A method to describe the metal-insulator transition (MIT) in disordered systems is presented. For this purpose the statistical properties of the eigenvalue spectrum of the Anderson Hamiltonian are considered. As the MIT corresponds to the transition between chaotic and nonchaotic behavior, it can be expected that the random matrix theory enables a qualitative description of the phase transition. We show that it is possible to determine the critical disorder in this way. In the thermodynamic limit the critical point behavior separates two different regimes: one for the metallic side and one for the insulating side.

The local-density approximation is used to investigate the electronic and structural properties of ${\mathrm{Nb}}_3$${\mathrm{O}}_3$ and the reasons for its unusual defect rocksalt structure with 25% ordered vacancies. A tight-binding Hamiltonian is fitted to the energy bands which are compared to those of the hypothetical rocksalt phase. This analysis indicates that the ions in the ordered vacancy structure are nearly neutral, the Madelung energy is small, and electronic band effects contribute decisively to its stability. The tight-binding results are consistent with the interpretation that this structure favors a special hybridization between Nb d orbitals and their direct interaction gives rise to strong and nearly covalent metal-metal bonds. The formation of metallic bonds mainly between Nb p and Nb d orbitals around the O-vacancy site gives rise to the so-called ‘‘vacancy band.’’

The self-interaction correction (SIC) to the local-spin-density approximation (LSDA) is studied for a single impurity problem in metals. Although the impurity resonance state is a scattering state, we find that a localized orbital, for which the SIC should be considered, is obtained as in the case where the impurity state forms a real bound state. Model calculation shows that the exchange splitting of the virtual bound state is remarkably enhanced by SIC suggesting that the SIC-LSDA calculation achieves improvement of the LSDA results. A scheme to determine the occupation number of orbitals self-consistently in order to deal with the fractional occupation for metallic systems is also discussed.

Derivatives of the five principal extremal cross sections of the Fermi surface of gold with respect to uniaxial stress along various directions have been determined from simultaneous measurements of the amplitudes of magnetostriction and torque quantum oscillations. The accuracy of the measurements is limited by a spurious magnetostriction signal caused by the crystal bending in response to the torque it experiences when its symmetry axis is misaligned with the magnetic field direction. The results are self-consistent and agree with the results of the direct tension measurements and with stress derivatives calculated from the absolute magnetostriction amplitude measurements. Discrepancies with results from other experimental work are discussed. Angular and tetragonal shear strain derivatives deduced from experiment agree within the combined range of uncertainty with theoretical values. The differences between experimental and theoretical values for gold are similar to those for copper.

The optical study of trivalent 3d transition-metal-oxide compounds (RM${\mathrm{O}}_3$) with the perovskitelike structure has revealed the variation of their electronic structure with the 3d element (M) as well as the A-site rare-earth element (R). The crossover of the gap nature from the Mott type to charge-transfer (CT) type with increasing atomic number of M is observed to occur around M=Cr. The variation of Mott and CT gaps with M species is quantitatively consistent with the tendency expected from an ionic model. However, for the low-energy electronic structures for the narrow-gap (or metallic) compounds (M=Ti,Co,Ni), the effects of the M-3d–O-2p hybridization must be included.

We report the results of a first-principles study of the structural properties, stability, and band structure of Ge in two complex tetrahedral phases: the ST12 phase (simple tetragonal with 12 atoms per unit cell) and the BC8 phase (body-centered cubic with 8 atoms per cell). We focus mainly on the ST12 phase, which is well established as being formed in Ge after release of pressure from the β-tin phase. Local-density-functional calculations of total energies, forces, and stresses were carried out using the plane-wave-pseudopotential method. At each fixed volume considered, the structures were fully relaxed with respect to their structural degrees of freedom (five for ST12, one for BC8). A comparison with the β-tin and diamond phases allows us to study the phase stability and pressure-induced phase transformations of these structures. For the ST12 and BC8 phases we find zero pressure volumes in good agreement with the experimental values. The ST12 phase turns out to be metastable, which is consistent with its formation on release of pressure from the β-tin phase. The BC8 phase is found to be less stable than the ST12 phase over the entire pressure range investigated (0–200 kbar). ST12 Ge is found to have a direct band gap, whereas BC8 is found to be semimetallic in our calculations.

