Calculations employing the local density approximation combined with static and dynamical mean-field theories (LDA+U and LDA+DMFT) indicate that the metal-insulator transition observed at 32 GPa in paramagnetic LaMnO3 at room temperature is not a Mott-Hubbard transition, but is caused by orbital splitting of the majority-spin eg bands. For LaMnO3 to be insulating at pressures below 32 GPa, both on-site Coulomb repulsion and Jahn-Teller distortion are needed.PACS numbers: 71.30.+h 71.20.-b 71.27.+a Since the discovery of colossal magnetoresistance (CMR) [1], manganites have been intensively studied. The key to understand CMR is the high-temperature paramagnetic insulating-like phase, which is characterized not only by an increase of resistivity with decreasing temperature, but also by unusual dynamical properties, such as low spectral weight at the Fermi level for a wide range of doping [2,3,4]. Theoretical understanding of this hole-doped paramagnetic phase remains incomplete, and CMR transition temperatures are lower than technologically desirable.In this Letter, we shall focus on the pressure-induced insulator-metal (IM) transition in the undoped parent compound LaMnO 3 with configuration t 3 2g e g . This transition occurs at room temperature, well above the magnetic ordering temperature (T N =140 K) , well below the cooperative Jahn-Teller (JT) temperature (T oo =740 K at 0 GPa), and at a hydrostatic pressure of 32 GPa where the JT distortion appears to be completely suppressed [5]. The IM transition thus seems to be a bandwidthdriven Mott-Hubbard transition of the e g electrons and points to the dominating importance of the Coulomb repulsion between two e g electrons on the same site. This is supported by recent self-interaction-corrected local density approximation (LDA) calculations, performed, however, for the cubic structure and magnetically ordered states at low temperature [6]. Structural distortions at 0 K as functions of pressure were recently calculated with the LDA+U method [7]. On the theoretical side, it has been an issue of long debate whether the JT distortion or the Coulomb repulsion is responsible for the insulating behavior of LaMnO 3 at normal pressure. The highpressure experiment [5] seems to favor the latter.Here, we shall study the room-temperature electronic structure of LaMnO 3 at normal pressure and the pressure-induced IM transition by means of LDA + U [8] and LDA + dynamical mean field theory (DMFT) [9] calculations. Upon going from the insulating to the metallic, high-pressure phase, we shall find that the orbital polarization and the concomitant splitting of the two majority-spin e g bands are gradually reduced. The IM transition takes place when the bands start to overlap. Since this occurs within the (orbitally) symmetry-broken phase, this IM transition is not a Mott-Hubbard transition. The Coulomb interaction, as well as the JT distortion are needed for a proper description of this transition and the insulating nature of LaMnO 3 .The orthorhombic crystal structure of LaMnO ...

No abstract

The calculation of imaginary time displaced correlation functions with the auxiliary field projector quantum MonteCarlo algorithm provides valuable insight (such as spin and charge gaps) in the model under consideration. One of the authors and M. Imada [1] have proposed a numerically stable method to compute those quantities. Although precise this method is expensive in CPU time. Here, we present an alternative approach which is an order of magnitude quicker, just as precise, and very simple to implement. The method is based on the observation that for a given auxiliary field the equal time Green function matrix, G, is a projector: PACS numbers: 71.27.+a, 71.10.Fd For a given Hamiltonian H = x,y c † x T x,y c y + H I and its ground state |Ψ 0 , our aim is to calculateHere, c † x creates an electron with quantum numbers x, c x (τ ) = e τ (H−µN ) c x e −τ (H−µN ) , and the chemical poten-H I corresponds to the interaction. Within the projector Quantum Monte Carlo (PQMC) algorithm, this quantity is obtained by propagating a trial wave |Ψ T function along the imaginary time axis [2-4]:The above is valid provided that: Ψ 0 |Ψ T = 0.To fix the notation, we will briefly summarize the essential steps required for the calculation of the right hand side (rhs) of the above equation at fixed values of the projection parameter Θ. A detailed review may be found in [5]. The formalism -without numerical stabilization -to compute time displaced correlation functions follows Ref.[6]. The first step is to carry out a Trotter decomposition of the imaginary time propagation: e −2ΘH = e −∆τ Ht/2 e −∆τ HI e −∆τ Ht/2 m + O((∆τ ) 2 ).Here, H t (H I ) denotes the kinetic (interaction) term of the model and m∆τ = 2Θ. Having isolated the interaction term, H I , one may carry out a Hubbard Stratonovitch (HS) transformation to obtain:where s denotes a vector of HS fields. For a Hubbard interaction, one can for example use various forms of Hirsch's discrete HS decomposition [7,8]. For interactions taking the form of a perfect square, decompositions presented in [9] are useful. The imaginary time propagation may now be written as:e −∆τ Ht/2 e x,y c † x Dx,y( sn)cy e −∆τ Ht/2 .The HS field has acquired an additional imaginary time index since we need independent fields for each time increment.The trial wave function is required to be a Slater determinant:Here N p denotes the number of particles and P is an N s × N p rectangular matrix where N s is the number of single particle states. Since U s (2Θ, 0) describes the propagation of non-interacting electrons in an external HS field, one may integrate out the fermionic degrees of freedom to obtain:where we have omitted the (∆τ ) 2 systematic error produced by the Trotter decomposition. In the above equation,

Ab initio calculation of the electronic properties of materials is a major challenge for solid-state theory. Whereas 40 years' experience has proven density-functional theory (DFT) in a suitable form, e.g. local approximation (LDA), to give a satisfactory description when electronic correlations are weak, materials with strongly correlated electrons, say d- or f-electrons, remain a challenge. Such materials often exhibit 'colossal' responses to small changes of external parameters such as pressure, temperature, and magnetic field, and are therefore most interesting for technical applications. Encouraged by the success of dynamical mean-field theory (DMFT) in dealing with model Hamiltonians for strongly correlated electron systems, physicists from the bandstructure and many-body communities have joined forces and developed a combined LDA+DMFT method for treating materials with strongly correlated electrons ab initio. As a function of increasing Coulomb correlations, this new approach yields a weakly correlated metal, a strongly correlated metal, or a Mott insulator. In this paper, we introduce the LDA+DMFT method by means of an example, LaMnO(3). Results for this material, including the 'colossal' magnetoresistance of doped manganites, are presented. We also discuss the advantages and disadvantages of the LDA+DMFT approach.

We consider the Kondo lattice model in two dimensions at half filling. In addition to the fermionic hopping integral t and the superexchange coupling J the role of a Coulomb repulsion U in the conduction band is investigated. We find the model to display a magnetic order-disorder transition in the U -J plane with a critical value of Jc which is decreasing as a function of U . The single particle spectral function A( k, ω) is computed across this transition. For all values of J > 0, and apart from shadow features present in the ordered state, A( k, ω) remains insensitive to the magnetic phase transition with the first low-energy hole states residing at momenta k = (±π, ±π). As J → 0 the model maps onto the Hubbard Hamiltonian. Only in this limit, the low-energy spectral weight at k = (±π, ±π) vanishes with first electron removalstates emerging at wave vectors on the magnetic Brillouin zone boundary. Thus, we conclude that (i) the local screening of impurity spins determines the low energy behavior of the spectral function and (ii) one cannot deform continuously the spectral function of the Mott-Hubbard insulator at J = 0 to that of the Kondo insulator at J > Jc. Our results are based on both, T = 0 Quantum Monte-Carlo simulations and a bond-operator mean-field theory.