhattacharyaa, Y. Lei, and S. Ghosh, Appl. Phys. Lett., 2004, 85, 973. 249. Y. L. Soo, S. W. Huang, Z. H. Ming, Y. H. Kao, and H. Munekata, Phys. Rev., B, 1996, 53, 4905. 250. Y. Nishikawa, A. Tackeuchi, M. Yamaguchi, S. Muto, and O. Wada, IEEE J. Sel. Topics Quantum Electron., 1996, 2, 661.
251. S. Koshihara, A. Oiwa, M. Hirasawa, S. Katsumoto, Y. Iye, C. Urano, H. Takagi, and H. Munekata, Phys. Rev. Lett., 1997, 78, 4617. 252. A. Oiwa, T. Slupinski, and H. Munekata, Appl. Phys. Lett., 2001, 78, 518. 253. H. Ohno, J. Cryst. Growth, 2003, 251, 285. 254. Y. Shon, W. C. Lee, Y. S. Park, Y. H. Kwon, S. J. Lee, K. J. Chung, H. S. Kim, D. Y. Kim, D. J. Fu, T. W. Kang, X. J. Fan, Y. J. Park, and H. T. Oh, Appl. Phys. Lett., 2004, 84, 2310. 255. R. Shioda, K. Ando, T. Hayashi, and M. Tanaka, Phys. Rev., B, 1998, 58, 1100. 256. H. Ohno, A. Shen, F. Matsukura, A. Oiwa, A. Endo, S. Katsumoto, and Y. Iye, Appl. Phys. Lett., 1996, 69, 396. 257. J. Sadowski, P. Mathieu, P. Svedlindh, J. Z. Domagala, J. Bak Misiuk, K. Swiatek, M. Karlsteen, J. Kanski, L. Ilver, H. sklund, and U. Sodervall, Appl. Phys. Lett., 2001, 78, 3271. 258. T. Kuroiwa, F. Matsukura, A. Shen, Y. Ohno, H. Ohno, T. Yasuda, and Y. Segawa, Electronic Lett., 1998, 34, 190. 259. H. Munekata, Intern. Conf. on Crystal Growth 13, ICCG 13 (Kyoto, Japan, August, 2001), 2001. 260. A. van Esch, L. van Bockstal, J. de Boeck, G. Verbanck, A. S. vas Steenbergen, R. J. Wellman, G. Grietens, R. Bogaerts, F. Herlach, and G. Borghs, Phys. Rev., B, 1997, 56, 13103. 261. S. J. Potashnik, K. C. Ku, R. Mahendiran, S. H. Chun, R. F. Wang, N. Samarth, and P. Schiffer, Phys. Rev., B, 2002, 66, 012408. 262. H. Shimizu, T. Hayashi, T. Nishinaga, and M. Tanaka, Appl. Phys. Lett., 1999,74, 398. 263. T. Hayashi, Y. Hashimoto, S. Katsumoto, and Y. Iye, Appl. Phys. Lett., 2001, 78, 1691. 264. D. V. Baxter, D. Ruzmetov, J. Scherschligt, Y. Sasaki, X. Liu, J. K. Furdyna, and C. H. Mielke, Phys. Rev., B, 2002, 65, 212407. 265. T. Hayashi, M. Tanaka, T. Nishinaga, H. Shimoda, H. Tsuchiya, and Y. Otsuka, J. Cryst. Growth , 1997, 175, 1063. 266. Y. Satoh, N. Inoue, Y. Nishikawa, and J. Yoshino, Proc. 3rd Symp. Physics and Applications of Spin Related Phenom ena in Semiconductors (Sendai, Japan, November 1997), Eds H. Ohno, J. Yoshino, and Y. Oka, 1997, p. 23. 267. T. Hayashi, M. Tanaka, T. Nishinaga, and H. Shimada, J. Appl. Phys., 1997, 81, 4865. 268. T. Hayashi, M. Tanaka, K. Seto, T. Nishinaga, and K. Ando, Appl. Phys. Lett., 1997, 71, 1825. 269. M. Tanaka, J. Vac. Sci. Technol. B, 1998, 16, 2267. 270. H. Ohno, J. Magn. Magn. Mater., 2004, 272276, 1. 271. K. Ando, T. Hayashi, M. Tanaka, and A. Twardowski, J. Appl. Phys., 1998, 83, 65481. 272. H. Ohno, F. Matsukura, T. Owiya, and N. Akiba, J. Appl. Phys., 1999, 85, 4277. 273. B. Beschoten, P. A. Crowell, I. Malajovich, D. D. Awschalom, F. Matsukura, A. Shen, and H. Ohno, Phys. Rev. Lett., 1999, 83, 3073. 274. A. Shen, F. Matsukura, S. P. Guo, Y. Sugawara, H. Ohno, M. Tani, A. Abe, and H. C. Liu, J. Cryst. Growth, 1999, 201/202, 379. 275. A. Twardowski, Mat. Sci. Eng., B, 1999, 63, 96.
