Topographical changes on vicinal Si(111) have been followed as a function of submonolayer Au coverage. Nucleation at step edges leads to subsequent growth of a single-domain overlayer of the Au-induced 5×2 structure. Phase boundaries between step, 7×7 and 5×2 regions have been studied, as well as the detailed structure of the step and 5×2 regions. Gradual changes in facet angle on the surface occurred, driven by 5×2 terrace growth. An inhomogeneous coverage of Au was observed, with a pattern extending over distances on the order of micrometers, indicating large-scale surface diffusion of both Au and Si atoms.

The commonly assumed quasiequilibrium particle distribution with the same quasi-Fermi-level for all quantum-dot carriers in the same energy (conduction or valence) band is found not to be valid for a wide range of temperatures at the inversion populations and bound energy separations (greater than a LO phonon energy) used in the literature. Bound state occupation factors obtained from the steady state solution of rate equations describing the ionization balance in room-temperature 100-Å-radius GaAs quantum dots whose centers are separated by 400 Å are found to be very different from the quasiequilibrium distribution used in an example from the literature. In such quantum dots, bound state transitions result from collisions between charged particles via the Coulomb interaction, and from interband and intraband radiative processes. The critical free electron concentration above which collisional processes can establish a quasiequilibrium in the conduction band is found to exceed ${10}^{19}$ ${\mathrm{cm}}^{\mathrm{-}3}$. Our numerical solution is in good agreement with Pitaevskii’s model from atomic physics of an electron random walk in energy as modeled by a Fokker-Planck equation. In our simple model, electrons are captured into a bound conduction band state via three-body recombination and phonon emission, and drop into lower energy bound states via a series of collisional deexcitations before combining with a valence band hole. Solution of the rate equations is standard in numerical studies of stimulated emission in atomic plasmas, but our present discussion is, to our knowledge, the first in the literature on semiconductor quantum-dot lasers.

We present an application to quantum dots of formulas from atomic physics for the rates of collisional ionization (binary encounter model) and dipole allowed collisional excitation (Van Regemorter and classical path models) occurring via the Coulomb interaction. The Van Regemorter, classical path, and binary encounter cross sections, which are known to be very successful in modeling atomic collisions, are shown to have accuracies comparable to (and are much easier to use than) the methods currently being used to study collisional rates in bulk excitons. Collisional excitation and ionization cross sections for a 150-Å-radius InSb quantum dot surrounded by CdTe barriers are shown to be much smaller than the geometric area π${\mathit{R}}^2$, where R is the quantum-dot radius. The reason that these cross sections are so small is that the Coulomb energy associated with the collision is typically much smaller than the collisional energy exchange. Explicit formulas are given for quantum-dot intraband oscillator strengths.

Optical-absorption experiments have been performed on short-period GaAs/AlAs superlattices of type-II in a waveguiding configuration at low temperature. We measured the absolute absorption coefficients of optical transitions involving electrons confined in X states in AlAs layers and holes confined in Γ states in GaAs layers. Complementary photoluminescence and photoluminescence-excitation experiments show clearly the origin of the transitions and support our results. Theoretical calculations of the effective dielectric tensor and absorption coefficient in the exciton resonance region are presented. The Γ-X mixing of electron states is explicitly taken into account and the dependence on the parity of layers is discussed. From comparison between experimental and theoretical values of the absorption coefficient we deduce the value of Γ-X coupling coefficient. The study of the photoluminescence decay time yields radiative exciton lifetimes and allows us to calculate the excitonic localization area in our structures.

