Table of contents

The annual General Conference of the Condensed Matter Division of the European Physical Society has grown into a leading forum for discussion and exchange of the latest scientific developments. The conference series started seven years ago at the initiative of J T Devreese with the Antwerp Conference; it continued with the conferences in Manchester (1982), Lausanne (1983), The Hague (1984), Berlin (1985) and Stockholm (1986). The 7th General Conference was held in Pisa from 7 to 10 April 1987; it was attended by about 900 participants, coming from more than 30 countries of Europe and overseas.

The present volume, in two issues T19A and T19B, collects the papers orally presented at the Pisa Conference. The scientific programme contained 90 invited speakers, distributed in plenary lectures, symposia and individual invited talks. The presentation of the papers in this volume follows the pattern of the Conference as closely as possible.

The recent explosion of interest in High Temperature Superconductivity motivated, besides the plenary lecture in the programme, the last minute inclusion of an open symposium on this subject. More than forty research groups contributed to this event. The titles of the contributions and the affiliation of the speakers have been included at the end of volume 19B for the benefit of interested readers.

Numerous contributions (more than 500), ranging over the entire field of Condensed Matter Physics, were presented in poster sessions. Several poster contributions at the conference (as well as a number of the contributions to the Symposium on High Temperature Superconductivity) will appear as regular articles in Physica Scripta. These articles will later be collected in a Reprint Series of Physica Scripta.

We wish to thank the Conference Committees and the individual people who helped in the choice of such stimulating material, and all the contributors to this volume for their cooperation in the timely preparation of manuscripts. A special acknowledgement is also due to the Sponsors, whose generous help has made this manifestation possible.

Without abstract, see full text

Julia sets are strange repellers which arise from mapping problems involving analytic functions of comples variables. Those sets are multifractals and have a kind of scale invariance. Relationships between various methods of calculating the scaling properties of these sets are discussed.

The exceptional interest in the new class of oxide superconductors and the importance of these materials are pointed out. Concepts that led to their discovery are then exposed. The discovery itself and its early confirmation are summarized, including the work until the beginning of 1987.

In this paper I present the progress that have been made in the theory of spin glasses (and related problems) in last years; the physical meaning of replica symmetry breaking is explained and some of the open problems are briefly discussed.

Computational methods permit the detailed study of microscopic properties and their macroscopic consequences in a host of problems in physics which may be inaccessible to direct experimental study and too complex for theoretical analysis. Reliable calculations from first principles, however, require enormous computing resources. In this talk we describe how parallel computing can provide a practical and cost-effective solution to this problem, illustrating the key ideas with examples from Monte Carlo and molecular dynamics simulations, electronic structure calculations and the analysis of experimental data.

The inelastic scattering of helium atoms from single crystal metal surfaces has been developed into a reliable and accurate method for measuring surface phonon dispersion curves. Recent results on aluminium, the noble metals, the transition metals Ni, Pd and Pt and W(001) (the latter of which shows a soft mode leading to reconstruction) are discussed and compared with theory. Al(110) differs from a recent ab initio frozen phonon calculation but neither Al(110) nor Al(111) show significant anomalies, seen in all the other metals. This anomaly is attributed to a lateral softening of force constants in the first layer and the observed trends in the force constant suggest a strong connection to a reduction in sp-d hybridization at the surface.

The neutron scattering technique has been applied in earnest to condensed matter and materials research following the construction of reactor neutron sources in the mid-1940s. Today reactor sources are used to obtain results that pervade these subjects and applications in biology and physical chemistry are growing rapidly. The development of accelerator-based spallation sources, in which protons liberate neutrons from a heavy metal target, opens new frontiers through both the enhancement of available neutron flux and the inherent time structure of the beams. We survey the particular advantages of advanced pulsed sources with illustrations taken from recent work using the spallation source ISIS, sited at the Rutherford Appleton Laboratory. Topics of work include surface science, structure determination and molecular and neutron-electron spectroscopy. Finally we sketch the current development of new instrument technology and foreseeable applications.

Atomic-Force Microscopy (AFM) is a promising new method for studying the surface structure of both conductors and insulators. Up to now, atomic resolution has been achieved on graphite.

