Table of contents

The inclusion of spectral broadening due to carrier scattering is an essential element in the calculation of optical gain and radiative recombination spectra of a semiconductor under high injection. We have examined various methods of incorporating this broadening and find that gain and emission spectra obtained by merely broadening the optical transitions do not satisfy the necessary balance which must exist between them. The problem is removed by weighting the spectral contribution by the value of the Fermi functions of the electron and hole at the energies that would be involved in this transition in the absence of spectral broadening. We have also investigated a more exact method which uses separate electron and hole line functions, and is equivalent to broadening the electron and hole quantum states then calculating the newly available transitions, rather than broadening the transitions themselves. Both the latter methods introduce a small amount of optical gain in the bandgap at low injections and we show that this is due to the use of a Lorentzian lineshape which tails off too slowly. We comment on the most appropriate approach to the calculation of gain and emission spectra and gain-current characteristics.

The intrasubband electron-polar optical phonon scattering rates for interface and confined phonons are calculated. The AlAs/GaAs/AlAs double heterostructures with independent confinement of electrons and phonons as well as structures containing a phonon wall (a phonon-reflecting barrier transparent to electrons) embedded in an electron quantum well (QW) are considered. It is shown that, because of the independent electron and phonon confinement in the double heterostructure, the scattering rate is lower than that obtained in the case of electron confinement alone. The total scattering rate by confined and interface phonons in the QW with a phonon wall is reduced significantly as compared with the rate of confined electron scattering by bulk phonons. Thus, the phonon wall within the electron QW is a powerful means of reducing electron scattering and enhancing, correspondingly, the electron mobility in two-dimensional (2D) heterostructures.

We present a model for calculating exciton states in quantum wells (QWs) in which one of the two band discontinuities is large whereas the other one is arbitrary (positive, negative or zero). Exciton binding energies are calculated for ZnSe/ZnS_xSe_1-x, In_xGa_1-xAs/GaAs and GaAs/Al_xGa_1-xAs QWs. Good agreement with experimental results is found for all these systems. For ZnSe/ZnS_xSe_1-x, in which the conduction band offset is almost vanishing, the observed increase in the exciton binding energy is found to be due to the effective mass mismatch between the two materials.

A series of Zn_0.93Mn_0.07Te/ZnTe multiple-quantum well (MQW) structures of different quantum well widths has been studied at 77 K using the conventional modulation technique of photoreflectivity (PR) with above-bandgap light providing the modulation as well as a novel version of the technique which employs below-bandgap excitation. Photoluminescence (PL) spectra of the structures are also presented. The appearance of the PR spectra obtained with below-bandgap excitation differs from that of the PR spectra with above-bandgap excitation, but fitting the spectra with appropriate lineshapes gives the same spectroscopic information about the energy positions of the excitonic transitions in the well and the bandgaps of the ZnTe buffer layer and the Zn_0.93Mn_0.07Te barriers. The fits to the PR spectra yield energy positions of the lowest excitonic transitions which are consistent with the PL data. From the transition energies calculated with a one-dimensional transfer matrix model in which strain effects are taken into account, we conclude that the structures are to a good approximation strained to the ZnTe buffer layers.

The energy spectrum of two-dimensional electron states localized in the surface quantum well in gapless p-HgCdTe is investigated by tunnelling spectroscopy in a quantizing magnetic field. We observe two types of tunnelling conductivity oscillation in a magnetic field, which arise from the density of 2D states oscillations at the energies E_F and E_F+eV, where E_F is the semiconductor Fermi level and V the voltage applied to the semiconductor electrode. To interpret the experimental data the energy spectrum of such a system in a magnetic field and tunnelling probabilities are calculated. It is shown that two situations take place in our experimental conditions: either one or both spin states are localized in the surface quantum well in dependence on the applied bias and magnetic field. The calculation shows a large difference between the probabilities of tunnelling into different spin states. This explains why the spin splitting of the oscillations is not observed in our experiments.

We report the temperature dependence of Al_0.46In_0.54As photoluminescence (PL) transition energies and Al_0.46In_0.54As/InP interface staggered line-up luminescence (SLL) energy. The S shape appearing from 4 to 90 K on the energy versus temperature curves of these PL energies is due to extrinsic recombinations. In particular, the S shape of the SLL energy curve versus temperature is probably due to acceptor impurities localized in AlInAs, at the interface (on-edge impurities). The binding energy of on-edge impurities is lower than its value in the bulk material. This explains why the S shape is less pronounced on the SLL than on the AlInAs PL curve. The band offsets were determined by solving the Schrodinger-Poisson equation system with a self-consistent calculation program. At 4.5 K, the conduction and valence band offsets of the Al_0.46In_0.54As/InP interface were respectively 0.384 eV and 0.295 eV. This is in agreement with the already reported value of 410 meV for the conduction band offset of the lattice-matched Al_0.48In_0.52As/InP heterostructure. The temperature dependence of the conduction and valence band offsets is shown to be important: respectively 35 meV and 23 meV between 4.5 and 300 K. The Van Vechten-Malloy model (following a thermodynamic approach) for the temperature dependence of the band offsets is compared with our results. The comparison shows only a qualitative agreement.

