Electronic, magnetic and transport properties of rareearth monopnictides

Chun-gang Duan,1,2 R. F. Sabirianov,2,3 W. N. Mei,2,3 P. A. Dowben,1,2 S. S. Jaswal,1,2 and E. Y. Tsymbal1,2

Department of Physics, University of Nebraska-Lincoln, Lincoln, Nebraska 68588 Nebraska Center for Materials and Nanoscience, University of Nebraska-Lincoln, Lincoln, Nebraska 68588 Department of Physics, University of Nebraska at Omaha, Omaha, Nebraska 68182

E-mail: [email protected]

Abstract The electronic structures and magnetic properties of many rare-earth monopnictides are reviewed in this article. Possible candidate materials for spintronics devices from the rareearth monopnictide family, i.e. high polarization (nominally half-metallic) ferromagnet and antiferromagnet, are identified. We attempt to provide a unified picture of the electronic properties of these strongly correlated systems. The relative merits of several ab initio theoretical methods, useful in the study of the rare-earth monopnictides, are discussed. We present current understandings on the possible half-metallicity, semiconductor-metal transitions, and magnetic orderings in the rare-earth monopnictides. Finally, based on these studies, we propose some potential strategies to improve the magnetic and electronic properties of these candidate materials for spintronics devices.

Contents 1 Introduction ...................................................................................................................................2 2 Early experiments and theoretical studies.......................................................................................3 3 Recent theoretical and experimental progresses..............................................................................5 3.1. SIC-LSD ...............................................................................................................................5 3.2. LSDA+U...............................................................................................................................6 3.3. GW Approximation (GWA) ..................................................................................................8 3.4. Dynamical Mean-Field Theory (DMFT) ...............................................................................8 3.5. Other theoretical approaches .................................................................................................9 3.6. Experimental Studies...........................................................................................................10 4 Electronic, magnetic and transport properties...............................................................................10 4.1. Gd monopnictides ...............................................................................................................14 4.2. Eu monopnictides................................................................................................................19 4.3. Er monopnictides ................................................................................................................21 5 Half-metallicity and other interesting phenomena ........................................................................22 5.1. Origins of half-metallicity ...................................................................................................22 5.2. Metal-insulator transitions ...................................................................................................22 5.3. Magnetic ordering ...............................................................................................................24 5.4. Spin-orbit coupling..............................................................................................................26 6 The Future Possibilities................................................................................................................27

6.1. Possible half-metallic antiferromagnets ...............................................................................27 6.2. Enhancement of the transition temperatures and potential device applications .....................27 Conclusions .................................................................................................................................28 Acknowledgments .......................................................................................................................29 References ...................................................................................................................................29

Introduction

Half-metallic ferromagnets, which were first so named by de Groot et al. in 1983 [1], recently have attracted growing interest from both experimentalists and theorists [2,3]. Currently, most of the nominally half-metallic ferromagnets found are compounds that involve transition metal elements, e.g., CrO2, Fe3O4, Co2MnSi [3], and suffer from a number of deficiencies that limit their application as high polarization materials in spintronics. The electronic structures of nominally half-metallic systems are of considerable interest as a mechanism for studying the interplay between high polarization and band structure. There attraction remains in spite of the growing recognition that true half metallic character is unlikely to be ever demonstrated at finite temperatures, due to magnons [4,5], as well as zero temperature interactions [6,7]. It is natural, therefore, to explore the rare-earth compounds, as rareearth elements generally have much larger magnetic moments and demonstrate some fascinating phenomena [8,9]. Rare-earth elements are chemically very similar due to an almost identical outer electron arrangement [10]. It remains, however, difficult to obtain impurity free single crystals of the rareearths or rare-earth compounds, and this may be responsible for some of the long standing controversies concerning their electronic structure, transport properties and magnetic properties. The rare-earths do, however, have different occupation numbers on the shallow inner 4f shell, ranging from 0 to 14 through the series La to Lu. This changing 4f occupation means that the rareearth elements and their compounds have wide range of different magnetic properties and electronic structures. Due to the unfilled 4f shells of rare-earth atoms, it is a challenging problem to obtain accurate theoretical description of the electronic structure of rare-earth compounds [11]. In spite of the fact that the 4f energy levels often overlap with the non-4f broad bands of the system, they generally form very narrow resonances, and are often treated as core states in the theoretical efforts. Due to the highly localized nature of the 4f electrons, the direct f-f interactions between neighboring rare-earth atoms are generally considered to be nearly negligible. However, there is evidence that this general belief has to be modified. For instance, in the cases of cerium (uranium) compounds [12,13,14,15,16], 4f (5f) level dispersions were observed experimentally, suggesting smaller f level localization in these systems. The unoccupied f states certainly will adopt all the trappings of band structure in the every conventional sense. In addition, the f orbital moments generally cannot be quenched by the crystal field. Therefore the total magnetic moments have both orbital and spin components, and spin-orbital interactions are particularly strong for many of the rare-earth elements and compounds. These inner shell magnetic moments are largely aligned through intra-atomic s(d)-f exchange interaction and weaker inter-atomic s-s (d-d) exchange interactions, thus their magnetic transition temperatures are generally much lower than those of 3d transitional metal elements or compounds. In spite of the increasingly compelling evidence for band structure [12], the 4f bands are generally very narrow, significantly different from the bands dominated by s, p and d states. Therefore there exist strong on-site Coulomb repulsions between the highly localized f electrons [17,18]. It makes the independent particle approximation no

longer valid and calculations based on local spin density approximation (LSDA) fail to describe the rare-earth 4f electrons correctly. Thus to explain the behaviors of rare-earth 4f electrons, many-body effects must be taken into account and more accurate approximations or calculations beyond LSDA are absolutely necessary. With the availability of better quality single crystals and thin films, together with the tremendous theoretical efforts in combining many-body theory and density functional theory (DFT) in the last twenty years, a better picture of electronic structure and magnetic properties of rare-earth elements and their compounds has taken shape. It is the major objective of this review to provide a detailed account of the progress made in the field of rare-earth monopnictides, RX (X=N, P, As, Sb, Bi). This particular family is chosen because there are more than fifty members that crystallize in the simple NaCl-type structure, making the rare-earth monopnictides excellent candidates for both experimental and theoretical analysis. In what follows, we first review early experiments and theoretical studies on rare-earth monopnictides, most of which were carried out between 1960s ~ 1990s. Then we report recent theoretical and experimental progress. In particular, several novel theoretical techniques are briefly introduced for handling strongly correlated systems. Then we discuss several interesting phenomena in rare-earth monopnictides. In providing an overview of the electronic, magnetic and transport properties of rare-earth monopnictides, we have given some preference to GdN and EuN. GdN and EuN are considered to be the most promising nominally half-metallic ferromagnets in the RX family. Attracting our attention are the mechanisms for magnetic ordering, the pressure/strain and impurity effects. Finally, we discuss the potential device applications and the possibilities for enhancing the transition temperatures of rare-earth monopnictides.

Early experiments and theoretical studies

Studies of the rare-earth elements and their compounds can be traced back more than seventy years. The agreement between experiment and calculations made by Van Vleck and Frank [19] on the effective magneton number of the rare-earth ions was regarded as one of the most successful applications of quantum mechanics to magnetism [20]. The surge in the study of rare-earth compounds in the 1960 was mainly motivated by a search for new ferromagnetic semiconductors s [21,22,23,24,25,26,27]. Researchers began realizing that it was too complicated to interpret the experimental results if crystal distortions or impurities were considered. Thus, investigations on rareearth monopnictides gradually became one of the most important branches in rare-earth studies, mostly because these binary compounds crystallize into the simple NaCl-type structure and were supposed to be more amenable to theoretical analyses [21]. Nevertheless, even for rare-earth compounds, with such simple structure, the full understanding of rare-earth monopnictides electronic structure and magnetic properties is still not complete. Rare-earth monopnictides demonstrate rich magnetic orderings and their electronic structures are sensitive to external pressure and impurities. Due to the poor computational power and limited theoretical methods, earlier theoretical studies were restricted to analytical model calculations, such as crystalline field [23], effective-point-charge [28], d-f Coulomb interaction [29] and p-f mixing [30]. These calculations could explain some phenomena for certain rare-earth monopnictides, but the overall agreement with experiments for the whole RX family was not satisfactory. The earliest ab initio band structure calculation of the rare-earth monopnictides was carried out by Hasegawa and Yanase in 1977 [31]. They adopted augmented planewave method with Slater X

exchange potential [32] to calculate the energy bands of GdX (X=N, P, As, Sb), in a self-consistent way. To deal with the 4f electrons, they used frozen-core approximation, i.e. the 4f electrons are treated as core electrons. They claimed that GdN is semiconductor with an adjusted band gap of 1 eV, and other Gd monopnictides are semimetallic, in their calculations. But they also realized that if spin polarization is taken into account, then the 5d-4f exchange splitting, which is estimated also to be around 1 eV, could make GdN exhibits metallic character as well [31]. In 1988, using a tight binding model with parameters chosen to fit the band structures calculated by Hasegawa and Yanase [31], Narita studied the magnetic susceptibilities and spin structures of GdX [33]. This was the first attempt to understand the magnetic properties of rare-earth monopnictides from band structure calculations, though he otained wrong magnetic ground states for these compunds. Using similar method, Xia et al. [34] studied the electronic structure of GdAs/GaAs supperlattice in 1991. The first spin-polarized ab initio calculations on rare-earth monopnictides was provided by Petukhov et al. in 1994 [35]. Using the linear-muffin-tin-orbital (LMTO) method within LSDA and treating the Er 4f states as localized corelike states with fixed spin occupancies, they calculated the electronic structure of ErAs and ErxSc1-xAs alloys. Later in 1996, they studied extensively the electronic structures, equilibrium lattice constants, cohesive energies, bulk moduli and magnetic moments of GdX and ErX (X=N, P, As) [36]. They found that the corelike treatment is quite satisfactory for most purposes not involving 4f electrons directly. They also found that the electron exchange splittings of the nitrides are significantly stronger than those for the arsenides and phosphides, rendering GdN to be metallic for one spin channel and semiconductive for the other. Thus, they were actually the first to claim that GdN could be half-metallic, within their calculation scheme which did not involve the 4f states explicitly. There were other theoretical studies on Gd compounds, e.g., Kasuya and Li tried f-d mixing and f-d exchange interactions to explain the strong ferromagnetism in GdN In 1997 [37]. The early experimental reports on rare-earth monopnictides are widely scattered in the literature. For reviews on experimental studies before 1970s, one can refer to Refs. [25], [38] and [39], but focusing on the experimental studies on the electronic, transport and magnetic properties, in the period from 1970s to 1990s, provides an insight into the parallel growth of materials science and electron spectroscopy. It should be pointed out that Gd monopnictides have been investigated most extensively among the rare-earth monopnictide family. This is because Gd is located in the middle of the rareearth group in the periodical table, and the Gd 4f orbitals are exactly half occupied, hence the orbital angular momentum is zero, as the ground state of Gd3+ is 8S7/2. Thus the spin-orbit, multipole, p-f and d-f mixing interactions are small [33], greatly simplifying the problems associated with electronic structure. In addition, the Gd monopnictides (and Gd metal) generally have much higher transition temperatures than those of their counterparts of other rare-earths, thus attracting much more practical interest [38]. Kaldis reported the first successful growth of large single crystals of GdP in 1974 [40]. This opened the door to a reliable determination of the electronic structure of GdP. Based on optical investigation [41], G ntherodt, Kaldis and Wachter deduced that the crystal field splitting of the Gd 5d states is about 1.8 eV and the positions of occupied Gd 4f levels in GdP are about 7 eV below EF. They also found that GdP exhibits metal-like conductivity. Later experiments on GdN were controversial: GdN was reported as semiconductor [42] and semimetal [43]. Studies on the magnetic exchange interaction on Gd compounds clearly demonstrated that the free carriers contribute to a RudermanKittel-Kasuya-Yosida (RKKY) [44,45,46] indirect exchange interaction [47,48]. Urban et al. [ 49] observed a large variation of the thermal broadening of the electron spin