We have developed a theory of the alternating-current (ac) relaxation-type conductivity due to small bipolaron (SB) hopping in amorphous semiconductors and insulators that possess deep centers of the dangling-bond type (D centers) with a negative two-electron correlation energy ${\mathit{U}}_{\mathrm{eff}}$. Unlike small po- larons, SB’s were treated as essentially three-level systems in the framework of a two-site approximation. To calculate, both numerically and analytically, the real part ${\mathrm{\sigma }}_1$∝${\mathrm{\omega }}^{\mathit{s}}$${\mathit{T}}^{\mathit{n}}$ of the ac hopping conductivity for different temperatures T in a wide range of audio and low radio frequencies ω, the dynamic polariz- ability, and the SB hopping rates for dangling bond pairs have been determined. When the electron tunneling integral corresponding to the smallest intersite separations is greater than the doubled polaron shift, both the polarizability and the hopping rate strongly depend upon the shape and parameters of the ground-state adiabatic potential of a small-size pair of strongly interacting D centers. This intimate pair can be viewed as a stretched or weakened bond. A classification of possible regimes of relaxation-type SB hopping (adiabatic and nonadiabatic, as well as tunnel and activation) has been proposed. Each of these corresponds to a specific temperature dependence of the exponents s and n. A comparison to experimental data on ac losses in chalcogenide glasses and a-${\mathrm{SiO}}_2$ has been made.

It demonstrates that, in these materials, a tunneling adiabatic (TA) regime of SB hopping with s≊1 and n≪1 has been observed at low temperatures provided that there are (i) strong Coulombic correlations in the intersite occupations and (ii) strong scattering of one-electron energies. As the temperature is increased, the TA regime seems to be replaced by the activation adiabatic regime with s<1 being a decreasing function and n>1 being an increasing function of T. It was found that in the negative-${\mathit{U}}_{\mathrm{eff}}$ amorphous systems the stretched bonds of size of 3.5–4.5 Å can be responsible for ac losses within the wide frequency range from ${10}^2$ to ${10}^8$ Hz.

The observation, via optically detected magnetic resonance, of shallow donors in GaAs is reported. A resonance having ‖g‖=0.43±0.01 was observed as a change in the long-wavelength photoluminescence from samples doped with approximately ${10}^{16}$ ${\mathrm{cm}}^{\mathrm{-}3}$ Te and approximately 4×${10}^{16}$ ${\mathrm{cm}}^{\mathrm{-}3}$ Cr. The large resonance linewidth of 930 mT is attributed to exchange interactions with randomly distributed recombination partners.

The Overhauser shift of the conduction-electron spin resonance (CESR) in InSb is detected via coherent spin-flip Raman scattering. This shift is proportional to the spin polarization of the nuclei and the probability ‖Ψ(${\mathbf{r}}_{\mathit{k}}$)${\mathrm{\Vert }}^2$ of finding an electron at the nuclear site ${\mathbf{r}}_{\mathit{k}}$. By using combined CESR and NMR techniques, the effective g factor in InSb was determined at different magnetic fields with the high precision available from NMR experiments. The nuclear-spin relaxation time ${\mathit{T}}_1$ was obtained as a function of magnetic field by two methods: Either by dynamically polarizing the nuclei or by observing the relaxation of the nuclei after field cycling from high to lower magnetic field. A measurement after buildup of high dynamic nuclear polarization yields information on the leakage factor of the nuclear-spin systems which describes the contribution of processes to the total relaxation mechanism other than via hyperfine interaction. Line shapes of the spin-flip Raman gain curve were studied with respect to the value of the nuclear-spin polarization. The probability ‖Ψ(${\mathbf{r}}_{\mathit{k}}$)${\mathrm{\Vert }}^2$ of finding an electron at the In or at the Sb nucleus was determined from a measurement of the Overhauser shift in thermal equilibrium of the nuclei. Since the Overhauser shift is proportional to the component of the nuclear-spin polarization in the direction of the external magnetic field, the method allows the observation of the transient mutation of the nuclei while applying rf pulses with the nuclear Larmor frequency. The results yield the effective transverse spin-relaxation time ${\mathit{T}}_2^{\mathrm{*}}$ by optically detected pulsed NMR.

