Table of contents

The informal meeting of Nordic semiconductor scientists and engineers, more than 30 years ago in Finland, has developed into a biannual Nordic Semiconductor Meeting. With increasing number of participants and broadening fields of interest, the Nordic Semi-conductor Meeting has become international; first with participants from Baltic countries and from St. Peterburg, and then from all over the world. The present 17th Nordic Semiconductor Meeting has record high 161 partici- pants and 30 invited speakers from 21 countries.

The scientific programme of the 17th Nordic Semiconductor Meeting is well balanced between science and technology, between basic research and industrial application, and between materials fabrication, experimental characterization, theoretical analysis and device design. The 30 invited lectures, 60 oral contributions, and 46 poster contributions cover important topics in theory, experiments, devices and sensors, materials, and modelling. A particular feature of the present Meeting is the special session "Microelectronic Technologies" with 4 invited speakers to give overview of the semiconductor industrial research activities in the Nordic countries. The 17th Nordic Semiconductor Meeting has delivered undoubtedly the message that today's science, which is based on today's technology, is tomorrow's technology.

We thank all the participants for their contributions, and we are grateful to the sponsors Nordita and NorFA. The opening lecture of the Meeting was given by Professor Hermann G. Grimmeiss, to whom the 17th Nordic Semiconductor Meeting was dedicated for his important contribution to the development of semiconductor sciences and technologies in the Nordic countries.

Trondheim, November 7, 1996

We present a semiclassical study of a transport process, the tunneling, in the presence of a magnetic field and a dissipative environment. We have found that the problem can be mapped onto an effective one-dimensional one, and the tunneling rate is strongly affected by the magnetic field, such as a complete suppression by a large parallel magnetic field, an example of the dynamical localization. In such case a small perpendicular component of the field, or the dissipation, can enhance the tunneling rate. In the small parallel field and finite temperatures the tunneling rate is finite. Explicit expressions will be presented in those cases. If viewing the tunneling in the presence of a magnetic field as a dissipative tunneling process, by varying the magnetic field and the potential one can obtain the dissipative spectral function between the subohmic s = 0 and the superohmic s = ∞. In combination with a real dissipative spectral function, the effect of the magnetic field can map the spectral function from s to 2 − s, with s > 2 mapping to s = 0, revealing a dual symmetry between the friction and the Lorentz force. Two cases relevant to experiments, the edge state tunneling in a Hall bar and the tunneling near the dynamical localization will be discussed in detail.

We investigate the autocorrelator of conductance peak heights for quantum dots in the Coulomb blockade regime. Analytical and numerical results based on Random Matrix Theory are presented and compared to exact numerical calculations based on a simple dynamical model. We consider the case of preserved time-reversal symmetry, which is realized experimentally by varying the shape of the quantum dot in the absence of magnetic fields. Upon a proper rescaling, the correlator becomes independent of the details of the system and its form is solely determined by symmetry properties and the number of channels in the leads. The magnitude of the scaling parameter is estimated by a semiclassical approach.

The current status of GaInAsSb/AlGaAsSb and InAsSb/InAlAsSb strained quantum-well (QW) lasers grown by molecular beam epitaxy is described. At ∼2 μm, GaInAsSb/AlGaAsSb lasers grown on GaSb have room-temperature pulsed threshold current density of 143 A/cm², cw power of 1.3 W, and a maximum cw operating temperature of 130 °C. Similar lasers emitting at ∼2.7 μm have operated pulsed up to 60 °C and cw up to −39 °C. At ∼3.5 μm, InAsSb/InAlAsSb QW lasers grown on InAs have exhibited cw power of 215 mW/facet at 80 K, with pulsed operation up to 225 K and cw up to 175 K. Grown on GaSb, the InAsSb/InAlAsSb QW lasers emitting at ∼3.9 μm have maximum pulsed and cw operating temperatures of 165 and 128 K, respectively. The performance at longer wavelengths is limited by Auger recombination.

Short period lateral superlattices are fabricated on AlGaAs/GaAs-Heterostructures by electron beam lithography. The samples are experimentally investigated by low temperature magneto transport experiments. For small values of the lateral potential modulation, the amplitude of the Shubnikov-de Haas oscillation is modulated, reflecting the oscillating band width of the Landau bands. For increasing values of the potential modulation the Landau band width increases, and the internal structure of the Landau bands is resolved. In particular, a splitting of the maxima of SdH oscillation is observed and attributed to the dominant energy bands and gaps within the Hofstadter Butterfly energy spectrum. These experiments constitute the experimental observation of band structure effects, arising from an artificially imposed superlattice potential.

We consider important modeling issues accentuated by the scaling of field effect transistors (FETs) into the submicron regime. High electric fields enhance non-ideal phenomena such a channel length modulation, hot-electron effects, and bias dependent mobility. In addition, improper scaling of device dimensions and doping levels give rise to so-called short-channel effects resulting from a weakening of the gate control over the channel charge, causing threshold voltage shift, increased leakage current, and ultimately punchthrough. Our modeling approach is based on a unified charge-control model (UCCM), which allows us to combine accuracy and good convergence properties with physically based model parameters. The various effects associated high fields and short channels in FETs are contained in a description of the linear conductance, the saturation current, and the subthreshold current, which are combined in unified and continuous expressions for the drain current and the intrinsic capacitances using UCCM.

Books as well as numerous articles have been written about hydrogen passivation in classical semiconductors such as Si and GaAs. The subject has gained a renewed interest recently since hydrogen is widely considered to saturate the hole conductivity of the wide bandgap semiconductors GaN and ZnSe which are currently most promising for blue light emitting devices. Other group-I impurities are capable of compensating the electrical conductivity of semiconductors both through directly neutralising (passivating) the impurity or providing space charge of polarity opposite to that of the dominating one. The paper reviews the similarities and differences between hydrogen and its light metallic neighbour in the periodic table, lithium. Also we provide a comparison with the heavier interstitial copper which is known for its ability to passivate shallow acceptors. Finally fundamental differences between shallow-level and deep level passivation will be addressed.

The paper summarizes a few basic properties of SiGe showing that SiGe is an interesting material for high speed electronics. The advantage of using heterostructures in silicon-based technologies is demonstrated by taking SiGe heterobipolar transistors (HBTs) as an example. First results obtained with very fast and low-noise HBTs are briefly mentioned. The paper is concluded by a short discussion of a few optoelectronic properties observed in various Si/Ge, Si/Si_1−xGe_x and Si_1−x−yGe_xC_y strained-layer superlattices and quantum wells with particular emphasis on electroluminescence properties and the benefit of these materials for new buffer concepts.

