Table of contents

The dynamics of multiple Raman scattering in a long path liquid nitrogen (LN₂) cell is analysed by using a long pulse unfocused Nd-YAG laser. This medium in this configuration is the most suitable to evaluate the window and mixing effects in the dynamics of the Raman process. In this experiment the window reflections have been evidenced in the time behaviour and they have a calculable effect to lower the Raman threshold. Moreover, some four wave mixing effects have also been calculated and observed both reducing the threshold for the second Stokes scattering and producing a pump regeneration.

New closed-form expressions to calculate frequency-dependent distributed inductance and the associated distributed series resistance of a single interconnect on a lossy silicon substrate (CMOS technology) are presented. The proposed analytical model for series impedance is based on a self-consistent field method and the vector magnetic potential equation. It is shown that the calculated frequency-dependent distributed inductance and the associated resistances are in good agreement with the results obtained from rigorous full-wave solutions and CAD-oriented equivalent-circuit modelling approach.

Satellite solar arrays are likely to be charged by the space environment up to a discharging threshold level, resulting in electrostatic discharges. Electrostatic discharges are phenomena known to cause temporary anomalies, mainly electromagnetic disturbances. Such discharges are studied experimentally in this paper on laboratory samples made of zinc or silver, and on real solar cell samples. Spectroscopic data show for all these samples the presence of eroded metal vapours, and microscopic photographs confirm the local melting of the electrode, even for low values of the capacitors (10 nF and 330 pF) used in the electrical circuit.

The wide-ranging impact of III-Nitride semiconductors is well known, encompassing high-brightness UV/blue/green/amber LEDs, UV/blue diode lasers and high-power microwave transistors (>11 W/mm). Alongside these impressive devices remain a surprising number of fundamental questions regarding the material properties of III-N structures. This cluster of articles addresses a variety of topics of current importance for the understanding and application of the III-Nitrides, mostly originating from groups within the UK.

Research into III-N materials and devices is widespread within the UK and has the UK nitrides consortium (UKNC) as its premier forum (http://www.uknitrides.org). From its origin in 1994, and with support from the DTI/EPSRC from 1998 to 2001, the UKNC has grown rapidly with membership now exceeding 40 separate groups with a healthy mix of academic and industrial institutions. The second UKNC conference was held at Strathclyde University in September 2001 with a programme including excellent invited presentations from Nicolas Grandjean (CHREA-CNRS, Valbonne) and Oliver Brandt (Paul Drude Institut, Berlin). The `flagship' paper of this cluster is a manuscript submitted from the second of these, addressing the highly important issue of the determination of the strain and composition within InGaN structures. This paper gives an authoritative, novel and insightful description of both the strengths and limitations of high-resolution x-ray diffraction when applied to InGaN quantum well structures.

Within the UK a widespread infrastructure for the growth of III-Nitride structures has developed including MOCVD reactors at Qinetiq Ltd, and the Universities of Strathclyde, Bath, Sheffield and Cambridge and MBE/CBE reactors at Sharp Laboratories of Europe and the Universities of Nottingham, Cardiff and Liverpool. The MOCVD teams feature in this cluster through papers from Strathclyde and Sheffield. The light-emitting properties of InGaN quantum wells are probed using series of samples emitting over the entire visible spectrum, by control of the growth temperature of the well, and with variations in optical properties induced by control of the growth temperature of the barrier. A second Sheffield paper changes the focus to the electronic applications of AlGaN/GaN structures, looking at surface passivation effects in transistors. The other main growth technique is represented in the paper by Huang et al discussing CBE of GaN bringing in the important dimension of substrate choice, in this case silicon (111). Such factors also feature in the paper from Oxford University describing deposition and in situ investigation of novel InN structures. The large body of UK-based research into III-nitrides not directly associated with material growth is exemplified by papers within this cluster applying the powerful technique of muon spin resonance to hydrogen in III-nitrides, analysing novel phonon replica structure in sub-band gap absorption and calculations of the heterostructure band offsets.