resonance (ESR) linewidth of Gd across the pnictides in 1978. They then claimed that the conduction electrons exhibit almost d character, thus the exchange interaction depends on the amount of overlapping between d wave functions. Palmstr et al. [50] successfully grew single crystal ErAs m films on GaAs in 1988. Later magnetotransport measurements showed that ErAs is a low-density semimetal that magnetically orders at low temperature (4.5 ~ 4.8 K) [51]. In 1990, Degiorgi, Bacsa and Wachter reported on results obtained from large single crystals of cubic and stoichiometric YbN [52]. They claimed that YbN is a self-compensated semimetal with the occupied f states about 6 eV below EF and the empty f states about 0.2 eV above EF. Their later studies confirmed the semimetallic nature of YbP and YbAs [53]. Chattopadhyay et al. carried out highpressure magnetization and neutron-diffraction experiments on CeSb in 1994 [54]. They found that the magnetic ordering of CeSb is very sensitive to hydrostatic pressure. Waldfried et al. used angleresolved photoemission to study the electronic structure of dissociatively chemisorbed nitrogen on Gd(0001) in 1995 [55]. Xiao and Chien found that GdN films are insulating [56]. Li et al. reported the growth of large single crystal of GdX (X=P, As, Sb, Bi) in 1996 and found them all to be wellcompensated semimetals and order antiferromagnetically [ 57 , 58 ]. The photoemission studies of Yamada et al. showed that the occupied 4f states in GdP lie around 8.4 eV below EF, and are shifted down from GdP to GdBi [59]. The experiments on polycrystal GdN confirmed that stoichiometric GdN is ferromagnetic with a transition temperature 58 K [60].

Recent theoretical and experimental progresses

Recent theoretical progresses on the rare-earth monopnictides studies follow the development of electronic structures calculation in solids. For strongly correlated systems, the accurate evaluations require going beyond the LSDA scheme. The following described methods have been developed to overcome the deficiency of LSDA calculations. 3.1. SIC-LSD Despite the impressive successes of the local spin density approximation (LSDA) in the past decade or so [61], LSDA calculations have an intrinsic deficiency, i.e. the self-interaction energy problem [62]. The self-Coulomb and self-exchange interactions in LSDA do not cancel completely, as in the case of Hartree-Fock approximation. This means that LSDA will fail to correctly describe systems with strong electron-electron interactions. Numerous efforts have been made to address this problem [62,63,64,65]. In the self-interaction-corrected local-spin density (SIC-LSD) approximation [65], the non-physical electron self-interactions, including both SIC Coulomb and corresponding SIC exchange-correlation terms, are subtracted from the LSDA Hamiltonian, and the energy functional is written as
occ n (r )ni (r') 3 3 ESIC = ELSDA i d rd r ' + ni (r ) xc ( ni (r ), 0)d 3r , r r' i

(1)

where ELSDA is the energy functional in LSDA, ni(r) is the charge density corresponding to the ith solution of the SIC-LSD equation, and xc (n , n ) is exchange-correlation energy density of a homogeneous systems with spin densities n and n . The SIC approach generates a orbital-dependent potential which can be significant for localized states, yielding a much-improved description of the

static Coulomb correlation effect compared to that provided by LSDA. The inclusion of SIC are often quite successful [66,67,68,69]. Another advantage of the SIC-LSD method is that the minimization of total energy, with respect to the number of localized electrons, leads to a determination of the nominal valence defined as the integer number of electrons available for band formation N v = Z N core N SIC , (2)

where Z is the atomic number, Ncore and NSIC are the number of core and localized (SIC) electron states, respectively. This information is important to the analyses of chemical properties of rare-earth compounds [67]. Application of the SIC-LSD methodology to the rare-earth monopnictides has been carried out by Aerts et al. [70], Horne et al. [71], Szotek et al. [72] in 2004. Together, these studies represent a systematic study of the electronic structure of thirteen rare-earth nitrides, from CeN to YbN. The calculations show that the rare-earths are trivalent in the mono-nitride ground state, with the exception of Ce which is tetravalent in CeN. On the basis of the SIC-LSD calculations, the claim is that these rare-earth nitrides display wide range of electronic properties, despite the same structure and similar lattice constants. Specifically, TbN, DyN and HoN are found to be narrow gap insulators, CeN, ErN, TmN and YbN are metallic in both spin channels, while PrN, NdN, PmN, SmN, EuN and GdN are half-metallic ferromagnets in these ground state calculations. They found that the f-band manifold was split by the SIC into localized and bandlike f electrons. Because of the different degree of hybridization between the rare-earth bandlike f and the nitrogen p states, in the vicinity of the Fermi level, these compounds demonstrate different electronic properties. This is consistent with the electronic structures of SmX (X=N, P, As, Sb, Bi) which have been studied by Svane et al. [73]. Svane and coworkers found that the occupied f bands are formed in the vicinity of the Fermi level in all the SmX compounds. 3.2. LSDA+U In the LSDA, the potential is treated as an averaged orbital-independent one-electron potential. This may be a reasonable approximation for weakly correlated system, which corresponds to an extreme case of a Hubbard model [74] where the on-site Coulomb repulsion (Hubbard U) is approaching zero. For strongly correlated systems like Mott insulators, however, ignoring the effective Coulomb parameter U results in totally wrong prediction of the energy gap. The LSDA+U [75,76,77,78] method was initially proposed to describe Mott insulators correctly. Following the Anderson model [79], electrons are separated into two subsystems: localized d or f electrons and delocalized s or p electrons. 1 The d-d (f-f) interaction is described by Hubbard term U ni n j , where ni is the d(f)-orbital 2 i j occupancy, instead of the LSDA averaged term (approximately) E = UN ( N 1) / 2 . Then an new energy functional can be written as

1 1 E LSDA+U = E LSDA UN ( N 1) + U ni n j , 2 2 i j
and the orbital energies

(3)

are given by

E LSDA+U 1 LSDA (4) = i + U ( ni ) . i ni 2 This new formula splits the orbital energies of occupied localized electrons (ni = 1) and unoccupied localized electrons (ni = 0) by U, thus reproducing the qualitatively correct physics for Mott-Hubbard insulators [77]. The above analysis is, however, only the simplified overview. The realization of LSDA+U scheme in practice requires much more theoretical work [75] which is not addressed here. In passing, it is worth noting that a rotationally invariant form of the LSDA+U functional is given by
LSDA+U

E LSDA+U = E LSDA +

UJ 2

[Tr

Tr( )] ,

(5)

in which is the spin, U and J are the (spherically averaged) screened Coulomb electron-electron interaction (Hubbard parameter U) and Hund rule exchange parameter (Stoner parameter I), s respectively [80]. In practice, parameters U and J can be taken as adjustable parameters (e.g., obtained through comparison with photoemission and inverse photoemission experiments). U and J can also be evaluated from the LSDA through the supercell LSDA approach [81]. Specifically, U = F0, J = (F2+F4)/14 for d electrons, J = (286F2+195F4+250F6)/6435 for f electrons [78], where Fk (k = 0, 2, 4, 6) are Slater integrals [82]. The first LSDA+U calculation on the rare-earth monopnictides was carried out by Liechtenstein et al. [83] in 1994. They studied the electronic structure and magneto-optical effects in CeSb. The Ce 4f level is found to be 2 eV below the Fermi level and is hybridized with Sb p bands. Using LMTO method, in 2003 Komesu et al. [84] studied the electronic structure of ErAs and obtained band dispersions qualitatively similar to experiments. In 2004, Duan et al. [85] considered the spin-orbit coupling in the LSDA+U calculation of bulk ErAs, using the full-potential linear-augmented-planewave (FLAPW) method [86]. The theoretical 4f multiplet structure agrees well with photoemission experiments. In addition, clear evidence of 4f-5d hybridization was found [86]. Later, with the same theoretical methodology, the electronic structure and magnetic ordering of ErN and ErAs was investigated [87]. In 2005, Duan et al. reported on the effect of strain on the electronic and magnetic properties of GdN [88]. The results of LSDA+U indicate that GdN is nominally a ground state half-metallic at the experimental lattice constant, but may undergo phase transition to a semiconductor with lattice expansion. Soon thereafter, these studies were extended to the electronic structure and magnetic ordering of the GdX monopnictides [89]. Duan et al. provided a quantitative analysis of the RKKY and superexchange interaction [90] of the GdX monopnictide systems. As previously indicated, a number of rare earth pnitcides appear to resemble ground state half metallic ferromagnets. Johannes and Pickett studied the electronic structure and magnetic exchange interactions in EuN and EuP also using FLAPW method [91]. They found EuN is a ground state halfmetallic ferromagnet within conventional LSDA+U band theory. Ghosh et al. studied the electronic, magnetic and optical properties of GdX in 2005 [92], with LMTO in the LSDA+U scheme. They also found GdN is half-metallic if no further adjustment is implemented. Using LMTO within the LSDA+U approach, Larson and Lambrecht [93] found that by putting the Hubbard U also on the d orbital, GdN is found to be a semiconductor, otherwise it is half-metallic, as noted by others [70,88].

3.3. GW Approximation (GWA) Eigenvalues of the Kohn-Sham equations [94] are often interpreted as single-particle energies and are compared with photoemission spectra. This is not really justified and in many cases leads to incorrect predictions. A proper way to interpret the photoemission spectra is to use quasiparticle concepts [95]. The quasiparticle energies Ei can be obtained [96] from: 1 [ 2 (r ) + V (r )] i (r ) + d 3r (r, r; Ei ) i (r) = Ei i (r ) , (6) 2 where the self energy addresses the effect of exchange and correlation and is intrinsically a nonlocal operator. The calculation of is, however, generally very difficult and requires some approximations. In the so called GW approximation (GWA) [97], is obtained by the Green function method: s i (r, r, E ) = d G (r, r; E )W (r, r; ) , (7) 2 here G stands for the Green function, W is the screened Coulomb interaction and can be evaluated by s the random phase approximation (RPA) [98], i.e. assuming the electrons are non-interacting when they respond to the external and induced field. The application of GWA to realistic materials was first carried out by Hybertsen and Louie in the 1980s [ 99 ]. Godby et al. [100 ] also obtained similar results using same approach. Due to the complexity in calculating the self energy, the application of GWA to rare-earth compounds has not been undertaken (for a recent review on GWA calculations, see Ref. [101]) until recently. Van Schilfgaarde et al. [102] have drawn attention to their quasiparticle self-consistent GW (QSGW) calculations on Gd, GdP and GdAs. The QSGW calculations overestimate the position of the minority Gd f shell by ~ 4 eV. The details of these calculations are not yet available. In 2000, using the LMTO method and assuming the quasi-particle energy gap corrections scale inversely with the dielectric constant, Lambrecht [103] predicted that GdN is a narrow indirect band gap insulator. He also claimed that by applying a magnetic field, the band gaps can be tuned by aligning the Gd 4f magnetic moments. 3.4. Dynamical Mean-Field Theory (DMFT) Dynamical mean-field theory (DMFT) is a more modern, non-perturbative method that has proven to be successful in investigating strongly correlated systems with local Coulomb interactions [104]. This theory was initially developed from the local impurity self-consistent approximation [105], which is the natural generalization of quantum many-body problems of the Weiss mean-field theory [106]. It is believed to be a major step towards the reunion of two theoretical approaches, i.e. the DFT and manybody model Hamiltonian of condensed matter physics. The spirit of DMFT is to map a lattice problem with many degrees of freedom onto an effective single-site (impurity) problem with less degrees of freedom. DMFT becomes exact in the limit of high lattice coordination numbers [105]. The underlying physical idea is that the dynamics, at a given site, may be considered to be the interaction of the degrees of freedom at this site with an external bath created by all other degrees of freedom on other sites, a dynamical mean-field approximation [106]. This impurity problem has to be solved self-consistently together with the k-integrated Dyson equation connecting the frequency dependent self energy ( ) and the on-site Green function G at frequency :

G ( ) =

1 1 3 0 d k + H LDA (k ) ( ) , VB

(8)

0 where is the chemical potential, H LDA is the one-particle Hamiltonian without the local Coulomb interaction, VB is the volume of the Brillouin zone. Many techniques have been applied to solve the Anderson impurity model, such as quantum Monte Carlo, iterative perturbations, the fluctuation exchange approximation, the mean-field slave boson approach and the numerical renormalization group [104]. With these efforts, it is now possible to combine DMFT with modern electronic structure calculations to carry out LDA+DMFT [107] or spectral DFT [108] calculation. The DMFT method has been successfully used to explain the transition in cerium, the phase of plutonium and several Mott insulators [104]. To our knowledge, however, there is no report of using DMFT to study the rare-earth monopnictides; but as there is no substantive technical barrier such calculations are only matter of time.