There is a lot of interest in obtaining efficient infra-red (IR) light emission from Si-based structures for use in optoelectronics. Although it has been theoretically predicted that Si_m/Ge_n atomic layer superlattices can have a quasi-direct bandgap, the experimental studies have not yet given very high luminescence intensities, particularly at room temperature, from such structures. So far, the most efficient method to have IR light emission at room temperature is to process Si/Si_1-xGe_x superlattices or quantum well structures into narrow (<60nm diameter) columnar structures. After planarization with insulating material it has been possible to fabricate LEDs using these columns. While the results are very promising there are also a number of unsolved problems concerning the mechanism allowing for efficient light emission and concerning the passivation of the surfaces of the columns to have a long-term stability of the emission. Another way to have IR light emission at room temperature and possibly obtain a Si-based laser is to use Er-doped material. For Er-doped LEDs, most of the work has been done on ion-implanted structures. It has been found that to have the Er-related emission at 1.54 μm it is necessary to also have co-dopants like O or F to activate the Er. Since a high temperature step is necessary to anneal out implantation damage it has been difficult to have high concentrations of Er/O without precipitation, as the required concentration for useful devices is far above the solid solubility of Er in Si. Low temperature growth using MBE is a promising method to achieve high Er/O or Er/F concentrations without precipitation and intense room-temperature electroluminescence has very recently been reported from a reverse biased Er/O-doped LED grown by MBE.

We first discuss the work on strain and critical thickness of large area lattice mismatched epilayers of GeSi, InGaAs and II-VI semiconductors. A summary of Finite Element (FE) calculations of stresses in stripe-substrate heterostructures is then given. Calculated stresses for novel stripe designs are also given. The results are then applied to interpret experiments. Raman and luminescence experiments on stripes of GaAs/Si, GeSi/Si and InGaAs/GaAs, Raman experiments on LOCOS structures and on GeSi quantum wires and quantum dots and luminescence experiments on InGaAs quantum wires are included in the discussion.

The cyclotron resonance absorption spectrum in a Wigner crystal is calculated. Due to band-structure effects, the electrons at different sites have different cyclotron resonance frequencies depending on their spin orientation. To calculate the absorption line shape we consider effects of Coulomb (dipole-dipole) interaction between the excitations at different sites. The averaging over different realizations of the spin arrangement is done within the coherent potential approximation. The results show a crossover from a double-peak spectrum at small filling factors where the dipole-dipole interaction is weak, to a single-peak spectrum at filling factors larger than about 1/6 where the interaction effects are completely dominating. As the electron filling factor is increased further, the absorption line becomes broader due to radiation damping. The results are in good agreement with experiments.

The volume 1/f noise in GaAs, Si and 6H-SiC appears as a result of fluctuations of the occupancy of those levels that form the density-of-state tails near the band edges. Tails of this kind are due to the presence of imperfections (defects, local stresses, impurities, clusters, etc.) and exist in any real crystal. A model of the volume 1/f noise in semiconductors is put forward.

The model predicts the existence of several physical effects associated with the same density-of-state tails near the band edge which are responsible for the volume 1/f noise. These effects are: a nonmonotonic dependence of the 1/f noise on the intensity of the band-to-band illumination (minority carrier concentration), a specific mechanism of a persistent photoconductivity, an increase of the 1/f noise intensity with deteriorating of the structural perfection of a semiconductor.

All these effects have been observed experimentally in GaAs and Si.

A convergent renormalized perturbation series in powers of the electron-electron coupling constant λ is proposed for calculating the energy of a quantum dot with hard wall confining potential. The renormalization procedure is based on the standard perturbation series, the scaling properties of the system and the asymptotic energy expansion valid in the λ → ∞ region. To obtain that asymptotic expansion the vibration excitations of the electron ring are considered. As the asymptotic expansion shows nonanalytical behavior a special procedure for matching the perturbation series with an arbitrary asymptotic function is developed. The method is illustrated by calculating the ground and some excited states of 2, 3 and 4 electrons in the dot. The comparison of the obtained results with the available exact numerical results shows the great accuracy of the proposed method over the whole range of λ values.

A review is given of recent experimental studies of Si-Ge interdiffusion and boron and antimony diffusion in relaxed and biaxial-tensile and biaxial-compressive strained Si_1-xGe_x for 0 ⩽ x ⩽ 0.50.

We have measured photon-assisted tunneling through a quantum dot with zero dimensional (0D) states. When the photon energy exceeds the separation between 0D-states we observed photon induced excited state resonances, as well as photon sideband resonances. We study the strength of these resonances as a function of the applied microwave field, and compare them to calculations.

AlGaN/GaN Heterostructure Field Effect Transistors (HFETs) are expected to have a high transconductance and a high saturation current because of the relatively high electron mobility and high saturation velocity of the two dimensional electron gas (2DEG) in AlGaN/GaN modulation doped structures. Our experimental results revealed that, in the fabricated devices, the transconductance is limited by a large source series resistance, R_s, and by a small sheet carrier concentration, n_s, of the 2DEG in the device channel. In order to increase n_s, we recently proposed to use a Doped Channel HFET (DC-HFET) design, similar to that used in AlGaAs/GaAs DC-HFETs. In AlGaN/GaN structures, an additional impurity scattering caused by the channel doping has relatively less of an effect on the electron mobility than in AlGaAs/GaAs devices, due to a much higher electron effective mass in GaN. One micron gate AlGaN/GaN Doped Channel HFETs reached the cutoff frequency f_T of 18.3 GHz. Short channel AlGaN/GaN Doped HFETs with 0.25 micron gates demonstrated the cutoff frequency f_T up to 36.1 GHz and the maximum oscillation frequency, f_max ≈ 97.3 GHz. These values of f_T and f_max are the highest value reported for any wide band gap semiconductor device. Our simulations of AlGaN/GaN Doped Channel HFETs show that transconductances in access of 1000 mS/mm may be obtained for very thin AlGaN layers, a large gate voltage swing, and for a large surface sheet density of the 2D electron gas.

We report successful experimental synthesis of pure crystalline C-N film on silicon and nickel substrates by bias-assisted hot filament chemical vapor deposition (bias-HFCVD). The obtained films have been firmly characterized by energy dispersive x-ray (EDX) and Auger electron microscopy for the chemical composition, x-ray diffraction (XRD) and selective-area electron diffraction (SAED) for the lattice structure, and scanning electron microscopy (SEM) and transmission electron microscopy (TEM) for the crystal morphology and symmetry. Our experimental lattice constants of α-C₃N₄ (a = 6.38 Å, c = 4.648 Å) and β-C₃N₄ (a = 6.24 Å, c = 2.36 Å) with a relative N/C composition of 1.30-1.40 on nickel are in good agreement with the ab initio calculations by less than 1.3% and 2.5%, respectively. A new tetragonal C-N phase with a = 5.65 Å and c = 2.75 Å was identified for the first time in the laboratory. Furthermore, a specific C₃N₄ phase growth was studied by adding a fraction of hydrogen. A "lattice match selection" for the growth of C₃N₄ crystals on silicon is proposed.