I am sure that within this selection of papers you will find interesting insights into the remarkable world of III-Nitride physics and applications.

Dr Robert Martin, Guest Editor

April 2002

We present an accurate model for the dynamical x-ray diffraction analysis of highly mismatched materials systems. Our approach features an exact incidence parameter as well as an exact calculation of the changes of Bragg angle and lattice plane inclination due to strain. We show that the commonly used approximations for these parameters are, in general, unable to describe diffraction profiles of different reflections consistently, and utterly fail for a strain on the order of 1% or more. The model as presented explicitly considers crystals with zincblende and wurtzite structure of arbitrary orientation and the resulting distortions. The validity of our model is demonstrated by various experimental examples, including closely lattice-matched GaAs/AlAs structures on GaAs(001) and GaAs(113), as well as highly mismatched GaN layers, (Al, Ga)N/GaN and GaN/(In, Ga)N heterostructures on 6H-SiC(0001) and γ-LiAlO₂(100).

This paper reviews recent work on the properties of hydrogen defect centres in two group III nitrides, AlN and GaN, and relevant studies by µSR spectroscopy, i.e. muon spin rotation, relaxation and resonance. We highlight, especially, results obtained by a form of nuclear quadrupole resonance. Implanted positive muons are used to mimic and model the behaviour of interstitial protons. The resultant defect centres exhibit both metastability and bistability. In AlN, they remain as positive ions but partition themselves between a highly mobile species and one that is trapped and immobilized to temperatures as high as 800 K in cage-like sites adjacent to nitrogen. The barrier to escape from the cage is 0.86 eV. In n-type GaN, the cage-site positive ions are stable only up to 200 K; above this temperature they capture electrons to convert to negatively charged centres, analogues of hydride ions, relocating to sites antibonding to gallium. These latter escape from the cage sites around 600 K with an activation energy of 1.5 eV to join more mobile negative ions diffusing via channel sites with an activation energy of 0.65 eV. Data on the neutral paramagnetic centre suggest that hydrogen can act as a shallow-donor in at least one other member of this family of materials, namely InN.

We present a theoretical investigation of band offsets in GaN-based group III-nitride ternary/binary heterostructures with zinc blende structure. The band offsets in InGaN/GaN and AlGaN/GaN heterostructures were studied using the recently proposed extended sp³ tight binding model (Ünlü H 2001 Phys. Status Solidi. b 223 195). Simulations show a small valence band offset in AlGaN/GaN heterostructure and large valence band offset in the InGaN/GaN heterostructure. Furthermore, the tensile strain in AlGaN leads to positive bowing of the conduction band offsets in AlGaN/GaN heterostructure, and the compressive strain in InGaN leads to negative bowing of conduction band offsets in InGaN/GaN heterostructure as a function of composition. Good agreement is found between predictions and experiment.

Different dielectrics were used for post-processing surface passivation of AlGaN/GaN heterostructure field-effect transistors (HFETs) and the resulting electrical characteristics examined. An increase in the maximum drain current of approximately 25% was observed after Si₃N₄ and SiO₂ deposition and ~15% for annealed SiO on AlGaN/GaN HFETs. In all cases, the passivation was found to increase the gate leakage current with an observed reduction in the leakage activation energy. However, the rise in gate leakage current was least for SiO. The plasma enhanced chemical vapour deposition method was found not to contribute to the passivation mechanism, whilst the presence of Si appears to be an important factor.