3.5. Other theoretical approaches There are also several calculations of rare-earth monopnictides using other approaches. Kalvoda et al. studied the cohesive properties of GdN clusters using a quantum chemical (Hartree-Fock) method in 1998 [109 ]. They obtained a reasonable lattice constant for GdN (1.3 ~ 2.0 % larger than the experiment values), but the details concerning the electronic structure were not presented. Combining a many-body model (multi-band Kondo lattice model) and initioband structure ab calculations (LMTO), Santos, Nolting and Eyert studied ferromagnetism and temperature-dependent electronic structure of hcp Gd in 2004 [110]. Later, using the same approach, Sharma and Nolting studied temperature dependant electronic correlation effects in GdN in 2006 [111]. Assuming GdN to be a semiconductor, they obtained its quasi-particle spectral densities and density of states and found that the correlation effects were strongly temperature dependant. Using s-f model, Bhattacharjee and Jaya, in 2006 [112], studied correlation and temperature effects on the electronic structure of bulk and thin film GdN. Bhattacharjee and Jaya also found a red shift of the GdN conduction bands with respect to the temperature. Without any special treatment of the 4f states, Landrum calculated the electronic structure of CeN in 1999 using LMTO method [113]. In summarizing the various theoretical approaches, discussed above, all have their advantages and drawbacks. For instance, because of the localized nature of the strongly correlated electrons, most of the theoretical methods, except for GWA, were first developed using the LMTO method, which is based on local orbital basis, as a starting point. For 4f systems, SIC+LSD method sometimes overestimates the separation between the occupied and unoccupied f states [70]. The LSDA+U method now has been implemented into several band structure schemes, due to its relatively simple physics, and thus is much more widely accepted. The HF approximation treatment in the LSDA+U scheme to the on-site Coulomb interactions is, however, too crude for strongly correlated systems [107]. This approach is successful in describing the long-range ordered insulating states of correlated systems, while it fails to describe correctly the strongly correlated paramagnetic states. GWA and DMFT methods are believed to be more accurate, than many other strategies, in describing the strongly correlated systems. But at this stage, the GWA and DMFT approaches are still too cumbersome to be used by the whole community. Nevertheless, concepts like the Hubbard U and the self energy are now being widely adopted [73,107]. In addition, the Wannier functions make it possible to set up localized orbital through plane

wave basis [114,115]. Thus we expect that there will be many more ab initio studies on strongly correlated systems over the next few years. 3.6. Experimental Studies A series of ESR measurements have been carried out on single crystals of GdAs [116], GdP [117] and GdBi [118] since 2000. The magnetic structure of these compounds were confirmed to be AFM-II type. The Fermi surface and magnetic properties of TbSb have been investigated by Nakanishi et al. in 2004 [119] using de Hass-van Alphen and high-field magnetization measurements. In 2005 Leuenberger et al. [120] reported electronic and magnetic properties of high-quality thin films of GdN. They found that a 500 thick GdN film already shows very close physical and magnetic properties to that of bulk GdN. The element specific x-ray magnetic circular dichroism (XMCD) measurements clearly demonstrated that the N p states are magnetically polarized. This agrees with theoretical predictions, as will be shown later. Leuenberger and coworkers also found that the electrical conductivity of GdN is thermally activated down to the ferromagnetic Tc, below which GdN exhibits metallic character. They attribute the origin of this semiconductor-metal transition to a nonstoichiometric GdN film, similar to that found for EuO [121,122]. Later experiments demonstrated a significant reduction of Tc as a result of the influence of lattice expansion [123], as predicted by theory [88]. More recently, Granville et al. [124] claimed that GdN is semiconducting in both paramagnetic and ferromagnetic states (Tc =68 K). They think that the N vacancies may be responsible for the conductivity in GdN. As a final note, although tangential to GdN, Leuenberger and coworkers [125] also found that Fe layers induce long range magnetic order in GdN layer at temperatures above the Tc of GdN in GdN/Fe multilayers.

Electronic, magnetic and transport properties

There are many similarities among the rare-earth monopnictide family, and some stunning differences with regard to metallicity and magnetic ordering. This is summarized in Table 1 which lists the experimental lattice constants of all the rare-earth monopnictides, together with their magnetic ordering, transition temperatures and metallicity, i.e., insulating (semiconducting) or (semi)metallic, for most of the more than sixty compounds in the rare-earth monopnictide family. Most of the rareearth monopnictide family have the NaCl-type structure (space group Fm 3m ). EuAs, EuSb and EuBi do not adopt the NaCl-type structure, and have been omitted in Table 1. There is no experimental report on the Pm monopnictides, most probably due to the fact that Pm must be artificially prepared and we also unable to find any report on YbBi. From Table 1, we can see that the lattice constants of rare-earth monopnictides, as a general trend, increase from N to Bi but decrease from La to Lu: the lattice constant typically decreases with increasing 4f occupancy, as indicated in Figure 1. This trend in the lattice constant can be simply explained by the increase of anion sizes and the decrease of cation sizes with the increase of atomic number. Thus the smallest lattice constant belongs to LuN (4.766 ), while the largest to LaBi (6.580 ): a difference of about 38%. We note that there are exceptions to this general trend, e.g., CeN, SmP. As shown in the experiments of Olcese [126] and theoretical calculations by Aerts et al. [70], the jump in lattice constants from CeN to PrN is due to the fact that the ground state for Ce in CeN is tetravalent

10

instead of trivalent. The bonding between the rare-earth atoms and pnictogen atoms cannot be simply described as ionic or covalent, otherwise all rare-earth monopnictide family would be expected to be insulators (semiconductors). However, from N to Bi, we can see the gradually increasing metallicity in these compounds. The typical carrier concentrations in these compounds are around 1019 ~ 1021 cm-3 [103], making them semimetals or highly doped n-type semiconductors. In listing the known energy gaps for those rare-earth monopnictide compounds that are regarded as semiconductors, we note that the light pnictides, like nitride, tend to be larger energy gap semiconductors than the heavy pnictogen compounds. The changes of energy gaps from La to Lu are rather scattered and it is hard to find a clear trend. We should point out that those data for energy gaps are not that reliable, since the electronic and transport properties are strongly affected by impurity dopants, thus a lot of controversies exist in the comparison of various experiments. For example, GdP, GdAs and GdSb were thought to be semiconductors [39], but later experiments on high-quality samples clearly showed that they are actually semimetals [47,57].

Figure 1. Lattice constants of rare-earth monopnictides: RN (squares), RP (cirlces), RAs (stars), RSb (down triangles), RBi (up triangles).

The magnetic ordering of rare-earth monopnictides is also quite complicated, ranging from ferromagnetic (FM), flip-flop FM [22], to different antiferromagnetic (AFM) configurations. Three typical AFM configurations (also see Ref. [127]) are depicted in Figure 2. As we can see from Table 1, most monopnictides have AFM ground states with exceptions among the nitrides and phosphides. Overall, we can see that the magnetic interactions gradually increase from La to Gd (FM, 58 K), then decrease again from Gd to Lu. With the increase of anion sizes, the FM pnictides become AFM and the N temperature increases. Apparently, HoP and DyP are in the intermediate region between FM el and AFM, as flip-flop FM can be regarded as combination of a FM and an AFM configuration [22]. This balance between AFM and FM ordering may also be the reason that CeN and TmN have no apparent net magnetic order.

11

Figure 2. Three antiferromagnetic ordering configurations in rare-earth monopnictides: (a) AFM-I, (b) AFM-II, (c) AFMIII. Arrows indicate magnetic moment orientations on rare-earth atoms.

The electronic structure calculations on three typical rare-earth pnictides, based on LSDA+U scheme with FLAPW method [86], confirm some of these trends and also can illustrate some of the affect of changing lattice constant, without changing the pnictide, which is not easily done in experiment. For the convenience, the Brillouin zone of the face-centered cubic (fcc) lattice is shown in Figure 3, together with definitions of some high-symmetry k-points and k-lines.

Figure 3. The Brillouin zone of the face-centered cubic lattice.

12

Table 1. Physical and magnetic properties of rare-earth monopnictides. Unless explicitly indicated, the data are collected from Ref. [39]. Numbers in the Metallicity are band gaps in eV derived from optical absorption of thin films if the material is insulator or semiconductor. SC means semiconductor, SM means semimetal.

N La Ce
a () Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Ordering TN(C) (K) Metallicity a () Metallicity 5.295 a SC, 0.82 5.013 a No order down to 1.5 K SM f 5.135 g

P
6.035 b 5.90 AFM-I 6~7, 8.5, 9 SC, 1.1 5.893

As
6.137 c 6.06 AFM-I 5~7.5, 7.5 SC, 0.7 6.018

Sb
6.490 d SC, 0.7-0.8 6.40 AFM(complicated) 16, 18 SM f 6.364

Bi
6.577 e 6.49 AFM(complicated) 25, 25.5, 26 6.448

Pr

Nd

Sm

Eu

SC, 1.03 5.123 a FM 27.6, 29, 32, 35 SC, 0.8 5.035 AFM <2, 13, 18 SC, 0.7 5.017 h SC, 0.76 4.974 j FM j 58.0 j SC, 0.98, 0.85 4.922 FM, AFM k TC: 34, 39, 42 TN: 31 k SC, 0.8 4.895 FM, AFM TC: 17, 17.6, 26 TN: 15 SC, 0.91,0.95 4.865 FM 11, 13.3, 18 SC, 0.73, 1.05 4.835 a FM 3.4, 4.5, 5, 6 SC, 1.2, 1.3 4.80 No order down to 1.3 K SC, 1.1 4.775 AFM 0.75 m SM n o 4.766 SC, 1.55, 1.6

5.826 AFM-I 7, 11 5.75 AFM 1.65 SC, 1.09 5.756 i

SC, 1.0 5.958 AFM-I 10.6, 12.5, 13 SC, 1.04 5.91 AFM 1.8 SC, 1.03 --

SC, 0.66 6.31 AFM-I 15.5, 16 6.26 AFM 2.1, 2.8 SC, 0.59 --

6.41 AFM-I 24, 25 6.35 AFM 8.6 --

Gd

Tb

5.709 j AFM-II j 15.9 j SM j 5.675 AFM-II 8, 9 5.643 Flop-side FM 4.5, 8 SC, 25 5.615 Flop-side FM 5.5 SC, 1.1-1.2 5.595 AFM-II 2.3, 3.1 SM l 5.56

Dy

Ho

Er

5.864 j AFM-II j 18.7 j SM j 5.815 AFM-II 10.5, 12 SC, 0.7 5.79 Flop-side FM 8.5 SC, 0.9,1.0 5.76 AFM-II 4.8 SC, 0.8 5.734 AFM-II 2.9, 3.5 5.71 SC, 1.18 5.69 AFM 0.55 m SM n p 5.68

6.219 j AFM-II j 23.4 j SM j 6.17 AFM-II 14, 15, 16.5 SC, 0.35 6.14 AFM-II 9.5, 12 6.12 AFM-II 5, 9 SC, 0.45 6.095 AFM-II 3.55, 3.7 6.08

6.295 j AFM-II j 25.8 j SM j 6.27 AFM-II 17, 18 6.24 AFM-II 12, 13 6.215 AFM 6.7 6.19 AFM 3.5, 3.9 6.18

Tm

Yb

Lu
a

5.543 AFM 0.66 m SM n p 5.533

6.067

--

6.056

6.156

[126]. b[128]. c[129]. d[130]. e[131]. f[132]. g[133]. h[134]. i[135]. j[60]. k[136]. l[51]. m[137]. n[53]. o[138]. p[139].