Flip chip mounting of bare dice is gaining widespread use in microelectronics packaging. The main drivers for this technology are high packaging density, improved performance at high frequency, low parasitic effects and potentially high reliability and low cost. Many companies have made significant efforts to develop a technology for bump processing, bare die testing and underfill encapsulation to gain the benefit of all potential advantages. We have focussed on low cost bumping of fully processed silicon wafers to develop a flexible scheme for various reflow requirements. The bumping process is based on galvanic plating from an alloy solution or, alternatively, from several elemental plating baths. Sputtered Mo/Cu or Cr/Cu is used as a wettable base for electroplating. Excess base metal is removed by using the bumps as an etching mask. Variation of the alloy composition or the layer structure, allows the adjustment of the bump reflow temperature for the specific requirements of the assembly. Using binary tin-lead and ternary tin-lead-bismuth alloys, reflow temperatures from 100 °C (bismuth rich alloys) to above 300 °C (lead rich alloys) can be covered. The influence of the plating current density on the final alloy composition has been established by ion beam analysis of the plated layers and a series of reflow experiments. To control the plating uniformity and the alloy composition, a new cup plating system has been built with a random flow pattern and continuous adjustment of the current density. A well-controlled reflow of the bumps has been achieved in hot glycerol up to the eutectic point of tin-lead alloys. For high temperature alloys, high molecular weight organic liquids have been used. A tensile pull strength of 20 g per bump and resistance of 5 mΩ per bump have been measured for typical eutectic tin-lead bumps of 100 μm in diameter.

This paper describes an optical gas concentration measurement system, which is based on an electrically tuneable micromachined Fabry-Perot interferometer (FPI). The operating principle of the system is NDIR Single-Beam Dual-Wavelength measurement. The FPI is tuned so that the pass band coincides with the absorption band of the measured gas; a detector records the strength of the signal getting through the measurement chamber. The pass band of the FPI is then shifted to either side of the absorption band; the detected signal constitutes the reference signal. The ratio of these two signals indicates the degree of light absorption and so the gas concentration.

Properties of the FPI are discussed, as well as the performance of the measurement system.

Results of an experimental investigation of possibility to use laser-induced centres as a nuclei for silicon oxide (SiO₂) precipitation processes in silicon are reported. The additional centres are generated during the laser annealing of silicon wafer by the double frequency of Q-switched Nd: YAG laser (I = 0.8-2.1 J/cm², λ = 0.53 μm, τ = 15 ns) without introduction of any additional impurity. It was shown, that the additional centres introduced during the laser annealing can act as nuclei for oxide precipitates. On the base of this phenomenon the elaboration of a method of controllable SiO₂ precipitates formation is possible.

It is shown that there is a regime below the critical current, I_c, for breakdown, where the longitudinal Hall resistivity grows essentially exponentially with the transport current. According to the present theory this subcritical behavior determines the value of I_c.

The laser-annealing effects on 1000 Å thick SiO₂ layers deposited by APCVD (Atmosphere Pressure Chemical Vapor Deposition) in a-Si/SiO₂ and SiO₂/a-Si structures were investigated. In both structures, the permeation of Si and H atoms into SiO₂ layer was observed, which may resulted in the degradation of breakdown field of SiO₂ layer. The pre-breakdown field of SiO₂ in SiO₂/a-Si structure was deteriorated considerably from 3.8 MV/cm to 2.3 MV/cm at the 300 mJ/cm² of laser energy density, and the breakdown occurred at 2.2 MV/cm while the breakdown field slightly changed in a-Si/SiO₂ structures.

The effects of two-step annealing on the device characteristics of polysilicon thin film transistors are investigated. The device characteristics such as drain current, mobility and threshold voltage of the low temperature furnace annealed polysilicon thin film transistors is improved considerably due to the reduction of in-grain defects by post-laser annealing. The post-high temperature furnace annealing (900°C) on the low temperature furnace annealed devices does not improve the device performance of polysilicon thin film transistors, due to the facts the annealing temperature of 900°C is much lower than the melting temperature (⩾ 1300°C) of polysilicon. We have also found that the activation energy and hydrogen passivation rate in two-step annealed polysilicon thin film transistors is also improved significantly by the reduction of tail states due to the elimination of in-grain defects.

An experimental technique using optical excitation by a YAG laser pulse for studying avalanche injection in power devices is demonstrated. This technique enables the creation of high uniform excess carrier concentrations in an optically defined device volume, involving very little heating. A method for determining the onset of avalanche multiplication, by studying the time integral of the reverse recovery current, is proposed. A PiN diode is observed to turn off from avalanching at a dissipated power density of more than 200 kW/cm².

We use perturbation theory for calculating the energy levels of a circular, hard-wall confined quantum dot, and present results for the lowest-lying levels for 2, 3 and 4 electrons. The method may give qualitative information about the energy levels of ultrasmall quantum dots, and is an essential ingredient for constructing interpolating expressions for the energy valid for dots of arbitrary size.

Radiation effects in small-geometry MOS devices are analysed using an UCCM based model. Parameters dependence upon the total dose are obtained by fitting experimental dc device characteristics, using appropriate parameter extraction procedures. The simulated current-voltage characteristics fit the measured ones accurately. Large and small-signal parameters model, for n and p-channel devices, are given as a function of the radiation dose.

Langmuir-Blodgett films of polyaniline have been fabricated and electrically characterized. The conductivity of the doped films was studied as a function of film thickness, temperature (77 K < T < 300 K) and electric field (0 < F ⩽ 10⁵ V/cm). The low-field conductivity obeyed the behaviour σ(T) = σ₀ exp [-(T₀/T)^1/2] over the whole temperature range measured, and ln (F) ∝ F^{1/(1 + ν)} at higher fields. The result indicate variable-range hopping in a parabolic Coulomb quasi-gap.

We investigate the effects of compressible and incompressible states on the FIR-absorption of quantum wires and dots in a homogeneous perpendicular magnetic field. The electron-electron interaction is treated in the Hartree approximation at a finite low temperature. The calculated dispersion of the collective excitations reproduces several experimental results.

The preparation of CdTe single-crystalline monodisperse powders in cadmium chloride flux modified by adding elemental tellurium was studied for production different particle sizes (30-300 μm). The growth of monograin powders was carried out in isothermal conditions in closed evacuated ampoules. It was found that at lower temperatures (500-650 °C) after maintaining the equilibrium conditions primary crystals grow mainly by deposition of material from saturated solution onto smooth surfaces of crystals. In higher temperature region (700-800 °C) there occurs remarkable sintering of particles and grains loose their round shape. The activation energy for volume growth ΔE_vol = 0.9 eV was found. Te in flux was found to prevent the formation of aggregates through sintering and to increase the crystal growth velocity.