We present a method for interpreting scanning tunneling microscope (STM) images of surfaces, and apply this method to the Si(111)2 × 1 surface. We demonstrate how the STM does not directly image the atoms on a surface. Rather, the images reveal the spatial distribution of those electronic states lying near the Fermi-level. For semiconductors, a given wave-function is often preferentially localized on specific atoms or bonds. A lowest order interpretation of the images thus requires a description of the surface wave-functions, and a simple method for qualitatively constructing the wave-functions is illustrated here. Using this method, we demonstrate that STM images of the Si(111)2 × 1 surface are consistent with the π-bonded chain model for this surface, and inconsistent with the π-bonded molecular model and the buckled model.

The interaction between tip and flat surface in the STM and AFM is reviewed and a new, AC method to determine the absolute value of area of contact and interaction is presented. Evidence that the tip-surface interaction can be purely elastic down to the near single atom scale is obtained, and has been confirmed by TEM on Al₂O₃. Strengths close to that of an ideal lattice are observed. Applied to the Atomic Force Microscope this AC method will give operation at a controlled, drift free level of interaction. The effects of attractive (surface energy) forces in the STM geometry are discussed, and the equilibrium configurations and extent of inelastic processes for tip and flat are described. A new imaging mechanism in STM is proposed for the cases (e.g., graphite, point contact microscope) where significant tip-surface contact occurs.

Electrons in two dimensions in a high magnetic field can condense into an incompressible quantum fluid state when the electron density is equal to certain rational fractions of the flux density. Much progress has been made in understanding the nature of this state based on the ansatz wavefunction of Laughlin. Here, we show that the incompressible nature of the quantum fluid can be derived from a semiclassical evaluation of the imaginary time propagator. The dominant classical paths consist of cooperative ring exchange events. When these condense, they mediate a logarithmic interaction between density fluctuations, and produce a precise fractional quantization condition for the quasi-particle charges. We further show that the essential long wavelength properties of Laughlin's ground state and quasi-particle wavefunctions can be derived constructively from this approach. We conclude with some speculations concerning the nature of the off-diagonal order in the quantum Hall liquid.

A description is given of the occurrence of fractional quantum Hall effects, and how these are influenced by the presence of disorder. It is shown that when electrons are photoexcited into a GaAs-GaAlAs heterojunction, the fractional state existing at a Landau level occupancy of 7/5 can dominate over that at 4/3. Activation energy measurements of the resistivity show how the energy gap of the fractional states is reduced by the presence of disorder. A new method of analysis of the resistivity minima is presented which is used to give information on the correlation length of the many body ground state.

Quantized values of the Hall conductivity exist in the two-dimensional electron and hole gases of GaAs/AlGaAs heterostructures at frequencies as high as 30 GHz. This is a proof that a low-frequency breakdown does not occur and that the intrinsic plateau width of σ_xy is non-zero. The microwave measurement also allows to separate effects connected with the boundary of the 2D carrier gas; no significant contribution was observed.

In mesoscopic metal films (size smaller than the inelastic diffusion length for the electrons), ensemble averaging is incomplete. Direct interference between the electron waves gives rise to a sample specific magnetoresistance. The magnetoresistance of a mesoscopic loop oscillates with flux-period h/e. In series arrays of mesoscopic loops, the h/e amplitude is inversely proportional to the square-root of the number of loops. This is in agreement with the stochastic "self-averaging" of the direct interference in larger samples. In long metal cylinders, impurity averaging is complete and only h/2e oscillations caused by the weak electron localization, are present.

The conductance of small electronic devices is affected by random quantum interference of a surprisingly universal character. The quantum phase information of electron waves is not destroyed by elastic scattering, so that diffusing electrons interfere in random patterns determined by the specific microscopic configuration of scatterers throughout regions of size comparable to the inelastic diffusion length, typically of order 0.1-1 μm at liquid helium temperatures. The conductance of each such quantum subunit is changed by random amounts with an rms average deviation of just e²/h. In larger devices, the total effect can be determined by considering the series/parallel combination of these subunits. New interference patterns can be created by changing the magnetic flux through each subunit, or by changing the chemical potential (gate voltage). In our experiments, we have measured the conductance of patterned silicon inversion layers (MOSFETs) as small as 40 nm wide with voltage probe at spacings as small as 150 nm, demonstrating agreement with the "universal conductance fluctuation" theory over a wide range of device sizes, shapes, and conductivities. Measurements "inside" a single quantum subunit show that each voltage probe independently measures a fluctuating potential associated with the nonlocal response of the entire quantum subunit.

