Table of contents

Using the lattice gas model and the transition-type-dependent Monte Carlo method, we calculated the chemical surface diffusion coefficients on a stepped surface. We assume that the step exerts an attraction or repulsion on adsorbed particles (adparticles) that occupy the up or down step sites, but no interactions between adparticles. Two kinds of activation energy, calculated from the harmonic potential and from the difference between the saddle-point and single-site energy, are used in our calculation. The calculated results show that perpendicular diffusion decreases greatly with increase in step repulsion and attraction at all coverages. However, for diffusion parallel to the steps, completely different results are obtained for these two calculation methods. If the energy barrier is calculated by the harmonic potential, diffusion parallel to the steps is both coverage and step independent. If the energy barrier is calculated by the second method, diffusion parallel to the steps is greatly enhanced with increase in the step repulsion and attraction at middle or high coverages but decreases slightly at low coverages. The calculated results explain the chemical diffusion anisotropy on a stepped surface. The results also show that the popular harmonic potential method may be not suitable for explaining the experiments where diffusion along step edges may be more rapid than on a flat surface.

The static relaxation and the lattice dynamics of rectangular edges of quadratic bars of four rock-salt-structured alkali halides is calculated using a shell model. The relaxation includes a large-scale contraction of the bar as well as tip-localized microscopic static displacements. The phonon spectra of relaxed quadratic bars are presented and modes localized at the edges are identified.

Observations of both nanoscale and microscale images were carried out on a NaCl(001) surface in water using atomic force microscopy (AFM). The observed higher density of steps and the observed convex and concave steps in water result from the forward motion of monatomic steps and crystal growth of the so-called step-flow mode. The non-equilibrium structure having a unit cell of structure has been observed for the first time on terraces. Since every on terraces is located at the same height as , each as well as each have been observed as protrusions by AFM. The metastable configuration at the interface between the NaCl(001) surface and water was transformed, at steps, into the equilibrium structure of NaCl(c), about 3 h after the NaCl(001) surface had been covered with its own saturated water. After about 30 min, however, still remains at the same height as at steps, or sodium ions leave the surface at steps. The reason why the forward motion of monatomic steps takes place at the interface between the NaCl(001) surface and water is the decrease of about 0.05 nm in the distance of like nearest neighbours in the non-equilibrium structure having a unit cell of structure on terraces after its structural transformation to the equilibrium structure of NaCl(c) with the aid of water molecules.

The structure and growth of ultrathin Au and Pd films on Cu(001) have been studied by low-energy ion scattering (LEIS) and low-energy electron diffraction (LEED). Due to the existence of a kinetic pathway for intermixing, both Au and Pd are incorporated into the Cu(001) surface at room temperature, forming a surface alloy at a coverage of 0.5 monolayer (ML). At Au coverage near 1.2 ML, the surface layer is pure Au; this is caused by Au-induced segregation from the underlying layer to the surface. No second- and/or third-layer Au is evident. In contrast to this, for the Pd/Cu(001) case a clock reconstruction is formed after deposition of 1 ML Pd. The LEIS data, through comparisons with 3-D computer simulation, reveal a 0.25 Å lateral clockwise - anticlockwise displacement of the first-layer Pd atoms. A simple model for this clock reconstruction is proposed. The driving force behind the dealloying for the Cu - Au alloy and Pd-induced reconstructing for the Cu - Pd surface is also discussed.

A new type of collective electromagnetic excitation, namely surface polaritons (SP) - in a 2D electronic layer in a high magnetic field, is predicted. We have found the spectrum, damping, and polarization of the SP over a wide range of frequencies and wavevectors k. It is shown that near the cyclotron resonance (CR) the phase velocity of the SP is drastically slowed down and the group velocity undergoes fundamental steps defined by the fine-structure constant . In the vicinity of a CR subharmonic , negative (anomalous) dispersion of the SP occurs. The relaxation of electrons in the 2D layer gives rise to a new dissipative collective threshold-type mode of the SP. We suggest a method for calculating the kinetic coefficients for the 2D electronic layer in a high magnetic field, using the Wigner distribution function formalism, and determining their spatial and frequency dispersions. Using this method we have calculated the lineshapes of the CR, which are in good agreement with experimental data.

Electron-correlation effects at metal surfaces can be studied qualitatively within the framework of the Hubbard model for a semi-infinite lattice. We propose a moment approach for an approximate determination of the local electronic self-energy that is applicable for systems with reduced translational symmetry. For the strong-correlation regime a one-pole ansatz for the self-energy can be motivated. All a priori unknown parameters in the ansatz for the self-energy are calculated self-consistently by exploiting the equality between two alternative but exact representations for the first four moments of the spectral density. With the resulting expression for the self-energy at hand, the many-body problem reduces to the problem of finding the local density of states (LDOS) for an energy-dependent effective one-particle Hamiltonian. We determine the LDOS for the semi-infinite system using a straightforward generalization of the standard tight-binding recursion method that is suitable for treating an energy-dependent Hamiltonian. The resulting energy dependence of the recursion coefficients is studied in detail for the (100) surface of a bcc crystal and an interpolation procedure for a numerically feasible evaluation of the theory is suggested.