Fiber lenses were made using a short piece of single mode fiber fusion spliced to the end of a graded index multimode fiber Ø = 125 μm and the end of the single mode fiber is rounded with fusion splicing. Nearfields, farfields and reflections from flat aluminium plate were measured. The fiber lenses were also modelled.

In this paper we present a comparative study of the optical DC and transient properties of the Si- and GaAs-metal-semiconductor-metal photodetectors (MSM-PD's). The interdigitated metal fingers for the detectors were fabricated on n-type Si and semi-insulating GaAs substrates by using optical lithography and lift-off technique. The width and spacing of the interdigitated metal fingers varied between 1 and 2 μm, and the detector area varied between 100 × 100 μm² and 500 × 500 μm². For Si-MSM-PD's we used 150 nm thick aluminum and for GaAs-MSM-PD's 280 nm thick Ti/Pt/Au multilayer metallization for the Schottky barrier formation. The Schottky diode capacitance vs. bias voltage as well as dark and illuminated I-V characteristics as a function of optical power have been measured. Also the photoresponsivity vs. wavelength has been measured for various bias voltages. The speed of the detectors was studied by using a colliding pulse mode-locked laser. The experimental results show that the leakage currents in GaAs-MSM detectors are much smaller than in Si detectors, but still a reasonable photoresponsivity can be achieved also in Si detectors. The speed of the Si detectors could be improved by fabricating them on an epitaxial n⁺ - n⁻ -structure, which results in a fast impulse response 625 ps (FWHM), 292 ps (fall time).

Direct energy conversion from sun light to electricity using crystalline silicon based solar cell modules has been commercially available for several years. Above 20% conversion efficiencies have been realised in laboratory scale since 1990 [1]. The more important issue in the present solar cell fabrication development is to adopt such procedures and equipment that allow practical module efficiencies with less material and processing costs. One way to cut the material costs is to use thinner silicon wafers, which gives more wafers per length of a silicon log.

In a high conversion efficiency cell the thickness has to be well below electron diffusion length for collection of the energy from long wavelength part of the solar spectrum. The diffusion length is an indicator of the material quality. Thin crystalline silicon solar cell technology gives the opportunity to choose cheaper, lower quality silicon as a starting material for the cell process without severe reduction in the conversion efficiency.

Optimisation of a thin crystalline silicon solar cell design includes efficient light trapping and passivation of the cell surfaces. The proportion of a cell surface area to the cell volume increases as thickness is reduced. This magnifies the relative importance of surface recombination of the charge carriers as compared to the volume recombination in a solar cell. Although the quality demand on the cell material is relaxed, high conversion efficiency thin solar cell technology puts more weight on the quality of the surface passivation.

In this paper, the fabrication procedure of thin crystalline silicon solar cell samples used in H.U.T. Electron Physics Laboratory is presented. The sample material is thinned using a 20% hot KOH bath to a nominal 1007 μm thickness. The standard cell sample is a 24 mm × 24 mm square with a p-type substrate, aluminium doped back surface field and phosphorous front diffusion. Aluminium evaporation through a stainless steel mask is adopted as a quick and simple metallisation scheme for both front and back contacts [2].

The current-voltage-characteristics together with spectral response data of a typical thin crystalline silicon solar cell is also illustrated for the evaluation of the process. The measurement data is evaluated using the procedures and software developed by the author [2, 3]. The software includes diode modelling of the solar cell current-voltage data and breaking the spectral response to contributions from the emitter, depletion region and the base with numerical estimates on diffusion length, surface recombination velocities and charge carrier lifetimes.

We study the transresistivity ρ₂₁ (or equivalently, the drag rate) of two Coulomb-coupled quantum wells in the presence of a perpendicular magnetic field, using semi-classical transport theory. Elementary arguments seem to preclude any possibility of observation of "Hall drag" (i.e., a nonzero off-diagonal component in ρ₂₁). We show that these arguments are specious, and in fact Hall drag can be observed at sufficiently high temperatures when the intralayer transport time τ has significant energy-dependence around the Fermi energy ε_F. The ratio of the Hall to longitudinal transresistivities goes as T²Bs, where T is the temperature, B is the magnetic field, and s = [δτ/δε](ε_F).

In this paper we present unified models for the main noise contributions of the MOSFET drain current spectral density: the flicker noise and the thermal noise. The new noise model is consistent with a unified drain current model. All equations present an infinite order of continuity through all operating regimes. This makes the model very useful to study the noise in analog circuits, especially in low-power applications.

We theoretically study free electron light absorption for a sample which is placed in a strong, time-dependent uniform electric field. In the case of static fields one observes the Franz-Keldysh effect: finite absorption for photon energies below the band gap. We refer to this phenomenon as the Franz-Keldysh tail. Modulation of the spectra above the band gap is also found. These static effects are observed in both 3D, 2D and 1D systems. Our analysis, based on nonequilibrium Green function techniques, shows that an analogous effect takes place in time-dependent fields: the dynamical Franz-Keldysh effect. Specifically, we relate the nonequilibrium absorption coefficient to the generalized density of states and analyze how it is affected by an external harmonically oscillating electric field. The dynamical Franz-Keldysh tail should be experimentally observable in the THz regime in 1D, 2D and 3D systems. Modulation above the band gap is weak in the 3D case but considerable fine structure is predicted for 1D and 2D systems. A characteristic feature of the dynamical Franz-Keldysh effect is that the absorption edge will be shifted up by an amount corresponding to the average kinetic energy of an free electron placed in the oscillating external electric field.

Effects of the drift region doping on the device characteristics in SOI BMFET are investigated by two-dimensional device simulation. The simulation results show that higher blocking capability with RESURF principle, larger current gain in high drain current and improved transient behaviour can be obtained by optimizing the drift region doping concentration. It is found that the SOI BMFET with n-resurf layers exhibits not only the excellent on-state characteristics but also higher breakdown voltage.

We consider the Luttinger Hamiltonian for the degenerate Γ₈ valence bands. Based on the bulk Hamiltonian we develop a Hermitian Hamiltonian for heterostructures, where the Luttinger parameters are spatially varying. The correct boundary conditions are derived from the minimal mathematical restrictions on the resulting coupled set of second order differential equations. The equations can be generally written to k_αD^αβ_jj'k_βF_j' + E_νF_j = εF_j' with the effective mass tensor fulfilling the symmetry property (D^αβ_jj')^† = D^βα_j'j. This requires F_j' and D^αβ_jj'k_βF_j' continuous in the effective mass equation. These boundary conditions automatically imply the conservation of current in the sample. This strict mathematical treatment of the boundary conditions for the multiband effective mass equation for heterostructures can reconcile the controversy on this subject. In an Appendix we comment on the new envelope function method for heterostructures, developed by Burt.