A discussion of the theoretical models trying to explain the semiconductor interfaces formation is presented. Metal-semiconductor and semiconductor-semiconductor junctions are discussed. While the metal-semiconductor interface is worse known due to the uncertainties in the chemisorption sites of the metal on the semiconductor surface, heterojunctions are better understood. We discuss the results obtained with different methods used to calculate heterojunctions band offsets: the induced density of interface states model (IDISM), the selfconsistent tight-binding model (SCTBM) and the selfconsistent local density method (SCLDM) are considered; all of them yield results in good agreement with the experimental evidence. The band offset accuracy for low ionicity semiconductors is higher in the SCLDM (better than 0.05 eV) and lower in the IDISM (better than 0.20 eV), while the SCTBM yields results with an intermediate accuracy, better than 0.1 eV.

The motion, the excitations and the dielectric response of carriers parallel to the two-dimensional electron layers of modulation doped heterojunctions and multi-quantum wells are characterised by collective excitations (density oscillations = plasmons) and single particle excitations (acceleration of single electrons). These excitations correspond to spin density and charge density fluctuations. They determine the response of layered electron systems to electromagnetic fields, such as screening of the Coulomb interaction, e.g., of impurities, or the conduction processes. We show experimentally, that a layered electron system with N layers has N discrete plasmon eigenmodes. Measuring these modes provides a method to study a range of interactions in the system: the Coulomb interaction between layers, and interaction of the plasmons with other excitations. The discrete layer plasmon modes fan out into a band due to the Coulomb interaction between electrons in different layers. We have measured the dispersion of the discrete layer plasmon eigenmodes as a function of the in-plane wave vector k_|| for modulation doped GaAs/AlGaAs quantum wells with a small number of layers (N = 2... 20). These measurements allow to test the large number of theoretical calculations of the plasmon dispersion in layered systems. They also provide a technique to characterise electron densities in layered electron gas systems. Thus measurement of the 3D plasmon frequency is currently used to determine the carrier concentration in doped bulk semiconductors. We have measured the wavevector and temperature dependence of the single particle excitation (SPE) spectra. The SPE spectra can be well fitted by a Lindhard-Mermin dielectric response function, and we determine the temperature dependence of the single particle relaxation time for our sample.

In semiconductor devices that are the backbone of modern technology the conducting electrons are often confined by field effect or doping near an interface potential barrier and form a quasi-two-dimensional electron system. High resolution lithography and refined etching techniques now make it possible to laterally confine such originally two-dimensional electron systems to narrow channels with widths comparable to the de-Broglie wavelengths of the electrons. The transition from two-dimensional to one-dimensional electronic behavior that occurs as lateral confinement lengths are lowered to about 100 nm is directly observed with infrared spectroscopy of the electronic excitations in laterally microstructured metal-oxide-semiconductor and semiconductor-heterojunctions devices.

The optical properties of one-side modulation-doped quantum wells are described and compared with theory in the Hartree approximation. The electron areal density is modified by voltage application or illumination and a measurement of the Band Gap Renormalization is obtained. The role of excited states, when populated, is emphasized.

After a brief summary of the key aspects of molecular beam epitaxy (MBE) of III-V compound semiconductors we present two examples for the application of MBE to modify the bulk properties of semiconductors through bandgap engineering in artificially layered structures. In both the GaAs/AlAs short-period and ultrathin-layer superlattices as well as in the monolayer or δ-doping in GaAs the concept of artificially layered semiconductor structures is scaled down to its ultimate physical limit normal to the crystal surface, as the constituent layers have a spatial extent of less than the lattice contant of the respective bulk material.

Superlattices, multiquantum and single quantum- well structures of IV-VI compounds are of interest because of their electronic properties as narrow gap materials with a many valley band structure. In addition, also their structural properties are quite unique and they offer the possibility to investigate structural phase transitions in artificially layered structures. Experimental data on PbTe/Pb_1-xGe_xTe are presented in this context. Electronic properties of superlattices are described within the frame work of the envelope function approximation and experimental data on interband magneto optical transitions in both geometries, namely B parallel and perpendicular to the layers, are presented. Coherent anti Stokes Raman scattering is used to obtain information on the electronic g-factors in PbTe/Pb_1-xSn_xTe (x = 0.175) MQW structures. The experimental data are analyzed with calculations of superlattice Landau states in both magnetic field configurations.