2402
Ivanov et al.
276. B. Grandidier, J. P. Hys, C. Delerue, D. Stievenard, Y. Higo, and M. Tanaka, Appl. Phys. Lett., 2000, 77, 4001. 277. R. K. Kawakami, E. Johnson Halperin, L. F. Chen, M. Hanson, N. Guebels, J. S. Speck, A. C. Gossard, and D. D. Awschalom, Appl. Phys. Lett., 2000, 77, 2379. 278. D. Chiba, N. Akiba, F. Matsukura, Y. Ohno, and H. Ohno, Appl. Phys. Lett., 2000, 77, 1873. 279. T. Hayashi, M. Tanaka, and A. Asamitsu, J. Appl. Phys., 2000, 87, 4673. 280. N. Akiba, D. Chiba, K. Natata, F. Matsukura, Y. Ohno, and H. Ohno, J. App. Phys., 2000, 87, 6436. 281. Y. Nagai, T. Kurimoto, K. Nagasaka, H. Nojiri, M. Motokawa, F. Matsukura, T. Dietl, and H. Ohno, Jpn J. Appl. Phys., 2001, 40, 6231. 282. S. J. Potashnik, K. C. Ku, S. H. Chun, J. J. Berry, N. Samarth, and P. Schiffer, Appl. Phys. Lett ., 2001, 79, 1495. 283. G. M. Schott, W. Faschinger, and L. W. Molenkamp, Appl. Phys. Lett., 2001, 79, 1807. 284. K. M. Yu, W. Walukiewicz, T. Wojtowicz, I. Kuryliszyn, X. Liu, Y. Sasaki, and J. K. Furdyna, Phys. Rev., B, 2002, 65, 201303. 285. J. Masek, J. Kudrnovsky, and F. Maca, Phys. Rev., B, 2003, 67, 153203. 286. T. B. Goennenwein, T. A. Wassner, H. Huebl, M. S. Brandt, J. B. Philipp, M. Opel, R. Gross, A. Koeder, W. Schoch, and A. Waag, Phys. Rev. Lett., 2004, 92, 227202. 287. E. Ohya, H. Shimizu, Y. Higo, J. Sun, and M. Tanaka, Jpn J. Appl. Phys., 2002, 41, 24. 288. T. Slupinski, H. Munekata, and A. Oiwa, Appl. Phys. Lett., 2002, 80, 1592. 289. Th. Hartmann, M. Lampalzer, W. Stolz, K. Megges, L. Lorberth, P. J. Klar, and W. Heimbrodt, Thin Solid Films, 2000, 364, 209. 290. W. Heimbrodt, Th. Hartmann, P. J. Klar, M. Lampalzer, W. Stolz, K. Volz, A. Schaper, W. Treutmann, H. A. Krug von Nidda, A. Loidl, T. Ruf, and V. F. Sapega, Physica E, 2001, 10, 175. 291. D. Chiba, K. Takamura, F. Matsukura, and H. Ohno, Appl. Phys. Lett., 2003, 82, 3020. 292. A. M. Nazmul, S. Sugahara, and M. Tanaka, Phys. Rev., B, 2003, 67, 241308. 293. S. Haneda, M. Yamaura, Y. Takatani, K. Hara, S. Harigae, and H. Munekata, Jpn J. Appl. Phys., 2000, 39, L9. 294. Y. L. Soo, G. Kioseoglou, S. Huang, S. Kim, Y. H. Kao, Y. Takatani, S. Haneda, and H. Munekata, Phys. Rev., B, 2001, 63, 195209. 295. T. Shono, T. Hasegawa, T. Fukumura, F. Matsukura, and H. Ohno, Appl. Phys. Lett., 2000, 77, 1363. 296. N. Theodoropoulou, A. F. Hebard, S. N. G. Chu, M. E. Overgerg, C. R. Abernathy, S. J. Pearton, R. G. Wilson, and J. M. Zavada, J. Appl. Phys., 2002, 91, 7499. 297. M. E. Overberg, B. P. Gila, G. T. Thaler, C. R. Abernathy, S. J. Pearton, N. Theodoropoulou, K. T. McCarthy, S. B. Arneson, A. F. Hebard, S. N. G. Chu, R. G. Wilson, J. M. Zavada, and Y. D. Park, J. Vac. Sci. Technol. B, 2002, 20, 969. 298. N. Theodoropoulou, A. F. Hebard, M. E. Overberg, C. R. Abernathy, S. J. Pearton, S. N. G. Chu, and R. G. Wilson, Phys. Rev. Lett., 2002, 89, 107203.