13

4.1. Gd monopnictides Though Gd monopnictides were studied most extensively in the RX family, there remains intense controversy regarding the electronic structure and transport properties, especially on GdN [31,36,43,58,70,88,93,103,120,124]. Various calculations have been undertaken on different magnetic phases (FM and AFM) with different schemes (LSDA and LSDA+U) of GdX (X=N, P, As, Sb, Bi). The electronic structures, typical for rare-earth monopnictides, are presented in the discussion that follows. Figure 4 shows band structure of GdN in FM state with and without on site 4f Coulomb 4 interactions. Here we used an effective Hubbard U efff = U J = 9.2 eV for Gd compounds in the LSDA+U scheme. As we will show later, this value gives much better agreement of the 4f energy levels found in experiment than those of the values used in previous studies on Gd metal or Gd compounds [140,141,142,88,89], i.e. U = 6.7 and J = 0.7 eV. As can be seen from Figure 4(a), the LSDA+U strategy has no significant impact on the whole LSDA band structure of GdN, except upon the energy levels of the occupied and unoccupied 4f states. In the pure LSDA scheme, the occupied 4f states are placed about 3.2 eV below EF, strongly hybridized with N 2p states, while the unoccupied 4f states locate about 2 eV above EF, mixed with Gd 5d states. There is no direct photoemission data on GdN single crystal, but experiments on GdP (As,Sb,Bi) show that the occupied and unoccupied 4f levels are situated 8 ~ 10 and 5 ~ 6 eV below and above EF, respectively [59]. Furthermore, the x-ray photoelectron spectroscopy (XPS) experiments on GdN films show that the Gd 4f levels are 7.8 eV below EF [120]. Thus the LSDA calculation fails to give correct the 4f energy level binding energies. This situation is much improved by the LSDA+U method [Figure 4(b)]. In fact, now the unoccupied 4f states remain around 6.6 eV above EF, and the occupied 4f states are 7.8 eV below EF, in very good agreement with experiments.

Figure 4. Electronic structure of ferromagnetic GdN on experimental lattice constants (a = 4.974 ) : (a) LSDA, (b) LSDA+U with Ueff = 9.2 eV. Solid and dotted lines represent spin majority and spin minority states, respectively.

14

Figure 4(b) actually shows the typical features of the band structure of most rare-earth monopnictides. The pnictogen p derived states (here N 2p) dominate the top of the valence bands. Precisely speaking, these pnictogen p states are actually hybridized with rare-earth (here Gd) 5d (6s) states, which dominate the bottom of conduction band. This hybridization results in a hole pocket at point and an electron pocket at X point in metallic pnictides. The numbers of electrons and holes are the same [31]. Even if the pnictides are determined to be semiconductors, the close proximity (and likely Fermi level crossing) of the unoccupied rare-earth 5d states near the X point would make the energy gap to be indirect. A significant exchange splitting can be found in the band structure, and this is indeed the origin of the nominal half-metallicity in rare-earth monopnictides, as we will discuss in detail later.

Figure 5. The l-projected DOS of GdN. Spin-minority states are of negative values.

The l-projected density of states (DOS) plot (Figure 5) provides an even clearer picture of the electronic structure of GdN. As one can see, those top valence states right below the Fermi level are predominantly N 2p states. The Gd d states, however, also make noticeable contribution. Note here, in our calculation both 5d eg and 5d t2g states participate in the hybridization with N 2p states, though the contributions to the bottom of the conduction band(s) are dominated by 5d t2g states. Not only does the calculated 4f levels agree well with experiment, but the Gd 5p states are located at the same binding

15

energy position (about 21 eV below EF) found in experiment [120]. The 1.8 eV separation between the Gd 5p majority and minority states is caused by the exchange splitting. We would like to point out that the Gd 5d-X 2p hybridization is important in establishing the physical properties of GdX. This hybridization results in magnetic moments on pnicogen sites and is responsible for the intriguing magnetic orderings and transport properties. Band structures for other Gd monopnictides in the ferromagnetic state, using LSDA+U, are shown in Figure 6. Note that the magnetic ground states of these compounds are AFM-II type. However, we can consider these structures as the saturated limit of the paramagnetic states in a magnetic field [36].

Figure 6. Electronic structure (LSDA+U with Ueff = 9.2 eV) of (a) GdP (aexp = 5.709 ), (b) GdAs (aexp = 5.864 ), (c) GdSb (aexp = 6.219 ), (d) GdBi (aexp = 6.295 ) in ferromagnetic phase.

16

The band structures seen for the rare earth pnictides from GdN to GdBi are quite similar, except that for GdSb and GdBi there are no gaps between the Gd 5d states and pnicogen p states at the 4 point. We find that, with a single universal U efff , the f levels in these compounds, both occupied and unoccupied, are all in excellent agreement with experimental observations [59,120], vindicating the application of the LSDA+U scheme. This is clearly shown in the XPS and X-ray bremsstrahlung isochromat spectroscopy (X-BIS) spectra of GdX (Figure 7). There exists a monotonic energy down shift from GdN to GdBi [59], which is clearly represented in LSDA+U band structures of GdX.

Figure 7. The XPS and X-BIS spectra of GdX (X=P, As, Sb, Bi) (from [59]). Red lines are added for GdN according to Ref. [120], assuming same Uff for GdN.

From the LSDA+U calculations, all the Gd monopnictides are seen to be semi metallic except for GdN which is nominally half-metallic with a half-metallic gap about 0.5 ~ 0.6 eV. Experimentally, the GdP (As, Sb, Bi) pnictides are found to be metallic [58]. For GdN, as noted previously, the experimental temperature dependent resistivity of GdN film [124] shows a picture (Figure 8) that differs from other GdX compounds (Figure 9, Ref. [58]). The high temperature behavior of the resistivity of GdN is more like that of a semiconductor, apparently different from other Gd monopnictides. The peak around the transition temperature can be explained by the long-range magnetic ordering which significantly decreases the spin-dependent scattering of charge carriers. The broad range of this peak may be attributed to the imperfect of periodicity or the Kondo-lattice model [143,144]. Though the resistivity in the very low temperature is not huge (5 cm), it is still more like a semiconductor. If GdN is really a semiconductor, with an optical gap about 1.5 ~ 2 eV [124], then it means that

17

there still is something important missing in the initiocalculations. Lambrecht [103], Ghosh et al. ab [92], Larson and Lambrecht [93] used different approaches to shift Gd 5d bands and managed to make GdN semiconducting. However, these empirical treatments of the theory do not give the correct positions of Gd 4f levels. Apparently, more efforts is needed to fill the gap between theory and experiment.

Figure 8. Temperature-dependent resistivity of a 200 nm thick GdN film. The pronounced peak at 68 K corresponds to the measured Curie temperature of the film. The inset shows the low temperature behavior of the resistivity (from [124]).

Figure 9. Temperature-dependent resistivity of stoichiometric GdX (X=P, As, Sb, Bi) single crystals measured at zero field (from [58]).

Since the ground states of GdX (X=P, As, Sb, Bi) are all of AFM-II type [see Figure 2(b)], it is interesting to compare the difference between the AFM and FM electronic structures. Figure 10 shows both AFM and FM bands of GdP with the same rhombohedral cell (where the Brillouin zone is shown in Figure 11) and adopting the experimental lattice constants. As we can see, the AFM bands show an average effect of the FM spin majority and minority bands and this leads to a small reduction in total energy. The essential feature of the electronic structure, however, remains the same. From the above analysis, apparently GdN is on the border between semiconductor and semimetal states. Thus the GdN physical properties are very sensitive to nonstoichiometry, impurities, pressure, temperature, and magnetic field, etc. In addition, exotic phenomena, such as electron-hole liquid [145],
18

may occur in the region between semiconductor and semimetal states. Again, if it is true that GdN is a semiconductor and GdP is a semimetal, then most probably there would exist some GdN-GdP alloys that would be half-metallic.

Figure 10. Electronic structure of GdP with rhombohedral cell (a) antiferromagnetic phase (AFM-II), (b) ferromagnetic phase. Note that in antiferromagnetic phase there is no difference between the spin majority and minority states.

Figure 11. The Brillouin zone of the rhombohedral lattice.

4.2. Eu monopnictides In the europium pnictide family only EuN and EuP crystallize with fcc structure. According to Hund s 3+ 6 7 3+ rules [146], the ground state of Eu (4f ) is F0, i.e. S = 3, L = 3, J = 0. Thus isolated Eu is supposed

19

to be nonmagnetic. In the crystalline environment, however, the orbital moment generally will be suppressed by the crystal field, thus net magnetic moments may exist in EuX (X=N, P). In addition, as in the case of Gd monopnictides, there could be induced magnetic moments on pnicogen atoms. Taking into account the spin-orbit coupling, bulk Eu3+ could exhibit significantly different magnetic properties from that of isolated Eu3+. Experimental studies on the electronic structure of the EuX (X=N, P) compounds are very rare. Theoretical calculations are also limited. So far there has only two published calculations on Europium monopnictides, to our knowledge. As previously noted, Horne et al. reported EuN to be half metallic based on the SIC+LSD method [71] and was confirmed by Johannes and Pickett [91] using the LSDA+U scheme. Following Johannes and Pickett [91], using the LSDA+U approach with U and J set as 8 and 1 eV respectively, we have obtained similar band structures for EuN and EuP, as shown in Figure 12. The spin-orbital effects are not considered in these calculations and will be discussed in next chapter. Johannes and Pickett calculations showed that the ground state of EuN and EuP are FM and AFM-II, s respectively. Here we only present results of FM states of two compounds for direct comparison. The difference between the FM and AFM states of EuP is very similar to the case of GdP (Figure 10).

Figure 12. Electronic structure (LSDA+U with Ueff = 7 eV) of (a) EuN (aexp = 5.017 ) and (b) EuP (aexp = 5.756 ) in ferromagnetic phases.