The Schottky-barrier heights ϕ_B of epitaxial NiSi₂ and CoSi₂ layers on Si(111) have been measured as a function of growth conditions. For type B NiSi₂ interfaces both I-V and photoelectric methods yield a high ϕ_B = 0.75 ± 0.03 eV for layers grown below 400°C, dropping below 0.7 eV for layers annealed above 550°C or grown by MBE above 550°C. ϕ_B of type A interfaces is always found to be low, i.e., ϕ_B = 0.62 ± 0.03 eV. Pinhole-free ultrathin (14-70 Å) CoSi₂ layers could be grown for the first time by using a novel solid phase epitaxy (SPE) technique, in which Co and Si are coevaporated in their stoichiometric ratio near room temperature (R.T.). These layers become perfectly epitaxial with type B orientation upon annealing to only 250°C and remain continuous up to T_s ≈ 600°C. From photoelectric measurements CoSi₂ layers grown at T_s < 350°C are found to exhibit a high ϕ_B = 0.75 ± 0.04 similar to B-type NiSi₂.

Thin epitaxial and continuous Si layers can be grown on CoSi₂ at T_s = 350°C. The prevalence of 2-dim. growth is established from the observation of the XPS Co 2p_3/2 core level peak intensity as a function of Si coverage. Kikuchi patterns observed by RHEED as well as high resolution TEM images from cross-sections show, that the orientation is the same as that of CoSi₂. CoSi₂/Si superstructures have been grown with individual layer thicknesses as small as 20 Å. Overgrowth of CoSi₂ by thick Si layers (> 1000 Å) requires ramping up of T_s to the usual growth temperatures for Si MBE. Without any optimisation preliminary experiments indicate, that single type growth of Si can be achieved. This is determined by ion channeling of an 1150 Å thick Si layer on 30 Å of CoSi₂ yielding minimum yields of 17% for the [111] and [110] directions and 40% for the [114] direction.

Investigation of the time evolution of extremely nonequilibrium charge carriers photoexcited into a semiconductor by a femtosecond laser gives valuable information about various scattering processes in the semiconductor. We describe here (i) an excite-and-probe experiment in GaAs quantum wells which measures the time evolution of the athermal distribution produced by the laser and thus gives information about carrier-carrier scattering processes and (ii) a luminescence experiment in GaAs and GaAs quantum wells which gives information about intervalley scattering of electrons in the conduction band.

In this paper I will review the experimental technique of hot electron spectroscopy. Using this measurement technique we have been able to measure the non-equilibrium distribution of hot electrons injected into GaAs. When hot electrons were injected into structures having narrow n-type transit region widths (< 850 Å) ballistic electron transport is prominent and in sample having wider transit regions (> 1700 Å) near diffusive transport occurs. These results can be explained by considering the interaction of the hot electrons with the coupled electron/phonon system in the transit region. In addition I will also discuss measurement of the interaction of hot electrons with an equilibrium hole gas.

Three topics on electron tunnelling through heterostructure barriers are examined. First, LO phonon-related oscillations under reverse bias in the J(V) and C(V) curves of single barrier n⁺GaAs/(AlGa)As/n^-GaAs/n⁺GaAs and n⁺(InGa)As/InP/n⁻(InGa)As/n⁺(InGa)As heterostructures are explained by the variation of the impedance of the n⁻ layers. For the GaAs/(AlGa)As/GaAs structure, a magneto-impurity resonance at 5T is observed in the oscillation amplitude, indicating that the impedance variation is due principally to impact ionisation of donors, rather than to LO phonon ionisation. The apparently anomalous behaviour of the magnetocapacitance in these devices in reverse bias is also accounted for. In the reverse-biased (InGa)As/InP/(InGa)As structures, the oscillations in J(V) have period ΔV = 33 meV, indicating that the electron energy relaxation occurs almost exclusively via Ga-As mode LO phonons of the n⁻(InGa)As layer.

Secondly, the oscillatory structure in the J(V) and C(V) characteristics of the forward biased (InGa)As/InP/(InGa)As heterostructures in the presence of a magnetic field (B||J) is investigated and used to determine the electron concentration in the accumulated two-dimensional electron gas (2DEG) as a function of forward bias voltage.