We have determined the effective radii of donors (interstitial Na) and of acceptors (Na on a Zn substitutional site) in ZnSe:Na. These were determined from the time decay of relatively close pairs. We feel prior determinations of such impurity radii, generally done by using the decay of the distant-pair peak, are mostly inaccurate, due to neglect of excitation transfer; the effect of such transfer is minimized by use of the decay of the close pairs. Our results agree closely with the values obtained from scaled effective-mass theory.

The x-ray rocking-curve broadening accompanying the transfer to and from the metastable state of EL2 and DX centers in GaAs and ${\mathrm{Al}}_{\mathit{x}}$${\mathrm{Ga}}_{1\mathrm{-}\mathit{x}}$As has recently been observed experimentally [Leszczynski et al., Semicond. Sci. Technol. 6, B66 (1991)]. This paper gives a more quantitative analysis of the experimental results. Computer simulations of rocking curves based on the dynamical theory of x-ray diffraction for various models of the real crystal structure made it possible to evaluate the conditions in which the lattice relaxation could be observed in experiment. The general conclusion is that in all the materials examined, the inhomogeneities played a decisive role. The possible range of the inhomogeneities and the strains around EL2 and DX centers is discussed in relation to their microscopic models.

Theoretical studies on the states of conduction electrons in ferromagnetic semiconductors have been made above the Curie temperature (${\mathit{T}}_{\mathit{C}}$). The conduction electron states are strongly affected by the magnetic ordering of localized moments (f spins), the direction of which is completely random in the high-temperature limit and is strongly correlated between different sites around ${\mathit{T}}_{\mathit{C}}$. We take the self-energy diagram with an electron propagator to second order in the exchange interaction between the conduction electron and f spins, together with the static two-spin correlation function, to calculate the density of states and the self-energy Σ(k,ω) for the conduction electron by the Green’s function technique, where the k dependence of the self-energy is directly taken into account. The effect of the sum rule for a spin-correlation function is also considered. Comparing the results of the present theory and those obtained with the coherent-potential approximation, it is revealed that the present method is suitable for (IS/W)≲0.1, where W is the bandwidth of the conduction band, and the product IS is the exchange interaction energy. As the temperature approaches ${\mathit{T}}_{\mathit{C}}$, the bottom of the conduction band in the paramagnetic region is lowered due to the exchange scattering which depends on the f-spin correlation between different sites. The effect of the sum rule for the spin-spin corelation function reduces the lowering of the bottom of the band, especially at T=${\mathit{T}}_{\mathit{C}}$. The result compares reasonably well with the temperature dependence of the absorption edge of EuO reported by Schoenes and Wachter [Phys. Rev. B 7, 3097 (1974)]. The present results are also compared with the results reported by Sinkkonen [Phys. Rev. B 19, 6407 (1979)].

