Table of contents

Many magnetotransport experiments in two-dimensional electron systems have been explained in recent years within the so-called edge-state model. Starting from a historical context, the model is described. Experiments in Hall-bar geometries with and without gate barriers are reviewed and discussed in terms of the edge-state picture.

This work provides an experimental and theoretical insight into the physical mechanisms involved in the co-diffusion of As and B in polysilicon/monocrystalline Si bilayers, during the formation of shallow N⁺ emitters for bipolar-CMOS technology. The RTA-induced redistribution of As and B successively implanted in a 380 nm LPCVD polysilicon layer has been studied by SIMS measurements. A strong retardation in the diffusion was observed for B, which was attributed essentially to grain growth in the polysilicon layer, detected by X-ray diffraction. A weak retardation effect was also noticed for As in the presence of B, but with no significant consequence in the doping profile of the monocrystalline emitter region. The electrical activation of As in the co-implanted structures is satisfactory from a RTA temperature of 1100 degrees C, although slightly lower than in wafers without B. First results of the authors process modelling have permitted the authors to fit diffusivities of the dopants.

Depth profiles of thermal donors formed at 450 degrees C in oxygen-rich n-type silicon have been studied through capacitance-voltage measurements of Au Schottky diodes in the oxygen concentration range 8.1*10¹⁷ to 15.3*10¹⁷ cm^-3. Thermal donor concentrations increase with depth and their depth profiles are empirically fitted to the error function equation. The obtained diffusivity decreases with annealing time and is higher for samples with lower initial oxygen concentrations. Combining with the infrared absorption measurements, it is found that the curves of the diffusivity as a function of loss of interstitial oxygen coincide with each other. A similar tendency is observed for the oxygen diffusivity evaluated with secondary-ion mass spectrometry. These results indicate that depth profiles of thermal donors are caused by both the out-diffusion and clustering of oxygen.

An optical study of Si-doped GaAs grown on (311)A-, (111)A-, (111)B- and (100)-oriented substrates by molecular beam epitaxy is reported. Raman measurements on these samples indicate good crystalline properties. Electrical characterization reveals an n-type doping of the (111)B and (100) samples and a p-type doping of the (111)A and (311)A samples. Detailed assessment of the low-temperature photoluminescence spectra suggests that the (311)A sample is more compensated than the (111)A sample and that the (111)B is more compensated than the (100) sample.

In this paper a theoretical study of the formation of a heterojunction between a Si (001) substrate and a very thin film of the more atomically dense beta -SiC (001) is presented. The structures that describe the deposition of the first layer of carbon and then the subsequent deposition of two further layers, one of silicon and the other carbon, to form the polar SiC film are discussed. In each case the atomic structure is determined from minimizing the total energy with respect to variations in the atomic coordinates. The formation of the interface between the polar SiC film and the silicon substrate is found to give rise to a distinct dipole within the top layers of the silicon substrate. Valence electron charge is transferred from the first silicon layer to the second silicon layer.

Raman spectroscopy has been used for non-destructive control of the crystal structure perfection of GaAs layers grown by atmospheric pressure metalorganic vapor phase epitaxy. The lineshape parameters have been determined by means of least-squares fits between the experimental Raman spectra and the theoretical ones. An attempt has been made to use these parameters in order to examine the influence of some growth conditions on the crystal structure perfection of the epitaxial layers. The results are supported by independent SEM and microhardness measurements. It has been shown that from Raman spectra analysis the effect of the substrate status on the epilayer structure can be studied along the layer thickness.

A way to control the dispersion law is suggested for an interface plasmon polariton propagating in a polariton-active metal-semiconductor structure with a Schottky barrier. An electrical voltage applied across the structure leads to a change in the thickness of the depleted 'insulator' layer at the interface and, as a result, to a change in the polariton dispersion law. Similar considerations are applicable to heterojunctions and metal-insulator-semiconductor structures. It is shown that a low-persistence effective modulator, which is tunable over a wide frequency range, or a controlled reflector, can be constructed on the basis of this effect. Some other possible applications are also discussed, namely light-wave filters, space light modulators, optical spectroscopy of contact potential difference, etc.