As we can see from the above Figure, the band structures of Eu monopnictides are similar to those of Gd monopnictides (Figure 4 and Figure 6). The major differences arise from the unoccupied 4f majority states. In EuN, this state is about 1 eV above EF, whereas in EuP this state crosses the Fermi

20

level. In both cases, the unoccupied 4f states are strongly hybridized with Eu 5d and pnicogen p states. Figure 12(a) shows that, in this conventional LSDA+U band calculation, EuN is half-metallic with a 0.5 eV gap in minority spin channel. Other Eu monopnictides are semimetallic, like EuP, as illustrated in Figure 12(b). 4.3. Er monopnictides Erbium monopnictides have been chosen for investigation because Er3+ represents another type of rare-earth ion, which has more than a half-filled 4f electron shell. This provides a contrast with Gd (with a half-filled 4f electron shell) and Eu (with a less than half-filled 4f electron shell). The ground state of isolated Er3+ (4f11) is 4I15/2, i.e. S = 3/2, L = 6, J = 15/2. Only ErAs has been studied extensively experimentally [50,51,84], whereas theoretical studies of all of the Er monopnictides have been undertaken [35,36,70,84,85,87]. Except for ErN, all Er monopnictides have AFM ground states and have similar electronic structures and physical properties. Here we present our LSDA+U band calculations for ErN and ErP in the ferromagnetic state. Following Komesu et al. [84], U and J are chosen to be 8.6 and 0.75 eV, respectively.

Figure 13. Electronic structure (LSDA+U with Ueff = 7.85 eV) of (a) ErN (aexp = 4.835 ), (b) ErP (aexp = 5.595 ) in ferromagnetic phases.

In Er monopnictides, all the 4f majority states and four of the seven 4f minority are occupied. Because of the Coulomb repulsion, the unoccupied 4f states reside 3.2 ~ 3.6 eV above EF in ErN. A zero half-metallic gap is found for ErN in the LSDA+U band calculations, whereas EuP is semimetallic in both spin channels with the 4f bands shifted down about 1 eV.

21

Half-metallicity and other interesting phenomena

5.1. Origins of half-metallicity In band calculations, half-metallicity will not occur if there is no exchange splitting. While an exchange splitting is a necessary condition for half-metallicity, exchange splitting alone is not sufficient, even in the rare earth monopnictides. The exchange splitting must be large enough to form a gap between the unoccupied and occupied states in one spin channel and at the same time the other spin channel should be partially occupied. Thus the formation of half-metallicity strongly depends on the magnitude of exchange splitting as well as the details of the electronic structure. For perfect ionic or covalent crystal, the exchange splitting is zero due to the fully occupied outermost electron shell, which also makes the system insulating. For partially occupied electron shell, the exchange splitting is generally around 0.1 ~ 1.5 eV in most elements and compounds, whereas the width of unoccupied bands is generally above 3 ~ 4 eV, therefore half-metallic materials are rather rare. In some sense, the band structure of rare-earth monopnictides resembles another class of nominally half-metallic materials: the full-Heusler alloys [147]. In the band structure there is a hole pocket around the point and a electron pocket around the X point (Figure 4). The spin exchange splitting is of opposite sign for the top valence and bottom conduction bands, i.e., the exchange splitting pushes the majority states to the top of the valence band (at the point) and simultaneously pushes down the minority states (at the X point). Thus, under some conditions, a gap forms in the spin-down channel in some calculated band structures. Rare-earth systems are known to exhibit band and wave vector dependent exchange splitting [9,11,148,149]. Taking GdN as example, we see that the Gd 5d states are polarized by the 4f majority states, due to exchange interactions and 4f-5d orthogonality. This means that the Gd 5d majority spin states, which dominate the bottom conduction bands, are shifted to greater binding energies. At the same time, the Gd 5d states are hybridized with N 2p states. This hybridization shifts up the N 2p majority spin states, resulting in the exchange splitting at the top valence bands, which mainly consist of the N 2p electrons. This shift in weight from Gd 5d to N 2p does result in a wave vector dependent exchange splitting. We note that the exchange splitting is also Hubbard U dependent, as can be seen from the differences between Figure 4(b) and Figure 14(a). This is because the larger U of the f states means less f-d interactions, thus a smaller exchange splitting. Another byproduct of the above Gd 5d-N 2p hybridization is the induced magnetic moments on nitrogen atoms. Actually, the induced magnetic moment on pnicogen atom is a typical feature of rareearth monopnictides. This was first pointed out by Petukhov et al. [36] and has been confirmed by recent experiments [120]. This induced pnicogen magnetic moment is antiparallel to that of the rareearth atoms, with a magnitude in the range of 0.05 ~ 0.14 . Such kind of induced antiparallel B magnetic moment on the non-magnetic atom is also found in magnetic interface structures, e.g. Co/SrTiO 3 [150,151] and Fe/BTO [152], in which the non-magnetic Ti atom has a negative magnetic moment due to the Ti 3d-Fe 3d hybridization. 5.2. Metal-insulator transitions We can see from Table 1, Figure 4 and Figure 6 that the metallicity of rare-earth monopnictides gradually increases from N to Bi. Since experimentally most of rare-earth nitrides are semiconductors, we can reasonably expect a metal-insulators transitions in some rare-earth pnictide compounds or their alloys, due to impurity doping, temperature, pressure or even magnetic field.

22

Duan et al. [88] studied both volume and lattice strain effect on the electronic properties of GdN 4 using U efff = 6 eV. The results are shown in Figure 14. It is evident that, at larger volumes, the electron (hole) pockets become substantially shallower, meaning that the bottom (top) of the band approaches the Fermi energy and the area of the Fermi surface decreases. Both carrier density and carrier mobility decreases with an increase of the cell volume. Though it is hard to realize such dramatic volume change in practice, the trend should be correct in the regimes that prove to be experimentally accessible. This transition is close to resembling a Mott metal-insulator transition, and such models [153] may be used to explain the gap in GdN found experimentally. We note that there is another mechanism of metal-semiconductor transition, i.e. through d-f or s-f hybridization. In the famous Falicov-Kimball model [ 154 ], the single-electron states consist of extended Bloch functions and localized states centered at the metallic ions in the crystal. Thus at low temperatures, the quasiparticle excitations are either localized holes or itinerant electrons. The electron-hole interaction may be responsible for the semiconductor-metal phase transition. The same spirit is presented in the electron-hole liquid theory [145], which is used to explain the ground state of ScN. Using a novel many-body theory and Kondo lattice model (s-f model), Kreissl and Nolting predicted that the decrasing magnetic order induces a transition from half-metallic to semiconducting behavior in EuB6 [155]. If this can be confirmed experimentally, then similar examples should be found among the rare-earth monopnictides.

Figure 14. The LSDA+U (Ueff = 6.0 eV ) band structure of GdN in the vicinity of the Fermi energy for three volumes: (a) at the calculated equilibrium lattice parameter a = 4.92 , (b) at the lattice parameter increased by 5% (a = 5.16 ), (c) at the lattice parameter increased by 14% (a = 5.63 ). Solid and dotted lines represent spin majority and spin minority states, respectively. The change of conducting properties are indicated by the change of energy difference between the top (bottom) of the hole (electron) pockets and the Fermi energy (from [88]).

23

5.3. Magnetic ordering The magnetic properties are closely related to the electronic and transport properties in rare-earth monopnictides. For instance, dramatic change of electric resistance has been found near Tc in GdN [120,124]. Despite their rather simple fcc structure, the magnetic orderings of the rare-earth pnictides are very complicated [22,23,39,54]. Based on the experience with rare-earth metals [9,142], it is generally believed that there exists an indirect RKKY exchange interactions in the rare-earth pnictides. In 1997, Li et al. estimated the exchange parameters of GdX from their several experimental data [60]. They attributed the sign change of exchange parameters to RKKY oscillations. Several groups believe that the super-exchange interaction should also exist, and may dominate these compounds [88,89,91]. Duan et al. [88,89] proposed a systematic way to study the magnetic ordering in the rare-earth pnictides fcc structures. The basic strategy is to first deduce the exchange interaction parameters of GdX compounds by fitting the first-principles total energies of different magnetic configurations to those computed within the Heisenberg model,
H =
n =1 ,2,3

nth NN

Jn

i> j

Si S j .

(9)

Here S i is the unit vector in the direction of the magnetic moment at the i-th lattice site, Jn is the exchange parameter between the n-th NN magnetic atoms, and we limit our consideration to third NN interactions. In this case the difference between the energy of the three AFM states En (n = I, II, III) shown in Fig.1 and the energy of the FM state EFM are
EI E I EFM = 8 J1 + 16 J 3 , EII E II EFM = 6 J1 + 6 J 2 + 12 J 3 , EIII E III EFM = 8 J1 + 2 J2 + 8 J3 .

( 10 )

The so derived exchange parameters are then put into Monte Carlo simulations to calculate the transition temperatures of the compound. Results on GdX are shown in Table 2.
Table 2. Calculated total energy differences per formula between ferromagnetic and three antiferromagnetic configurations for Gd monopnictides En = En EFM (n = I, II, III). Values inside the parentheses are calculated using the theoretical lattice constants. Theoretical and experimental transition temperatures [14] are listed for comparison (from [89]).
a() Exp.(Theor.) GdN GdP GdAs GdSb GdBi 4.97 (4.92) 5.71 (5.63) 5.86 (5.77) 6.22 (6.11) 6.30(6.24) EI (meV) 6.7(7.1) -2.8(-3.3) -3.9(-4.3) -5.7(-6.6) -6.9(-7.3) EII (meV) 4.2(5.1) -6.5(-6.9) -8.4(-8.8) -11.1(-11.5) -13.4(-13.8) EIII (meV) 6.5(7.3) -3.5(-4.0) -4.7(-5.4) -7.2(-7.9) -8.8(-9.3) J1 (meV) 0.86(0.95) -0.17(-0.21) -0.22(-0.34) -0.51(-0.60) -0.66(-0.71) J2 (meV) -0.14(-0.04) -0.74(-0.74) -0.91(-0.93) -1.13(-1.10) -1.37(-1.38) J3 (meV) -0.01(-0.03) -0.09(-0.11) -0.13(-0.10) -0.10(-0.11) -0.10(-0.10) TN(C) (K) MC Exp. 34(38) 12(13) 14(16) 20(19) 22(23) 58.0 15.9 18.7 23.4 25.8

The lattice constant dependence of the exchange parameters J1 and J2 are also obtained (Figure 15). From the analysis of these data, Duan et al. [89] proposed an expression for the superexchange interaction:

24

super 2

= nd

t4 pd 3

( 11 )

where is the energy difference between the rare-earth 5d orbital and the outmost p state of pnicogen ion, nd is the induced d moment on rare-earth atom due to atomic 4f-5d exchange interactions, and the hopping parameter tpd can be evaluated according to Harrison [ 156 ]. We should note here the evaluation of Eq. (11) is quite empirical, since the accurate determinations of and other parameters are not available. Nevertheless, the estimated superexchange interactions in GdX are found to be AFM and have the same magnitude as the total effective magnetic interactions. These AFM superexchange interactions are strengthened when the size of pnicogen atoms increases and eventually dominate the magnetic interactions in rare-earth pnictides. This picture agrees well with experimental observations [60]. Using similar technique, Larson and Lambrecht [93] also obtained the exchange parameters of GdX in good agreement with experiment.

Figure 15. Calculated exchange parameters J1 (a) and J2 (b) for Gd pnictides as a function of the lattice strain: GdN (circles), GdP (stars), GdAs (diamonds), GdSb (squares), GdBi (triangles). The lattice strain is defined by the relative deviation of the lattice constant from the theoretical equilibrium lattice constant (from [89]).