Thirdly, a new type of tunnelling phenomenon in forward-biased single-barrier n⁺(InGa)As/n⁻(InGa)As/InP/n⁺(InGa)As heterostructures is described for the case when a quantising magnetic field is applied in the plane of the InP barrier (J ⊥ B). The occurrence of two distinct series of resonances in the voltage- or field-dependence of the current is interpreted in terms of electron tunnelling from a 2DEG in the n⁻ layer into interfacial Landau states in the n⁺(InGa)As. These states correspond to classical skipping orbits of an electron along the tunnel barrier interface.

Through the most recent technological advances, it has been possible to get into the subpicosecond and submicron regimes, where hot electron effects are very important. The purpose of this paper is to show how the Monte Carlo technique can be successfully applied to the study of ultrafast phenomena in semiconductors. Two main topics will be dealt with:

(1) the energy relaxation of photoexcited electrons in bulk GaAs and GaAs-AlGaAs quantum wells. The dynamical evolution of the polar optical phonon distribution is followed in the simulation. It is found that the presence of hot-phonons greatly reduces the cooling rate of the photo-excited electrons, in qualitative agreement with available experimental results.

(2) The energy and momentum losses of hot electrons injected into a doped region, as found for example in the planar doped barrier (PDB) and in the tunneling hot electron transfer amplifier (THETA) devices. The interaction between the injected electrons and the background of cold carriers is shown to be a very effective channel of dissipation.

Recent advances in quantum well (QW) tunneling devices are discussed. These include resonant tunneling (RT) bipolar and field effect transistors; infrared lasers and detectors based on sequential RT and superlattice effective mass filters.

Multigigabit fiber-optic systems for telecommunication, computer interconnections, microwave RF links in phase-array radars, and neural networks are described. The latest developments in high-speed lasers, detectors, fibers, and integrated optoelectronics are briefly discussed.

This paper reviews some of the properties of quantum well lasers (QWLs), with emphasis on the two-dimensional origins of these. It is shown that two main effects determine the lowering of threshold current, namely the diminished density of states (DOS) (favourable factor) and the diminished optical confinement (unfavorable factor). The good operation of GaAs-GaAlAs QWLs also relies on more subtle effects such as the square 2D DOS, the enhanced optical matrix element, the high quantum efficiency... The poor operation of GaInAs based quantum well lasers is due to the detrimental Auger effect which is larger than in 3D lasers because of the higher carrier densities at which QWLs operate. Several other useful properties of QWLs in the performance (high-frequency, narrow-line) and manufacturing fields are discussed. Problems and advantages of 1 and 0D quantum-wire and quantum-box lasers are briefly evaluated.

A review of recent investigations on tunable far infrared semiconductor lasers on hot holes in Ge are given. This laser operates due to inverted hot hole distributions caused by the accumulation of carriers in crossed E and B fields in special regions of the momentum space. A radiative transition responsible for the simulated emission is the transition between subbands of light and heavy holes or between light hole Landau levels. Some additional possibilities to obtain inverted distributions are discussed.

Polydiacetylenes are conjugated polymers which can be obtained as macroscopic single crystals of high perfection. With these samples a detailed picture of the electronic states of the extended polymer chains has been built up. The conjugated chains have a large third-order hyperpolarisibility and, hence, potential for application in non-linear optical devices. A brief review of these properties of polydiacetylenes is presented. Recent studies of the non-linear optical properties of mixed crystals by electric field induced second harmonic (EFISH) generation are described. SHG signals observed after removal of the applied field are used to follow the decay of trapped charge. These results are compared with direct measurements of photocurrent decay in polydiacetylenes.

Solitons and polarons in conducting polymers are strongly coupled electron-lattice excitations. The lattice relaxation theory generalized by us to include the self-consistency of multi-electron states with lattice symmetry-breaking is summarized. The discrete symmetries and corresponding selection rules for both radiative and nonradiative processes are discussed. Theoretically calculated probability of nonradiative decay of an electron-hole pair into a soliton pair and that of electron (hole) into polaron as well as the probability of soliton pair photo-generation is compared with results of numerical and laboratory experiments. The resonance Raman scattering data of cis-polyacetylene are interpreted in terms of a bipolaron model. The parameters involved are determined directly from experimental data. Other applications of lattice relaxation theory to conducting polymers are briefly mentioned.