Scattering of electrons in edge states of a magnetically quantized two-dimensional electron gas (2DEG) by nonequilibrium acoustic phonons has been studied both experimentally and theoretically. A phonoconductance imaging technique was used to probe the spatial and magnetic-field (B) dependence of the acoustic-phonon scattering of electrons in a GaAs device. It was discovered that a response was observed only when phonons were incident at the edge of the 2DEG. Giant oscillations in the phonoconductance with magnetic field were observed, these were periodic in 1/B and changed sign when the field was increased beyond 1.3 T. At low field the oscillations were negative, i.e., phonon scattering causes a decrease in conductance, whereas at high field the oscillations become positive. These results are described theoretically by considering the change in group velocity of the edge electrons upon absorption of a phonon which changes their energy and wave vector. Assuming a sharp edge potential, an important parameter in the theory turns out to be the product of the phonon momentum component parallel to the edge and the magnetic length. If this parameter is greater than unity, then interedge state transitions leading to a decrease in electron group velocity produce the dominant effect, a decrease in conductance. However, if it is less than unity, only intraedge state transitions can take place giving rise to an increase in group velocity and hence an increase in conductance. If the potential at the sample edge is considered to be smooth, which is a more realistic assumption, then the theoretical results remain qualitatively the same, except that the field at which interedge state transitions are no longer possible is reduced in line with the greater separation of the edge channels. This makes it possible to use our results to estimate both the potential gradient at the edge of the sample and the spatial separation of the edge channels. It is also demonstrated that the spatially resolved phonoconductance measurements as a function of magnetic-field strength can be used to probe the local electron concentration and the charge separation connected with the Hall voltage is determined for our sample.

The Green-function method was adopted to study magnetopolarons in lateral surface superlattices (LSSL’s), where lateral periodic potentials due to periodically structured interfaces are introduced into two-dimensional electronic systems. Numerical calculations were carried out for the energy dispersions of magnetopolarons in GaAs LSSL structures with periodic potentials along the lateral x direction, such as those produced by deposition of AlAs and GaAs fractional layers on (001) vicinal GaAs substrates. The degenerate magnetopolaron energy levels become dependent on the electron wave vector ${\mathit{k}}_{\mathit{y}}$ in the y direction, forming magnetopolaron bands. For weak magnetic field (${\mathrm{\omega }}_{\mathit{c}}$≪${\mathrm{\omega }}_{\mathrm{LO}}$), the effect of lateral periodic potentials on electron–LO-phonon interaction energies are negligible. The energy dispersions of magnetopolarons are determined mainly by the energy dispersions of free-electron Landau levels in LSSL’s. In the resonant region (${\mathrm{\omega }}_{\mathit{c}}$≊${\mathrm{\omega }}_{\mathrm{LO}}$), the effect of lateral periodic potentials on electron–LO-phonon interaction energies must be considered in order to obtain the correct magnetopolaron energy dispersions. The energy pinning level for the lowest branch of magnetopolaron energy levels is shifted to lower energy when compared with that in quantum wells with planar interfaces, which changes greatly the energy dispersions of magnetopolaron energy levels in the resonant region. New branches of magnetopolaron energy levels, not present in quantum wells with planar interfaces, were predicated in LSSL’s in the resonant region. The transition energies from magnetopolaron ground levels to excited levels form wide bands, which are expected to broaden greatly the cyclotron resonance absorption spectra of LSSL’s.

We consider cation interdiffusion in the calculation of far-infrared reflectivity (FIR) spectra of GaAs/AlAs superlattices grown onto n-doped GaAs substrates. Optimum substrate doping levels exist at which the sensitivity of FIR to intermixing is greatly enhanced, thereby allowing the determination of interface composition and thickness.

The abruptness of the semiconductor heterojunction has been dealt with by many past researchers by incorporating the single-band effective-mass equation with a connection matrix at the interface. This connection-matrix method has been used to predict the eigenenergies of the GaAs-${\mathrm{Al}}_{\mathit{x}}$${\mathrm{Ga}}_{1\mathrm{-}\mathit{x}}$As quantum wells. In this paper, the validity of this approach to quantum wells is examined by varying the location of the interface within a lattice constant to reveal the eigenenergy dependence on the location. The eigenenergy should be independent of the interface location if the single-band effective-mass equation combined with the connection matrix is a consistent theoretical entity. However, as the interface is displaced from -a to a (where a is the lattice constant) relative to the location where the numerical values of the connection matrices are usually given in the literature, the first and the second eigenenergies of the GaAs-${\mathrm{Al}}_{0.3}$${\mathrm{Ga}}_{0.7}$As quantum wells can deviate by about 6 and 20 meV, respectively. Therefore, it is concluded that the connection-matrix approach to quantum wells is not sufficiently consistent. However, the eigenenergy deviations become less signficant for the quantum well with lower barrier energy and wider well width.