299. M. Zajac, J. Gosk, M. Kaminska, A. Twardowski, T. Szyszko, and S. Podliasko, Appl. Phys. Lett., 2001, 79, 2432. 300. M. L. Reed, M. K. Ritums, H. H. Stadelmaier, M. J. Reed, C. A. Parker, S. M. Bedair, and N. A. El Masry, Mater. Lett., 2001, 51, 500. 301. M. L. Reed, N. A. El Masry, H. Stadelmaier, M. E. Ritums, N. J. Reed, C. A. Parker, J. C. Roberts, and S. M. Bedair, Appl. Phys. Lett., 2001, 79, 3473. 302. N. Theodoropoulou, A. F. Hebard, M. E. Overberg, C. R. Abernathy, S. J. Pearton, S. N. G. Chu, R. G. and Wilson, Appl. Phys. Lett., 2001, 78, 3475. 303. S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto, and H. Hoi, J. Crystal. Growth, 2002, 237, 1358. 304. T. A. Bither and W. H. Cloud, J. Appl. Phys., 1965, 36, 1501.W 305. G. T. Thaler, M. E. Overberg, B. Gila, R. Frazier, C. R. Abernathy, S. J. Pearton, J. S. Lee, S. Y. Lee, Y. D. Park, Z. G. Khim, J. Kim, and F. Ren, Appl. Phys. Lett., 2002, 80, 3964. 306. Y. L. Soo, G. Kioseouglou, S. Kim, S. Huang, Y. H. Kaa, S. Kubarawa, S. Owa, T. Kondo, and H. Munekata, Appl. Phys. Lett., 2001, 79, 3926. 307. N. A. Theodoropoulou, A. F. Hebard, S. N. G. Chu, M. E. Overberg, C. R. Abernathy, S. J. Pearton, R. G. Wilson, and J. M. Zavada, Appl. Phys. Lett., 2001, 79, 3452. 308. S. J. Pearton, M. E. Overberg, G. Thaler, C. R. Abernathy, N. Theodoropoulou, A. F. Hebard, S. N. G. Chu, R. G. Wilson, J. M. Zavada, A. Y. Polyakov, A. Osinsky, and Y. D. Park, J. Vac. Sci. Technol. A, 2002, 20, 583. 309. H. Akinaga, S. Nemeth, J. De Boeck, L. Nistor, H. Bender, G. Borghs, H. Ofuchi, and M. Oshima, Appl. Phys. Lett., 2000, 77, 4377. 310. M. Hashimoto, Y. K. Zhou, M. Kanamura, and H. Asahi, Solid State Commun., 2002, 122, 37. 311. S. E. Park, H. J. Lee, Y. C. Cho, Se Y. Jeong, C. R. Cho, and S. Cho, Appl. Phys. Lett., 2002, 80, 4187. 312. T. Endo, T. Slupi nski, S. Yanagi, A. Oiwa, and H. Munekata, Extended Abstracts, 7 th Symp. on the Physics and Application of Spin Related Phenomena in Semicon ductor s (Yokohama, Japan, December 1718, 2001) , 2001, p. 33. 313. M. Yamanouchi, D. Chiba, F. Matsukura, and H. Ohno, Nature, 2004, 428, 539. 314. S. G. Yang, A. B. Pakhomov, S. T. Hung, and C. Y. Wong, Appl. Phys. Lett., 2002, 81, 2418. 315. S. Y. Wu, H. X. Liu, Lin Gu, R. K. Singh, L. Budd, M. van Schilfgaarde, M. R. McCartney, D. J. Smith, and N. Newman, Appl. Phys. Lett., 2003, 82, 3047. 316. S. Yanagi, K. Kuga, T. Slupinski, and H. Munekata, Physica E, 2004, 20, 333. 317. G. A. Medvedkin, T. Ishibashi, T. Nishi, and K. Hiyata, Jpn J. Appl. Phys., 2000, 39, L949. 318. G. A. Medvedkin, K. Hirose, T. Ishibashi, T. Nishi, V. G. Voevodin, and K. Sato, J. Cryst. Growth, 2002, 236, 609. 319. P. G. Baranov, S. I. Goloshchapov, G. A. Medvedkin, and V. G. Voevodin, Pis ma v Zhurn. Eksp. Teor. Fiziki, 2003, 77, 686 [JETP Lett., 2003, 77, 582 (Engl. Transl.)].