A simple model for describing the family of quasi-one-dimensional halogen-bridged metal complexes is presented which takes both electron-phonon and electron-electron interactions into account. A competition between chemical bonding and charge disproportionation occurs in these MX chains. The charge-density wave ground state appears to be the rule, but minor modifications, e.g., replacing chlorine by iodine, can strongly change its character from a large amplitude mixed valence limit to that of a small amplitude Peierls distortion. Correspondingly, the polaronic defect states which are very localized in one regime become quite extended in the other. Experimental data for two specific materials are analysed in terms of the two limiting cases.

Finally the recently discovered quasi-two-dimensional high-T_c superconductors and the related three-dimensional material BaPb_1-xBi_xO₃ are considered as interpenetrating MX chains. It is proposed that in these compounds the oxygen plays a similar role as the halogen in the MX chain in delocalizing the electronic states at the Fermi energy.

Many of the highly anomalous properties of heavy electron metal can be explained by assuming that to gain hybridization energy a small number of electrons are promoted from an f-level well below the Fermi energy. The f-states form an almost localized Fermi liquid with a very large effective mass due to the many low energy spin fluctuations of the f-states. Spin-fluctuation mediated interactions generally favor unconventional rather than s-wave singlet superconductivity. A sure sign of an unconventional form of superconductivity would be to observe an anisotropy in the electronic or lattice properties.

Local heavy fermions are formed at the sites of independent Kondo ions as substantiated by the discovery of a pronounced low temperature anomaly in the thermal expansion of certain dilute and moderately concentrated Ce alloys. In periodic Kondo lattice systems like CeAl₃ a coherent heavy fermion band structure is developed at low T, one hallmark of which is a maximum in the Sommerfeld coefficient γ(T) near 0.4 K. This distinct phenomenon can even show up in the presence of long-range magnetic order as in CeCu₂Ge₂ and is found to have a strong influence on the superconducting properties of CeCu₂Si₂. According to its normal-state behavior, UBe₁₃ appears to be closely related to the Ce-based Kondo lattices. A number of the superconducting properties of pure and Th-doped UBe₁₃ can be well understood in terms of an anisotropic pairing interaction, which seems to track the anisotropy of the Fermi surface. This leads to a reinterpretation of the phase diagram of the superconducting states in U_1-xTh_xBe₁₃.

The tetragonal (1-2-2) intermetallic compound URu₂Si₂ represents the only heavy-electron material [γ(T > 18 K) ≈ 180 mJ/mol · K²] in which both antiferromagnetic (T_N ≃ 17.5 K) and superconducting (T_c ≃ 1.2 K) transitions occur. At low temperatures (T < T_c) there is a coexistence of these two ordered states. The purpose of this paper is to present the basic experimental features of URu₂Si₂ including the results of specific-heat, magnetization, susceptibility, critical-field, high-field magnetization, (magneto-) resistivity, pressure dependence, thermal expansion, neutron-scattering, NMR, Hall-effect, electron spectroscopy, and μSR. These properties suggest that the antiferromagnetic transition, i.e., the alternate plane order of unusually small 0.03μ_B/U moments along the c-axis, opens up an energy gap on some portions of the Fermi surface. The remaining nongapped electrons are able at a lower temperature to become superconducting. Hence it would seem that both the magnetism and the superconductivity are carried by the same hybridized 5f-U electrons.

Recent neutron diffraction experiments have shown structure at the intensity level 10^-5 very close to the 110 b.c.c. Bragg point in metallic potassium. Three possible explanations of these data have been proposed: (1) Charge Density Waves. (2) Martensitic Embyros. (3) Spectrometer resolution/double-scattering events.

These alternative interpretations are discussed.