We report an experimental study of the band-gap photoluminescence from strain-adjusted ${\mathrm{Si}}_{\mathit{m}}$${\mathrm{Ge}}_{\mathit{n}}$ (m=9,6,3; n=6,4,2) superlattices under applied hydrostatic pressure. The strain adjustment was achieved by a thick, step-graded ${\mathrm{Si}}_{1\mathrm{-}\mathit{x}}$${\mathrm{Ge}}_{\mathit{x}}$ buffer layer resulting in an improved quality of the superlattice with respect to dislocation density. The no-phonon (NP) lines shift linearly under applied hydrostatic pressure in all superlattices to lower energies. The stress dependence was modeled using an approach based on deformation potentials and effective-mass theory. Under the assumptions made a close resemblance between experiment and theory was found. From the slopes of the stress shifts and amplitudes under various pressures, further evidence is given to an earlier assignment of the NP lines as arising from band-to-band recombination.

Recent experiments on GaAs/AlAs double-barrier diodes incorporating ${\mathit{n}}^{\mathrm{-}}$/${\mathit{n}}^{+}$/${\mathit{n}}^{\mathrm{-}}$ spacer layers have shown that they exhibit two conduction curves that remain distinct across zero bias. Such devices can be reversibly switched between the two conduction curves and retain memory of the curve last switched to, even after short-circuit conditions. In this paper, we model the memory-switching phenomenon using a quantum kinetic formulation based on the Wigner distribution function. To obtain accurate Wigner distribution functions, an improved four-point difference scheme with an upwind bias is used to model the drift term in the equation of motion. The calculations result in two equilibrium (zero-voltage) states in double-barrier diodes incorporating ${\mathit{n}}^{\mathrm{-}}$/${\mathit{n}}^{+}$/${\mathit{n}}^{\mathrm{-}}$ spacer layers. Associated with each equilibrium state is a distinct and stable conduction curve. This work is a step toward an understanding of the more complex aspects of the phenomenon, such as state switching when the device is driven far from equilibrium.

In order to study the local regime in the alkali-metal-promoted oxidation of silicon, we consider the coadsorption of an oxygen molecule and a potassium atom on the dimerized Si(100)-(2×1) surface. Antiferromagnetic (singlet) spin correlations within the Si dimers are taken into account from the outset by working with a generalized Hubbard Hamiltonian. In this background, the adsorption of a K atom strongly polarizes the medium creating a local charge-spin bag, around the ${\mathrm{K}}^{+}$ ion, which sets the scenario for promoted oxidation. When an ${\mathrm{O}}_2$ molecule in its ground state ${(}^3$${\mathrm{\Sigma }}_{\mathit{g}}$) approaches this region of the surface, it is influenced by the attractive electrostatic field of the ${\mathrm{K}}^{+}$ ion, with the corresonding lowering of its affinity level. This eventually crosses the Fermi level and captures the excess electron charge in the bag. A superoxide ${\mathrm{O}}_2$ ion, the ${}_{}{}^2{}_{}{}^{}\mathrm{\Pi }_{\mathit{g}}^{}{}_{}{}^{}$ state, is thereby formed, ionicly bonded to ${\mathrm{K}}^{+}$ and covalently bonded to the Si dimer. In the absence of potassium, the oxygen molecule simply physisorbs in a state adiabatically connected to its gas-phase ground state.