We have studied the electronic states in a non-magnetic spacer layer of Co/Cu/Co sandwich layers using the magneto-optical Kerr effect (MOKE) and x-ray photoemission and have found that they are discrete owing to the quantum size effect and appear approximately every . This coincides with the oscillation period of the saturation field. We connect them with the quantum minority-spin interface states near the Fermi level . The magnetoresistance found in these sandwich layers has been calculated on the basis of the quantum size effect transport theory with the requirement that there exist quantum well states within individual layers or groups of layers. The results of the calculation and experiments have been discussed.

Equations for angle- and spin-resolved photoemission from core levels of ferromagnets are derived using the atomic model. They are applied to the n p subshells and to the particular geometries of experiment with the photoemission normal to the surface, which have been used already in several experiments. It is shown that for these geometries the spin-resolved spectra obtained with linearly polarized light are especially simple and contain the contribution of only one or two magnetic sublevels of the state, and of only one sublevel of the state, which allow one to resolve the magnetic splitting of core levels. The use of circularly polarized or unpolarized light gives a less transparent picture.

Thermoluminescence and optical absorption measurements have been carried out in KBr: crystals irradiated with monochromatic UV light (200 - 300 nm) and x-rays at room temperature. For UV- and x-irradiated crystals strong similarities between the thermoluminescence glow curves have been found, suggesting that the low-energy UV radiation produces the same defects as produced by x-irradiation in this material. The thermoluminescence glow curves are composed of six glow peaks located at 337, 383, 403, 435, 475 and 509 K. Thermal annealing experiments in previously irradiated crystals show clearly a correlation between the glow peak located at 383 K and the F-centre thermal bleaching process. Also, the excitation spectrum for each thermoluminescence glow peak has been investigated, showing that the low-energy radiation induces the formation of F centres.

The new solid solutions have been investigated by different experimental methods in the temperature range 10 K - 300 K. X-ray and calorimetric measurements show that the incommensurate phase characteristic of is also observed for . Raman spectra were obtained for and . Two low-frequency, temperature-dependent bands were observed over the whole temperature range investigated. These results are discussed in connection with structural disorder and the presence of precursor effects of an incommensurate phase transition.

One-dimensional systems generated by deterministic rules show interesting properties, such as a structure factor with peaks whose amplitudes scale with the size of the system according to a power law, with a scaling exponent in general dependent on the wavevector. I apply a wavelet transform to the structure factor of the Thue - Morse and the period-doubling chains and show how this technique succeeds in computing the scaling exponents and in finding the wavevector associated with them. This example shows that wavelet analysis could become an efficient tool for studying the Fourier transform of aperiodic systems.

The approximate Green's functions of the localized electrons, obtained by the cumulant expansion of the periodic Anderson model in the limit of infinite Coulomb repulsion, do not satisfy completeness even for the simplest families of diagrams, like the chain approximation. The idea that employing -derivable approximations would solve this difficulty is shown to be false by proving that the chain approximation is -derivable and does not satisfy completeness. After finding a family of diagrams with Green's functions that satisfy completeness, we put forward a conjecture that shows how to select families of diagrams with this property.

The method of the double-time - temperature Green function is used to study theoretically the magnetic properties of layered Heisenberg ferrimagnets. The sublattice magnetization is calculated for different interlayer coupling strengths in the whole range of temperatures. According to the extent to which the interlayer coupling suppresses the two-dimensional spin fluctuations, we divide the low-temperature regime into two parts and give the asymptotic expressions for sublattice magnetization in two low-temperature regimes and in the vicinity of the Curie temperature, respectively. We also discuss the dependence of the Curie temperature on the interlayer coupling strength.

We have carried out magnetic neutron diffraction studies of a single crystal of neodymium at high pressure, using clamp cells on the diffractometer D10 at the Institut Laue - Langevin, Grenoble. The magnetic ordering was investigated at pressures up to 14 kbar in the temperature range 1.4 - 25 K. A number of quite different magnetic phases were observed, including the coexistence of ferromagnetic and antiferromagnetic ordering of moments on the inequivalent crystallographic sites. The structures formed and the population of the different symmetrically equivalent domains appear to be affected by complex interactions between the cubic-site ferromagnetism, the antiferromagnetic hexagonal-site ordering and a uniaxial component of stress.

We examine the influence of an anisotropic interaction term of Dzyaloshinskii - Moriya (DM) type on the ground state ordering of the - spin- Heisenberg antiferromagnet on the square lattice. For the DM term we consider several symmetries corresponding to different crystal structures. For the pure - model there are strong indications for a quantum spin liquid in the region of . We find that a DM interaction influences the breakdown of the conventional antiferromagnetic order by (i) shifting the spin-liquid region, (ii) changing the isotropic character of the ground state towards anisotropic correlations and (iii) creating for certain symmetries a net ferromagnetic moment.