We report a study of the emission properties of InGaN/GaN quantum wells, grown by metalorganic vapour phase epitaxy, as a function of barrier-growth temperature, using continuous wave and time-resolved photoluminescence spectroscopy. We observe that high barrier-growth temperatures lead to a blue-shift of the photoluminescence, a reduction of the recombination lifetime, and a broadening of the emission linewidth. These effects are consistent with increased indium desorption during the pause preceding the growth of the barrier at higher temperatures. The blue-shift and the reduced lifetime are consistent with a reduction of the average indium concentration, while the increase of the linewidth is caused by the increase of the indium composition fluctuations due to the randomness of the desorption process. At the same time, the high barrier temperature samples have the highest photoluminescence efficiency at room temperature due to the reduced number of defects in the barrier.

In this paper, we describe the growth and characterization of InGaN single quantum wells with emission peaks in the blue, green, amber and red spectral regions, grown by metal-organic vapour phase epitaxy. Starting from the growth of a blue-emitting (peak ~430 nm) InGaN quantum well at 860°C the InGaN growth temperature was progressively reduced. The photoluminescence peak wavelength, measured at low temperature, shifts through the green and orange spectral regions and reaches 670 nm for an InGaN growth temperature of 760°C. This corresponds to an energy lower than the currently accepted band-gap of the binary compound, InN. Spectral characteristics of the luminescence peaks will be discussed, including an analysis of the phonon-assisted contribution. Low energy secondary ion mass spectrometry analysis provides information on the indium content and thickness of the `blue' and `red' quantum wells. The results are combined to discuss the origin of the `sub-band-gap' luminescence in terms of the combined influence of InN-GaN segregation and the effect of intense piezoelectric fields.

Optoelectronic modulation spectroscopy (OEMS) has been used to explore the sub band-gap responses of deep energy states in GaN prepared by MOVPE. Modulation of the channel current of an AlGaN HEMT revealed a replica structure centred on 2.16 eV which cannot be attributed to Fabry-Perot oscillations. This was superimposed on a broad background with an anti-phase response, suggesting a continuous spectrum of states to which electrons were excited from the valence band.

The phase of the replica responses confirmed that each peak was due to a transition to the conduction band. The average peak separation was 89 meV, close to the LO phonon energy in GaN, 92 meV, suggesting that charge in the state at 2.16 eV was strongly coupled to the lattice. This appears to be the first report of the phonon replica structure in an absorption spectrum. The observations are compared with a previously published luminescence replica spectrum. It is demonstrated using configuration coordinate diagrams that there should be a coincidence of lines, and this was confirmed experimentally. These observations provide confirmation that the electron state concerned is located 2.16 eV below the conduction band.

The geometry of the configuration coordinate diagram leads to the conclusion that the Frank-Condon energy of this defect was between 0.40 and 0.55 eV with a corresponding Stokes shift between 0.80 and 1.10 eV.

Heteroepitaxial growth of InN on GaN(0001) and on Si(111) has been studied using scanning tunnelling microscopy (STM) and atomic force microscopy (AFM) in order to elucidate the growth modes. Using cracked ammonia as group V source, InN/GaN grows in a two-dimensional mode, despite the fact that there is an 11.2% lattice mismatch. Only a pre-exposition of the GaN substrate to silicon enables a growth mode change from two-dimensional to three-dimensional. When we deposit indium in a cracked ammonia environment on Si(111), we again observe two-dimensional growth with small islands of varying size. Surface reconstructions on the islands have been observed with atomic resolution. However, we believe that atomic hydrogen, a cracking product of ammonia, prevents the growth of InN and the observed reconstructions are indium adsorbate structures on Si(111).

The chemical beam epitaxy of GaN on Si (111) using an AlAs buffer layer has been investigated. Transmission electron microscopy of cross-section samples reveals a nanoscale columnar structure consisting of the hexagonal GaN polytype. Selected area diffraction indicates an epitaxial relationship between the GaN and Si substrate which is described by GaN [0001] ∥ Si [111] and GaN (100) ∥ Si (11). Raman spectroscopy of the samples shows that the AlAs longitudinal optical (LO) and transverse optical (TO) modes are observed at 360 and 400 cm^-1, respectively, and that the GaN E₂ TO and A₁ LO modes occur at 562 and 733 cm^-1, respectively. Raman peaks between 616 and 673 cm^-1 are attributed to an interlayer of AlN formed by the nitridation of the AlAs buffer layer.