Spectroscopic and thermodynamic properties of some 4f and 5f compounds are studied using the Anderson impurity model. The large degeneracy of the f-level allows us to treat 1/N_f as a small parameter. Exact results are obtained for N_f = ∞ and T = 0, and even for N_f = 1 or 2 accurate results are obtained if corrections of order 1/N_f are included. Comparison with experimental core level photoemission spectra determines the parameters of the model. With these parameters the model has been used to study other electron spectroscopies, such as M_4,5XAS, 4f PES and BIS, for Ce compounds. The model gives a fairly good description of these spectroscopies. The values of the parameters place the known Ce intermetallics in the Kondo limit, where the 4f spectral function has a narrow resonance at the Fermi energy, which has been seen spectroscopically. This narrow resonance is related to a large density of low-lying excitations, which explains the large low temperature susceptibility and specific heat. The model has also been applied to the 4f photoemission spectra of some Pr and Nd compounds. The agreement with experiment is satisfactory. Finally the model has been used to study the 4f core spectrum of some actinide dioxides. This determines the importance of the hybridization between the 5f state and the O 2p states, and provides an interpretation of a satellite as well as the photon energy dependence of the spectrum close to the 3d → 5f resonance.

We compare a variety of approximation schemes, all predicting the excitation structure of a semiconductor. All schemes are placed within one common framework, in which the electron self energy is the central quantity. In arriving at this self energy it turns out to be essential to screen the Coulomb interaction between electrons both statically and dynamically. We subsequently discuss the Hartree, Hartree-Fock, local density functional and Slater Xα schemes, which all lead to incorrect excitation structures, and the GW scheme of Hedin which, in view of recent successes, appears to be a very promising scheme.

The purpose of this article is twofold: to define briefly the state of the art of theoretical approaches to electronic, structural and thermal properties of small aggregates of metals and semiconductor elements, and to present new (sometimes preliminary) results of computer simulations on silicon, sodium and gold clusters. In particular, these calculations use the unified-molecular-dynamics-density-functional (Car-Parrinello) method for clusters of very small size (Si_N and Na_N with N up to 10), and effective many-body potentials for the gold clusters of larger size (N up to 1000). The connection between experiments and theory is discussed critically.

Physical and chemical trends of the properties of semiconductors are studied on the basis of first-principles electronic structure calculations. Total-energy calculations for compounds under pressure and in various (hypothetical) crystal structures are performed within the local-density approximation. The self-consistent band structures are calculated by means of the LMTO method. By transformation of the orbitals to an orthogonal basis sp³ bond orders are derived, and a transformation to a tight-binding basis allows the derivation of ionicities, polarities, metallicities etc from the first-principles potential parameters. The theoretical ionicities relate fairly well to the empirical Phillips scale, and it is shown that a critical value close to 0.8 separates the fourfold and sixfold coordinated crystal structures. Optical-phonon and "absolute" hydrostatic deformation potentials are derived, and a simple scheme is suggested for the calculation of band lineups in semiconductor heterostructures. The band offsets are derived from the assumption of a partial alignment of the "dielectric midgap energy" (DME) levels of the two compound semiconductors. The DME is related to the charge-neutrality points introduced by Tejedor and Flores, and by Tersoff.

The progress of Density Functional (DF) calculations of optical, electrical and structural properties of deep defects in semiconductors has been remarkable over the past ten years. Time seems ripe for the extension of such a theory to finite temperature thermodynamical properties (notably free energies of formation and migration). This paper is a step towards that direction. The quasi-harmonic theory of Lattice Dynamics is proposed as a good candidate for such an extension. One important difficulty is the numerical effort required to extract the full dynamical matrix vs. volume from standard DF band calculations for large systems (30-100 atoms/cell), which are on the other hand needed for a thermodynamically consistent theory. To overcome this serious problem we point out some exact relationships existing between differential properties of trajectories and force constants, which may be used to extract accurate dynamical matrices from low-temperature Molecular Dynamics runs. The formalism is presented and successfully tested for a 32-atom argon crystal with and without a vacancy, for which exact results are available. Such a new formalism is proposed as a powerful tool to evaluate dynamical matrices, and thus phonon spectra and thermodynamical properties, of low-symmetry semiconductor systems like defects and amorphous materials, for which a few Carr-Parrinello Molecular Dynamics trajectories are much cheaper than the huge number of self-consistent calculations traditionally required by frozen-phonon or force-constant techniques.

Recent ideas to extend the scope and applicability of large-scale computer simulation of condensed phases are discussed. These include (i) the use of simulated annealing and related methods in first-principles calculations and (ii) the development "effective-medium" and similar approximate approaches to interatomic interactions in low-symmetry situations. Examples of applications to molecular dynamics simulations are presented.