This dissertation describes a combined computational and experimental study to understand the fundamental electrostatic, thermal, electromagnetic, and discharge related processes during the ageing of metallised polymer film capacitors. In the event of internal breakdowns, these capacitors are capable of "self-healing" through a controlled isolation of defects on the electrode surfaces by mosaic patterning the electrode. The objective of this project is to develop viable computer models to unravel electrothermally activated ageing processes in capacitors. To provide the necessary validation to any capacitor models developed, our work is supported by comprehensive experiments including industrial standard accelerated life tests and associated breakdown damage analyses of tested capacitors.

A novel passive antenna system, capable of discriminating specific electromagnetic signals is addressed. This antenna system will be able to detect signals of certain bandwidths, amplitudes and propagation directions. The philosophy behind this design was to maximise the signal discrimination at a stage prior to reception. The development of such systems could relieve the work involved in post detection discrimination, which may be time consuming and expensive. A major motivation of these studies lies in the difficulties inherent in signal detection for mobile radio communication systems operating at microwave frequencies. Such an antenna system consists of two components. They are the filter section and the detector array. The filter is designed in such a way that only the near normal signal to the locally flat area will be admitted and the rest reflected. The detector array will be at an appropriate position below the filter. Two types of filter structures have been studied for this angular filtering property. They are the Dielectric Multilayers (DML) and periodic arrays of slots as Frequency Selective Surfaces (FSS). DML are constructed by stacking layers of dielectric material whose permittivities vary in a near sinusoidal manner. Such a structure is known to have the ability to admit certain frequency bands of signals. The conventional transmission/reflection matrix method is used for its analysis. Also an optimisation procedure is carried out to minimise the loss of the signal in the DML. The characteristics of the DML as a beam-director and Beam-shaper have also been investigated. FSS exhibit the characteristics of band pass and band stop filters, depending upon the nature of the surface (periodic arrays of elements or slots) . Here the band pass nature is utilised by using arrays of slotted elements. These surfaces are tuned to admit narrow band signals. The well-known modal analysis method has been employed to study the FSS characteristics. The FSS have been studied in the context of frequency scanning, beam shaping, beam directing as well as angular scanning. A prototype has been constructed to simulate a multi signal environment in which the above structures have been experimentally assessed.

The purpose of the work described in this dissertation was to investigate the use of donor ion implantation into GaAs for the production of doped layers suitable for the manufacture of Schottky gated field effect transistors or MESFETs. The project had four main aims: (i) The establishment of a reliable method of encapsulation. (ii) A study of the use of short thermal pulse annealing to activate donor ions implanted into GaAs. (iii) An investigation of various ion implantation schedules to determine the most suitable for producing layers for MESFETs. (iv) The manufacture and characterisation of MESFETs on ion implanted material. This area of the project was performed at the Allen Clark Research Centre (Plessey Research, Caswell). Two sets of apparatus were constructed to achieve the rapid deposition of Si3N4 encapsulating layers. Both sets of equipment allowed the deposition of reliable encapsulating layers which protected the GaAs from decomposition during high temperature (900°C) annealing. With the second apparatus uniform layers of Si3N4 could be deposited over one inch diameter slices. The rapid thermal pulse annealing was also performed in the Si3N4 deposition equipment. The annealing temperature (900°C) was reached in less than 10 seconds. It was found, by examining the degree of electrical activation of implanted ions, that a total of 30 seconds annealing was sufficient to achieve the maximum number of implanted ions that could be activated at 900°C. Several species of ions, implanted at various energies, were studied as suitable donors, using thermal pulse annealing. These included selenium, tellurium, germanium, tin and silicon. Additionally the implantation of dual ion combinations, and the co-implantation of selenium ions at two energies was investigated. The results of Hall measurements and profiling achieved in this study compare favourably with those reported in the literature for similar implants annealed in more conventional furnace annealing schedules. The characterisation of MESFETs produced on implanted layers compare favourably with those manufactured on epitaxially grown material, both in terms of gain and noise figure. It was found also that characteristics of these MESFETs depended on the type of substrate used. The I-V characteristics of devices produced on bulk semi-insulating material were slightly inferior to those produced on epitaxial buffer layers, although the microwave performance was not significantly worse.