Spin-dependent photoconductivity is observed in (100) silicon films grown on sapphire by CVD and MBE. The CVD films are either in their as-grown state or have undergone single or double solid-phase epitaxial regrowth. For all samples a resonant decrease in photoconductivity is observed at a field of about 0.34 T for a microwave frequency of about 9.6 GHz and at about 3.3 mT when the frequency is about 92 MHz. The fractional change in photoconductivity at resonance is measured as a function of the magnetic field strength, microwave or radiofrequency power, temperature, light intensity and sample voltage. The results are interpreted in terms of a quantum mechanical treatment of the pair model of Kaplan, Solomon and Mott and values are extracted for the spin relaxation time, pair dissociation rate and singlet recombination rate. In some samples a resonant change in dark conductivity is also observed and interpreted.

A theoretical study of the electronic band structure and linear optical response of (Si)_n/(Ge)_m strained-layer superlattices (SLSs) grown on [110]and [111]-oriented Si_1-xGe_x substrates is presented. The dependence of the band structure on the superlattice periodicity, composition and strain has been studied. The calculations show that many SLSs grown along the [110] direction on substrates rich in Ge become direct-gap materials, for example the (Si)_n/(Ge)_m SLSs with n=m=2-6 or with n+m=10 grown on a Ge(110) substrate, but with very small oscillator strength across the gap. For growth along the [111] direction, the (Si)₂/(Ge)_n SLSs with n=2, 6, 8 and 10 grown on Ge are found to be direct gap materials. In this case, the lowest transitions at the Gamma point have very large oscillator strengths, making these SLSs candidates for optoelectronic applications. The linear optical response of these SLSs has also been investigated, and several structures as well as the anisotropy appearing in the dielectric function spectra have been studied.

The structural properties of short-period molecular beam epitaxy grown p-i-n doped Si₁₇Ge₂ superlattices with ten, five and two periods were investigated by X-ray triple-axis diffractometry and X-ray reflectivity at glancing incidence. The X-ray analysis of these structures shows that they are pseudomorphic and yields information in particular on the interface roughness parameter evaluated in terms of the electron density profile which follows from the analysis of the X-ray reflectivity.

The evolution of strain relaxation of molecular-beam-epitaxy-grown Ge_1-xSi_x layered structures deposited on Ge(001) substrates was investigated by employing high-resolution X-ray reciprocal-space mapping. In contrast to the commonly investigated SiGe alloy layers on Si(001) substrates, in this case, the Ge_1-xSi_x alloy layers experience a bi-axial tensile strain due to the larger lattice constant of the Ge substrate, rather than a bi-axial compressive strain. The structures consist of a linearly graded Ge_1-xSi_x alloy buffer (B1), followed by an alloy buffer layer (B2) with constant Si content. For the four samples investigated, the grading rates of B1 were varied in the range 13.2-32.5% Si mu m^-1, and the thicknesses were varied in the range 0.4-1.8 mu m. From our measurements, it turns out that the lower part of the graded buffer B1 close to the Ge(001) substrate is completely strain-relaxed, whereas the top part of the graded buffer and the uniform alloy buffer (B2) are only partly strain-relaxed, showing a linear increase in the in-plane strain towards the surface. In addition, the measured maximum in-plane strains at the surface depend on the grading rate of B1 as well as on the thickness of B2. These strains turn out to be much larger than those predicted by existing equilibrium theories for strain relaxation in epitaxial layers.

Magnetic circular dichroism (MCD) is demonstrated to allow the simultaneous characterization of the V²⁺ and V³⁺ ions in vanadium-doped Bridgman grown CdTe bulk samples. The identification of the MCD bands relies on the analysis of n- and p-type co-doped samples as well as on spectroscopic arguments. A direct correlation of the strength of certain MCD bands with the V³⁺ concentrations determined by electron paramagnetic resonance spectroscopy allowed the calibration of our spectra for V³⁺. Based on photo-MCD experiments, we also provide a calibration of MCD bands related to V²⁺. Finally, the [V²⁺]/[V³⁺] ratio can be derived simply from the shape of the MCD spectrum in the low-energy region (E<1.1 eV).