2403
320. P. G. Baranov, S. I. Goloshchapov, G. A. Medvedkin, S. B. Orlinskii, and V. G. Voevodin, Physica B, 2003, 340342, 878. 321. S. Cho, S. Choi, G. B. Cha, S. C. Hong, Y. Kim, Y. J. Zhao, A. J. Freeman, J. B. Ketterson, B. J. Kim, Y. C. Kim, and B. C. Choi, Phys. Rev. Lett., 2002, 88, 257203. 322. S. Choi, G. B. Cha, S. C. Hong, S. Cho, Y. Kim, J. B. Ketterson, S. Y. Jeong, and G. C. Yi, Solid State Commun., 2002, 122, 165. 323. S. J. Pearton, C. R. Abernathy, G. T. Thaler, R. Frazier, F. Ren, A. F. Hebard, Y. D. Park, D. P. Norton, W. Tang, M. Stavola, J. M. Zavada, and R. G. Wilson, Physica B, 2003, 340342, 39. 324. S. J. Pearton, M. E. Overberg, C. R. Abernathy, N. A. Theodoropoulou, A. F. Hebard, S. N. G. Chu, A. Osinsky, V. Zuflyigin, L. D. Zhu, A. Y. Polyakov, and R. G. Wilson, J. Appl. Phys., 2002, 92, 2047. 325. S. Cho, S. Choi, G. B. Cha, S. C. Hong, Y. Kim, A. J. Freeman, J. B. Ketterson, Y. Park, and H. M. Park, Solid State Commun., 2004, 129, 609. 326. K. Sato, G. A. Medvedkin, T. Ishibashi, S. Mitani, K. Takanashi, Y. Ishida, D. D. Sarma, J. Okabayashi, A. Fujimori, T. Kamatani, and H. Akai, J. Phys. Chem. Solids, 2003, 64, 1461. 327. Y. Ishida, D. D. Sarma, K. Okazaki, J. Okabayashi, J. I. Hwang, H. Ott, A. Fujimori, G. A. Medvedkin, T. Ishibashi, and K. Sato, Phys. Rev. Lett., 2003, 91, 107202. 328. T. Hwang, J. H. Shim, and S. Lee, Appl. Phys. Lett., 2003, 83, 1809. 329. V. M. Novotortsev, V. T. Kalinnikov, L. I. Koroleva, R. V. Demin, S. F. Marenkin, T. G. Aminov, G. G. Shabunina, S. V. Boichuk, and V. A. Ivanov, Zh. Neorg. Khim., 2005, 50, Iss. 4 [Russ. J. Inorg. Chem., 2005, 50, Iss. 4 (Engl. Transl.)]. 330. S. J. Pearton, Y. D. Park, C. R. Abernathy, M. E. Overberg, G. T. Thaler, J. Kim, F. Ren, J. M. Zavada, and R. G. Wilson, Thin Solid Films, 2004, 447448, 493. 331. V. N. Prigodin, N. P. Raju, K. I. Pokhodnya, and J. S. Miller, A.J.Epstein Adv. Mater., 2002, 14, 1230. 332. M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, and J. M. Tour, Science, 1997, 278, 252. 333. E. G. Emberley and G. Kirczenow, Chem. Phys., 2002, 281, 311. 334. Y. Masumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M. Kawasaki, P. Ahmet, T. Chikyow, S. Koshihara, and H. Koinuma, Science, 2001, 291, 854. 335. I. B. Shim, S. Y. An, C. S. Kim, S. Y. Choi, and Y. W. Park, J. Appl. Phys., 2002, 91, 7914. 336. S. A. Chambers, S. Thevuthasan, R. F. C. Farrow, R. F. Marks, J. Thiele, L. Folks, M. G. Samant, A. J. Kellock, N. Ruzycki, D. L. Ederer, and U. Diebold, Appl. Phys. Lett., 2001, 79, 3467. 337. R. J. Kennedy and P. Stampe, Proc. Amer. Phys. Soc. Meet ing, March 1822, 2002. 338. R. J. Kennedy, P. A. Stampe, E. Hu, P. Xiong, S. von Moln r, and Y. Xin, Appl. Phys. Lett ., 2004, 84, 2832.W 339. L. A. Balagurov, S. O. Klimonskii, S. P. Kobeleva, A. F. Orlov, N. S. Perov, and D. G. Yarkin, Pisma v Zhurn. Eksp. Teor. Fiziki, 2004, 79, 111 [JETP Lett., 2004, 79, 98 (Engl. Transl.)].
