We study the (001) surface of the semiconductors Si and Ge using the total energy algorithm due to Chadi. The total energy minimization theory is described together with the LCAO theory and the Special Points Method. The electronic bands are calculated using the Slater-Koster simplified LCAO theory in its first-near-neighbour approximation. The Special Points Method is used to compute the electronic contribution to the total energy. We describe and illustrate a method for deriving special points for a distorted lattice. Our method is quicker and easier than first principles derivation of special points for distorted lattices. The accuracy of the total energy scheme is tested by computing bulk phonon frequencies and elastic moduli. The calculated results are in reasonable accord with experimental values. The Chadi formalism is used to deduce the (2X1) reconstructed geometries of Si and Ge (001) surfaces. Techniques for calculating projected bulk bands and surface states are detailed and applied to the dimer geometry of the (2X1) reconstruction. The symmetric dimer is found to give metallic surface states. The asymmetric dimer has the lower energy and semiconducting surface states. It is found that the calculated positions and dispersions of the semiconducting states do not agree with photoemission results. High order reconstructions on the Si and Ge (001) surface are examined and found to be within a few meV in energy of the (2X1) reconstruction. The Si (001) - (2X1) surface structure resulting from the Chadi formalism is compared with that deduced from LEED analysis and from an energy minimization process using an ab initio pseudopotential calculation. The Chadi structure is found to be the lowest in energy but we argue, by considering other theories, that the Chadi formalism is qualitative at best. We propose some improvements of the Chadi formalism to make the theory more quantitative for surface structural geometries.

Although the technology of integrated passives in power electronics has been receiving a lot of attention recently, behavior prediction of these integrated passive structures is still not on an acceptable level for widespread applications. A lot of work has been done recently on accurate electromagnetic modelling of these structures, but the complex models investigated are not practical for the average engineer in power electronics to apply, and integrated passives remain a subject of interest in research and academia, but very infrequently applied in industry. The aim of this dissertaion is to provide a bridge between the mathematical models currently being investigated, and circuit level behaviour prediction which may be used by practicing engineers to design power electronics circuits which make use of integrated passives. The history of integrated passives is first investigated, along with historical modelling techniques and their shortcomings. Two similar modern distributed circuit theory models are investigated and aspects of both are combined to form the model that is used as a mathematical foundation for this dissertation. This model is analysed, and some methods are proposed for integrating the resulting differential equations. A transformation is proposed for transforming the transmission network representa- tion of the structure, which results from integrating the differential equations, into a network of admittances, which may be used for applying the technique of nodal analysis to a circuit containing an integrated passive structure. This admittance network model is used to implement a frequency domain simulation model in a practical circuit simulator. i In the integration of the circuit differential equations, the method of modal analysis is applied. In this analysis a system of wave equations is derived and solved in the frequency domain. By applying the inverse fourier transform to these wave solutions it is found that the modal wave propogation is a simple time shift in the time domain for each propogating mode in lossless structures. Applying this observation a transient model is implemented in the circuit simulator for lossless integrated passive structures. Although this is limited to the lossless case, the simulator still appears to be giving good results. The zero voltage switching two inductor boost converter was then investigated to construct a case study for the simulator. The topology was analysed, and a design method found. A discrete converter was constucted to verify the analysis. The design of an integrated passive structure for this converter is then presented, and the simulation results show that the simulator may is robust enough to be applied to practical problems. The integrated converter could unfortunately not be constructed due to materials processing limitations, and thus the simulation result remain to be experimentally verified. The results do however closely match those predicted by the widely used lumped element models, apart from some high order effects. / Prof. I.W. Hofsajer

Increases in the electrical power consumption of modern mobile vehicles, due to lighting, heating, motors and the various electronic loads have highlighted the need for efficient and reliable power generating equipment. The electronic loads impose stringent limits on the power supply and for this reason, it is important at the design stage, to be able to predict the system transient and steady-state performance using a mathematical model.

The effect of bismuth on the electrochemistry of the lead acid battery has been investigated using the techniques of linear sweep voltammetry, chronocoulometric and galvanostatic cycling, potentiostatic pulse experiments and corrosion measurements. Optical and scanning electron microscopy enabled a morphological examination of the electrodes to be carried out.

After an introduction to amplifiers in communications and an exposition of the literature specifically relevant to high linearity power amplifiers, this study investigates more thoroughly various aspects of envelope feedback as applied to Bipolar Tuned Power Amplifiers at HF and VHP. It is discovered that under the correct conditions a new mode of linear operation exists where gain compression, AM-PM conversion and input impedance are simultaneously linearised, and in this region DC-RF power efficiency is also improved. Spectral measurements are presented from an envelope feedback amplifier constructed to operate over this region. A computerised system is described for measuring accurately the gain and phase shift of the test amplifier against variation of collector supply, quiescent bias current and RF drive power. The results from these measurements are presented as 3-dimensional projections and as contour plots. Subsequently the stored data is used to re-construct two-tone spectra, which is then analysed to show contributions to the spectrum from the gain compression and AM-PM conversion mechanisms separately. Conclusions are drawn with respect to effects of bias on these two mechanisms. A mechanism has been discovered which gives a symetric spectra without requiring AM-PM conversion at the fundamental frequencies. An attempt is made to model the amplifier with a non-linear circuit transient analysis program (SPICE). Good correlation is obtained for some parameters and these results are also plotted in 3-D and in contour.