A comprehensive study of impurity-induced phonon disordering in ${\mathrm{Cd}}_{1\mathrm{-}\mathit{x}}$${\mathrm{Zn}}_{\mathit{x}}$Te alloys is reported for a variety of samples (with composition ranging from x=0.005–0.5, and 1) by using far-infrared reflectivity and Raman-scattering spectroscopy. Substantial differences were noted among the various published values for the optical-phonon frequencies versus x. Contrary to an earlier Raman study on molecular-beam-epitaxy grown ${\mathrm{Cd}}_{1\mathrm{-}\mathit{x}}$${\mathrm{Zn}}_{\mathit{x}}$Te/GaAs films, our results yield increases in the numbers of both the CdTe- and ZnTe-like TO phonons with x. A modified random-element isodisplacement model provides excellent fits to the optical phonons. The effects of impurity-induced phonon disordering are studied, within the band mode region, by using an average-t-matrix formalism. Unlike earlier speculations in which a gap mode in ZnTe:Cd lies near ∼140–145 ${\mathrm{cm}}^{\mathrm{-}1}$, our theory predicts it to be at a higher frequency, ∼153 ${\mathrm{cm}}^{\mathrm{-}1}$. Group-theoretical analysis suggests that the gap mode exhibits a triply degenerate vibrational state and it can be detected both by IR absorption and Raman-scattering spectroscopy.

The alloy disorder scattering potential can introduce large k-vector scattering transitions, which are responsible for intervalley scattering. Disorder-induced scattering in ${\mathrm{Al}}_{\mathit{x}}$${\mathrm{Ga}}_{1\mathrm{-}\mathit{x}}$As is investigated in reciprocal-lattice space by calculating the (momentum) wave-vector components of the self-energy correction to the virtual-crystal model, using the coherent-potential approximation. The imaginary part of the coherent potential (the self-energy) is responsible for disorder-induced energy-level broadenings as a result of the scattering. The calculated disorder-induced intervalley scattering potential for Γ-X transitions is found to be near the observed experimental value. The results for the disorder-induced L-X intervalley scattering potential are incorporated into an ensemble Monte Carlo simulation of a time-resolved optical pump-probe experiment in ${\mathrm{Al}}_{0.6}$${\mathrm{Ga}}_{0.4}$As. It is found that disorder-induced L-X intervalley scattering lifetimes are 3–5 times larger than the phonon-induced intervalley scattering lifetimes.

Angle-resolved photoemission data have been used to determine the conduction-band structure of GaAs along the ΓKX and ΓΔX directions. A structure plot of kinetic energy versus initial-state energy has been used to identify the valence bands involved in the observed transitions. The determination of values of k for the observed transitions was based on theoretical valence bands of GaAs from a linear-muffin-tin-orbital (LMTO) calculation. The experimentally determined final states are in reasonably good agreement with the LMTO conduction bands.

We have developed a semiempirical total-energy functional for silicon-hydrogen interactions in solids. This functional, which is based on a quantum-mechanical analysis, is general enough that it can be applied to a variety of Si-H systems. It contains eight Si-H parameters which are obtained by fitting to bond lengths and vibrational frequencies in ${\mathrm{SiH}}_{\mathit{n}}$ molecules (n=1,2,3,4). As tests of this functional, we have applied it to the calculation of various properties of ${\mathrm{SiH}}_{\mathit{n}}$ and ${\mathrm{Si}}_2$${\mathrm{H}}_6$ molecules and of the hydrogenated Si(111) surface. Our results are in very good agreement both with experiment and with first-principles calculations. Our Si-H functional and the associated parameters thus show promise for application to other silicon-hydrogen systems.