The chemistry of formation of copper sulfide and copper indium disulfide films by the chemical spray pyrolysis process has been investigated. It was established that the formation of copper sulfide films by the spray process of water solutions of CuCl₂ and SC(NH₂)₂ passes through the stage of formation of intermediate complex compound Cu(SCN₂H₄)Cl·H₂O in the initial solution. Thermal behaviour studies in air exhibit rearrangement of the complex at temperatures higher than 210 °C with formation of copper sulfide. Copper oxide is the decomposition product at temperatures higher than 700 °C in air. CuInS₂ is formed by the reaction between copper and indium sulfides.

We present a simple analytical model for the drain current dependent cutoff frequency f_T in submicron high frequency MOSFET's. The model combines the SPICE MOS3 drain current model and a modified McMacken-Chamberlain model for the current dependent transit time, which allows the description of the I-V characteristics and f_Tvs. current with the same set of SPICE parameters. The model is verified by comparing simulated results to measured data for 0.6 μm RF MOSFET's fabricated at VTT Electronics. The model explains well the observed behaviour. Furthermore, the comparison indicates that when the channel length is reduced to a half-micron range and below, most of the charging delay is caused by the transit time.

We present evidence from current transient spectroscopy which strongly indicates a non-correlation between the well known Cu_A and Cu_B levels in GaAs, situated at 0.15 eV and 0.44 eV above the valence band, respectively. The fact that both levels are passivated following the introduction of lithium or hydrogen has been previously established. After reverse-bias annealing at moderate annealing temperatures Cu_B is reactivated independent of Cu_A. The reactivation is shown to follow an Arrhenius relationship with a thermal activation energy around 1.30eV for both passivating species. In partially Li-passivated material, a new level is introduced at 0.12 eV above the valence band. This level is passivated by further introduction of Li and reactivated during reverse-bias annealing with a characteristic activation energy of 1.11 eV. We speculate that Cu_A might be due to a double Cu_Ga acceptor which is singly or doubly passivated by Li and doubly passivated by hydrogen under the diffusion conditions of this work while Cu_B arises from a different defect. Although only one ionization level of the double acceptor is observed, this interpretation is consistent with results previously presented in the literature.

The effects of conservation of in-plane momentum and doping levels on the low-temperature characteristics of GaSb/AlSb/InAs/GaSb/AlSb/InAs broken-gap interband tunneling structures are investigated. A three-band k · p model incorporating the coupling effects among electrons, heavy holes and light holes is used to calculate the tunneling current-voltage characteristics. It is found that the kink and the small peaks in the low-temperature characteristics are results of the conservation of in-plane momentum. With a 30 Å-thick InAs well, the low-temperature characteristics change from the ones with a shoulder at low doping levels to the ones with a kink at high doping levels. In addition, the peak current density decreases at high doping levels. With a 120 Å-thick InAs well, the peak current density increases with increasing doping levels. Besides, the small peaks at low and medium voltages and the main current peak become more significant at high doping levels. The detailed carrier transport can be interpreted with the tunneling factors and the conservation of the in-plane momentum.

Phonon-assisted resonant magneto-tunneling has been considered within the coherent tunneling model. The magneto-tunneling case is particularly interesting because the effective electron-phonon coupling increases with the increasing magnetic field as g_eff ∝ B^η, where η = 1/2 for the LO-phonon coupling and 1 < η < 3/2 in the case of deformational acoustical phonons. Under the magneto-LO-phonon resonance condition, the mixing of tunneling channels leads in the case of phonon emission to a sharp splitting of the transmission peak, the splitting beeing proportional to g_eff. A smooth smearing of the peak is on the contrary seen in the case of incoherent phonon absorption processes. For the acoustical deformational interaction, we focus on the coupling to non-equilibrium phonons. It is shown that dc-current profile is very sensitive to the presence of non-equilibrium phonons, as well as to their frequency distribution. In particular, we believe that the resonant magneto-tunneling device can be useful for detection and spectroscopy of the phonons and electron-phonon interactions.

The excess carrier concentrations and currents in EBIC and LBIC measurement situations are calculated in a solar cell structure. The surface recombination velocities and following boundary conditions are taken into account by using extra deltafunction like generation or recombination points located outside the layer.

The changes of the thicknesses of plasma oxidized ultrathin silicon dioxide layers have been studied after they have been kept in room air. The thickness increases faster in the beginning and saturates then, so that the total increasement is typically no more than about 1 nm. The surface quality, produced by the oxidation process, is likely to be the main factor which affects the amount of increase. To produce very stable layers with nonreactive surfaces one has to adjust the oxidation temperature and the native oxide removal very carefully.

Papers [1, 2] by the authors of the present work propose and develop a new approach to modeling thermal (or carrier-velocity-fluctuation) noise in semiconductor devices and integrated circuits. The approach combines theory of Itô's stochastic differential equations and device/circuit theory. This innovative combination enables to form consistent mathematical basis of noise modelling and simulation. The present work is the next step in this development. It proposes a method to reduce high-frequency values of spectral densities of terminal currents in low-current silicon bipolar transistors (BTs) including modern high-speed (multigigahertz) devices. This method presents the "anti-resonance" structural modification of two-junction devices. The corresponding numerical results are calculated by the QuAnT software intended for computer-assisted learning. These data demonstrate 2-5-order reduction in the high-frequency spectral densities for a 24-GHz silicon BT at only 12% deterioration in its speed performance. The method points out the triple-diffusion-or-implantation (TDI) structure and its optional silicon-on-insulator-and-trench-isolation (SOI/TI) version as the device structures which enable to implement the proposed method. It is noticed that optimization of the conventional (rather than the above special) SOI/TI structure for high speed disagrees with the optimization for low level of high-frequency noise. High-frequency-noise minimization and speed-performance maximization are in general contradictory strategies in design of silicon BTs. An appropriate balance between them can be achieved by means of the proposed method and application of the relevant quantative-analysis tools like the QuAnT software or any other software of more capabilities.

We explore the effects of the proximity to a superconductor on the level density of a billiard for the two extreme cases that the classical motion in the billiard is chaotic or integrable. In zero magnetic field and for a uniform phase in the superconductor, a chaotic billiard has an excitation gap equal to the Thouless energy. In contrast, an integrable (rectangular or circular) billiard has a reduced density of states near the Fermi level, but no gap. We present numerical calculations for both cases in support of our analytical results. For the chaotic case, we calculate how the gap closes as a function of magnetic field or phase difference.

Differential reflection measurements are performed on a n-i-p-i modulator with superlattice in the active layer. We use a modulation ellipsometer with pulsed laser excitation of the sample, avoiding electrical contacting of the device. We present for the first time a density matrix model which can be used for calculating the dielectric function of both superlattices as well as multiple quantum wells. The model includes the off-diagonal Coulomb coupling potentials, which are crucial for superlattice structure modeling, and the applied electric field which is required in a modulator device. An optical transfer-matrix model is used to obtain the differential reflection spectra of the device, using the dielectric function of the superlattice calculated with the density matrix model. Thus a direct and quantitative comparison with the experimental spectrum is shown.