Duan et al. [87] also investigated the magnetic dipole interactions and magnetic anisotropic energies to the magnetic properties in rare-earth monopnictides. They found that dipole-dipole interactions could play an important role in the determining the magnetic behavior of the rare-earth compounds, and are especially instrumental in the magnetic anisotropy, but they cannot solely account for stabilizing the magnetic ordering in a fcc structure [51]. It should be pointed out that although the calculated Curie temperatures, using theoretical exchange coupling parameters within the Heisenberg model, agree reasonably well with experiment, it is still difficult to uniquely attribute the mechanism of exchange interactions in these compounds to the RKKY or super-exchange mechanisms. The relative contribution from RKKY interaction varies considerably due to the change in metallicity across the monopnictides. For example, if the system is insulating, the RKKY interactions should vanish. Therefore, in addition to the RKKY exchange and

25

superexchange contributions, other exchange coupling mechanisms are possible in the rare-earth monopnictides, such as higher order exchange interactions [37], direct exchange, and double exchange interactions. Actually, there is experimental evidence that direct exchange coupling could be responsible for the higher Curie temperature found in insulating GdN films [124]. 5.4. Spin-orbit coupling The spin-orbit coupling is important for the band structure and many other properties, e.g. magnetocrystalline anisotropy, of compounds containing heavier elements like RE. For Gd ions, the spin-orbit term L (here s and L are spin and orbital angular momentum operators, respectively) may be ignored since the 4f shells are exactly half occupied. For other rare-earth ions with nonzero total orbital angular momentum, the spin-orbit term may be significant. There are very few ab initio calculations on the rare-earth pnictides involving spin-orbit coupling. Duan et al. studied spin-orbit effect in ErAs [85], and Johannes and Pickett also considerd this effect in their studies on EuN and EuP [91]. Leuenberger et al. mentioned the presence of an important spin-orbit interaction in the final N p states of GdN [120]. Figure 16 presents the theoretical band structure of ErAs that takes into account both Hubbard U term and spin-orbit interaction. In this calculation, the inclusion of spin-orbit interaction is realized by a fully relativistic approach for the core electrons and a second variational method for the valence states [157]. In contrast to the LSDA+U result [84], the occupied 4f bands are split further due to spinorbit coupling, leading to better agreement with experiments [85].

Figure 16. The calculated LSDA+U+SO band structure of bulk ErAs. In the spin-orbit calculation, spin up and spin down states are mixed (from [85]).

26

Spin-orbit coupling is at the heart of magnetooptical effects such as Kerr and Faraday polarization rotation. CeSb has been claimed to have a 90 degrees Kerr rotation [158]. While the intrinsic value of the Kerr rotation of CeSb is still debated, agreement exists that in any case CeSb has among the largest Kerr rotation, possibly the largest. A series of calculations of the magnetooptical properties of the rareearth monopnictides reproduced experiment results with various success [83,92,159].

The Future Possibilities

While the rare-earth monopnictides have the simplest structure of the rare-earth compounds, substantive changes can be easily undertaken by altering the constituents. Actually, by mixing rareearth pnictides in different ratios, we can tune the physical properties, e.g. the Tc, carrier density, resistance, of rare-earth pnictides to desired one [24,38,43,47,35,162]. Impurity doping, thus nonstoichiometry, also has significant influence on the properties on rare-earth pnictides. These subjects are too complicated and are beyond the scope of our current review. We just want to mention that there are many choices for improving the physical, magnetic and transport properties of rare-earth pnictides. For binary rare earth pnictide compounds, there are various formulas with different structures. For example, there are at least five structure types in the RX2 phases, i.e. LaP2, NdAs2, SmSb2, HoSb2, ZrSi2 [39]. Of particular interest we mention anti-Th3P4-type pnictides, which have attracted much interest recently [160,161]. As will show later, some of them demonstrate very large Tc. For ternary compounds, the situation is much more complicated [39]. Active experimental studies are being carried out on these systems [162,163,164]. 6.1. Possible half-metallic antiferromagnets In 1995, Van Leuken and de Groot proposed another type of half-metallic material, i.e. the halfmetallic antiferromagnet [165]. This material is calculated to be 100% spin polarized at the Fermi surface but has zero net magnetization, thus could be used as a tip material in spin-polarized scanning tunneling microscopy. There are several theoretical predictions concerning the existence of the halfmetallic antiferromagnet [165,166,167,168], but none of them has been confirmed experimentally. Half-metallic antiferromagnets cannot be found with the simple pure rare-earth monopnictides. As Figure 10(a) shows, the symmetric arrangement of Gd ions forms exactly the same averaged potentials for both spin-up and spin-down electrons, thus their DOS plots are the same. In considering Gd3+ and Eu2+, which have the same electron configuration (8S7/2), and that both EuO and GdN are potential half-magnetic materials, it may be very possible to make an EuO-GdN compound, in which Eu2+ and Gd3+ are aligned antiparallel, yet two spin channels may show difference at Fermi level and results in nominal half-metallicity. This idea needs further verifications, but if it is true, then this material would be a half-metallic antiferromagnet with the simplest structure (double fcc). 6.2. Enhancement of the transition temperatures and potential device applications The rare-earth monopnictides compounds have generally rather low transition temperatures. GdN has the largest Tc (58 ~ 68 K), but it is still far below the room temperature. How to improve the Tc in these compounds is of interest.

27

In 1971, Kuznietz reported the effect of anion substitution in GdN and UN [169]. It was found that carbon substitution can significantly improve the Tc, while O substitution works the other way. An example is GdN0.88C0.12 (Gd4N3.5C0.5 in Ref. [38]), whose Tc was reported to be as high as 190 K. This can be explained by the reduction of Gd-Gd distance, which may lead to increase in 5d-5d overlap. Considering that Gd metal has somewhat higher Tc = 293.2 K [170], efforts in this direction are very promising. Wachter and Kaldis also confirmed this kind of effect of Gd-Gd distance [43]. Actually, Ref. [38] also reported that some other GdN derived alloys have Tc even higher than room temperature (340 K). The other possibility for increasing the Tc in the rare-earth pnictides is impurity doping. Magnetic impurities are helpful in assisting in the alignment of neighboring magnetic moments and strengthen the magnetic interactions or even change the RKKY interaction to a long-range ferromagnetic interaction, thus dramatically enhancing the Tc. Examples can be found in Ref. [171], which reported that by adding few percent of magnetic ions such as Mn, the Tc of GaN or ZnO well exceeds the room temperature. Again mentioning the experiments on GdN/Fe multilayers [125] in this context, it is worth noting that the Fe layers are found to be responsible for the long range magnetic order in GdN layer at temperatures largely above Tc. In addition, we noticed that some rare-earth pnictides of the Th3P4 type (or its anti-type) have very high Tc values, even higher than that of Gd metal. For example, the Tc of Gd4Bi3 is 340 K [172], for Gd4Sb3 is 260 K [173], for Tb4Bi3 is 200 K. The origin of these high Tcs needs further investigations. Despite their low transition temperatures, applications for the rare earth pnictides do exist. For example, ErAs has been used to study magnetotransport [51] and resonant tunneling [174]; EuS has been used in hybrid spin filter devices [175,176]. These materials are of special interest for using as barriers in magnetic tunnel junctions [177]. This is due to the exchange splitting of the conduction band which makes the barrier height spin-dependent. Thus, given the exponential dependence of the tunneling current on barrier height, a highly spin-polarized current is expected. In a series of spindependant tunneling experiments, Moodera et al. found a tunneling spin polarization of approximately 80% in Al/EuS/M junctions, where M was Ag, Au, or Al [178]. Using a related Eu chalcogenide, EuSe, they were able to demonstrate essentially 100% spin polarization [179]. New hybrid devices utilizing the spin filtering properties of rare-earth monopnictides are expected in the near future.

Conclusions

After more than forty yearsefforts, stimulated first by the search for magnetic semiconductors and now by spintronics materials, we now have a better understanding of the electronic, magnetic and transport properties of rare-earth monopnictides. Rare-earth monopnictides are generally semiconductors or semimetals. Despite their simple rock salt structure, they demonstrate various types of magnetic ordering generally with low transition temperatures. Their electronic structures and magnetic properties are sensitive to the temperature, pressure (strain) and impurity effects. The rare-earth 4f-5d interactions and the hybridizations between the rare-earth non-4f and pnictogen p states are responsible for many fascinating phenomena that occur in rare-earth monopnictides. Different from the rare-earth metals, both super-exchange interaction and indirect RKKY-type interaction coexist in rare-earth monopnictides, and the former is dominant in heavy rare-earth monopnictides. It is possible that other exchange coupling mechanisms, such as direct exchange and double exchange, also could be found in rare-earth monopnictides. Based on the discussion in this paper, we conclude that the high polarization, nominally half-

28

metallic, rare-earth monopnictides are most likely to be found in rare-earth nitrides or nitrogen involved rare-earth alloys. This is because those nitrides are on the border between semiconductor and metal due to their anion sizes. Impurity doping, structural distortion, lattice constraint effects have significant influence on the electronic, magnetic and transport properties of rare-earth nitrides and might be helpful to modify these compounds into desired products. During the past decades, various theoretical and experimental studies have been made to understand the intriguing properties of these materials. Theoretically, much more powerful tools have been developed to solve the complicated yet fascinating many-body problem existing in these strongly correlated systems, such as SIC, LSDA+U, GWA and DMFT. With the unbelievable increase of the computational power and the maturity of massively parallel computing technology, we now have enough confidence to deal with large scale calculation involved in these theoretical studies. Experimentally, state-of-the-art crystal and film growth techniques make it possible to determine the electronic structures, e.g. the Fermi surfaces, and physical properties of these compounds. Modern techniques like low energy electron-diffractions, scanning tunneling microscopy, angle-resolved (inverse) photoemission, neutron-diffraction are helpful to make clear the geometric, electronic and magnetic structures of the film and crystal. In addition, now element specific XMCD experiment can study the properties of individual atoms in the system. There are still many unresolved issues related to the half-metallicity in rare-earth monopnictides and many blanks that need to be filled in Table 1. To address these issues or fill in the blanks, a close collaboration between experimentalists and theorists is absolutely necessary. Nevertheless, it just takes a little imagination to conclude that, in the future, we should expect further exciting results to emerge from the study of electronic structure of rare-earth compounds. Acknowledgments The authors acknowledge insightful discussions with P.W. Anderson, D.S. Wang, A.G. Petukhov, W. Nolting and A. Sharma. Special thanks to S. Granville, S. Suga, D.X. Li for giving permissions to use the original figures in their publications. This work was supported by NSF (Grants DMR-0203359, MRSEC-DMR-0213808, and EPS-9720643), the Nebraska Research Initiative and Department of the Army Grants DAAG 55-98-1-0273 and DAAG 55-99-1-0106 and the Office of Naval Research (Grant No. N00014-06-1-0616). The computational part of this work was completed utilizing the Research Computing Facility of the University of Nebraska-Lincoln.

References
[1] R.A. de Groot and F.M. Mueller, P.G. van Engen and K.H.J. Buschow, Phys. Rev. Lett. 50, 2024 (1983). [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 294, 1488 (2001). [3] W.E. Pickett and J.S. Moodera, Phys. Today 54, May 39 (2001). [4] P.A. Dowben and R. Skomski, J. Applied Physics 95, 7453 (2004). [5] P.A. Dowben and S.J. Jenkins, Frontiers in Magnetic Materials, ed. by Anant Narlikar (Springer Verlag, 2005), p. 295. [6] L. Chioncel, M.I. Katsnelson, R.A. de Groot, and A.I. Lichtenstein, Phys. Rev. B 68, 144425 (2003). [7] V.Y. Irkhin and M.I. Katsnelson, J. Phys. C: Solid State Phys. 18, 4173 (1985). [8] For a complete review, see the series: Handbook on Physics and Chemistry of Rare Earth, ed. by K.A. Gschneidner and L. Eyring (North Holland, Amsterdam, 1979). [9] P.A. Dowben, D.N. McIlroy and Dongqi Li, Handbook on the Physics and Chemistry of Rare Earths, ed. by K.A. Gschneidner and L. Eyring (North Holland, Amsterdam, 1997), Vol. 24, Chap. 159.

