This work investigates the performance limitations of selected CMOS sensor technologies comparing them with an interline CCD imager. The derived performances are used to evaluate whether the claimed benefits mentioned above apply now and will continue to do so over the next decade. The digital-camera and multimedia markets are widely predicted to expand rapidly within the next few years. In these markets, an essential function of an imaging system is analogue-to-digital conversion (ADC). Consequently, ADC is considered as an example of the on-chip integration of support circuitry. Analysis demonstrates the potential system-power saving of parallel focal-plane signal processing and discusses the most suitable converter architectures. A test-chip, designed to validate conclusions from the theoretical analysis, is described. Finally, the results from the test-chip are compared to the developed theory. From this work, conclusions are drawn as to the viability and future development of CMOS imagers; specifically, applications are highlighted where the CMOS imager is most likely to be preferable to the CCD sensor. A number of suggestions for future research are made throughout this work.

This thesis investigates the colour imaging performance that can be obtained by combining the inherent spectral dependence of light absorption depth in silicon with a suitable colour filter array. This colour filter array can contain less colours of wider bandwidth, which makes more efficient use of the illumination. A novel active pixel is developed which contains two photo-junctions stacked vertically in the silicon. Device simulation of the spectral response, along with quantum efficiency measurement, is used to assess the raw spectral selectivity of the double junction structure in two standard CMOS technologies. The response with various colour filter combinations is also presented showing that colour imaging with only two filters should be possible. Active pixel circuitry suitable for such a photo-diode structure is discussed, and the calculated noise performance for the double junction pixel is compared with the standard 3-transistor active pixel. This analysis shows that comparable performance can be achieved using the new pixel type. The design of a simple analogue output 184 x 154 pixel image sensor using the double junction pixel is described. A full characterisation of the performance of the sensor is presented and results compared with a standard sensor. In order to complete a camera system, the processing needed to obtain colour images is implemented in software. Example images from the new camera are given and the colorimetric accuracy of the sensor with different filter combinations is discussed. Results show that the colour reproduction, though acceptable, is not yet as accurate as with a standard RGB sensor. Test patterns are used to examine the aliasing artifacts of the camera. Again, comparison with the standard approach is used to highlight the relative strengths and weaknesses.

Exploiting the advantageous properties of Ferroelectric Liquid Crystals (FLCs) in Liquid-Crystal-on-Silicon (LCoS) microdisplay devices has proved very challenging for several reasons. Means of controlling the small cell gap required for optimum electro-optical performance (typically around 0.8 µm) even across the small active area of such displays had to be developed. Improving the compatibility of the silicon chip with this particular liquid crystal configuration and its intrinsically high susceptibility to cosmetic defects was also required. This thesis presents some process development work aimed at solving these issues. An advanced post-processing procedure for the preparation of silicon backplanes relying on the use of chemical mechanical polishing (CMP) has been employed to prepare realistic sample surfaces for studying the resulting topography on the liquid crystal layer. A process sequence for producing integrated peripheral spacer structures on silicon backplanes is presented and its compatibility with ferroelectric liquid crystals assessed. The use of thin films deposited on the back of silicon wafers for flattening the silicon chip in order to improve the cell gap uniformity across the device was demonstrated. It is also shown that patterning of this stress compensation layer offers the possibility of controlling the symmetry of its flattening effect. Such option is advantageous in terms of the additional latitude it provides in terms of IC design.

No description available.

Traditionally, colour images have been captured by image sensors with a trichromatic checkerboard colour filter using the Bayer Pattern. This filter exhibits non-optimum colour aliasing. The following thesis introduces a novel way of processing an image using a hexagonal array with trichromatic or quadrochromatic filters. This form of processing ensure that all colours are equally sampled, and that the resolution in all directions for all colour channels is practically the same. This leads to a minimisation of aliasing if the lens is optimised for the array. A CMOS camera was manufactured with a standard OS process. Using raw data captured by the camera in association with colour filters, the trichromatic and quadrochromatic colour filter array options were evaluated. The implementation of the colour processor using switched capacitors is investigated alongside the more conservative digital route. A switched capacitor readout buffer is designed and simulated to test the overall performance of the circuit. Using this information, the design based on a switched capacitor implementation for the colour processor is evaluated and rejected. The digital resources required by the new system are investigated by programming an FPGA capable of processing raw data from the hexagonal CMOS camera into a colour picture in real-time. The resulting digital resources required to reconstruct an image from a camera with hexagonal pixels is compared to those needed by the Bayer pattern. Finally since the processing of the image is going to be performed on the same die as the sensor, the supplies are going to be affected by the switching elements, whether the colour processor is implemented in switched capacitor or digital form. This thesis also investigates how the pixels react to varying supplies, and how to render them more resistant to variations in the supplies, so as to deliver a better picture at a minimum cost. Various solutions to improve the power supply rejection of the camera are investigated and simulated. The more interesting options are laid out and tested.