We have fabricated a poly-Si TFT using a novel oxidation method, which improves the surface roughness at the interface between the poly-Si layer and the gate oxide layer. The surface roughness at the interface is observed by transmission electron microscopy (TEM) and is also evaluated by atomic force microscopy (AFM) after the oxide layer is removed using buffered HF acid.

Compared with the poly-Si TFTs fabricated by the conventional oxidation method, the proposed poly-Si TFT exhibits the remarkable enhancement of the electrical parameters, such as the subthreshold swing and the threshold voltage. It is observed that the proposed poly-Si TFT has a higher dielectric strength and the device characteristics are not degraded significantly after an electrical stress. The improvement of the surface roughness at oxide/poly-Si interface is found to be critical to enhance the device performance.

The most popular supported catalysts in semiconductor gas sensors are noble metals such as Pd, Ag, Au and Pt which are used in order to increase both the selectivity and sensitivity of sensors. The mechanism responsible for changes in the conductance of the semiconductor support due to reactions on the catalyst is in many cases Fermi energy control. The gas changes the Fermi energy of the catalyst and the Fermi level of the semiconductor equilibrates with that of the catalyst. This, in turn, means a well-controlled dispersion of the catalyst particles on the semiconductor support. Here we study the agglomeration properties of ultra-thin sputtered noble metal layers on oxide semiconductor surfaces during annealing up to 400 °C. Ag, Au and Pd layers were sputtered on tin-dioxide thin-film surfaces and AFM (Atomic Force Microscopy) was used to study the agglomeration properties of the layers. Obtained results show that this new tool for the surface research is also very useful in the study of diffusion and agglomeration properties of noble metals on oxidic semiconductor surfaces.

Laser chemical vapor deposition (LCVD) of low resistivity copper from Cu(hfac)tmvs (copper(I)-hexafluoroacetylacetonate trimethylvinylsilane) is presented. The deposition was performed on Si₃N₄ and SiO₂ passivated ICs. The deposition of Cu lines was carried out by a focused Ar⁺ laser beam. The deposition speed and the laser power were varied in the range of 24 to 400 μm/s and 40 to 1200 mW, respectively. The precursor gas partial pressure was 0.3 mbar buffered in 10 mbar He or H₂. The morphology of the deposited lines was investigated by AFM, optical microscopy and LIMA. The LIMA analysis showed the deposited copper surface was contaminated but the contamination level decreased when the layer was depth profiled. The electrical resistivity was found to be 3.7 μΩ cm as determined by four point probe measurement. The effect of the laser beam focus location on the substrate surface is described. Low resistivity laser direct-write Cu lines were utilised in specific IC customisation cases.

A charge conserving capacitance model for GaAs MESFETs for Computer Aided Design (CAD) applications is presented. Simple analytical expressions for node charges were developed by integrating the sheet charge density over the gate length, which guarantees the charge conservation. The capacitances were defined by taking derivatives of the charges with respect to terminal voltages at the three nodes. The resulting, so-called transcapacitances are nonreciprocal in nature. Thus, we have defined a new equivalent circuit based on a total of nine transcapacitances, which can be organized as a 3 × 3 matrix. The model is valid both above and below threshold. Good agreement has been demonstrated with the experimental data.

An account is presented on the pressure-sensitive impurity centres in hard silicon carbide crystals, characterized by relatively large hydrostatic pressure shifts of the narrow photoluminescence lines belonging to these centres. Two examples of such centres are considered: a quasimolecular centre comprising Ti atom and a centre formed by an impurity atom from the Sc subgroup, most likely Y. In both cases the individual emission lines in the corresponding ABC and abc spectra exhibit considerable blue shifts amounting to 1.08-1.25 meV/kbar for ABC spectrum and 1.68-1.80 meV/kbar for abc spectrum, thus exceeding the red shifts of the well-known R emission lines in ruby (−0.094 meV/kbar) by more than an order of magnitude. For both impurity centres in SiC crystals the pressure-induced quenching of the luminescence is predicted.

Electrical properties of different low Schottky barrier junctions on high quality strained-Si_1-xGe_x layers with 0 ⩽ x ⩽ 0.24 were studied. The investigated metals were Ir, Pt, Pd and Fe. In addition, PtSi/p-Si_1-xGe_x junctions were also produced and characterized; the silicide was deposited by co-sputtering. High-resolution multi-crystal x-ray diffraction (HR-MCXRD) was used to investigate the sample quality and strain state of the MBE grown p-Si_1-xGe_x layers and to accurately determine the Ge fraction in the fabricated junctions. The Schottky barrier height of the studied junctions were found to decrease with the Ge fraction (x) in the Si_1-xGe_x layer. Different junctions showed the same dependence on the Ge fraction. The reduction of the barrier heights with increasing Ge fraction was found to be related to the valence band discontinuity in the underlying Si_1-xGe_x/Si heterojunction. It was then concluded that the cut-off wavelength of infrared detector fabricated on Si_1-xGe_x layer can be tuned in a wide range in the infrared region by changing x. A cut-off wavelength of 13.8 μm was shown to be feasible in this work.

Two different applications of PS layers in the solar cell technology were demonstrated. In the first case, microporous layers, which are formed on silicon surface after relatively long anodical etching, roughen surface and enhance the optical confinement in the wafer increasing quantum efficiency in the spectral region close to 1 μm. Such a light-trapping effect was verified by spectral transmittivity and photoresponse measurements. In the second case, thin nanoporous layers formed by a short anodical treatment on already manufactured solar cells were investigated. The increase of the short-circuit current and efficiency by nearly 30% has been observed. Such an improvement was explained by antireflective and surface passivating actions of the porous silicon layer.

A short survey is given of our computational approach for the study of the chemical surface sensitivity of some oxide and sulphide semiconductors. Many oxides and sulphides behave as n-type semiconductors with donors originating from nonstoichiometric defects. Therefore, n-type semiconductors with negligible hole concentrations are considered here. Chemical activity of n-type semiconductor surfaces is based on adsorbate interactions both with conduction-band bulk electrons and with localized electrons trapped at surface states. Some effects may also origin from a possible ionic response of the semiconductor to changing internal electric fields. A dynamical approach with expressions similar to those for the generation-recombination processes (Schockley-Hall-Read theory) is used to describe adsorbate interactions with conduction-band bulk electrons. As practical examples we consider SnO₂(110) − 1 × 1 and wurtzite CdS(1010) − 1 × 1 surfaces, both being the most stable faces of their structures, respectively, and nonpolar in the bulk derived geometry. With the SnO₂(110) surface we use clusters in calculations of localized electronic surface states originating from different defects, while a slab method is used to calculate the atomic relaxation at the CdS(1010) surface.