29

B.D. Cullity, Introduction to magnetic materials (Addison Wesley, 1972), p. 178. I.N. Yakovkin, T. Komesu, and P. A. Dowben, Phys. Rev. B 66, 035406 (2002). C.G. Olson, S.J. Chase, P. Canfield and D.W. Lynch, J. Electron Spectrsoc. Rela. Phenom. 93, 175 (1998). A.B. Andrews, J.J. Joyce, A.J. Arko, J.D. Thompson, J. Tang, J.M. Lawrence, J.C. Hemminger, Phys. Rev. B 51, R3277 (1995). [14] E. Guziewicz, T. Durakiewicz, M.T. Butterfield, C.G. Olson, J.J. Joyce, A.J. Arko, J.L. Sarrao, D.P. Moore, L. Morales, Phys. Rev. B 69, 045102 (2004). [15] E. Guziewicz, T. Durakiewicz, C.G. Olson, J.J. Joyce, M.T. Butterfield, A.J. Arko, J.L. Sarrao, A. Wojakowski, Surf. Sci. 600, 1632 (2006). [16] E. Guziewicz, T. Durakiewicz, P.M. Oppeneer, J.J. Joyce, J.D. Thompson, C.G. Olson, M.T. Butterfield, A Wojakowski, D.P. Moore, A.J. Arko, Phys. Rev. B 73, 155119 (2006). [17] P.W. Anderson, Phys. Rev. 115, 2 (1959). [18] P.W. Anderson, Rev. Mod. Phys. 50, 191 (1978). [19] J.H. Van Vleck and A. Frank, Phys. Rev. 34, 1494 (1929). [20] J.H. Van Vleck, Rev. Mod. Phys. 50, 181 (1978). [21] N. Sclar, J. Appl. Phys. 33, 2999 (1962); 35, 1534 (1964). [22] H.R. Child, M.K. Wilkinson, J.W. Cable, W.C. Koehler and E.O. Wollan, Phys. Rev. 131, 922 (1963). [23] G.T. Trammell, Phys. Rev. 131, 932 (1963), and references therein. [24] F. Holtzberg, T.R. McGuire, S. Methfessel and J. C. Suits, Phys. Rev. Lett. 13, 18 (1964); J. Appl. Phys. 35, 1033 (1964); J. Appl. Phys. 37, 976 (1966). [25] G. Busch, J. Appl. Phys. 38, 1386 (1967), and references therein. [26] R.F. Brown, A.W. Lawson and G.E. Everett, Phys. Rev. 172, 559 (1968). [27] T.R. McGuire, R.J. Gambino, S.J. Pickart and H.A. Alperin, J. Appl. Phys. 40, 1009 (1969). [28] R.J. Birgeneau, E. Bucher, and J.P. Maita, L. Passell, K.C. Turberfield, Phys. Rev. B 8, 5345 (1973). [29] A. Narita and T. Kasuya, J. Magn. Magn. Mater. 52, 373 (1985). [30] H. Takahashi and T. Kasuya, J. Phys. C: Solid State Phys. 18, 2697 (1985); 18, 2709 (1985). [31] A. Hasegawa and A. Yanase, J. Phys. Soc. Jpn. 42, 492 (1977). [32] J.C. Slater, Phys. Rev. 81, 385 (1951). [33] A. Narita, J. Phys. Soc. Jpn. 57, 251 (1988); 57, 3961 (1988). [34] J.-B. Xia, S.-F. Ren, and Y.-C. Chang, Phys. Rev. B 43, 1692 (1991). [35] A.G. Petukhov, W.R.L. Lambrecht, and B. Segall, Phys. Rev. B 50, 7800 (1994). [36] A.G. Petukhov, W.R.L. Lambrecht, and B. Segall, Phys. Rev. B 53, 4324 (1996). [37] T. Kasuya and D.X. Li, J. Mag. Mag. Mat. 167, L1 (1997). [38] J.J. Rhyne and T.R. McGuire, IEEE Trans. Magn. 8, 105 (1972). [39] F. Hulliger, Handbook on the Physics and Chemistry of Rare Earths, ed. by K.A. Gschneidner and L. Eyring (North Holland, Amsterdam, 1979), Vol. 4, Chap. 33, p. 153. [40] E. Kaldis, J. Cryst. Growth 24/25, 53 (1974). [41] G. G ntherodt, E. Kaldis and P. Wachter, Solid State Commun. 15, 1435 (1974). [42] E. Kaldis and C. Zurcher, Helv. Phys. Acta 47, 421 (1974). [43] P. Wachter and E. Kaldis, Solid State Commun. 34, 241 (1980). [44] M.A. Ruderman and C. Kittel, Phys. Rev. 96, 99 (1954). [45] T. Kasuya, Progr. Theoret. Phys. (Kyoto) 16, 45 (1956). [46] K. Yosida, Phys. Rev. 106, 893 (1957). [47] P. Wachter, E. Kaldis, and R. Hauger, Phys. Rev. Lett. 40, 1404 (1978). [48] P. Wachter, Physics Reports 44, 159 (1978). [49] P. Urban, K. Jansen, G. Sperlich and D. Davidov, J. Phys. F: Met. Phys. 8, 977 (1978). [50] C.J. Palmstr N. Tabatabaie, and S.J. Allen, Appl. Phys. Lett. 53, 2608 (1988). m, [51] S.J. Allen, N. Tabatabaie, C.J. Palmstr G.W. Hull, T. Sands, F. DeRosa, H.L. Gilchrist, and K.C. Garrison, Phys. m, Rev. Lett. 62, 2309 (1989). [52] L. Degiorgi, W. Bacsa, and P. Wachter, Phys. Rev. B 42, 530 (1990). [53] L. Degiorgi, S. Teraoka, G. Compagnini, and P. Wachter, Phys. Rev. B 47, 5715 (1993). [54] T. Chattopadhyay, P. Burlet, J. Rossat-Mignod, H. Bartholin, C. Vettier and O. Vogt, Phys. Rev. B 49, 15096 (1994). [55] C. Waldfried, D.N. McIlroy, Dongqi Li, J. Pearson, S.D. Bader, and P.A. Dowben, Surf. Sci. 341, L1072 (1995). [56] J.Q. Xiao, and C.L. Chien, Phys. Rev. Lett. 76, 1727 (1996). [57] D.X. Li, Y. Haga, H. Shida, and T. Suzuki, T. Koide and G. Kido, Phys. Rev. B 53, 8473 (1996).

[10] [11] [12] [13]

30

[58] D.X. Li, Y. Haga, H. Shida, and T. Suzuki, Y.S. Kwon, Phys. Rev. B 54, 10483 (1996). [59] H. Yamada, T. Fukawa, T. Muro, Y. Tanaka, S. Imada, S. Suga, D.-X. Li, and T. Suzuki, J. Phys. Soc. Jpn. 65, 1000 (1996). [60] D.X. Li, Y. Haga, H. Shida, T. Suzuki, Y.S. Kwon and G. Kido, J. Phys.: Cond. Matt. 9, 10777 (1997). [61] R.O. Jones and O. Gunnarsson, Rev. Mod. Phys. 61, 689 (1989). [62] J.P. Perdew, A. Zunger, Phys. Rev. B 23, 5048 (1981), and references therein. [63] R.D. Cowan, Phys. Rev. 163, 54 (1967). [64] R.A. Heaton, J.G. Harrison, and C.C. Lin, Phys. Rev. B 28, 5992 (1983). [65] A. Svane, O. Gunnarsson, Phys. Rev. Lett. 65, 1148 (1990). [66] A. Svane and O. Gunnarsson, Phys. Rev. B 37, 9919 (1988). [67] P. Strange, A. Svane, W.M. Temmerman, Z. Szotek and H. Winter, Nature 399, 756 (1999). [68] L. Petit, A. Svane, Z. Szotek and W.M. Temmerman, Science 301, 498 (2003). [69] A. Svane, Phys. Rev. Lett. 68, 1900 (1992). [70] C.M. Aerts, P. Strange, M. Horne, W.M. Temmerman, Z. Szotek and A. Svane, Phys. Rev. B 69, 045115 (2004). [71] M. Horne, P. Strange, W.M. Temmerman, Z. Szotek, A. Svane and H. Winter, J. Phys.: Condens. Matter 16, 5061 (2004). [72] Z. Szotek, W.M. Temmerman, A. Svane, L. Petit, P. Strange, G.M. Stocks, D. K dderitzsch, W. Hergert and H. Winter, J. Phys.: Condens. Matter 16, S5587 (2004). [73] A. Svane, V. Kanchana, G. Vaitheeswaran, G. Santi, W.M. Temmerman, Z. Szotek, P. Strange and L. Petit, Phys. Rev. B 71, 045119 (2005). [74] J. Hubbard, Proc. Roy. Soc. London Ser. A 276, 238 (1963); 277, 237 (1964); 281, 401 (1964). [75] V.I. Anisimov, J. Zaanen and O.K. Andersen, Phys. Rev. B 44, 943 (1991). [76] V.I. Anisimov, I.V. Solovyev, M.A. Korotin, M.T. Czyzyk, and G.A. Sawatzky, Phys. Rev. B 48, 16929 (1993). [77] A.I. Liechtenstein, V.I. Anisimov, J. Zaanen, Phys. Rev. B 52, R5467 (1995). [78] V.I. Anisimov, F. Aryasetiawan and A.I. Lichtenstein, J. Phys.: Condens. Matter 9, 767 (1997). [79] P.W. Anderson, Phys. Rev. 124, 41 (1961). [80] S.L. Dudarev, G.A. Botton, S.Y. Savrasov, C.J. Humphreys, A.P. Sutton, Phys. Rev. B 57, 1505 (1998). [81] V.I. Anisimov and O. Gunnarsson, Phys. Rev. B 43, 7570 (1991). [82] J.C. Slater, Quantum Theory of Atomic Structure (McGraw-Hill, New York, 1960), Vol. 1. [83] A.I. Liechtenstein, V.P. Antropov, and B.N. Harmon, Phys. Rev. B 49, 10770 (1994). [84] T. Komesu, H.-K Jeong, J. Choi, C.N. Borca, P.A. Dowben, A.G. Petukhov, B.D. Schultz, and C.J. Palmstr Phys. m, Rev. B 67, 035104 (2003). [85] C.-G. Duan, T. Komesu, Hae-Kyung Jeong, Jaewu Choi, C.N. Borca, Wei-Guo Yin, Jianjun Liu, W.N. Mei, P.A. Dowben, A.G. Petukhov, B.D. Schultz, and C.J. Palmstr Surface Review and Letters 11, 531 (2004). m, [86] P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka and J. Luitz, WIEN2k, An Augmented Plane Wave + Local Orbitals Program for Calculating Crystal Properties (Karlheinz Schwarz, Techn. Universit Wien, Austria), t 2001. ISBN 3-9501031-1-2. [87] C.-G. Duan, R.F. Sabiryanov, Jianjun Liu, W.N. Mei, P.A. Dowben, J.R. Hardy, J. Appl. Phys. 97, 10A915 (2005). [88] C.-G. Duan, R.F. Sabiryanov, Jianjun Liu, W.N. Mei, P.A. Dowben, J.R. Hardy, Phys. Rev. Lett. 94, 237201 (2005); Phys. Rev. Lett. 96, 139901 (2006). [89] C.-G. Duan, R.F. Sabiryanov, W.N. Mei, S.S. Jaswal, P.A. Dowben, E.Y. Tsymbal, Appl. Phys. Lett. 88, 182505 (2006). [90] P.W. Anderson, Phys. Rev. 79, 350 (1950); Phys. Rev. 115, 2 (1959). [91] M.D. Johannes and W.E. Pickett, Phys. Rev. B 72, 195116 (2005). [92] D.B. Ghosh, Molly De, and S.K. De, Phys. Rev. B 72, 045140 (2005). [93] P. Larson and Walter R. L. Lambrecht, Phys. Rev. B 74, 085108 (2006). [94] W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 (1965). [95] L.D. Landau, Sov. Phys. JETP 3, 920 (1957); 5, 101 (1957); 35, 70 (1959). [96] F. Aryasetiawan and O. Gunnarsson, Rep. Prog. Phys. 61, 237 (1998). [97] L. Hedin, Phys. Rev. 139, A796 (1965). [98] D. Pines, D. Bohm, Phys. Rev. 85, 338 (1952). [99] M.S. Hybertsen, S.G. Louie, Phys. Rev. Lett. 55, 1418 (1985); Phys. Rev. B 37, 2733 (1988). [100] R.W. Godby and M. Schl L.J. Sham, Phys. Rev. Lett. 56, 2415 (1986). ter, [101] G. Onida, L. Reining, A. Rubio, Rev. Mod. Phys. 74, 601 (2002). [102] M. van Schilfgaarde, T. Kotani, and S. Faleev, Phys. Rev. Lett. 96, 226402 (2006).