The formation of circular cone-shaped Ge islands (12° cones) has been observed for the growth of eight monolayers of Ge on Si(001) at 700 °C using Sb as a surfactant. The Ge cones are strained and grow pseudomorphically, adopting the Si lattice constant. They have a tilt angle of 12° and are composed of [117]-, [105]-type, and all intermediate facets. The island-size distribution is peaked around a typical size of ∼300–400 Å, which results from a formation process under equilibrium conditions for diffusion of the Ge atoms. Growth at lower temperatures down to 300 °C with Sb as a surfactant results in epitaxial but very rough Ge films which show a high degree of disorder.

We report results of a combined study of in situ x-ray-photoemission spectroscopy (XPS) and low-energy-electron diffraction (LEED) for silicon nitride films fabricated at room temperature. A variety of nitride films of ${\mathrm{SiN}}_{\mathit{x}}$ (0.2≤x≤1.33) were formed by bombarding low-energy nitrogen ions (E≤600 eV) onto a Si(001) surface without thermal treatment. The results reveal that the Si-N bond nature in the ion-deposited layers (IDL’s), derived from the characteristic XPS spectra, strongly resembles that of a typical thermally prepared silicon nitride, β-${\mathrm{Si}}_3$${\mathrm{N}}_4$. We find the chemical shift per Si-N bond to be 0.62 eV, and a shift of Fermi level due to nitridation less than 0.1 eV. We estimate the thickness of the unannealed IDL’s to be less than 18 Å. Upon annealing an IDL, progressive changes of XPS peaks and LEED patterns suggest that nitrogen atoms migrate from initial metastable defect sites to thermally stable sites, and tend to coalesce to form locally ordered microcrystallites. The thermal activation energy barrier in an IDL of E=400 eV, x=1.33 is found to be about 0.21 eV, in agreement with theoretical predictions.

We present a simple method to compute the optical properties of quantum wells of any shape, using microscopic calculations of the electronic states. Near the exciton resonances simple expressions for the polariton states and the radiative lifetimes are obtained as functions of the in-plane wave vector ${\mathit{K}}_{\mathrm{?}}$. The cases of a GaAs/${\mathrm{Al}}_{\mathit{x}}$${\mathrm{Ga}}_{1\mathrm{-}\mathit{x}}$As/AlAs asymmetric well and of double wells are considered; the results relate to those of the symmetric well. The computed splitting between the Z mode and the T mode is shown to depend mostly on the thickness, in fair agreement with experiments by Fröhlich et al. Polaritons associated with crossed excitons involving pairs of subbands forbidden in the symmetric case are found, and are shown to give rise to additional structure in the reflectivity.

We report a Monte Carlo study of phase transitions in a finite two-dimensional (2D) system of charged classical particles which are confined by a circular parabolic or hard-wall well. The ground-state configurations are found by static energy calculations and their structures are analyzed using the Voronoi constructions. A Mendeleev table for these classical 2D-like atoms is obtained. We calculate the radial and angular components of the displacements of the particles as functions of temperature and determine the critical temperatures for the order-disorder phase transitions. The intershell rotation and intershell diffusion transitions are investigated. The results are compared with Wigner crystallization in the infinite 2D system.

We study the Landau level localization and scaling properties of a disordered two-dimensional electron gas in the presence of a strong external magnetic field. The impurities are treated as randomly distributed scattering centers with parametrized potentials. Using a transfer matrix for a finite-width-strip geometry, we calculate the localization length as a function of system size and electron energy. The finite-size localization length is determined by calculating the Lyapunov exponents of the transfer matrix. A detailed finite-size scaling analysis is used to study the critical behavior near the center of the Landau bands. The influence of varying the impurity concentration, the scattering potential range and its nature, and the Landau level index on the scaling behavior and on the critical exponent is systematically investigated. Particular emphasis is put on studying the effects of finite range of the disorder potential and Landau level coupling on the quantum localization behavior. Our numerical results, which are carried out on systems much larger than those studied before, indicate that pure δ-function disorder in the absence of any Landau level coupling gives rise to nonuniversal localization properties with the critical exponents in the lowest two Landau levels being substantially different. Inclusion of a finite potential range and/or Landau level mixing may be essential in producing universality in the localization.

