Search

We present a method for calculating the electronic structure of correlated materials based on a truly first-principles LDA+U scheme. Recently we suggested how to calculate U from first-principles, using a method which we named constrained RPA (cRPA). The input is simply the Kohn-Sham eigenfunctions and eigenvalues obtained within the LDA. In our proposed self-consistent LDA+U scheme, we calculate the LDA+U eigenfunctions and eigenvalues and use these to extract U. The updated U is then used in the next iteration to obtain a new set of eigenfunctions and eigenvalues and the iteration is continued until convergence is achieved. The most significant result is that our numerical approach is indeed stable: it is possible to find the effective exchange and correlation interaction matrix in a self-consistent way, resulting in a significant improvement over the LDA results, regarding both the bandgap in NiO and the f-band exchange spin-splitting in Gd, but some discrepancies still remain., Comment: 11 pages, 4 figures

The metal to insulator transition in the charge transfer NiS{2-x}Se{x} compound has been investigated through infrared reflectivity. Measurements performed by applying pressure to pure NiS2 (lattice contraction) and by Se-alloying (lattice expansion) reveal that in both cases an anomalous metallic state is obtained. We find that optical results are not compatible with the linear Se-alloying vs Pressure scaling relation previously established through transport, thus pointing out the substantially different microscopic origin of the two transitions., Comment: Accepted for publication in Phys. Rev. B

Using a combined local density functional theory (LDA-DFT) and quantum Monte Carlo (QMC) dynamic cluster approximation approach, the parameter dependence of the superconducting transition temperature Tc of several single-layer hole-doped cuprate superconductors with experimentally very different Tcmax is investigated. The parameters of two different three-band Hubbard models are obtained using the LDA and the downfolding Nth-order muffin-tin orbital technique with N=0 and 1 respectively. QMC calculations on 4-site clusters show that the d-wave transition temperature Tc depends sensitively on the parameters. While the N=1 MTO basis set which reproduces all three $pd\sigma$ bands leads to a d-wave transition, the N=0 set which merely reproduces the LDA Fermi surface and velocities does not.

The outstanding discrepancy between the measured and calculated (local-density approximation) Fermi surfaces in the well-characterized, paramagnetic Fermi liquid Sr2RhO4 is resolved by including the spin-orbit coupling and Coulomb repulsion. This results in an effective spin-orbit coupling constant enhanced 2.15 times over the bare value. A simple formalism allows discussion of other systems. For Sr2RhO4, the experimental specific-heat and mass enhancements are found to be 2.2., Comment: 4 pages, 2 figures

Due to the orthorhombic distortion of the lattice, the electronic hopping integrals along the $a$ and $b$ diagonals, the orthorhombic directions, are slightly different. We calculate their difference in the LDA and find $t_{a}^{\prime}-t_{b}^{\prime}\approx 8 $meV. We argue that electron correlations in the insulating phase of La$_{2-x}$Sr$_{x}$CuO$_{4}$, i. e. at doping $x\leq 0.055,$ dramatically enhance the $(t_{a}^{\prime}-t_{b}^{\prime}) $-splitting between the $a$- and $b$-hole valleys. In particular, we predict that the intensity of both angle-resolved photoemission and of optical absorption is very different for the $a$ and $b$ nodal points.

We present the results of a crystal structure determination using neutron powder diffraction as well as the superconducting properties of the rare-earth sesquicarbide La2C3 (Tc ~ 13.4 K) by means of specific heat and upper critical field measurements. From the detailed analysis of the specific heat and a comparison with ab-initio electronic structure calculations, a quantitative estimate of the electron-phonon coupling strength and the logarithmic average phonon frequency is made. The electron-phonon coupling constant is determined to \lambda ~ 1.35. The electron-phonon coupling to low energy phonon modes is found to be the leading mechanism for the superconductivity. Our results suggest that La2C3 is in the strong coupling regime, and the relevant phonon modes are La-related rather than C-C stretching modes. The upper critical field shows a clear enhancement with respect to the Werthamer-Helfand-Hohenberg prediction, consistent with strong electron-phonon coupling. Possible effects on the superconducting properties due to the noncentrosymmetry of the crystal structure are discussed., Comment: 12 pages, 11 figures, submitted to Phys. Rev. B