The temperature dependences of domain wall pinning field and of anisotropy field in magnetic garnets and amorphous ribbons are reviewed and compared on the basis of our own experiments and data taken from the literature. The results are discussed in the frame of a modified classical model of the domain wall coercive field of magnetic materials with a stationary distribution of magnetic defects. It is shown that - independently of the substantial difference between the two investigated types of materials - the temperature dependence of the domain wall pinning field is predominantly determined by the temperature dependence of the anisotropy field and only modified by the defects shape and distribution. This result shows that from the two major sources of the coercive properties - the magnetic parameters and the material structural characteristics - the temperature dependence of coercivity is determined mainly by the former one. The experimental results made it also possible to get a quantitative insight into the temperature dependence of the efficiency function\it of these materials, which describe the interaction between the moving domain wall and the defects.

Intense electroluminescence in the range 77-300 K has been observed from interface transitions in type II P-Ga_0.84In_0.16As_0.22Sb_0.78/ n-In_0.83Ga_0.17As_0.82Sb_0.18 single heterojunctions grown by liquid phase epitaxy from an In-rich melt. The quaternary epitaxial layers forming the P-n heterojunction were unintentionally doped and grown lattice-matched onto a (100)-oriented n-type InAs substrate. The electroluminescence and photoluminescence emission spectra from the heterostructure were investigated in detail and are discussed below. The luminescence spectra contained two interface-induced emission bands: hν_A which was identified with radiative transitions between electron and hole quantum well sub-bands across the P-Ga_0.84In_0.16As_0.22Sb_0.78/n-In_0.83Ga_0.17As_0.82Sb_0.18 heterointerface, while the emission band hν_B originates from radiative transitions involving acceptor states in the narrow gap In_0.83Ga_0.17As_0.82Sb_0.18 near the type II heteroboundary.

Based on the characteristic potentials previously developed for multi-terminal diffusive mesoscopic conductors by Sukhorukov and Loss, the formulas are presented of both the short-circuit noise currents including thermal noise and excess noise (1/f noise and generation-recombination noise) currents and the open-circuit noise voltages of macroscopic multi-terminal homogeneous semiconductor resistors with arbitrary geometry under dc bias. The formulas are of neat volume integral form and they give the information about how much contribution a certain region of a device makes to the terminal noise currents or voltages. A comparison between the characteristic potential method and the impedance field method developed by Shockley and co-workers is presented. A direct experimental verification of the derived formulas has been provided by measuring thermal and 1/f noise of three-terminal homogeneous polycrystalline silicon resistors and carbon resistors.

A deformation of a short intensive laser pulse temporal profile by optical field ionization (OFI) processes is under consideration. In one-dimensional analysis at a comparatively long distance of the pulse propagation the deformation is ensured by optical field absorption in addition to the dispersion. Three-dimensional effects lead to a modulation of the temporal profile at a laser pulse penetration depth less than the Rayleigh length. The modulation originates from interaction between the pulse and the inhomogeneous dynamic plasma channel created by OFI during the laser pulse propagation. A criterion of the laser pulse `smooth' penetration into a gas undergoing ionization at a distance about Rayleigh length is offered.

The fluid equations for a one-dimensional, unmagnetized and collisionless plasma are numerically integrated in a finite domain to obtain sheaths at the absorbing wall system boundaries. The type of boundary conditions required for the formation of and for sustaining the long-time evolution of sheaths is discussed in detail. In contrast to earlier specialized fluid approaches, this treatment utilizes a more general fully non-linear fluid model, incorporating both ion and electron dynamics, to generate time evolutionary sheaths. In particular, we avoid restrictions such as the use of Boltzmann distributions for one or more components, the use of an artificial two-section plasma fluid model or the use of the quasi-neutrality assumption.