340. Z. Wang, W. Wang, J. Tang, L. D. Tung, L. Spinu, and W. Zhou, Appl. Phys. Lett., 2003, 83, 518. 341. N. H. Hong, J. Sakai, and A. Hassini, Appl. Phys. Lett., 2004, 84, 2602. 342. H. Akinaga, M. Mizuguchi, K. Ono, and M. Oshima, Appl. Phys. Lett., 2000, 76, 357. 343. R. Engel Herbert, J. Mohanty, A. Ney, T. Hesjedal, L. Dweritz, and K. H. Ploog, Appl. Phys. Lett., 2004, 84, 1132. 344. K. Morita, K. M. Itoh, J. Muto, K. Mizoguchi, N. Usami, Y. Shiraki, and E. E. Haller, Thin Solid Films, 2000, 369, 405. 345. N. R. S. Farley, C. R. Staddon, L. Zhao, K. W. Edmonds, B. L. Gallagher, and D. H. Gregory, J. Mater. Chem., 2004, 14, 1087. 346. K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett., 2001, 79, 988. 347. J. H. Kim, H. Kim, D. Kim, Y. E. Ihm, and W. K. Choo, Physica B, 2003, 327, 304. 348. S. W. Lim, D. K. Hwang, and J. M. Myoung, Solid State Commun., 2003, 125, 231. 349. D. P. Norton, S. J. Pearton, A. F. Hebard, N. Theodoropoulou, L. A. Boatner, and R. G. Wilson, Appl. Phys. Lett., 2003, 82, 239. 350. S. G. Yang, A. B. Pakhomov, S. T. Hung, and C. Y. Wong, IEEE Trans. Magn., 2002, 38, 2877. 351. P. Sharma, A. Gupta, F. J. Owens, K. V. Rao, R. Sharma, R. Ahuja, J. M. O. Guillen, B. Johannson, and G. A. Gehring, Nature Mater., 2003, 2, 673. 352. Y. W. Heo, M. P. Ivill, K. Ip, D. P. Norton, S. J. Pearton, J. G. Kelly, R. Rairigh, A. F. Hubard, and T. Steiner, Appl. Phys. Lett., 2004, 84, 2292. 353. T. Fukumura, Z. Jin, M. Kawasaki, T. Shono, T. Hasegawa, S. Koshihara, and H. Koinuma, Appl. Phys. Lett., 2001, 78, 958. 354. S. Kolesnik, B. Dabrowski, and J. Mais, J. Supercond., 2002, 15, 251. 355. H. Saeki, H. Tabata, and T. Kawai, Solid State Commun., 2001, 120, 439. 356. A. S. Risbud, N. A. Spaldin, Z. Q. Chen, S. Stemmer, and R. Seshadri, Phys. Rev., B, 2003, 68, 205202. 357. Z. Jin, T. Fukumura, M. Kawasaki, K. Ando, H. Saito, T. Sekiguchi, Y. Z. Yoo, M. Murakami, Y. Matsumoto, T. Hasegawa, and H. Koinuma, Appl. Phys. Lett., 2001, 78, 3824. 358. V. A. L. Roy, A. B. Djuriic , H. Liu, X. X. Zvang, Y. H. Leung. M. H. Xie, J. Gao, H. F. Lui, and C. Surya, Appl. Phys. Lett., 2004, 84, 756. 359. P. V. Radovanovic and D. R. Gamelin, Phys. Rev. Lett., 2003, 91,157202. 360. S. Ramachandran, A. Tiwari, and J. Narayan, Appl. Phys. Lett., 2004, 84, 5255. 361. D. A. Schwartz, N. S. Norberg, Q. P. Nguyen, J. M. Parker, and D. R. Gamelin, J. Am. Chem. Soc., 2003, 125, 13205. 362. K. M. Hanif, R. W. Meulenberg, and G. F. Strouse, J. Am. Chem. Soc., 2002, 124, 11495. 363. Y. H. Jeong, S. J. Han, J. H. Park, and Y. H. Lee, J. Magn. Magn. Mater., 2004, 272276, 1976. 364. F. V. Mikulec, M. Kuno, M. Bennati, D. A. Hall, R. G. Griffin, and M. G. Bawendi, J. Am. Chem. Soc., 2000, 122, 2532.
2404
Ivanov et al.