Magnetic self-trapping of an exciton in a quantum well is investigated by a variational method in the adiabatic approach. It is shown that the exciton magnetic polaron (EMP) can be formed in a quantum well at temperatures up to the critical temperature (T_c), which is found to depend on the characteristics of the structure. T_c decreases with increasing external magnetic field. The EMP binding energy in the CdTe/Cd_1-xMn_xTe quantum well structure is calculated. Primary localization of the exciton at a monolayer fluctuation of the interface is taken into account. The conditions for EMP formation in one-, two- and three-dimensional (1D, 2D and 3D) structures are analysed on a comparative basis.

The tunnelling current, the conductance and the second derivative of the current through single-barrier heterojunction diodes of GaAs/Al_0.5Ga_0.5As/GaAs are calculated including electron-phonon interaction in the electrodes. The effects of interface phonons are taken into account by using the dielectric continuum model. The conductance decreases when the applied voltage is greater than the value corresponding to the optical phonon energy. The role of the interface phonons in the decrease is studied, and it is found that the symmetric interface mode originating in GaAs electrodes plays the most significant role. The magnitude of the decrease and the onset are also discussed.

Deep levels have been characterized in Si-doped and undoped InAlAs layers lattice matched to InP. At ambient pressure and low temperature, the Schottky junction capacitance shows a small persistent photocapacitance effect, related to photoionization thresholds at 0.6 and 1.1 eV. Deep-level transient spectroscopy spectra show a dominant, low-density electron trap at around 320 K, with an emission energy very dependent on the bias conditions and the hydrostatic pressure (0.5 to 0.9 eV). This is attributed to an interaction of a high density of interface states with the trap emission process. Under hydrostatic pressure, neither the PPC per cent nor the DLTS spectra shape changes. Photoluminescence spectra in both Si-doped and undoped InAlAs layers show two weak peaks at 0.80 and 0.95 eV, that the authors suggest are due to a native defect involving an As antisite. They conclude that the detected traps have no relation with the DX centres.

The hydrostatic pressure dependence of both the transport and quantum mobility observed in n-GaAs heavily doped with Si is calculated by taking into account spatial correlations due to Coulomb interactions. The only mechanism included in the calculations is ionized impurity scattering in which the charged centres are ionized donors occupied DX centres and ionized acceptors. The correlation functions that reflect the spatial correlations are obtained by Monte Carlo simulations which enable the authors to account for the DX freeze-out temperature. It is shown that it is possible to simulate absolute values of experimental mobility data up to 20 kbar for samples with different compensation ratios. The results confirm the negative-U model for DX centres.

The authors report on Monte Carlo simulations of low-field hole transport at 77 K in InGaAs-AlGaAs quantum wells of different widths and alloy compositions. The valence subband structure is obtained using a k.p method within the infinite well approximation, which accounts for mixing between heavy and light hole states. The effect of alloy, impurity and phonon scattering are included in the transport simulations. Although the infinite well approximation is only expected to be reliable for barriers with an aluminium fraction greater than about 0.4, for which the heavy hole well is sufficiently deep, the results show good agreement with experimental measurements for a finite 90 AA In_0.18Ga_0.82As-GaAs quantum well. A study of hole transport in 90 AA In_xGa_1-xAs wells (0.10<or=x<or=0.25) predicts a mobility which increases with indium concentration since the reduction in the effective mass of the highest HH1 subband due to strain more than compensates for the greater alloy scattering rate. An analysis of wells with 18% indium content and widths in the range 50-150 AA indicates a general increase in hole mobility with well width but with a local minimum around 90 AA due to intersubband scattering from the HH1 subband to the heavier HH2 subband.

The authors have studied AlGaAs grown by molecular beam epitaxy (MBE) before and after post-growth hydrogen plasma treatment by means of temperature-dependent photoluminescence (PL), capacitance-voltage (C/V), and electron-beam-induced current (EBIC) analysis. The excitonic and extrinsic luminescence features of n-type AlGaAs samples subjected to a hydrogen plasma reveal the neutralization effect of vacancies and shallow acceptors. The hydrogen incorporation provides an increase of the non-radiative lifetime of the minority carriers. This behavior was checked by investigations of the minority carrier diffusion and of the quantum yield, which increase due to the hydrogen incorporation. The increase of the non-radiative lifetime is probably caused by the passivation of non-radiative recombination channels involving vacancies.