31

[103] W.R.L. Lambrecht, Phys. Rev. B 62, 13538 (2000). [104] For a recent review, see G. Kotliar, S.Y. Savrasov, K. Haule, V.S. Oudovenko, O. Parcollet, C.A. Marianetti, Rev. Mod. Phys. 78, 865 (2006). [105] W. Metzner and D. Vollhardt, Phys. Rev. Lett. 62, 324 (1989). [106] A. Georges,G. Kotliar,W. Krauth and M.J. Rozenberg, Rev. Mod. Phys. 68, 13 (1996). [107] V.I. Anisimov, A.I. Poteryaev, M.A. Korotin, A.O. Anokhin and G. Kotliar, J. Phys.: Condens. Matter 9, 7359 (1997). [108] R. Chitra and G. Kotliar, Phys. Rev. B 62, 12715 (2000). [109] S. Kalvoda, M. Dolg, H.-J. Flad, P. Fulde, and H. Stoll, Phys. Rev. B 57, 2127 (1998). [110] C. Santos and W. Nolting, V. Eyert, Phys. Rev. B 69, 214412 (2004). [111] A. Sharma and W. Nolting, J. Phys.: Condens. Matter 18, 7337 (2006). [112] S. Bhattacharjee and S.M. Jaya, Eur. Phys. J. B 49, 305 (2006). [113] G.A. Landrum, R. Dronskowski, R. Niewa, F.J. DiSalvo, Chem. Eur. J. 5, 515 (1999) [114] N. Marzari and D. Vanderbilt, Phys. Rev. B 56, 12847 (1997). [115] M. Stengel and N.A. Spaldin, Phys. Rev. B 73, 075121 (2006). [116] K. Koyama, M. Yoshida, H. Nojiri, T. Sakon, D.X. Li, T. Suzuki and M. Motokawa, J. Phys. Soc. Jpn. 69, 1521 (2000). [117] K. Koyama, M. Yoshida, D.X. Li, and M. Motokawa, J. Phys. Soc. Jpn. 70, 2774 (2001). [118] T. Tomimatsu, K. Koyama, M. Yoshida, D.X. Li, and M. Motokawa, Phys. Rev. B 67, 014406 (2003). [119] Y. Nakanishi, T. Sakon, M. Motokawa, M. Ozawa and T. Suzuki, Phys. Rev. B 69, 024412 (2004). [120] F. Leuenberger, A. Parge, W. Felsch, K. Fauth, and M. Hessler, Phys. Rev. B 72, 014427 (2005). [121] M.R. Oliver, J.A. Kafalas, J.O. Dimmock, and T.B. Reed, Phys. Rev. Lett. 24, 1064 (1970). [122] Y. Shapira, S. Foner, and R.L. Aggarwal, T.B. Reed, Phys. Rev. B 8, 2316 (1973). [123] F. Leuenberger, A. Parge, W. Felsch, F. Baudelet, C. Giorgetti, E. Dartyge, and F. Wilhelm, Phys. Rev. B 73, 214430 (2006). [124] S. Granville, B.J. Ruck, F. Budde, A. Koo, D.J. Pringle, F. Kuchler, A.R.H. Preston, D.H. Housden, N. Lund, A. Bittar, G.V.M. Williams, and H.J. Trodahl, Phys. Rev. B 73, 235335 (2006). [125] F. Leuenberger, A. Parge, and W. Felsch, T. Neisius, O. Mathon, J. Appl. Phys. 100, 033905 (2006). [126] G.L. Olcese, J. Phys. F: Met. Phys. 9, 569 (1979). [127] J.S. Smart, Effective field theories of magnetism (W. B. Saunders Company, Philadelphia & London, 1966), p. 76-77. [128] S. Ono, K. Nomura, H. Hayakawa, J. Less Comm. Metals 38, 119 (1972). [129] A. Iandelli, E. Botti, Atti della Accademia Nazionale dei Lincei, Classe di Fisiche, Matematiche e Naturali, Rendiconti 254, 498 (1976). [130] J.M. Leger, D. Ravot and J. Rossat-Mignod, J. Phys. C: Solid State Phys. 17, 4935 (1984). [131] K. Nomura, H. Hayakawa and S. Ono, J. Less Comm. Metals 52, 259 (1977). [132] J. Schoenes and W. Reim, J. Less Comm. Metals 112, 19 (1985). [133] P. Ettmayer, J. Waldhart, A. Vendl, Monatshefte fuer Chemie 110, 1109 (1979). [134] H. Jacobs, U. Fink, Z. Anorg. Allgem. Chem. 438, 151 (1978). [135] K.E. Mironov, G.P. Brygalina, Izvestiya Akademii Nauk SSSR, Neorganicheskie Materialy 10, 787 (1974). [136] F. Bommeli, L. Degiorgi and P. Wachter, J. Mag. Mag. Mat. 140, 1159 (1995) [137] D.X. Li, A. Oyamada, K. Hashi, Y. Haga, T. Matsumura, H. Shida, T. Suzuki, T. Kasuya, A. D and F. Hulliger, nni J. Mag. Mag. Mat. 140, 1169 (1995). [138] W. Klemm, G. Winkelmann, Z. Anorg. Allgem. Chem. 288, 87 (1956). [139] A. Iandelli, Atti della Accademia Nazionale dei Lincei, Classe di Fisiche, Matematiche e Naturali, Rendiconti 37, 160 (1964). [140] B.N. Harmon, V.P. Antropov, A.I. Liechtenstein, I.V. Solovyev, and V.I. Anisimov, J. Phys. Chem. Solids 56, 1521 (1995). [141] A.B. Shick, A.I. Liechtenstein, and W.E. Pickett, Phys. Rev. B 60, 10763 (1999). [142] Ph. Kurz, G. Bihlmayer and S. Bl J. Phys.: Cond. Matt. 14, 6353 (2002). gel, [143] F. Steglich, U. Rauchschwalbe, U. Gottwick, H.M. Mayer, G. Sparn, N. Grewe, U. Poppe, J.J.M. Franse, J. Appl. Phys. 57, 3054 (1985). [144] T.M. Rice and K. Ueda, Phys. Rev. Lett. 55, 995 (1985). [145] R. Monnier, J. Rhyner, and T.M. Rice, D. D. Koelling, Phys. Rev. B 31, 5554 (1985). [146] N.W. Ashcroft and N.D. Mermin, Solid State Physics (Holt, Rinehart, and Winston, New York, 1976), p. 652. [147] I. Galanakis and P.H. Dederichs, N. Papanikolaou, Phys. Rev. B 66, 174429 (2002).

32

[148] [149] [150] [151] [152] [153] [154] [155] [156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [177] [178] [179]

C. Waldfried, T. McAvoy, D. Welipitiya, T. Komesu, P.A. Dowben and E. Vescovo, Phys. Rev. B 58, 7434 (1998). T. Komesu, C. Waldfried and P.A. Dowben, Phys. Lett. A 256, 81 (1999). I.I. Oleinik, E.Y. Tsymbal, and D.G. Pettifor, Phys. Rev. B 65, R020401 (2002). J. Velev, K.D. Belashchenko, D. Stewart, M. van Schilfgaarde, S.S. Jaswal, and E.Y. Tsymbal, Phys. Rev. Lett. 95, 216601 (2005). C.-G.Duan, S.S. Jaswal, E.Y. Tsymbal, Phys. Rev. Lett. 97, 047201 (2006). P.A. Dowben, Surf. Sci. Reports 40, 151 (2000). L.M. Falicov, J.C. Kimball, Phys. Rev. Lett. 22, 997 (1969). M. Kreissl and W. Nolting, Phys. Rev. B 72, 245117 (2005). W.A. Harrison, Electronic structure and the properties of solids (Freeman, San Francisco, 1980), pp. 451, 481 and 505. D.J. Singh, Planewaves, Pseudopotentials and the LAPW Method (Kluwer Academic, Boston, 1994), p. 86. R. Pittini, J. Schoenes, O. Vogt, P. Wachter. Phys Rev Lett. 77, 944 (1996). D.B. Ghosh, M. De, and S. K. De, Phys. Rev. B 67, 035118 (2003). J. Szade and M. Drzyzga, J. Alloys Compounds 299, 72 (2000). M. Drzyzga, J. Szade, J. Deniszczyk and T. Michalecki, J. Phys.: Condens. Matter 15, 3701 (2003). F. Weber, A. Cosceev, A. Nateprov, C. Pfleiderer, A. Faisst, M. Uhlarz, H.V. Lohneysen, Physica B: Condens. Matt. 359, 226 (2005). Y. Mozharivskyj, H.F. Franzen, J. Solid Stats Chem. 152, 478 (2000). L.F. Schneemeyer, R.B. van Dover, and E.M. Gyorgy, J. Appl. Phys. 61, 3543 (1987). H. van Leuken and R.A. de Groot, Phys. Rev. Lett. 74, 1171 (1995). W.E. Pickett, Phys. Rev. B 57, 10613 (1998). B.I. Min, M.S. Park and J.H. Park, J. Phys.: Condens. Matter 16, S5509 (2004). H. Akai and M. Ogura, Phys. Rev. Lett. 97, 026401 (2006). M. Kuznietz, J. Appl. Phys. 42, 1470 (1971). H.E. Nigh, S. Legvold, and F.H. Spedding, Phys. Rev. 132, 1092 (1963). J.M. Coey, M. Venkatesan, and C.B. Fitzgerald, Nature Materials 4, 173 (2005). S. Methfessel and E. Kneller, Appl. Phys. Lett. 2, 115 (1963). F. Holtzberg, T.R. McGuire, S. Methfessel, and J.C. Suits, J. Appl. Phys. 35, 1033 (1964). D.E. Brehmer, K. Zhang, C.J. Schwarz, S.P. Chau, and S.J. Allen, Appl. Phys. Lett. 67, 1268 (1995). P. LeClair, J.K. Ha, H.J.M. Swagten, J.T. Kohlhepp, C.H. van de Vin, and W.J.M. de Jonge, Appl. Phys. Lett. 80, 625 (2002). A.T. Filip, P. LeClair, C.J.P. Smits, J.T. Kohlhepp, H.J.M. Swagten, B. Koopmans, and W.J.M. de Jonge, Appl. Phys. Lett. 81, 1815 (2002). E.Y. Tsymbal, O. Mryasov and P.R. LeClair, J. Phys.: Condens. Matter 15, R109 (2003). J.S. Moodera, X. Hao, G.A. Gibson, and R. Meservey, Phys. Rev. Lett. 61, 637 (1988). J.S. Moodera, R. Meservey, and X. Hao, Phys. Rev. Lett. 70, 853 (1993).

33