The results of an x-ray-scattering study of the (3×1)-to-disordered phase transformation of the Si(113) surface are reported. A continuous commensurate-solid to incommensurate-fluid transformation at ${\mathit{T}}_{\mathit{c}}$=950±40 K is observed. At the transformation, the reconstructed layer becomes unaxially incommensurate along the cubic (11¯0) direction (x direction). It remains commensurate along the (332¯) direction (y direction). Critical scattering shows power-law behavior over nearly two decades of reduced temperature [t=(T-${\mathit{T}}_{\mathit{c}}$)/${\mathit{T}}_{\mathit{c}}$] with exponents β¯=0.66±0.05 for the incommensurability (ε), ${\nu }_{\mathit{x}}$=0.65±0.07 for the inverse correlation length in the incommensurate direction (${\mathrm{\kappa }}_{\mathit{x}}$), ${\nu }_{\mathit{y}}$=1.06±0.07 for the inverse correlation length in the commensurate direction (${\mathrm{\kappa }}_{\mathit{y}}$), and γ=1.56±0.13 for the susceptibility (χ). Below ${\mathit{T}}_{\mathit{c}}$ the variation of the square of the order parameter, proportional to the peak intensity at the commensurate position (${\mathit{I}}_0$), varies with an exponent 2β=0.22±0.04.

It is noteworthy that the correlation lengths in the disordered phase scale anisotropically, that is, ${\nu }_{\mathit{x}}$≠${\nu }_{\mathit{y}}$, and that the collected exponents do not conform to those of any previously known universality class. In addition to the critical exponents of the transformation, two universal constants have been measured. The ratio of the incommensurability and the inverse correlation length along the incommensurate direction in the disordered phase is found to be independent of temerature, i.e., β¯=${\nu }_{\mathit{x}}$, consistent with predictions for a two-dimensional chiral melting universality class, and to have the value ${\mathit{w}}_0$=1.6±0.2. Also, the combination ${\mathit{R}}_{\mathit{s}}$=χ${\mathrm{\kappa }}_{\mathit{x}}$${\mathrm{\kappa }}_{\mathit{y}}$/${\mathit{I}}_0$${\mathit{V}}_{\mathit{r}}$, where ${\mathit{V}}_{\mathit{r}}$ is the two-dimensional resolution volume, is independent of the reduced temperature, consistent with the derived hyperscaling relationship ${\nu }_{\mathit{x}}$+${\nu }_{\mathit{y}}$=γ+2β. According to the hypothesis of two-scale-factor universality, ${\mathit{R}}_{\mathit{s}}$ is a universal constant, which we find takes the value ${\mathit{R}}_{\mathit{s}}$=0.07±0.03.

Results of a spot-profile-analysis low-energy-electron-diffraction study of the 3×1 order-disorder phase transition of the Ge(113) and Si(113) surfaces are reported. For Ge(113) agreement with predictions for chiral melting with isotropic scaling is found. For Si(113) we compare our findings to those of other LEED and x-ray-scattering studies.

We analyze collective excitations of carriers in three- and two-dimensional crystals made up of cluster-type units (fullerenes, quantum dots, etc.). This class of compounds is unique in that the electrons are essentially confined to the individual clusters, while the latter are both large in size and close packed. We derive a general expression for the dispersion law of conduction-band plasmons, and apply it to alkali-metal-doped fullerides and quantum-dot arrays. We find that the dispersion relation is modified from the case of a homogeneous electron gas: it becomes measurably steeper and acquires a downward slope. This change is due to the appearance of a term which is determined by the density distribution of electrons within a unit cluster. Related electron-energy-loss spectroscopic data on the doped fullerides are discussed.

A Green’s-function technique is presented that describes the electronic properties of surfaces and interfaces in the framework of multiple-scattering theory as based on localized structure constants. Results of self-consistent calculations are presented for the Cu (111), (110), and (100) surfaces. The obtained surface densities of states and work functions are in good agreement with previous calculations.