An undoped cuprate with apical fluorine and inner (i) and outer (o) CuO2-layers is a 60 K superconductor whose Fermi surface (FS) has large n- and p-doped sheets with the SC gap on the n-sheet twice that on the p -sheet (Y. Chen et al.). The Fermi surface is not reproduced by the LDA, but the screening must be substantially reduced due to electronic correlations, and oxygen in the o-layers must be allowed to dimple outwards. This charges the i-layers by 0.01|e|, causes an 0.4 eV Madelung-potential difference between the i and o -layers, quenches the i-o hopping, and localizes the n-sheets onto the i-layers, thus protecting their d-wave pairs from being broken by scattering on impurities in the BaF layers. The correlation-reduced screening strengthens the coupling to z-axis phonons., Comment: 4 pages, 3 figures

The Hubbard \emph{U} of the \emph{3d} transition metal series as well as SrVO$_{3}$, YTiO$_{3}$, Ce and Gd has been estimated using a recently proposed scheme based on the random-phase approximation. The values obtained are generally in good accord with the values often used in model calculations but for some cases the estimated values are somewhat smaller than those used in the literature. We have also calculated the frequency-dependent \emph{U} for some of the materials. The strong frequency dependence of \emph{U} in some of the cases considered in this paper suggests that the static value of \emph{U} may not be the most appropriate one to use in model calculations. We have also made comparison with the constrained LDA method and found some discrepancies in a number of cases. We emphasize that our scheme and the constrained LDA method theoretically ought to give similar results and the discrepancies may be attributed to technical difficulties in performing calculations based on currently implemented constrained LDA schemes., Comment: 24 pages, 13 figures; Submitted to Phys. Rev. B

Using the full potential linear muffin-tin orbitals (FP-LMTO) method we examine the pressure-dependence of superconductivity in the two metallic phases of Boron: bct and fcc. Linear response calculations are carried out to examine the phonon frequencies and electron-phonon coupling for various lattice parameters, and superconducting transition temperatures are obtained from the Eliashberg equation. In both bct and fcc phases the superconducting transition temperature T$_c$ is found to decrease with increasing pressure, due to stiffening of phonons with an accompanying decrease in electron-phonon coupling. This is in contrast to a recent report, where T$_c$ is found to increase with pressure. Even more drastic is the difference between the measured T$_c$, in the range 4-11 K, and the calculated values for both bct and fcc phases, in the range 60-100 K. The calculation reveals that the transition from the fcc to bct phase, as a result of increasing volume or decreasing pressure, is caused by the softening of the X-point transverse phonons. This phonon softening also causes large electron-phonon coupling for high volumes in the fcc phase, resulting in coupling constants in excess of 2.5 and T$_c$ nearing 100 K. We discuss possible causes as to why the experiment might have revealed T$_c$'s much lower than what is suggested by the present study. The main assertion of this paper is that the possibility of high T$_c$, in excess of 50 K, in high pressure pure metallic phases of boron cannot be ruled out, thus substantiating the need for further experimental investigations of the superconducting properties of high pressure pure phases of boron., Comment: 16 pages, 8 figures, 1 Table

Choi et al. [Phys. Rev. B 66, 020513 (2002)] recently presented first principles calculations of the electron-phonon coupling and superconductivity in MgB2, emphasizing the importance of anisotropy and anharmonicity. We point out that (1) variation of the superconducting gap inside the sigma- or the pi-bands can hardly be observed in real samples, and (2) taking the anisotropy of the Coulomb repulsion into account influences the size of the small gap, Delta_pi., Comment: 3 pages, 2 color figures

A recent experiment by Shimizu et al. has provided evidence of a superconducting phase in hcp Fe under pressure. To study the pressure-dependence of this superconducting phase we have calculated the phonon frequencies and the electron-phonon coupling in hcp Fe as a function of the lattice parameter, using the linear response (LR) scheme and the full potential linear muffin-tin orbital (FP-LMTO) method. Calculated phonon spectra and the Eliashberg functions $\alpha^2 F$ indicate that conventional s-wave electron-phonon coupling can definitely account for the appearance of the superconducting phase in hcp Fe. However, the observed change in the transition temperature with increasing pressure is far too rapid compared with the calculated results. For comparison with the linear response results, we have computed the electron-phonon coupling also by using the rigid muffin-tin (RMT) approximation. From both the LR and the RMT results it appears that electron-phonon interaction alone cannot explain the small range of volume over which superconductivity is observed. It is shown that ferromagnetic/antiferromagnetic spin fluctuations as well as scattering from magnetic impurities (spin-ordered clusters) can account for the observed values of the transition temperatures but cannot substantially improve the agreeemnt between the calculated and observed presure/volume range of the superconducting phase. A simplified treatment of p-wave pairing leads to extremely small ($\leq 10^{-2}$ K) transition temperatures. Thus our calculations seem to rule out both $s$- and $p$- wave superconductivity in hcp Fe., Comment: 12 pages, submitted to PRB