This thesis examines the radiation pattern characteristics of microstrip linear travelling-wave array antennas. Initial experimental investigations indicate that these structures can, in some circumstances, suffer from severe polarisation impurity particularly in the off-axis planes. These performance limitations have not been reported in the literature prior to this work, and are not amenable to calculation by existing theories. With the objective of accurate radiation pattern calculation of generalised periodic microstrip travelling-wave array structures, a new, unified, theoretical treatment is developed. This theory is based on a Green's Function formulation in an infinite array environment. In this way, higher order effects such as mutual coupling, surface wave generation and hybrid-mode propagation are included. Additionally, the analysis is structured to enable a wide variety of antenna geometries to be considered. Each non-identical period of a practical array is considered as being in an infinite array of identical periodic geometry. This geometry is then specified in terms of a single period (or "unit cell"). The unit cell geometry of the array to be studied is specified as a set of non-equal rectangular segments whose dimensions and positions are variable. In a numerical implementation of the analysis, this allows the geometry of the antenna to be specified at run-time. Application of the infinite array results to practical, non-infinite structures is discussed, and the resultant radiation patterns calculated via a suitable Green's Function. This theoretical approach is validated by comparison between experimental and predicted results. Very good correlation is obtained although in short (10 wavelength) structures, direct radiation from the transition effects the experimental results. In both the general applicability and the accuracy of the results, the theory appears to be a significant advance in the analysis of this class of radiating structures.

Many doses of ions have been implanted through near-surface AlGaAs/GaAs double-barrier diodes. The first objective of this work was the creation of a resistive layer beneath the diodes in selected areas of the wafer. It is shown that if the damage within the double-barrier diodes could be annealed without removing the resistive layer, the three-dimensional integration of the diodes with a second level of devices beneath the resistive layer could be attained. Implantation-and-annealing to create either a damaged or a chemically-compensated resistive layer has been attempted, where, during both types of process, the damage within the doublebarrier diodes was much less than that below them. After implantation of 5.0x1018 2.0MeV B+ ions cm-2, and anneals at 600° C, near-surface Al0.4Ga0.6As/GaAs double-barrier diodes still had good quality negative differential-resistance. It is shown that if (the smaller and less damaging) 1.2MeV Be+ ions were implanted instead of the 2.0MeV B+ ions, an n+-doped layer beneath the diodes can, in principle, be chemically compensated without destroying the diodes irreparably. This work was the first to successfully carry out the anneal-induced recovery of an ion-implanted electronic device having quantum-length-scale layers. The second objective of this work was the elucidation of the electronic and structural characteristics of the same implanted-and-annealed double-barrier diodes. Before annealing, electron conduction through the ion-implanted diodes was limited primarily by field-enhanced emission of electrons from defect states within the lightly-doped spacer layers. The current of ballistic electrons through the as-grown double-barrier structures was suppressed by implantation-and-annealing; this was probably caused by scattering of these electrons by the remaining defect states. The suppression of the ballistic-electron current within implanted-and- annealed double-barrier diodes is proposed to be the primary cause of their larger-than-as-grown 5K and 77K peak-to-valley current ratios. Multi-stage annealing of defects within the double-barrier diodes has been investigated by electrical measurements. The anneal-induced creation of defect clusters within the device mesas was confirmed by both DC and AC measurements, where these clusters were surrounded by percolation paths of as-grown material. Single-electron switching and resonant tunnelling through donor states have been observed within the percolation paths at 4.2K; these observations indicate that the typical diameter of the paths was probably less than five microns, and possibly less than one micron.