Various magnetic properties of amorphous Fe - Y alloys at finite temperatures such as the densities of states, local magnetic moments, susceptibilities, and the magnetic phase diagram have been investigated on the basis of the finite-temperature theory of amorphous magnetic alloys. It is found that the calculated magnetization, distribution of local moments, susceptibilities, and magnetic phase diagram in the most random atomic configuration show a spin-glass phase, a re-entrant spin-glass phase, ferromagnetism, and paramagnetism with reducing Fe concentration. These results describe qualitatively the experimental data for the melt-spun Fe - Y alloys. It is verified that the nonlinear magnetic couplings between Fe local moments and the large-amplitude fluctuations of local moments due to structural disorder result in the spin glass in the Fe-rich region. The decrease of the average coordination number of the Fe atom due to the large difference in atomic size between Fe and Y atoms, on the other hand, stabilizes the ferromagnetism at lower Fe concentrations. The difference in magnetic phase diagram between the melt-spun and the sputtered Fe - Y alloys is interpreted in terms of that between the atomic short-range orders.

We study the Drude weight D and the total optical weight K for a t - t' - J model on a square lattice that exhibits a metallic phase-modulated antiferromagnetic ground state close to half-filling. Within a suitable 1/N expansion that includes leading quantum-fluctuation effects, D and K are found to increase linearly with small hole doping away from the Mott metal - insulator transition point at half-filling. The slow zero-sound velocity near the latter transition identifies with the velocity of the lower-energy branch of the twofold excitation spectrum. At higher doping values, D and K eventually saturate and then start to decrease. These features are in qualitative agreement with optical conductivity measurements in doped antiferromagnets.

Magnetization is measured in plastically deformed alloy in the temperature range from 4.5 to 80 K.The temperature dependence of spontaneous magnetization has been analysed according to the self-consistent renormalization theory of spin fluctuation. The value of decreases from to and the Curie temperature declines from 57.8 to 50.1 K with 50% strain. The decreases in and can be explained by the introduction of the antiphase boundary, which reveals the existence of an atomically ordered structure in the Ni - Cu alloys.

The coercivity behaviour of bonded magnets made from sieved inert-gas-atomized Nd(Dy) - Fe - B powders has been investigated. For magnets with particle sizes , the coercivity is controlled by a nucleation mechanism as in the sintered magnet. For the magnet with particle sizes of , the hardening mechanism is mainly controlled by nucleation as in the sintered magnet but the hardening mechanism of domain wall pinning or the nucleation of a single domain, which usually appears in melt-spun materials, may also be involved. The coercivity behaviour of bonded magnets made from gas-atomized powders seems to be between those of the sintered and melt-spun magnets depending on the particle size. Meanwhile it has been found that the demagnetizing field of bonded magnets made from the inert-gas-atomized powders is very small.

We investigate the physical meaning of the longitudinal muon spin-relaxation rate measured for a paramagnet at temperatures high with respect to the magnetic phase transition temperature. The depolarization rate is shown to be a function of the relaxation rates of the quasi-elastic and inelastic magnetic excitations. Using this analysis, we consider recently published data recorded on the strongly correlated electron systems CeNiSn and .

The reflectance of nickel-pigmented anodized thin alumina films on aluminium has been measured in the infrared range between 5 and . The strong absorption of plain porous alumina, found at around for p-polarized light at an oblique angle of incidence and originating from the longitudinal optical phonons interacting with light, is reduced for pigmented alumina. Electrochemical deposition of nickel in the pores of anodic alumina starts from the bottom of the pores, forming a nickel-pigmented bottom layer as the time of pigmentation increases. It is shown from p-polarized reflectance measurements at an oblique angle of incidence that with increasing thickness of the nickel-filled bottom layer the LO mode absorption is decreasing. This is consistent with effective-medium calculations which have been performed using the Maxwell Garnett model. It is found to be an effect of the geometrical shape and orientation of the nickel particles, which for these samples appear as rods with their long axes perpendicular to the sample surface.

In this paper we report self-consistent all-electron local-density calculations using the linear combination of Gaussian orbitals method for cuprous oxide. We calculated the directional Compton profile and their anisotropic effect. The spherical average Compton profile has also been calculated and compared with the experimental results. We also calculated the density of states of cuprous oxide.

In the first line of table 1 on page 1164, Γ₃ should read Γ₁.

The following corrections should be noted:

In the title, `small' should read `rather'.

On page L165, line 8, `small' should read `large'.

On page L165, line 10, `neat' should read `heat'.

On page L165, lines 18 and 23, `small' should read `rather'.

On page L169, line 6, `small' should read `rather'.

On page L170, penultimate line, `small' should read `rather'.