Many failure theories have been postulated to predict the behaviour of glass reinforced plastics (GRP) under complex stresses. However, the efficient use of these theories for design purposes is inhibited because the experimental complex stress data needed to verify and discriminate between them is not available. This thesis presents some of the experimental data required. Uniaxial stress test results from flat laminates, and biaxial stress test results from thin-walled tubes under combined internal pressure and axial force, are presented for various damage states under both static and fatigue loading for a plane isotropic material and an orthotropic material. The data were then used to establish which failure theory provided the most acceptable prediction of the observed behaviour. For ultimate strength, the Norris Failure theory gave adequate predictions, but not for resin cracking. The most generally acceptable predictions, for both materials, were given by those theories whose equations contain a constant which is derived from complex stress data, these being the modified Marin and the strength tensor theories. For the orthotropic material, three-dimensional representations of the plane stress ultimate strength and damage failure surfaces are presented for both static and fatigue loading. However, even though a considerable number of specimens were tested, only very few failure surface sections were experimentally established. It was found for both materials that fatigue loading was more damaging than static loading, for both uniaxial and biaxial stress conditions, and that the behaviour of the materials was dependent upon the ratio of the biaxial stresses. Jointed reinforcement layers severely affect the zero-tension fatigue behaviour of the plane isotropic material, but the effect on the orthotropic material decreases with increasing fatigue life. The effect of macro-voids on crack initiation should be acknowledged if damage is used as a design criterion.

The distillation point efficiencies for the alcohol-water binary, ternary and quaternary systems were measured using a modified Oldershaw column. This column is expanded above the tray to separate the newly formed bubbles from the column wall, thus eliminating the surface tension induced wall effects for positive systems and discouraging wetted wall effects. The excessive and recirculating foam and froth found in the conventional Oldershaw column is due to these wall effects and does not represent conditions in large scale distillation. The point efficiencies measured using this column for the system methanol/water were lower than the point efficiencies deduced from the composition profiles across a large and narrow rectangular distillation column using an eddy diffusion model. The narrow rectangular column had a liquid flow path length of about one meter, thus avoiding stagnant zones and flow non-uniformities. The lower efficiencies were due to the shorter contact time between the gas and the liquid. This contact time was increased markedly by fitting an outlet weir to the modified Oldershaw column, thus increasing the tray liquid hold-up and the point efficiencies. These point efficiencies were about 10 per cent lower than those on the large tray at a similar value of the F. Factor. The eddy diffusion model predicted rectangular tray efficiencies about 10 to 20 per cent lower than those measured, when using the improved modified column point efficiencies. Using a suitable model, the improved point efficiencies were scaled-up to the conditions existing on the rectangular tray. This resulted in the large tray values of 2 to 4 per cent lower tray efficiencies than those measured. The surface tension effect on the point efficiencies of the binary systems MeOH/n.PrOH using the original modified Oldershaw column in the absence of wall effects using the concept of the Marangoni stabilising index. The surface tension of these systems were measured using a glass thermometer. The system MeOH/H20 had the highest Marangoni index and showed the highest point efficiencies throughout the composition range, with the EtOH/H2O/n.PrOH, with low values of the Marangoni index, showed comparable point efficiencies throughout the composition range. These systems demonstrate all the possible types of surface tension behaviour. The effects of the outlet weir height and hole size on the point efficiencies in the rectangular column operating under similar hydrodynamic conditions were also investigated using the system MeOH/H2O. There was an increase in point and tray efficiencies on increasing the outlet weir height from 2 mm to 12.7 mm. There was also small increase in point and tray efficiencies on decreasing the hole size from 6.4 mm to 1 mm at the expense of higher pressure drops. The point efficiencies of these trays under different hydraulic conditions were in the range 85 to 95 per cent, with subsequent high tray efficiencies. This provides further evidence of the high tray efficiencies available to the design engineer if the detrimental effects of stagnant zones and flow non-uniformities were eliminated. Two highly non-ideal ternary systems and quaternary system were also studied using the original modified Oldershaw and the rectangular columns. Considerable differences between the individual component point efficiencies were observed. These differences are probably caused by the interactive nature of the mass transfer in these systems. These systems also exhibited equal component point efficiencies in parts of the composition range, which illustrates the composition dependency of these systems. The individual component tray efficiencies for these systems were noticeably different, even with equal component point efficiencies operating across the tray. These differences were simulated using the eddy diffusion model, highlighting the effects of limited liquid back mixing on the tray. The composition profile for the system MeOH/EtOH/H20 were predicted and compared with the measurements across the rectangular column using three methods derived from the original Maxwell and Stephan mass transfer equations. These predictions were in good agreement with the measurements. However, as the comparison is only based on a one meter flow path length, the actual design of distillation column using these methods would be conservative. The prediction of the composition profiles using the point efficiencies from the original version of the modified 0ldershaw column yielded a similar observation for both the ternaries and the quaternary system. An expanded aluminium tray (Expamet 607A) was also subject to preliminary efficiency tests in the rectangular column. This material has corrugated angled holes, thus encouraging the liquid flow across the tray by using the vapour momentum. This material showed much lower pressure drops, due to its high open area compared with conventional sieve trays, and discourages weeping and entrainment.