The temperature dependence of the carrier recombination lifetime in a gold-doped p-type silicon is analysed on the basis of the Shockley-Read-Hall recombination theory. The analysis indicates that the temperature dependence of the recombination lifetime does not practically give an energy level of a recombination centre locating at an energy level deeper than about 0.35 eV from either the silicon conduction-band edge or the valence-band edge. The recombination lifetime in a gold-doped p-type silicon is measured at various temperatures using a microwave photoconductive decay method. The observed temperature dependence of the recombination lifetime indicates that the lifetime is controlled by the gold donor. However, the observed recombination lifetime is about one-eighth of the recombination lifetime calculated using the gold-donor concentration as determined by deep-level transient spectroscopy. Some causes for this discrepancy are discussed.

The incorporation of erbium from a solid source into molecular beam epitaxy (MBE) Si and Si/Ge alloys grown at substrate temperatures of 500 degrees C and 700 degrees C has been studied by photoluminescence, electrical measurements, secondary-ion mass spectrometry (SIMS), Rutherford backscattering (RBS) and transmission electron microscopy (TEM). Erbium concentrations between 10¹⁸ and 10²² cm^-3 were obtained but the maximum photoluminescence intensity was from samples with an erbium concentration of 2*10¹⁸ cm^-3. Above this concentration the onset of erbium precipitation could just be observed by TEM. The authors found no shallow donors or acceptors attributable to erbium but they observed a high concentration of deep acceptors with an activation energy of 360 meV; these may be due to impurities in the erbium source rather than being directly related to the rare earth. Implantation with oxygen is found to enhance the Er³⁺-related photoluminescence signal when measured at temperatures greater than 77 K but to have little effect on the low-temperature luminescence. A detailed study of the temperature dependence of the luminescence reveals tree quenching mechanisms with average activation energies of approximately 5, 20 and 130 meV. The authors attribute the first two to de-excitation effects in the matrix, and the last to processes competing with the internal 4f transition.

Using Airy functions an analytical expression is obtained for the transmission coefficient of double-barrier quantum well structures under bias. The function which determines the resonance energies follows from this result. It allows one to calculate the variations of the shift of the resonance energies with applied field (Stark effect). A number of examples are given of such variations for electrons and holes in the ground state and in the first excited state. Comparison is also made with results obtained when using some approximate expressions for the Stark shift.

A simple numerical method for studying the effect of an applied magnetic field on the energy spectrum of non-periodic superlattice structures is presented. The magnetic field could be either parallel or perpendicular to the growth direction. The authors method is based on the transfer matrix technique and on the effective-mass approximation. They discuss the advantages and disadvantages of the proposed approach using several examples. In particular, they study the perturbation to the energy spectrum of a periodic superlattice induced by the introduction of an enlarged well. They found that these perturbations are negligible to B//z but significant for B perpendicular to z. Preliminary results for Fibonacci superlattices in magnetic fields are also presented. In these quasi-periodic structures the energy levels become strongly dispersive in the presence of a perpendicular magnetic field.

The influence of a lateral confinement on the current-voltage characteristic (CVC) of resonant tunnelling devices is considered for both classical and quantum confinement. In the ideal structure quantum confinement causes a non-monotonic appearance of the transconductance rather than of the CVC itself. To observe a peak structure in the CVC some additional perturbation, e.g. a single Coulomb impurity, must exist near a conducting channel. The number, position and amplitude of peaks are determined by the position of the impurity.

The response of a quantum well to an ultrafast optical pulse is examined theoretically using the Liouville equation in the electric dipole approximation. In particular a comparison is made between a system with a valence subband and a single conduction subband, and one with an additional conduction subband positioned to provide extra contributions to the resonant nonlinear polarization. The various contributions to the nonlinear polarization are identified and a simple physical picture is presented. It is shown that interference between the individual contributions is important in determining the overall nonlinear response, and that there are situations for which the addition of the second conduction subband is not beneficial.