The results of first principles calculations of the electronic structure and the electron-phonon interaction in MgB2 are used to study theoretically the temperature dependence and anisotropy of the magnetic field penetration depth. The effects of impurity scattering are essential for a proper description of the experimental results. We compare our results with experimental data and we argue that the two-band model describes the data rather well., Comment: submitted to Phys. Rev. B

A large number of experimental facts and theoretical arguments favor a two-gap model for superconductivity in MgB$_{2}$. However, this model predicts strong suppression of the critical temperature by interband impurity scattering and, presumably, a strong correlation between the critical temperature and the residual resistivity. No such correlation has been observed. We argue that this fact can be understood if the band disparity of the electronic structure is taken into account, not only in the superconducting state, but also in normal transport.

The heat capacity anomaly at the transition to superconductivity of the layered superconductor MgB$_2$ is compared to first-principles calculations with the Coulomb repulsion, $\mu^\ast$, as the only parameter which is fixed to give the measured $T_c$. We solve the Eliashberg equations for both an isotropic one-band and a two-band model with different superconducting gaps on the $\pi$ and $\sigma$ Fermi surfaces. The agreement with experiments is considerably better for the two-band model than for the one-band model., Comment: final published version

A theoretical model for quasiparticle and Josephson tunneling in multiband superconductors is developed and applied to MgB2-based junctions. The gap functions in different bands in MgB2 are obtained from an extended Eliashberg formalism, using the results of band structure calculations. The temperature and angle dependencies of MgB2 tunneling spectra and the Josephson critical current are calculated. The conditions for observing one or two gaps are given. We argue that the model may help to settle the current debate concerning two-band superconductivity in MgB2., Comment: minor corrections, published in Phys. Rev. B 65, 180517(R) (2002)

For the 40K-superconductor MgB2 we have calculated the electronic and phononic structures and the electron-phonon interaction throughout the Brillouin zone ab initio. In contrast to the isoelectronic graphite, MgB2 has holes in the bonding sigma-bands, which contribute 42 per cent to the density of states: N(0) =0.355 states/(MgB2 eV spin). The total interaction strength, lambda =0.87 and lambda,tr=0.60, is dominated by the coupling of the sigma-holes to the bond-stretching optical phonons with wavenumbers in a narrow range around 590 cm^{-1}. Like the holes, these phonons are quasi two-dimensional and have wave-vectors close to Gamma-A, where their symmetry is E. The pi-electrons contribute merely 0.25 to lambda and to lambda,tr. With Eliashberg theory we evaluate the normal-state resistivity, the density of states in the superconductor, and the B-isotope effect on Tc and Delta0, and find excellent agreement with experiments, when available. Tc=40 K is reproduced with mu*=0.10 and 2Delta0/kB Tc=3.9. MgB2 thus seems to be an intermediate-coupling e-ph pairing s-wave superconductor., Comment: 5 pages, 3 ps figures, Accepted for Phys. Rev. B Rapid Commun

By calculation and analysis of the bare conduction bands in a large number of hole-doped high-temperature superconductors, we have identified the energy of the so-called axial-orbital as the essential, material-dependent parameter. It is uniquely related to the range of the intra-layer hopping. It controls the Cu 4s-character, influences the perpendicular hopping, and correlates with the observed Tc at optimal doping. We explain its dependence on chemical composition and structure, and present a generic tight-binding model., Comment: 5 pages, Latex, 5 eps figures

We have studied the electronic structure of two Chevrel-phase compounds, Mo$_6$Se$_{7.5}$ and Sn$_{1.2}$Mo$_6$Se$_{7.5}$, by combining photoemission spectroscopy and band-structure calculations. Core-level spectra taken with x-ray photoemission spectroscopy show systematic core-level shifts, which do not obey a simple rigid-band model. The inverse photoemission spectra imply the existence of an energy gap located $\sim 1$ eV above the Fermi level, which is a characteristic feature of the electronic structure of the Chevrel compounds. Quantitative comparison between the photoemission spectra and the band-structure calculations have been made. While good agreement between theory and experiment in the wide energy range was obtained as already reported in previous studies, we found that the high density of states near the Fermi level predicted theoretically due to the Van Hove singularity is considerably reduced in the experimental spectra taken with higher energy resolution than in the previous reports. Possible origins are proposed to explain this observation., Comment: 8 pages, 5 figures