NH₃ and NH_x<3 radicals are produced downstream a microwave discharge containing Ar-N₂-H₂ gas mixture. The chemical mechanism under investigation consists of heterogenous reactions between adsorbed species NH or NH₂ (denoted NH_s and NH_2s) and H or H₂ flowing downstream the discharge. NH_s is adsorbed on the stainless steel reactor wall and reacts with H or H₂ producing NH_2s or . Then, part of NH_2s produced reacts with H atoms producing ; another part is desorbed from the tube wall: . We assume that NH₃ is spontaneously and totally desorbed. From the balance equations, we determine analytical relations for NH_2s, NH₂ and NH₃ concentrations. We then measure values of reaction rate constants and compare the numerical results to measurements performed in the afterglow by means of mass spectrometer versus the %H₂ injected in the discharge. We measure values in two different initial gas mixtures, 98.7% Ar-1.3% N₂ and 66.6% Ar-33.3% N₂. In the first gas mixture, k₁, k₂(NH_s), k₃(NH_s) and k_sg range between 1×10^-17 and 2×10^-17 m³ s^-1, 0.035 and 0.045 m s^-1, 9 and 11 m s^-1, and 0.30 and 0.35 m^-1 s^-1, respectively. In the second gas mixture, as expected, similar values are found for k₁ and k_sg but the other two values increase by a factor of 5. Such an increase for k₂(NH_s) and k₃(NH_s) is probably due to the increase of the (NH_s) concentration on the reactor wall. The recombination coefficient γ is deduced from the previous rate constant values. We find γ₁ = 4.12×10^-4, γ₂ = 4.91×10^-6 and γ₃ = 7.93×10^-4, using the mean values of reaction rate constants determined for k₁, k₂ and k₃, respectively, in the first gas mixture. To our knowledge, these results have never been published before. They are in good agreement with values given in the literature for other similar mechanisms. Finally, we conclude that the loss of H atoms on the reactor wall mainly results in producing NH_2s and NH₃.

A quadrupole mass spectrometer combined with a Li⁺ ion reactor setup was successfully used to qualitatively analyse and follow the evolution of the ionic and neutral species produced in an N₂/O₂ microwave discharge plasma. Mass spectrometric analysis showed that significant amounts of gas-phase neutral NO₂ and HNO₂ were produced in addition to the familiar NO and N₂O species, detected in the form of Li⁺ adducts. A number of ionic clusters due to gas-phase ion-molecule reactions were detected.

The column plasma of a dc glow discharge in neon is comprehensively studied by a new self-consistent hybrid method. The method is based on the solution of the steady-state fluid equations of charge carriers and the Poisson equation using electron transport coefficients and ionization frequencies which arise from a space-dependent kinetic treatment of the electron component. The column description comprises, in addition, a space-dependent treatment of the excited particles in the plasma, the treatment of the plasma-wall interaction processes of the charge carriers and the determination of the electric field. The hybrid method provides a very efficient self-consistent column description although it avoids rough approximations often applied in fluid methods. Results of the method for a neon column plasma are compared with those of probe measurements and spectroscopic studies. The results confirm the importance of the ionization of excited neon atoms for the charge carrier production and emphasize the necessity of the detailed description of the excited states included in the column model.

Direct current N₂ flowing discharges were generated and conditions for the pink afterglow were obtained. Emissions from N₂(B, C) and N₂⁺(B) radiative states were studied as a function of pressure, flow rate and post-discharge position. A one-dimensional kinetic model accounting for N₂(X, v), N₂( A, B, C, a, a', a''), N(⁴S), N_1-4⁺(X) and N₂⁺(B) species has been developed in order to describe the experimental observations. The analysis on the complete set of processes assumed in this paper has provided possible generation mechanisms for N atoms, N₂⁺(B) ions and N₂(B, C) electronically excited states as well as for metastable ones. It has been shown that ionization, excitation and dissociation processes occur simultaneously at the post-discharge region when the vibrational distribution function of N₂(X, v) states heats as resulting from the efficient V-V pumping mechanism, which is very sensitive to pressure conditions. Here, the pink afterglow is described as a non-equilibrium plasma, i.e. ambipolar diffusion for ions and the condition of charge neutrality are assumed.