The introduction gives consideration to the aspects of communication principles, power converters, PWM modulation techniques and the application of LSI intergrated circuits as regards their effect on the design of variable speed AC drives. Chapter 1 includes background material where several types of variable speed A.C. drives are compared. The theoretical possibilities of various PWM schemes show that certain techniques offer improved harmonic spectrum properties, more suited to the wide range speed control of induction motors. Chapter 2 introduces alternative approaches to the analysis of sinuisoidal or natural PWM and the regular sampled PWM waveforms. The latter waveform is analysed by expressing the function in terms of the time domain; conversion from the time to the frequency domain is then obtained using a Fast Fourier Transform (FFT) technique implemented on the 32K PET microcomputer. Chapter 3 describes a successful analogue rig designed to generate several, alternative PWM waveforms which were used to drive an induction motor via the McMurray bridge inverter. Practical performances of the differing PWM waveforms were evaluated, using this equipment and the regular sampled, asymmetric PWM waveform was shown to have some operational advantages - it was this technique which was then given further consideration for microprocessor implementation. Chapter 4 outlines the top-down, structured design philosophy necessary for the efficient successful realisation of the hardware and software of a microprocessor system. The detail of the complete unit is given in Chapter 5 which describes the hardware and software of a microprocessor design used to provide the switching drive to the induction motor via the inverter. Several sotware algorithms are described which, when run by the control program in the appropriate sequence, produce the three phase PWM output waveform. An analysis of results is provided in Chapter 6 which shows that both the analogue and microprocessor rigs can operate an induction motor over wide speed ranges. The micropocessor realisation was far more accurate in its emulation of the Regular Sampled Asymmetric PWM waveform and controlled the motor over a 100:1 speed range. The concluding chapter summarises the problems, state-of-art and further applications of the microprocessor implementation whose unique properties are hardly obtainable with other, more conventional forms of practical realisation.

The rapid development of high-speed, high-density integrated circuits has brought about a situation where the delay times and distortion of signals transmitted on the interconnections (microstrip lines) within these packages are now comparable with those of the devices in the circuit. Hence when designing high-speed digital systems the effect of signal delay, distortion, and attenuation, on these interconnections has now become a necessary part of integrated circuit design. Therefore accurate modelling and simulation of interconnects is a very important subject for research. These interconnections can create a number of problems such as signal delay, distortion, attenuation, and crosstalk between lines of close proximity. These microstrip lines have the same behaviour as exhibited by transmission lines and can therefore be described using the well-known Telegrapher's equations. A quasi-distributed equivalent circuit model describing the behaviour of such microstrip lines is implemented into the SPICE circuit simulator. This allows an investigation into lossless and lossy line characteristics and illustrates the importance of choosing the correct impedances for both the lines and devices within the integrated circuit package. The model is then extended to include crosstalk between neighbouring lines, by means of a transformation network. This study of crosstalk illustrates that logic functions lying in an intervening space between the pulse-activated lines are found to be affected more than the outside lines. The MATHEMATICA package is used for the calculation of capacitance, inductance, impedance, time delay and transformation network control parameters for any set of microstrip lines of a given geometry. The results obtained from these calculations are then used for further simulation runs using the SPICE software for different line configurations. The results obtained are seen to be consistent with all previous work conducted on coupled three line structures, and give good verification of both the Mathematica program and the SPICE equivalent circuit model.

The first part of this work gives a broad introduction on the origins of ZnO bismuthbased v aristor materials. The electrical properties of ZnO varistors follow, giving a detailed and up to date review of all the models published with an attempt to explain their complex conduction mechanisms. Stresses found in HV networks are also reviewed in detail, their knowledge being necessary to link the de gradation and fail ure mechanisms of ZnO varistors which are explained at depth. The manufacture of ZnO varistors on a laboratory scale is also presented: disc type as well as thick-film type varistors were manufactured and tested during this inv estigation. The parameters influencing the microstructure and electrical characteristics of varistors during sintering were reviewed. Testing sequences used by manufacturers as well as surge arrester testing standards were also assessed to derive tests for the laboratory produced samples.

No description available.

No description available.