The magnetophotoluminescence from thin InSb layers grown by molecular beam epitaxy (MBE) on InSb and GaAs substrates and from MBE InAs on GaAs substrates is studied at low temperatures. With undoped homoepitaxial InSb, the photoluminescence (PL) spectrum contains up to five emission bands. From the observed magnetic field and excitation level dependence of the bands it is inferred that their origin is related to radiative recombination of an exciton-impurity complex, to direct recombination of a free electron with an uncharged acceptor, to recombination at an A⁺ centre and at deep defects. By comparison with PL from the substrate alone it is shown that two of the defects responsible are unique to the epilayer. The PL spectrum of epitaxial layers of InSb on GaAs shows a single broad band due to band-to-band radiative recombination in the InSb film. The PL of high-mobility InAs layers grown heteroepitaxially at 490 degrees C on GaAs shows a single emission line. Samples grown at lower temperature, resulting in lower mobility, show a second line due to direct recombination of a free electron with an uncharged shallow acceptor, suggesting a correlation between the reduction of mobility and the presence of this acceptor.

The emission spectra, frequency characteristics and output power of single-heterostructure AlGaAs LEDs are studied experimentally and theoretically. The peak wavelength of spectra increases with increasing donor concentration, which is explained as being due to gradual changes of emission regions from the n to p region, based on the calculations of emission spectra and injection ratios of carriers. The cut-off frequency increases with increasing donor concentration but shows little change with acceptor concentration. To understand the phenomena, frequency response is derived taking injection of both types of carriers into account. The calculated cut-off frequency agrees with that of experiments on both donor and acceptor dependences. Considerable change in the ratio of emission from the two regions is shown to be the origin of the phenomena. The output power decreases with the increase of donor concentration and shows a mild trade-off against the cut-off frequency. These characteristics are explained provided that quantum efficiencies in the Te-doped n region decrease with increasing donor concentration, though ratios of the emission from the n region decrease concurrently.

Positron lifetime and Hall effect measurements have been performed to study vacancy defects in GaP. No Ga vacancies have been detected, while P vacancies have always been found in as-grown material. They show at least two different ionization levels within the upper half of the bandgap, which have been attributed to V_P⁺/V_P⁰ at E_C-E_D>400 meV and V_P⁰/V_P^- at E_C=E_D=250+or-30 meV, respectively. The concentration of the P vacancies in the as-grown samples was found to be in the order of some 10¹⁶ cm^-3. The specific positron trapping rate for negatively charged P vacancies was determined as mu _v^P=(1.9+or-0.5)*10¹⁵ s^-1.

The stability regions of Te-rich p-conducting Hg_1-xCd_xTe with x=0.20, 0.28 and 0.38 were investigated by means of high-temperature in situ Hall and conductivity measurements. The Te-rich limits of the p-side stability regions, measured in situ, correspond to higher electrically active native point defect concentrations than reported in the literature for quenched samples. In the temperature range 300<or=T<or=900 K, there is a linear relationship between the reciprocal temperature and the logarithm of electrically active native acceptor concentrations. The authors conclude that only one predominant kind of native point defect exists probably the mercury vacancy V_Hg. Its defect formation energies are calculated. The assumption of Hg vacancies as the predominant kind of defects is supported by precision measurements of lattice parameters, which decrease with increasing defect concentration. From these investigations the authors conclude that a considerably greater number of defects exist than are electrically measurable, especially at higher Cd contents.

A non-conventional technique of thermally detected adsorption (TD-OA) is used to study a strained InAs/InP quantum well with thickness between two and tree monolayers. The spectrometer and TD cell are briefly described and the TD-OA spectra are presented. Three signals due to the InAs well are identified. The simple envelope-function approach is sufficient for the interpretation of these, which are attributed to transitions involving the heavy-hole levels. However, for the light-hole-related peak, a more sophisticated theory, taking into account the coupling between the light hole and spin-orbit split-off states, is necessary to fit the result.

Measurements have been made of the carrier lifetimes in MOCVD and MBE InGaAs quantum wells in a waveguide configuration using a CW probe and a picosecond pump at a wavelength of 1.5 mu m. Values in the range 4-7 ns were obtained for excitation densities of the order of 3-6*10¹⁷ cm^-3.

Smooth, anisotropic dry etching of InN, AlN and GaN layers is demonstrated using low-pressure (1-30 mTorr) CH₄/H₂/Ar or Cl₂/H₂ ECR discharges with additional DC biasing of the sample. The etch rates are in the range 100-400 AA min^-1 at 1 mTorr and -150 V DC for Cl₂/H₂, while higher biases are needed to initiate etching in CH₄/H₂/H discharges. The presence of hydrogen in the gas chemistries is necessary to facilitate equi-rate removal of the group III and nitrogen etch products, leading to smooth surface morphologies.