A method for the characterization of oxide-polysilicon-oxide on silicon by monochromatic ellipsometry is presented. By refining a previous work which enabled an estimation of the thickness of the top oxide from a single ( Delta , psi ) measurement, the procedures we propose here allow the simultaneous determination of two unknowns in our samples. The practical process is illustrated by several examples, using a graphical approach which follows from our inversion methods.

Pb_0.2Sn_0.8Te thin films of different thicknesses have been deposited on glass substrates by flash evaporation. Electrical resistivity and thermoelectric power measurements have been carried out in the temperature range 300-500 K as a function of film thickness. The thickness dependence of the electrical resistivity and thermoelectric power observed has been explained using the effective mean free path model of classical size effect theory. The thermoelectric power factor was calculated from the measured values of electrical resistivity and thermoelectric power. The thermoelectric power factor is found to be dependent on thickness and temperature. By taking thermal conductivity values from the literature, the thermoelectric figure of merit was also calculated and it was found that this too was dependent on the temperature and thickness.

Deep level transient spectroscopy (DLTS) characterization was performed after forming iron silicides on silicon by a solid state reaction at 450-850 degrees C. No deep levels were discovered in n-type silicon when the reaction took place on cleaned previously unprocessed wafers. When the silicide formation was applied to samples given a standard silicon oxidation procedure, two defect levels, correlated to the presence of iron, were detected by DLTS. None of the levels could with any degree of certainty be identified. The levels observed in n-type oxidized silicon had activation energies of approximately 0.30 and 0.33 eV. In p-type silicon a defect level appeared with activation energy of 0.40 eV both in chemically cleaned previously unprocessed and in oxidized silicon samples. The observed defect levels at 0.33 eV in n-type silicon and at 0.40 eV in p-type silicon showed defect profiles with lower concentration towards the silicide/silicon interfaces. All observed traps both on n- and on p-type silicon were removed by prolonged annealing or increased annealing temperature. This suggests that the formation of iron silicides on silicon may be compatible with silicon integrated circuit technology when considering electrically active defects.

In this study the system Ti-Si-C was investigated in terms of two aspects, metallurgical and electrical, in order to understand the formation and the properties of Ti electrical contacts on n-type 6H-SiC. For the metallurgical investigation bulk diffusion couples were prepared from monocrystalline 6H-SiC and Ti and annealed between 700 and 1200 degrees C for different lengths of time. The reaction zones were investigated using a SEM (secondary electron and backscattered electron images as well as energy-dispersive X-ray analysis). For the investigation of the electrical properties Ti contacts were sputter-deposited onto 6H-SiC wafer stripes and annealed at similar temperatures. The contact properties were measured in terms of current-voltage characteristics. We discovered that, over the whole temperature range investigated, the reaction layer growth follows a parabolic growth law which is thermally activated. Above 1200 degrees C the diffusion path from SiC to Ti is SiC/Ti₃SiC₂/Ti₅Si₃/two-phase Ti₅Si₃+TiC_1-y/Ti₅Si₃/Ti. The contacts show ohmic behaviour. Between 1000 and 800 degrees C the diffusion path is: SiC/Ti₃SiC₂/Ti₅Si₃/two-phase Ti₅Si₃+TiC_1-y/Ti₃Si/Ti. The contacts are also ohmic. Below 700 degrees C the diffusion path is SiC/TiC_1-y/two-phase Ti₅Si₃+TiC_1-y/Ti₃Si/Ti. The contacts are of Schottky type. It is concluded that the phase in contact with the SiC determines the electrical properties of the junction and the possibility of manufacturing a complete Schottky diode using n-type 6H-SiC and Ti only is demonstrated.

The time dependence of hot carrier degradation in PMOSFETs is simulated on a 2D device simulator, MEDICI. The degradation is modelled by a spatially uniform negative fixed charge, introduced at the oxide-silicon interface. The effect on the device drain current, threshold voltage and transconductance is studied as a function of the extent and magnitude of the fixed charge. The extent of the fixed charge is identified with the drain extension due to the inversion, by the fixed charge, of part of the channel adjacent to the drain junction. The drain extension is experimentally determined from the gate-to-drain capacitance, whose stress time dependence therefore gives an indirect time dependence for the degradation of other device parameters. Simulation results show a logarithmic time dependence of degradation in a number of device parameters, confirming earlier experimental observations. Simulation of the electron injection current reveals two peaks, the first one near the drain junction and the second one near the edge of the degradation region. The second peak is considerably weaker and accounts for the progressively longer time needed to create additional degradation.