365. J. F. Suyver, S. F. Wuister, J. J. Kelly, and A. Meijerink, Nano Lett., 2001, 1, 429. 366. D. J. Norris, N. Yao, F. T. Charnock, and T. A. Kennedy, Nano Lett., 2001, 1, 3. 367. C. A. Stowell, R. J. Wiacek, A. E. Saunders, and B. A. Korgel, Nano Lett., 2003, 3, 1441. 368. X. Chen, M. Na, M. Cheon, S. Wang, H. Luo, B. D. McCombe, X. Liu, Y. Sasaki, T. Wojtowicz, J. K. Furdyna, S. J. Potashnik, and P. Schiffer, Appl. Phys. Lett., 2002, 81, 511. 369. S. B. Ogale, R. J. Choudhary, J. P. Buban, S. E. Lofland, S. R. Shinde, S. N. Kale, V. N. Kulkarni, J. Higgins, C. Lanci, J. R. Simpson, N. D. Browning, S. Das Sarma, H. D. Drew, R. L. Greene, and T. Venkatesan, Phys. Rev. Lett., 2003, 91, 077205. 370. J. De Boeck, W. van Roy, J. Das, V. F. Motsnyi, Z. Liu, L. Lagae, H. Boeve, K. Dessein, and G. Borghs, Semicond. Sci. Technol., 2002, 17, 342. 371. D. P. Young, D. Hall, M. E. Torelli, Z. Fisk, J. L. Sarrao, J. D. Thompson, H. R. Ott, S. B. Oseroff, R. G. Goodrich, and R. Zysler, Nature, 1999, 397, 412. 372. H. R. Ott, J. L. Gavilano, B. Ambrosini, P. Vonlanthen, E. Felder, L. Degiorgi, D. P. Young, Z. Fisk, and R. Zysler, Physica B, 2000, 281/282, 423. 373. H. J. Tromp, P. van Gelderen, P. J. Kelly, G. Brocks, and P. A. Bobbert, Phys. Rev. Lett., 2001, 87, 016401. 374. F. M. Zhang, Y. Zeng, J. Gao, X. C. Liu, X. S. Wu, and Y. W. Du, J. Magn. Magn. Mater., 2004, 282, 216. 375. H. K. Kim, D. Kwon, J. H. Kim, Y. E. Ihm, D. Kim, H. Kim, J. S. Baek, C. S. Kim, and W. K. Choo, J. Magn. Magn. Mater., 2004, 282, 244. 376. F. M. Zhang, X. C. Liu, J. Gao, X. S. Wu, Y. W. Du, H. Zhu, J. Q. Xiao, and P. Chen, Appl. Phys. Lett., 2004, 85, 786. 377. A. Oiwa, Y. Mitsumori, R. Moriya, T. Supinski, and H. Munekata, Phys. Rev. Lett., 2002, 88, 137202. 378. J. Wang, G. A. Khodaparast, J. Kono, T. Slupinski, A. Oiwa, and H. Munekata, J. Supercond: Incorporating Novel Magnetism, 2003, 16, 373. 379. J. Konig, J. Schlieman, T. Jungwirth, and A. H. MacDonald, in Electronic Structure and Magnetism of Com plex Materials, Springer Verlag, Berlin, 2002. 380. T. Dietl, F. Matsukura, and H. Ohno, Phys. Rev., B, 2002, 66, 033203. 381. M. Berciu and R. N. Bhatt, Phys. Rev. Lett., 2001, 87, 107203. 382. Yu. G. Semenov and S. M. Ryabchenko, Fizika Nizkikh Temperatur, 2000, 26, 1197 [Low Temp. Phys., 2000, 26, 1886 (Engl. Transl.)]. 383. V. K. Dugaev, V. I. Litvinov, J. Barnas, and M. Vieira, Phys. Rev., B, 2003, 67, 033201. 384. J. Inoue, S. Nonoyama, and H. Itoh, Phys. Rev. Lett., 2000, 85, 4610. 385. J. Inoue, S. Nonoyama, and H. Itoh, Physica E, 2001, 10, 170. 386. A. A. Abrikosov and L. P. Gorkov, Zhurn. Eksp. Teor. Fiziki, 1963, 16, 1575 [Sov. Phys. J. Exp. Theor. Phys., 1963, 16, 1575 (Engl. Transl.)]. 387. S. Sanvito, G. Theurich, and N. Hill, J. Superconductivity, 2002, 15, 85. 388. H. Akai, Phys. Rev. Lett., 1998, 81, 3002.
389. J. Park, S. Kwon, and B. Min, Physica B , 2000, 281/282, 703. 390. P. Mahadevan and A. Zunger, Phys. Rev., B, 2004, 69, 115211. 391. K. Sato and H. Katayama Yoshida, Jpn J. Appl. Phys., 2001, 40, L485. 392. P. Mahadevan and A. Zunger, Phys. Rev. Lett., 2002, 88, 047205. 393. P. Mahadevan and A. Zunger, Phys. Rev. Lett., 2002, 88, 159904 erratum. 394. L. Kronik, M. Jain, and J. R. Chelikowsky, Phys. Rev., B, 2002, 66, 041203 R. 395. M. van Schilfgaarde and O. N. Mryasov, Phys. Rev., B, 2001, 63, 233205. 396. E. Kulatov, H. Nakayama, H. Mariette, H. Ohta, and Y. A. Uspenskii, Phys. Rev., B, 2002, 66, 045203. 397. V. A. Telezhkin and K. B. Tolpygo, Fizika i Tekhnika Poluprovodnikov, 1983, 16, 1137 [Sov. Phys. Semiconduc tors, 1983, 16, 857 (Engl. Transl.)]. 398. S. Fujinaga, Molecular Orbitals Method, Ivanami Setan, Tokyo, 1980 (in Japanese) [Metod molekulyarnykh orbitalej, Mir, Moscow, 1983, 461 pp. (translated from Japanese)]. 399. Yu. M. Kagan and K. A. Kikoin, Pisma v Zhurn. Eksp. Teor. Fiziki, 1980, 31, 367 [Sov. Phys. JETP Lett., 1980, 31, 335 (Engl. Transl.)]. 400. A. Zunger, in Solid State Physics, Vol. 39, Eds H. Ehrenreich and D. Turnbull, Academic, Orlando, 1986, p. 276. 