The radial distribution of grown-in microdefects in eight Czochralski-grown silicon crystals was measured by counting the flow pattern (FP) defects revealed by preferential etching. At the center of the crystal, the FP-defect density increased from 5.2 to 6.7 × 10⁵ 1/cm³, when the pulling speed was increased from 0.8 to 1.1 mm/min. The magnitude of this effect was only about half as large, when the pulling speed was increased from 1.1 to 1.3 mm/min. Annealing at 1200 °C for 2 h in argon ambient was found to decrease the FP-defect densities significantly, but less than that in oxygen ambient.

Laser-assisted chemical vapor deposition (LCVD) of nickel from Ni(CO)₄ has been utilised for the restructuring of integrated circuit (IC) interconnections. Nickel lines were deposited on a SiO₂ passivated IC to achieve new local interconnections between integrated circuit structures. Depositions were carried out over the pressure range of 0.2 to 2.2 mbar of pure Ni(CO)₄ buffered in 0 to 800 mbar He. Argon ion laser wavelengths of 488 and 514.5 nm, laser power of 50-150 mW and a laser scan speed of 80 μm/s were utilised for the deposition. The morphology and chemical contents of the deposited interconnection microstructures was examined by AFM, optical microscopy and LIMA. The resistivity of the deposited lines was found to be close to the nickel bulk resistivity. The utilisation of Nd: YAG and XeCl excimer lasers in the cutting of Al and Mo conductor lines for integrated circuit modification is also described.

New interest has recently arisen on inductors integrated monolithically together with CMOS electronics. The reason for the current attention lies in the demand for higher level of integration and increased operation frequency. Earlier efforts to make advantage of inductors in addition to other passive elements led to conclusion that the high conductivity of silicon and the large area required for the device would restrict their use. General development of integrated circuit fabrication technology has now made it possible to reappraise the use of integrated inductors.

Two important process related aspects which affect the quality factor of inductors are the series resistance and parasitic capacitance of the structure. We have worked on reducing the resistance by using alternative metallization for the inductors and at the same time born in mind the demand for integration. We also have tried different oxide thicknesses between silicon substrate and the device to reduce parasitic capacitance. For alternative metallizations electroplated gold and thick sputter deposited aluminium have been used. Comparison between the quality of inductors as well as manufacturability of each process has been made.

Ion implantation and the associated annealing was investigated using luminescence and absorption measurements to demonstrate the efficacy of implantation for introducing various classes of dopants into GaN. A wide range of implantation and annealing studies were performed with several dopant species (Zn, Nd, Er, Ar). Room temperature ion implantation was performed on MOCVD grown GaN samples at energies between 200 and 400 keV with doses ranging from 1 × 10¹³ to 1 × 10¹⁵ cm⁻². Conventional furnace annealing in flowing NH₃ or N₂ resulted in good recovery of implantation damage at an annealing temperature of 1000°C for 90 min. However, surface degradation became evident for annealing in an NH₃ environment at temperatures above 1000°C. Low temperature photoluminescence showed that the optical activation of the Zn, and the rare earth elements, Er and Nd, was optimum in the NH₃ annealing environment at 1000°C.

Screen-printing technique has been used to fabricate α-SnWO₄ thick films on alumina substrates. α-SnWO₄ powder was prepared by heating an equimolar mixture of SnO and WO₃ powders both in vacuum-sealed silica tubes and in argon atmosphere at 600 °C for about 15 hours. Screenprinted thick films were sintered in air for an hour at different temperatures between 550 °C and 850 °C in order to get different mixed-oxide structures from the partial decomposition of α-SnWO₄. X-ray diffraction and EDS (Energy Dipersive Spectroscopy of X rays) have been used for the characterization of the phase structure and composition of the films. The electrical conductivity together with gas-response properties of the films was measured at different temperatures between room temperature and 500 °C. H₂S, H₂, CO, CH₄ and SO₂ at different concentrations in synthetic air were used as test gases in the measurements. The powder fused in vacuum had better gas-response properties than that fused in argon atmosphere. The sintering temperature of screen-printed thick films was found to affect their gas-response properties, with an optimum sintering temperature at 550 °C. The α-SnWO₄ phase alone in the films showed both high sensitivity and selectivity to low concentrations of H₂S at temperatures between 150 °C and 300 °C.

We present an expression for the impurity resistivity for a general number of subbands in singly and doubly δ-doped structures. The true wave functions of the carriers and the intraband scattering from interaction between carriers of different bands are considered. The calculations are performed within the generalized Drude approach and with the temperature dependent RPA screening matrix. Numerical results for general temperatures are presented for singly n-doped Si and doubly n-doped Si with varying distance between the doped layers. We propose an explanation for the observed enhancement of the mobility in doubly δ-doped structures, as compared to singly δ-doped systems.

The MicroMOS quasi-deterministic particle dynamics 3D Monte-Carlo program for submicron MOS transistors is now extended with electronphonon interaction models for intervalley scattering and with a carrier-lattice energy exchange model. The impact ionisation and Auger recombination models are also improved. The carrier transport and lattice heat transport problems are self-consistently solved. As example the results of simulation, the spatial distribution of electrons, electron-phonon scattering events, impact ionisation events, Auger recombination events and the lattice temperature are presented. A new, low drain-to-source voltage breakdown effect has been observed.

We have studied the fabrication and modelling of integrated planar inductors on silicon made by using a standard 0.8 μm BiCMOS process. In order to aid the design of circuits including integrated inductors we have developed and tested a semiempirical equivalent circuit model for the planar inductors. The model, which has been implemented in the circuit simulator APLAC, shows good agreement with the measured inductor values, quality factors and S-parameters. Although the measured Q-values are below 10 in the GHz frequency range, the integrated inductors can be utilized in many RF IC applications. As a first demonstration for this we have successfully applied an integrated inductor to a low voltage monolithic voltage-controlled oscillator circuit at 1.8 GHz.

We present a numerical Wigner function model for the simulation of mesoscopic semiconductor devices. An exact non-local formulation of the transport equation in the Wigner presentation and its discretization for numerical calculation are discussed. The capability of the developed Wigner simulator is demonstrated by applying it to a resonant tunneling diode where quantum interference effects are important.

A theory of the d.c. electric current induced in a quantum channel by a propagating surface acoustic wave (acoustoelectric current) is presented. The first observation of the acoustoelectric current in such a situation was reported by J. M. Shilton et al. [10]. In a quasi-one-dimensional channel defined in a GaAs-AlGaAs heterostructure by a split-gate depletion, the acoustoelectric current showed giant oscillations as a function of the gate voltage. Such a behavior was qualitatively explained by an interplay between the energy-momentum conservation law for the electrons in the upper transverse mode with a finite temperature splitting of the Fermi level. In the present paper, a more detailed theory is developed, and important limiting cases are considered.