Experiments have been performed on a fully-developed turbulent air flow in a square duct using the technique of Digital Particle Image Velocimetry (DPIV). Due to rapid refinement and development of this technique, a new system has been implemented and tested along with purpose-written software and a novel optical fibre delivery system. Fibre delivery is most desirable for PIV applications due to all-round flexibility, allowing measurements to be made in hazardous environments or where optical access is limited. Unfortunately, until now this was not possible with the Nd:YAG pulsed laser due to the ensuing high power densities produced causing irreversible damage at the input end of the fibre core. However, with the introduction of a Diffractive Optical Element to the launch conditions, it has been successfully shown it is possible to transmit laser light at power densities an order of magnitude higher than previously possible, thus providing sufficient illumination for the imaging of 1<I>μm</I> particles. Another advantage of fibre delivery is that it allows a volume to be traversed. Hence, 8 cross-sections of the flow were analysed using DPIV in order to demonstrate the behaviour of secondary motions which exist in the flow. These exist due to the non-circular geometry of the bounding conduit and the action of turbulence. This is not only a challenging application given the limited resolution of the recording media and the magnitude of such secondary circulations (which are approximately, 1-3% of the bulk velocity), but it is the first time that DPIV has been employed to investigate this complex flow regime which even to this day is not well understood. The statistics are most satisfactory and in good agreement with the measurements ascertained from numerical modelling and other experimental techniques. From a DPIV perspective, this demonstrates the ability of this technique with regards to resolving small-scale turbulent motions with an excellent degree of accuracy. Comparisons between the conventional methods of light generation and optical fibre delivery are also presented in terms of mean flow statistics. This provides a direct comparison to be made and therefore, allows future measurements to be carried out with confidence.

The potential of new erbium doped materials has been assessed with the aim of enhancing and tailoring the emission properties of ER^III ions (such as PL spectrum and lifetime) in pure and modified SiO₂ hosts through sol-gel methodologies. The main focus has been to investigate potential use of novel precursor, erbium triflate, in combination with different organophosphorus ligands, and aluminium co-dopants, for the synthesis of sol-gel thin films, bulk and power materials. This thesis describes two main approaches. In the first approach, sols were synthesised using standard sol-gel hydrolysis and condensation process. This was followed by the deposition of thin films by spin coating. After the deposition the films were thermally treated at high temperature to produce compact inorganic SiO­₂ networks. Finally, the optical characterisation of the films has been studied. The films were optically inactive. The second approach consisted of synthesis of sol-gel bulk materials in the form of glasses and glass ceramics co-doped with Al^III and P^V ions, transparent glasses and glass ceramics incorporating Er^III ions concentration as high as ~10 mol % were prepared containing active nanocrystallites of erbium phosphate in amorphous SiO₂ matrix. Optical characterisation of these glass ceramics has shown that broad and flat emission photoluminescence spectra of Er^III ions with FWHM greater than 100 nm are obtained only in the materials containing ErPO⁴ nanocrystals with average diameter of 4-5nm combined with a narrow and uniform size distribution. Visible upconversion was also detected in the Er-doped sol-gel SiO₂ powders under excitation at 800 and 980 nm. Excited state absorption (ESA) and energy transfer up-conversion (ETU) have been discussed as possible mechanism for visible up-conversion.

Solitons have been proposed as information carriers in next generation fibre optic networks. As the stable waveform in nonlinear optical fibres, solitons are resistant to a wide variety of perturbations from fibre effects, optical devices and other solitons. While they are a particularly robust waveform, solitons must be periodically amplified in order to compensate for fibre loss. This amplification causes random fluctuations in the soliton's position, thus limiting the potential data transfer rates in soliton-based fibre optic networks. As a result fibre optic networks utilising solitons will also have to be filtered in order to remove harmful background radiation. A variety of other devices may be necessary for soliton control or modification. One such device is a fibre compressor which is a section of dispersion decreasing fibre. When passed through a fibre compressor, a train of solitons will decrease in width and show a proportional increase in amplitude. This allows a higher data transfer rate. It is of interest to determine the evolutionary behaviour of solitons in fibre optic networks containing the above mentioned devices and fibre properties. The constituent equation modelling pulse evolution in a nonlinear optical fibre is the nonlinear Schrödinger (NLS) equation. The NLS equation possesses an exact inverse scattering solution. However the evolution to the steady state from an initial pulse is governed by an integral equation and so is difficult to determine. It is this evolutionary behaviour which is of interest. In addition, modelling the above mentioned optical devices and fibre effects requires adding perturbing terms to the NLS equation. These perturbed NLS equations do not possess inverse scattering solutions and so analytical solutions do not exist. Both of these factors lead to the use of approximate and computational techniques to analyse evolutionary pulse behaviour of perturbed NLS equations.

Applications using liquid crystal on silicon (LCoS) spatial light modulators (SLM) demand a high level of device flatness. The overall backplane and individual mirror flatness directly influences SLM optical quality in a number of interrelated ways. They affect the liquid crystal (LC) cell gap uniformity, optical efficiency and LC alignment characteristics directly. In turn, the LC thickness variation and alignment quality affects device contrast ratio. The final cell gap uniformity, and LC alignment quality are also influenced by surface morphology, and by the chemistry of the surface over which the LC flows during cell filling. It is therefore critical to minimise the surface morphology and have a good understanding as to the LC's interaction with the surface over which it flows. Chemical mechanical polishing (CMP) can be used to increase device flatness and reduce the problems associated with surface morphology. A technique to remove the layout and process dependent surface topography using CMP has been investigated. Problems were initially encountered due to the presence of the large, feature dense, pixel array. These were overcome by the development of a novel pre-CMP dielectric etch step, resulting in a greatly improved post-CMP dielectric uniformity and a sub-nanometer RMS surface finish. Conventional planarisation techniques leave the mirrors standing proud of the surrounding dielectric surface. Two methods of reducing this mirror step-height have been compared, namely mirror damascene and via-damascene. The mirror-damascene method resulted in mirrors that are co-planar with the dielectric surface. While this improved the LC flow uniformity during cell filling, it introduced new concerns such as mirror dishing and array erosion. A more attractive technique is that of via damascene which produces vias that are level with the dielectric surface. This approach allows the deposition of thin high quality aluminium mirrors. Initial problems were encountered with via dishing and CMP induced dielectric degradation; both of which were addressed using a post-damascene dielectric buff.

No description available.