401. K. A. Kikoin, Elektronnye svoistva primesei perekhodnykh metallov v poluprovodnikakh [Electronic Properties of Tran sition Metal Impurities in Semiconductors], Energoatomizdat, Moscow, 1991, 304 pp. (in Russian). 402. K. A. Kikoin and V. N. Fleurov, Transition Metal Impurities in Semiconductors, World Scientific, Singapore, 1994, 360 pp. 403. I. M. Lifshits, Usp. Fiz. Nauk, 1963, 83, 617 [Sov. Phys. Usp., 1963, 6 (Engl. Transl.)]. 404. Yu. A. Izyumov and M. V. Medvedev, Teoriya magnitouporyadochennykh kristallov s primesyami [Theory of Magnetically Ordered Doped Crystals], Nauka, Moscow, 1970, 271 pp. (in Russian). 405. B. I. Shklovskii and A. L. Efros, Elektronnye svoystva legirovannykh poluprovodnikov [Electronic Properties of Doped Semiconductors], Nauka, Moscow, 1979, 416 pp. (in Russian). 406. I. M. Lifshits, S. A. Gredeskul, and L. A. Pastur, Vvedenie v teoriyu neuporyadochennykh sistem [Introduction to Theory of Disordered Systems], Nauka, Moscow, 1982, 358 pp. (in Russian). 407. M. Lannoo and J. Bourgoin, Point Defects in Semicon ductors , Springer Verlag, BerlinHeidelbergNew York, 1981. 408. V. A. Ivanov, P. M. Krstajic , F. M. Peeters, V. N. Fleurov, and K. Kikoin, J. Magn. Magn. Mater., 2003, 258259, 237.W 409. P. M. Krstajic , V. A. Ivanov, F. M. Peeters, V. N. Fleurov, and K. Kikoin, Europhysics Lett., 2003, 61, 235. 410. P. M. Krstajic , V. A. Ivanov, F. M. Peeters, V. Fleurov, and K. Kikoin, J. Superconductivity: Incorporating Novel Mag netism, 2003, 16, 111. 411. V. A. Ivanov, P. M. Krstajic , F. M. Peeters, V. Fleurov, and K. Kikoin, Physica B, 2003, 329, 1282.
2405
412. V. Fleurov, K. Kikoin, V. A. Ivanov, P. M. Krstajic , and F. M. Peeters, J. Magn. Magn. Mater. , 2004, 272276, 1967. 413. V. A. Ivanov, V. Fleurov, and K Kikoin; V. M. Novotortsev, T. G. Aminov, S. F. Marenkin, G. G. Shabunina, L. I. Koroleva, and V. T. Kalinnikov; B. A. Aronzon, V. V. Rylkov, and S. V. Gudenko, Proc. 16th National Symp. on Condensed Matter Physics, SFKM 2004, Eds R. Zikic, Z. V. Popovic , M. Damjanovic , Z. Radovic , Sokobanja (Serbia and Montenegro), September 2023, 2004, p. 224227. 414. P. M. Krstajic , F. M. Peeters, V. A. Ivanov, V. Fleurov, and K. Kikoin, Phys. Rev. B, 2004, 70, 195215. 415. B. A. Aronzon, V. V. Rylkov, S. V. Gudenko, S. G. Mikhailov, V. Fleurov, K. Kikoin, and V. A. Ivanov, Materialy nauchnoi konf. ISFTT [Proc. Annual ISSSP Sci entific Conference] (Russian Scientific Center "Kurchatov In stitute", Moscow, Russia, March 1618, 2004), Moscow, 2004, p. 70 (in Russian). 416. a) S. F. Marenkin, V. M. Novotortsev, T. G. Aminov, V. A. Ivanov; V. Fleurov, K. Kikoin; L. I. Koroleva, V. A. Morozova, R. V. Temin, and R. Szymczak, 14th Intern. Conf. on Crystal Growth ICCG 14, Session "Growth of Mag netic Semiconductors for Spintronics", Book of Abstr. (0655) (Grenoble, France, 913 August 2004), Grenoble (France), 2004; b) V. Fleurov, V. A. Ivanov, and K. Kikoin, Frontiers in Condensed Matter, Russia Israel Conference (Shoresh (near Jerusalem), Tel Aviv, October 1927, 2003), p. 110, http://arXiv.org/cond mat/0311525. 417. Data in Science and Technology Semiconductors: Group IV Elements and III V Compounds, Ed. O. Madelung, Springer Verlag, Berlin, 1991, 164 pp. 418. H. sklund, L. Ilver, J. Kanski, J. Sadowski, and R. Mathieu, Phys. Rev., B, 2002, 66, 115319. 419. J. Schneider, U. Kaufmann, W. Wilkening, M. Baeumler, and F. Khl, Phys. Rev. Lett., 1987, 59, 240. 420. M. Linnarsson, E. Janzen, B. Monemar, M. Kleverman, and A. Thilderkvist, Phys. Rev., B, 1997, 55, 6938. 421. E. J. Singley, K. S. Burch, R. Kawakami, J. Stephens, D. D. Awschalom, and D. N. Basov, Phys. Rev., B, 2003, 68, 165204. 422. T. Graf, M. Gjukic, L. Grgens, O. Ambacher, M. S. Brandt, and M. Stutzmann, Appl. Phys. Lett ., 2002, 81, 5159. 423. V. T. Kalinnikov, Yu. V. Rakitin, and V. M. Novotortsev, Usp. Khim., 2003, 72, 1123 [Russ. Chem. Rev., 2003, 72, 995 (Engl. Transl.)]. 424. V. F. Sapega, T. Ruf, and M. Cardona, Phys. Stat. Solidi (b), 2001, 226.