Under pressure cesium undergoes a transition from a high-pressure fcc phase (Cs-II) to a collapsed fcc phase (Cs-III) near 4.2GPa. At 4.4GPa there follows a transition to the tetragonal Cs-IV phase. In order to investigate the lattice vibrations in the fcc phase and seek a possible dynamical instability of the lattice, the phonon spectra of fcc-Cs at volumes near the III-IV transition are calculated using Savrasov's density functional linear-response LMTO method. Compared with quasiharmonic model calculations including non-central interatomic forces up to second neighbours, at the volume $V/V_0= 0.44$ ($V_0$ is the experimental volume of bcc-Cs with $a_0$=6.048{\AA}), the linear-response calculations show soft intermediate wavelength $T_{[1\bar{1}0]}[{\xi}{\xi}0]$ phonons. Similar softening is also observed for short wavelength $L[\xi\xi\xi]$ and $L[00\xi]$ phonons and intermediate wavelength $L[\xi\xi\xi]$ phonons. The Born-von K\'{a}rm\'{a}n analysis of dispersion curves indicates that the interplanar force constants exhibit oscillating behaviours against plane spacing $n$ and the large softening of intermediate wavelength $T_{[1\bar{1}0]}[{\xi}{\xi}0]$ phonons results from a negative (110)-interplanar force-constant $\Phi_{n=2}$. The frequencies of the $T_{[1\bar{1}0]}[{\xi}{\xi}0]$ phonons with $\xi$ around 1/3 become imaginary and the fcc structure becomes dynamically unstable for volumes below $0.41V_0$. It is suggested that superstructures corresponding to the $\mathbf{q}{\neq}0$ soft mode should be present as a precursor of tetragonal Cs-IV structure., Comment: 12 pages, 5 figures

We review the simple linear muffin-tin orbital method in the atomic-spheres approximation and a tight-binding representation (TB-LMTO-ASA method), and show how it can be generalized to an accurate and robust Nth order muffin-tin orbital (NMTO) method without increasing the size of the basis set and without complicating the formalism. On the contrary, downfolding is now more efficient and the formalism is simpler and closer to that of screened multiple-scattering theory. The NMTO method allows one to solve the single-electron Schroedinger equation for a MT-potential -in which the MT-wells may overlap- using basis sets which are arbitrarily minimal. The substantial increase in accuracy over the LMTO-ASA method is achieved by substitution of the energy-dependent partial waves by so-called kinked partial waves, which have tails attached to them, and by using these kinked partial waves at N+1 arbitrary energies to construct the set of NMTOs. For N=1 and the two energies chosen infinitesimally close, the NMTOs are simply the 3rd-generation LMTOs. Increasing N, widens the energy window, inside which accurate results are obtained, and increases the range of the orbitals, but it does not increase the size of the basis set and therefore does not change the number of bands obtained. The price for reducing the size of the basis set through downfolding, is a reduction in the number of bands accounted for and -unless N is increased- a narrowing of the energy window inside which these bands are accurate. A method for obtaining orthonormal NMTO sets is given and several applications are presented., Comment: 85 pages, Latex2e, Springer style, to be published in: Lecture notes in Physics, edited by H. Dreysse, (Springer Verlag)

We have studied the Cu core level photoemission spectra in the Anderson impurity model for cuprates with different Cu-O networks, dimensionalities (zero, one, two and three) and Cu valencies (two and three). We focus on the shape of the leading peak and obtain very good agreement with the experimental data. We show how the shape of the spectrum is related to the valence electronic structure and the Cu-O network but also that other atoms can play a role., Comment: 5 pages, latex2e, 2 eps figure, accepted for publ. in PRL

We describe the screened Korringa-Kohn-Rostoker (KKR) method and the third-generation linear muffin-tin orbital (LMTO) method for solving the single-particle Schroedinger equation for a MT potential. The simple and popular formalism which previously resulted from the atomic-spheres approximation (ASA) now holds in general, that is, it includes downfolding and the combined correction. Downfolding to few-orbital, possibly short-ranged, low-energy, and possibly orthonormal Hamiltonians now works exceedingly well, as is demonstrated for a high-temperature superconductor. First-principles sp3 and sp3d5 TB Hamiltonians for the valence and lowest conduction bands of silicon are derived. Finally, we prove that the new method treats overlap of the potential wells correctly to leading order and we demonstrate how this can be exploited to get rid of the empty spheres in the diamond structure., Comment: latex2e, 32 printed pages, Postscript figs, to be published in: Tight-Binding Approach to Computational Materials Science, MRS Symposia Proceedings No. 491 (MRS, Pittsburgh, 1998)