A mathematical method to interpret spectral responses measured from solar cells has been developed. Taking an inverse Laplace transform from the spectral response of a solar cell the spatial dependent collection efficiency of the cell can be obtained. Several important material parameters of the solar cell can be extracted from this function. Applying this method the properties of the solar cell can be investigated without applying characterization methods to the cell itself. We have applied the method both to simulated solar cells andto real solar cells.

Thin SiO₂ films, which have been fabricated by ultra vacuum plasma oxidation, have been characterised by method based on simple I/V-measurement. Tunnelling current through thin SiO₂ film is observed to have a small oscillatory component with high electric field. This current oscillation is due to resonance between injected electron waves in the SiO₂ conduction band and waves reflected at the SiO₂/Si interface. The phase and the amplitude of the oscillation depends on the oxide film thickness [1]. Oxide thickness and electron effective mass in the SiO₂ conduction band can be evaluated by fitting theoretical formula for Fowler-Nordheim tunnelling current density to measured data.

The quenching mechanism of the self-activated (SA) and Cu-Cl photoluminescence (PL) bands and 1.05 eV PL band in Zn(SeTe) solid solutions are investigated. It is supposed that SA and Cu-Cl PL bands in ZnTe-rich Zn(SeTe) solid solutions are quenched due to the complex defects Te_iCl_Te and Te_i2Cl_Te, which formation decrease the V_ZnCl_Te and single Cl_Te concentration. This assumption is confirmed experimentally. By annealing under Zn vapour pressure the intensity of SA and Cu-Cl PL bands in ZnTe-rich Zn(SeTe) solid solutions are restored. It is supposed that 1.05 eV PL band in ZnTe is formed due to a Te excess in ZnTe and the V_Te-Te_i pair is proposed for this band.

The photoluminescence spectra and formation of the complex defects in the Cl doped CdTe p-type layers are investigated. It is supposed that at low Cl concentration complex defects Te_iCl_Te and V_CdCl_Te are responsible for the low p-type resistivity in CdTe. At higher Cl concentration formed neutral complexes Te_i2Cl_Te and V_Cd2Cl_Te increased the p-type resistivity. At high Cl concentration excess Cl_Te caused a high resistivity and photoconductivity of CdTe. It is proposed that Te_i and complex defects with Te_i caused the 1.4eV PL band quenching and the 1.0eV PL band formation. The unstable 1.25eV PL band is supposed to be due to V_Cd-V_Te pair. Acceptor impurities in initial material as Cu and its complexes can increase the p-type resistivity value in CdTe: Cl.

The efficiency of (ZnCd)S, (ZnCd)Se and Zn(SeTe) solid solutions, doped with copper and chlorine, as x-ray downconverting phosphors for photodetectors is investigated. It is found, that the phosphors based on the solid solutions of the group II-VI wide-bandgap compounds and prepared in oxygen-free medium, can increase the photodetectors sensitivity to x-ray radiation more than one order of magnitude and enhance the detectors photocurrent versus radiant flux characteristics linearity.

We present a microscopic calculation of transport in strongly doped superlattices where domain formation is likely to occur. Our theoretical method is based on a current formula involving the spectral functions of the system, and thus allows, in principle, a systematic investigation of various interaction mechanisms. Taking into account impurity scattering and optical phonons we obtain a good quantitative agreement with existing experimental data from Helgesen and Finstad (J. Appl. Phys. 69, 2689, (1991)). Furthermore the calculated spectral functions indicate a significant increase of the average intersubband spacing compared to the bare level differences which might explain the experimental trend.

Metamorphic In_xGa_1−xAs/InAlAs (x ⩾ 0.8) modulation doped heterostructures have been grown on GaAs using a linearly graded In(AlGa)As buffer layer, and their structural and electric properties have been investigated. Surface morphology was found to depend on growth temperature and graded buffer thickness. Low growth temperature resulted in a relatively smooth surface with a minimum root-mean-square roughness value of 4–7 nm. The In(AlGa)As graded buffer effectively prevented dislocations from threading into the top layers. The epilayer grown on the graded buffer was tilted and not fully relaxed. High electron mobility and sheet density were achieved. The highest mobility value was 13740 cm²/Vs with a carrier density of 1.9 · 10¹² cm⁻² at 300 K. These values are comparable with InP-based InGaAs/InAlAs modulation doped heterostructures.

Barium strontium titanate Ba_0.6Sr_0.4TiO₃ thin film capacitors with metal oxide La_0.5Sr_0.5CoO₃ electrodes fabricated by sol-gel processing have been studied. The preparation of stock solution, firing and annealing of the films are described. Structural, compositional and morphological analyses using x-ray diffraction and SEM/EDS are reported. The film properties are between those of crystalline cubic perovskite and amorphous films. At room temperature the dielectric constant and loss factor of films at 20kHz were 40 and 0.038 respectively. The films showed a Debye-type dielectric relaxation with increasing temperature. The leakage current density is about 1.1 μA/cm² at 0.10 MV/cm.

We have used the effective mass approximation in a transfer matrix formalism to describe both confined and unconfined (above barrier) states in strained InGaAs/GaAs multiple quantum well (MQW) structures. We deduce the transfer matrices for a pseudomorphic MQW structure from the envelope wavefunction boundary conditions, modified to account for the discontinuities in the lattice constant. In the calculations we also take into account the finite width of the cap-layer, and the finite height of the vacuum potential. The potential profiles of the InGaAs/GaAs MQWs are derived using the calculated positions of the band-edges, taken from the model-solid theory. Photoluminescence excitation (PLE) spectroscopy has been performed on a series of InGaAs/GaAs MQWs (4-6 periods) at 11 K. We observed excitonic peaks at energies both below and above the GaAs free exciton. The values of the calculated transition energies are in good agreement with the experimental peaks.

A new SOI lateral emitter switched thyristor structure with partial buried oxide, which suppresses the parasitic thyristor turn-on and improves a maximum controllable current density, is proposed and compared with conventional LEST and LIGBT with partial buried oxide by using 2-D numerical simulations. Simulation results show that the proposed LEST exhibits latch-free characteristics without sacrifice of the on-resistance and switching speed while the LIGBT on the partially bonded SOI does not improve latch-up capability.

Ballistic fluctuations due to coherent mode mixing dominate magneto-transport in the dot at milli Kelvin temperatures. As the temperature is increased and Ohmic behavior is established in the dot, the shape of the weak localization (WL) peak follows the linear dependence predicted by the semi-classical theory for regular electron billiards. This temperature sets a limit above which these predictions can be observed at a single device. The shape of the WL peak is quite robust to the rounding of the QPC openings, whereas a relatively small rounding of corners inside the dot (which is thought to be caused by varying of the gate voltage) is sufficient to destroy regular motion of the electrons in the nominally square dot. This may explain a transition from the linear to the Lorentzian dependence of the WL peak (the latter is characteristic for chaotic billiards in the semiclassical picture) which was detected in the recent experiments. The probability density distribution inside the dot is visualized and analyzed.