The physical processes occurring during the electrical explosion of metallic conductors has attracted interest for many years. Applications include circuit breakers, segmented lightning divertor strips for aircraft radomes, disruption of metallic shaped charge jets, plasma armatures for electromagnetic railguns and plasma generators for electrothermal-chemical guns. Recent work has cited the phenomenology of the fragmentation processes, particularly the development of a plasma around the lower resistance condensed fragments. An understanding of both the fragmentation process and the development of the accompanying formation of plasma is essential for the optimization of devices that utilize either of these phenomena. With the use of x-radiography and fast photography, this paper explores the wire explosion process, in particular the relationship between the fragmentation, plasma development and resistance rise that occurs during this period. A hypothesis is put forward to account for the development of plasma around the condensed wire fragments.

Experimental parameters used in this study are defined. Wires studied were typically copper, with a diameter of 1 mm and length in excess of 150 mm. Circuit inductance used were from 26 to 800 µH. This relatively high circuit inductance gave circuit rise times less than 180 MA s^-1, slow with respect to many other exploding wire studies. Discharge duration ranged from 0.8 to 10 ms.

In complex thermal material processes, it is essential to find the more effective processing parameters. Sensitivity is useful in estimating the effect of variation of individual parameters in mathematical models of interesting systems. By use of the analytical solution of the temperature analysis, a sensitivity analysis for laser surface treatment was established by the differentiation of the analytic solution with respect to the laser beam radius and beam scanning velocity. Both analytical and finite-difference gradients are employed to validate the effectiveness of the sensitivity analysis calculated in this work. This simple method can be applied to calculate the effectiveness of the processing parameters for a given condition. In addition, the results of sensitivity analysis imply a dominant influence of the laser beam size in the region of parameters used.

The new model of incoherent-to-coherent infrared (IR) image converter based on a GaAs photoconductor joined to an electro-optic (EO) Bi₁₂SiO₂₀ crystal has been analysed theoretically and experimentally. The possibility of field transfer from the photoconductor to the EO crystal under the IR radiation sufficient for realization of the EO (Pockels) effect in the EO crystal was assessed. Based on the electric field parameters and the parameters of the photoconductor and EO crystal, the threshold sensitivity of the converter was estimated. The experimental photoconductor-EO crystal structure by which IR radiation (0.9-1.5 µm) was converted into the coherent visible radiation was obtained on the basis of theoretical calculations. The measured threshold sensitivity of the converter, 5×10^-4 W cm^-2, was found to be in the limits of theoretical estimation.

The proof of the mathematical equivalence between Debye distributed processes in natural frequencies and non-Debye processes for arbitrary variations of temperature and electric field is re-stated and put on a firmer basis.

Electrostatic force microscopy was used to detect nanoscale dielectric constant variations in two different, carbon-doped oxide low-k dielectrics deposited by plasma-enhanced chemical vapour deposition and subjected to oxidizing isotropic plasmas and inert gas isotropic plasmas. Samples were polished at 10\r{} to the sample normal to enhance the through-the-thickness spatial resolution. We observed that the technique was able to detect k{}~{}0.1 variations of dielectric constant with ~10 nm spatial resolution. We also observed that the oxidizing isotropic plasma caused damage to a depth of approximately 10-50 nm and that one of the carbon-doped oxides was more susceptible to plasma-induced damage. The estimated increase of dielectric constant from the oxidizing plasma was from k~2.5-3 to k>4-5. The damage from the inert gas plasma was observed to be deeper but less severe.