We have investigated whether the electron-phonon interaction can support a d-wave gap-anisotropy. On the basis of models derived from LDA calculations, as well as LDA linear-response calculations we argue that this is the case, for materials with buckled or dimpled CuO2 planes, for the so-called buckling modes, which involve out-of-plane movements of the plane oxygens., Comment: 5pages, Latex2e, 6 Postscript figures

The equilibrium structure, energy bands, phonon dispersions, and s- and d-channel electron-phonon interactions (EPIs) are calculated for the infinite-layer superconductor CaCuO2 doped with 0.24 holes per CuO2. The LDA and the linear-response full-potential LMTO method were used. In the equilibrium structure, oxygen is found to buckle slightly out of the plane and, as a result, the characters of the energy bands near EF are found to be similar to those of other optimally doped HTSCs. For the EPI we find lambda(s)=0.4, in accord with previous LDA calculations for YBa2Cu3O7. This supports the common belief that the EPI mechanism alone is insufficient to explain HTSC. Lambda(x^2-y^2) is found to be positive and nearly as large as lambda(s). This is surprising and indicates that the EPI could enhance some other d-wave pairing mechanism. Like in YBa2Cu3O7, the buckling modes contribute significantly to the EPI, although these contributions are proportional to the static buckling and would vanish for flat planes. These numerical results can be understood from a generic tight-binding model originally derived from the LDA bands of YBa2Cu3O7. In the future, the role of anharmonicity of the buckling-modes and the influence of the spin-fluctuations should be investigated., Comment: 19 pages, 9 Postscript figures, Latex

Recent experimental results on two hcp phases of barium under high pressure show interesting variation of the lattice parameters. They are here interpreted in terms of electronic structure calculation by using the LMTO method and generalized pseudopotential theory (GPT) with a NFE-TBB approach. In phase II the dramatic drop in c/a is an instability analogous to that in the group II metals but with the transfer of s to d electrons playing a crucial role in Ba. Meanwhile in phase V, the instability decrease a lot due to the core repulsion at very high pressure. PACS numbers: 62.50+p, 61.66Bi, 71.15.Ap, 71.15Hx, 71.15La, Comment: 29 pages, 8 figures

We describe the LDA bandstructure of YBa_2Cu_3O_7 in the 2 eV range from the Fermi energy using orbital projections and compare with YBa_2Cu_4O_8. Then, the high-energy and chain-related degrees of freedom are integrated out and we arrive at two, nearest-neighbor, orthogonal, two-center, 8-band Hamiltonians, the even and odd bands of the bi-layer. Of the 8 orbitals, Cu{x2-y2}, O2x, O3y, and Cus have \sigma character and Cu{xz}, Cu{yz} O2z, and O3z have \pi character. The roles of the Cu_s orbital, which has some Cu{3z2-1} character, and the four \pi orbitals are as follows: Cu_s provides 2nd- and 3rd-nearest-neighbor (t' and t') intra-plane hopping, as well as hopping between planes (t_{perp}). The \pi -orbitals are responsible for bifurcation of the saddle-points for dimpled planes. The 4-\sigma-band Hamiltonian is generic for flat CuO_2 planes and we use it for analytical studies. The reduction of the \sigma-Hamiltonian to 3- and 1-band Hamiltonians is explicitly discussed and we point out that, in addition to the hoppings commonly included in many-body calculations, the 3-band Hamiltonian should include hopping between all 2nd-nearest-neighbor oxygens and that the 1-band Hamiltonian should include 3rd-nearest-neighbor hoppings. We calculate the single-particle hopping between the planes of a bi-layer. We show that the inclusion of t' is crucial for understanding ARPES for the anti-ferromagnetic insulator Sr_2CuO_2Cl_2. Finally, we estimate the value of the inter-plane exchange constant for an un-doped bi-layer in mean-field theory using different single-particle Hamiltonians., Comment: Stanford Superconductor Conference - March 1995. 17 pages (Postscript), 5 Postscript figures (3 figures left) The whole file (5.3 Mb) one could get via anonymous ftp: ftp://radix1.mpi-stuttgart.mpg.de/pub/outgouing/lowham.uu