The feasibility of numerical simulation of kinetic processes in narrow-gap semiconductors under heavy field emission current is demonstrated. The description of the conduction-band electron behaviour based on the kinetic approach to the process of electron field emission from semiconductors is carried out using the particle method. The three-dimensional character of electron scattering in the momentum space by the phonons and ionized impurities is taken into account. Impact ionization initiated by electrons is also considered as the essential electron energy dissipation mechanism for narrow-gap semiconductors. On the basis of the numerical results for a one-dimensional n-InSb-emitter structure it is found that at room temperature substantial electron gas cooling takes place near the n-InSb-vacuum interface and the emission process is governed by thermalized electrons from the Gamma valley.

In this article we present a theoretical investigation of the tunnelling current in resonant tunnelling diodes based on a realistic tight-binding approach. The hydrostatic pressure dependence of the tunnelling current is calculated and compared with available experimental results. An important contribution from the X valley of the barrier region is demonstrated, which becomes especially significant at high pressures. The level crossing between X and Gamma localized states is shown to modify the peak value of the tunnelling current.

Current-voltage characteristics of Pd₂Si/p-Si(111) Schottky barrier diodes studied over a wide temperature range (60-201 K) are shown to follow a thermionic emission-diffusion mechanism under both the forward and the reverse bias conditions. The barrier parameters as evaluated from the forward I-V data reveal a decrease of zero-bias barrier height ( phi _b0) but an increase of ideality factor ( eta ) and series resistance (R_s) with decrease in temperature. Moreover, the changes in phi _b0, eta and R_s become quite significant below ~100 K. An In(I_s/T²) versus 1/T plot is found to fit well with two straight lines in different temperatures regimes giving an activation energy of 0.33 eV (201-107 K) and 0.24 eV (below 107 K) and an effective Richardson constant of 33 A cm^-2 K^-2. However, the activation energy of 0.33 eV corresponds to the zero-bias barrier height at absolute zero. An In(I_s^f/T²) versus 1/ eta T plot is suggested to obtain the flat-band barrier height and the effective Richardson constant; the corresponding values obtained are 0.401 eV and 32.2 A cm^-2 K^-2 respectively. It is shown that the `T<sub>0</sub> effect` cannot account for the apparent increase in ideality factor and decrease of barrier height at low temperatures. Finally, the decrease of barrier height with voltage under the reverse bias condition is attributed mainly to interfacial layer effects with a small contribution due to image force lowering.

A photovoltaic cell was fabricated by sandwiching a monolayer of the pigment cyanidin adsorbed on nano-porous n-TiO₂ film (deposited on conducting tin oxide glass) within a transparent polycrystalline film of p-CuI, filling the intercrystallite pores of the porous n-TiO₂ film. Photoexcited dye is found to inject electrons into n-TiO₂ and holes into p-Cul, generating photocurrents and photovoltages that are impressively high for a dye-sensitized solid-state photovoltaic cell.

The differential resistance characteristics of junctions formed from annealed granular superconducting contacts and GaAs:AlGaAs heterostructures are reported. At low bias voltage there is a smooth resistance dip due to Andreev reflection at the interfaces and diffusive electron transport in the semiconductor. At higher bias there is a peak in the resistance corresponding to the onset of phonon-mediated processes, and further structure with an unusual temperature dependence. The width of the Andreev dip and its apparent insensitivity to temperature are due to a combination of the temperature dependence of the superconductor gap, the thermal carrier distribution in the normal regions, and the geometry of the contacts. Disorder near the interfaces is believed to be responsible for the high degree of Andreev reflection, and this increased probability persists in an applied magnetic field, as has been predicted.

An alternative process scheme for dry local oxidation is presented. It has several advantages when compared with the conventional wet process. For example, dry local oxidation of silicon reduces the bird`s beak length as well as retarding the appearance of field oxide thinning in submicrometre mask openings. Furthermore, it avoids detrimental gate oxide thinning and reduces the complexity of the process, because it can be performed at the same time as the well drive-in (DW-LOCOS). Preliminary results demonstrate that there is no degradation in the electrical characteristics of diodes and MOS capacitor test structures, thus permitting DW-LOCOS to be advantageously incorporated into CMOS-VLSI processing sequences.

A new method is proposed for the creation of a dynamic quantum superlattice in a two-dimensional electronic system in a semiconductor crystal at low temperatures. This method is based on the action of an intense piezoelectric field, produced by a standing transverse hypersonic wave, on a two-dimensional electron system placed near the crystal sample surface. The standing hypersonic wave is excited under the action of a microwave field in a prismatic beam resonator made from a dielectric crystal with strong piezoelectric coupling.