Several ${\mathrm{P}}_{\mathrm{In}}$ antisite structures are observed in InP and identified using optical detection of magnetic resonance (ODMR) and electron-nuclear double resonance (ODENDOR) via magnetic circular dichroism in near-band-gap absorption. One of these is the isolated ${\mathrm{P}}_{\mathrm{In}}$ antisite previously studied in as-grown and electron-irradiated p-type InP. Two others, clearly distinguishable in the ODENDOR, are perturbed antisites that are produced by electron irradiation and observed only in n-type samples. The isolated antisite is also detected by ODMR and ODENDOR in as-grown p-type materials in two competing photoluminescence bands, peaking at 0.8–0.9 and 1.1 eV. It is proposed that both bands arise from donor-acceptor pair recombination involving the isolated antisite defect, the 0.8–0.9-eV band originating from the ${\mathrm{P}}_{\mathrm{In}}^{+}$/${\mathrm{P}}_{\mathrm{In}}^{2+}$ donor level and the 1.1-eV band from the ${\mathrm{P}}_{\mathrm{In}}^0$/${\mathrm{P}}_{\mathrm{In}}^{+}$ donor level. With this identification, the energy position of the isolated antisite ${\mathrm{P}}_{\mathrm{In}}^{+}$/${\mathrm{P}}_{\mathrm{In}}^{2+}$ level is estimated to be at ${\mathit{E}}_{\mathit{V}}$+1.1±0.1 eV and that of its ${\mathrm{P}}_{\mathrm{In}}^0$/${\mathrm{P}}_{\mathrm{In}}^{+}$ level to be at ${\mathit{E}}_{\mathit{V}}$+1.39±0.03 eV. The perturbed antisite levels appear higher in the gap consistent with the presence of an acceptor in a near-neighbor shell, a likely candidate being the indium vacancy.

Infrared-absorption measurements have been made on epitaxial layers of GaAs grown by metal-organic molecular-beam epitaxy and doped with either ${}_{}{}^{12}{}_{}{}^{}\mathrm{C}$ or ${}_{}{}^{13}{}_{}{}^{}\mathrm{C}$ impurities to a concentration of 1.5×${10}^{19}$ ${\mathrm{cm}}^{\mathrm{-}3}$. After the introduction of either hydrogen or deuterium atoms from a radio-frequency plasma the localized vibrational modes (LVM) of H-${\mathrm{C}}_{\mathrm{As}}$ pairs have been observed and analyzed. Two ${\mathit{A}}_1$ modes and two E modes have been found for D-${\mathrm{C}}_{\mathrm{As}}$ pairs, and so this is the first hydrogen-passivated shallow-impurity pair for which all four LVM’s are present. Admixing of the two E modes explains an apparent anomaly when the spectra of samples containing H${\mathrm{-}}^{12}$${\mathrm{C}}_{\mathrm{As}}$ are compared with those containing D${\mathrm{-}}^{12}$${\mathrm{C}}_{\mathrm{As}}$ pairs. There is good agreement with published ab initio theory, and anharmonicity of the stretch modes is discussed. No evidence has been found for the presence of the other carbon complexes, nor for the amphoteric behavior of carbon.

We present qualitative and quantitative evidence that the H-glass model is applicable to metastability annealing in P-doped hydrogenated amorphous silicon and measure the principal parameter of the model. We study the effect of H content on the annealing of quenched-in, metastable, excess dark conductivity in amorphous silicon. The annealing temperature (${\mathit{T}}^{\mathrm{*}}$) is 355±20 °C in sputtered P-implanted amorphous silicon (a-Si) containing 0.1 at. % of H. After hydrogenation with about 10 at. % of H, ${\mathit{T}}^{\mathrm{*}}$ falls to 175±20 °C, much closer to the 130 °C observed in glow-discharge hydrogenated a-Si. We derive a quantitative prediction of the H-glass model: the H-diffusion coefficient at ${\mathit{T}}^{\mathrm{*}}$ is inversely proportional to the H content in a-Si samples. Our measurements of H-diffusion coefficient at ${\mathit{T}}^{\mathrm{*}}$ are in agreement with this prediction.

We present an analysis of the scattering of hot carriers by conduction electrons in heavily doped semiconductors within the random-phase approximation (RPA). Different approximations to the temperature-dependent RPA are considered: (i) the two-pole approximation developed by Rorison and Herbert, (ii) the plasmon-pole approximation, and (iii) the Lindhard dielectric function. We present a range of results for n-doped GaAs for different carrier energies, doping levels, and temperatures, and we examine the ranges of validity of the different approximations. As an extension of our theory we include, within the two-pole approximation, the coupling of optical phonons to the electron system.