This thesis presents a study of the potential for ion implantation to play a more significant role in the manufacture and fabrication of commercially available multilayer microwave devices. Two different applications for ion implantation in device manufacture are investigated. Firstly, implant isolation as an alternative to wet chemical etching for the planar doped barrier diode, and low and high power versions of the graded gap Gunn diode are attempted. It is demonstrated that the technique is an excellent method of lateral device isolation for the current generation of these devices, having little or no effect on the performance of the planar doped barrier diode, but with significant improvements in across wafer uniformity of device area, and hence, improvements in uniformity of device characteristics. Implant isolation of the graded gap Gunn diode has been met with mixed success. It has been shown that implant isolation has no detrimental effect on the ability of the device to emit microwaves at 77GHz, despite encapsulation of the active regions of the device in ion implanted GaAs. Problems have, however, been encountered with the geometry of the integral heat sink device (high power version), resulting in parasitic capacitance to the extent that the device experiences a shift in output frequency and power. Secondly, doping of GaAs by ion implantation of the dopants magnesium and zinc, for the production of p-type layers relevant to GaAs multilayer microwave device manufacture is studied. Attempts are made to emulate the MBE grown planar doped banier buried p-type spike using ion implantation of magnesium through a n-type contact region, and into a 'n-i-n' MBE grown structure. The result is a working bulk unipolar diode, with barrier height dependent on magnesium implanted energy and dose. A p-type dopant diffusion control experiment is also conducted whereby the depth of the diffusion-controlling phosphorus co-implant is varied to yield different p-type doping profiles. It is demonstrated that using this method it is possible to achieve a p-n junction with a gradually decaying p-type surface region, or an extremely abrupt junction. This technique is then used to produce and study varactor diodes, focusing on their rest capacitances, and their capacitance ratios. SIMS analysis and differential Hall effect measurements are also performed.

Flat panel displays based on electron field emission can provide the benefits of the high resolution of a cathode ray tube display while possessing the portability of a liquid crystal display. To date, the problem with a field emission flat panel display based on silicon is that it usually involves complex photolithography processing, making it too complex and expensive to be commercially viable. In this thesis, the emphasis of the research is to fabricate a three terminal silicon device for flat panel display based on field emission technology without using photolithography processes. Laser crystallised amorphous silicon is chosen for our material which creates a rough silicon surface whose roughness gives rise to field enhancement. Furthermore, this process is widely used in the display industry to fabricate silicon based display driver thin film transistors, which can be readily incorporated. It is important to understand the electron field emission mechanism from the laser crystallised amorphous silicon and to find optimum conditions for emission. In the course of our research, we established a regime for super sequential lateral growth or a hybrid sequential lateral solidification and super lateral growth in Nd:YAG crystallisation of amorphous silicon. Excimer laser crystallised amorphous silicon under optimum conditions gives emission currents of the order of 10-5A (current densities ~ 0.04 A/cm2) at threshold fields less than 15 V/mum in a diode configuration, without the need for a forming process. Through experiments, we concluded that the field emission mechanism from these samples is not controlled purely by surface phenomena, contrary to what was suggested by the Fowler Nordheim theory. Instead, it is the diffusion of the underlay metals into the silicon that create clusters of silicide that allow the electrons to become "hot" while travelling between the clusters. Lastly, a novel process illustrating that a three-terminal device based on laser crystallised amorphous silicon can be fabricated without the need for photolithography. However, the field emission data showed that some fine-tuning of the process is still required.

This work assesses the performance of reflector antennas under adaptive control. In this study, a conventional reflector antenna is given adaptive control by placing additional feeds in the reflector focal plane. The antenna effectively becomes a Multiple Beam Antenna (MBA). This configuration has received far less attention than the Sidelobe Canceller which achieves spatial discrimination by placing low gain elements in the periphery of the reflector. The applications that motivate this work involve rotating radar antennas subject to main beam interference or multiple targets that must be resolved within a 3-db beamwidth. The possibility of the interference being correlated with the wanted signal is also considered. Within the context of these applications the assessment addresses: The basic (main-beam and sidelobe) cancellation performance of this type of antenna. The influence of beam characteristics such as crossover levels, gain, sidelobe levels, etc. in performance. The Control Laws that avoid cancellation of a wanted signal while rejecting interference even if knowledge in signal direction is only approximate. Algorithms to solve adaptively the above Control Laws at speeds consistent with rotating antennas and rapidly changing interference. The use of the beamformer output to generate a reference signal to avoid cancellation of the wanted signal. An algorithm, so far used in the frequency domain, applied in this work to the space domain with the purpose of avoiding cancellation of the wanted signal in the presence of correlated interference such as multipath. The construction of an experimental antenna is also undertaken and it is demonstrated that adaptive nulling is readily achieved in practice.

No description available.

This thesis presents an experimental and theoretical investigation of electrical failure in MOS structures, with a particular emphasis on short-pulse and ESD failure. It begins with an extensive survey of MOS technology, its failure mechanisms and protection schemes. A program of experimental research on MOS breakdown is then reported, the results of which are used to develop a model of breakdown across a wide spectrum of time scales. This model, in which bulk-oxide electron trapping/emission plays a major role, prohibits the direct use of causal theory over short time-scales, invalidating earlier theories on the subject. The work is extended to ESD stress of both polarities. Negative polarity ESD breakdownis found to be primarily oxide-voltage activated, with no significant dependence on temperature of luminosity. Positive polarity breakdown depends on the rate of surface inversion, dictated by the Si avalanche threshold and/or the generation speed of light-induced carriers. An analytical model, based upon the above theory is developed to predict ESD breakdown over a wide range of conditions. The thesis ends with an experimental and theoretical investigation of the effects of ESD breakdown on device and circuit performance. Breakdown sites are modelled as resistive paths in the oxide, and their distorting effects upon transistor performance are studied. The degradation of a damaged transistor under working stress is observed, giving a deeper insight into the latent hazards of ESD damage.