425. K. W. Edmonds, K. Y. Wang, R. P. Campion, A. C. Neumann, C. T. Foxon, B. L. Gallagher, and P. C. Main, Appl. Phys. Lett., 2002, 81, 3010. 426. I. Kuryliszyn, T. Wojtowicz, X. Liu, J. K. Furdyna, W. Dobrowolski, J. M. Broto, M. Goiran, O. Portugal, H. Rakoto, and B. I. Raquet, Acta Phys. Pol. A, 2002, 102, 659. 427. K. C. Ku, 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 D. D. Awschalom, Appl. Phys. Lett., 2003, 82, 2302. 428. B. Gallagher, CECAM Workshop: Diluted Magnetic Semi conductors (Lyon, June 1214 2003), Lyon, 2003. 429. K. M. Yu, W. Walukiewicz, T. Wojtowicz, W. L. Lim, X. Liu, Y. Sasaki, M. Dobrowolska, and J. K. Furdyna, Appl. Phys. Lett., 2002, 81, 5. 430. S. Sonoda, S. Shimizu, T. Sasaki, Y. Yamamoto, and H. Hori, J. Appl. Phys., 2002, 91, 7911. 431. T. Okuno, T. Fujino, M. Shindo, M. Katayama, K. Oura, S. Sonoda, and S. Shimizu, Jpn J. Appl. Phys., 2002, 41, L415. 432. M. Sato, H. Tanida, K. Kato, T. Sasaki, Y. Yamamoto, S. Sonoda, S. Shimizu, and H. Hori, Jpn J. Appl. Phys., 2002, 41, 4513 433. H. Hori, S. Sonoda, T. Sasaki, Y. Yamamoto, S. Shimizu, K. Suga, and K. Kindo, Physica B, 2002, 324, 142. 434. I. Ya. Korenblit and E. F. Shender, Usp. Fiz. Nauk, 1978, 126, 233 [Sov. Phys. Usp., 1978, 21, 832 (Engl. Transl.)]. 435. Y. J. Zhao, W. T. Geng, A. J. Freeman, and T. Oguchi, Phys. Rev. B, 2001, 63, 201202. 436. Y. J. Zhao, S. Picozzi, A. Continenza, W. T. Geng, and A. J. Freeman, Phys. Rev., B, 2002, 65, 094415. 437. S. Cho, S. Choi, S. C. Hong, Y. Kim, J. B. Ketterson, B. J. Kim, Y. C. Kim, and J. H. Jung, Phys. Rev., B, 2002, 66, 033303. 438. A. Stroppa, S. Picozzi, A. Continenza, and A. J. Freeman, Phys. Rev., B, 2003, 68, 155203. 439. N. Bloembergen and T. J. Rowland, Phys. Rev., 1955, 97, 1679. 440. G. Bouzerar, Phys. Rev. Lett., 2004, 92, 069701. 441. V. I. Litvinov and V. K. Dugaev, Phys. Rev. Lett., 2004, 92, 069702. 442. E. Z. Meilikhov, Zhurn. Eksp. Teor. Fiziki, 2003, 124, 650 [J. Exp. Theor. Phys., 2003, 97, 587 (Engl. Transl.)]. 443. E. Z. Meilikhov and R. Z. Farzetdinova, Zhurn. Eksp. Teor. Fiziki, 2004, 125, 1367 [J. Exp. Theor. Phys., 2004, 98, 1198 (Engl. Transl.)]. Received August 30th, 2004