The electronic nose is a rudimentary replica of the human olfactory system. However there has been considerable commercial interest in the use of electronic nose systems in application areas such as environmental, medical, security and food industry. In many ways the existing electronic nose systems are considerable inferior when compared to their biological counterparts, lacking in terms of discrimination capability, processing time and environmental adaptation. Here, the aim is to extract biological principles from the mammalian olfactory systems to create a new architecture in order to aid the implementation of a nose-on-a-chip system. The primary feature identified in this study was the nasal chromatography phenomena which may provide significant improvement by producing discriminatory spatio-temporal signals for electronic nose systems. In this project, two different but complimentary groups of systems have been designed and fabricated to investigate the feasibility of generating spatio-temporal signals. The first group of systems include the fast-nose (channel 10 cm x 500 μm2), proto-nose I (channel 1.2 m x 500 μm2) and II (channel 2.4 m x 500 μm2) systems that were build using discrete components. The fast-nose system was used to characterise the discrete sensors prior to use. The proto-nose systems, in many ways, resembles gas chromatography systems. Each proto-nose system consists of two microchannels (with and without coating) and 40 polymer-composite sensors of 10 different materials placed along it. The second group of systems include the hybrid-nose and the aVLSI-nose microsensor arrays assembled with microchannel packages of various lengths (5 cm, 32 cm, 7lcm, 240 cm) to form nose-on-a-chip systems. The hybrid-nose sensor array consists of 80 microsensors built on a 10 mm x 10 mm silicon substrate while the aVLSI-nose sensor array consists of 70 microsensors built on a 10 mm x 5 mm silicon substrate using standard CMOS process with smart integrated circuitries. The microchannel packages were fabricated using the Perfactory microstereolithography system. The most advanced microchannel package contains a 2.4 m x 500 J.lm2 microchannel with an external size of only 36 mm x 27 mm x 7 mm. The nose-on-a-chip system achieved miniaturisation and eliminates the need for any external processing circuitries while achieving the same capability of producing spatio-temporal signals. Using a custom-designed vapour test station and data acquisition electronics, these systems were evaluated with simple analytes and complex odours. The experimental results were in-line with the simulation results. On the coated proto-nose II system, a 25 s temporal delay was observed on the toluene vapour pulse compared to ethanol vapour pulse; this is significant compared to the uncoated system where no delay difference was obtained. Further testing with 8 analyte mixtures substantiated that spatio-temporal signals can be extracted from both the coated proto-nose and nose-on-a-chip (hybrid-nose sensor array with 2.4 m long microchannel) systems. This clearly demonstrates that these systems were capable of imitating certain characteristics of the biological olfactory system. Using only the temporal data, classification was performed with principal components analysis. The results reinforced that these additional temporal signals were useful to improve discrimination analysis which is not possible with any existing sensor-based electronic nose system. In addition, fast responding polymer-composite sensors were achieved exhibiting response times of less than 100 ms. Other biological characteristics relating to stereolfaction (two nostrils sniffing at different rates), sniffing rate (flow velocity) and duration (pulse width) were also investigated. The results converge with the biological observations that stereolfaction and sniffing at higher rate and duration improve discrimination. Last but not least, the characterisation of the smart circuitries on the aVLSI-nose show that it is possible to achieve better performance through the use of smart processing circuitries incorporating a novel DC-offset cancellation technique to amplify small sensor response with large baseline voltage. The results and theories presented in this study should provide useful contribution for designing a higher-performance electronic nose incorporating biological principles.

EPR measurements have been carried out on a range of differently doped samples, and amongst the many systems observed, two previously unreported defects have been identified. They both incorporate hydrogen and are the first defects to be positively identified to contain hydrogen in the diamond lattice. The two defects have been identified as the negatively charged nitrogen-vacancy-hydrogen centre (NVH-), and the negatively charged vacancy-hydrogen centre (VH-). The NVH- centre has been identified as having trigonal (C3v) symmetry and an overall electron spin of S = ½ . The spin Hamiltonian parameters have been determined and explained in terms of the proposed model of the defect. The hydrogen atom of the defect is located in the vacancy of the nearest-neighbour nitrogen-vacancy defect and appears to be bonded to the nitrogen atom, thus maintaining the observed C3v symmetry. The VH- centre has also been identified as having C3v symmetry, but has an overall electron spin of S = 1. The hydrogen atom is bonded to one of the four carbon atoms surrounding the vacancy and produces a very small hyperfine interaction. This is explained with the aid of the model and by performing an extended dipole calculation between the hydrogen atom and the unpaired electron probability density localised on the three equivalent carbon neighbours. No reasonable predictions on this defect could be made from studying the same defect found in silicon. The well-documented NV- defect has also been studied and modified spin Hamiltonian parameters have been determined. They vary significantly from the previously accepted ones and for the first time can explain the accepted model of the defect.

This work is an attempt to develop inorganic gas separation membranes for the purposes of separating high temperature binary gaseous mixtures. Carbon dioxide and nitrogen mixtures are the focus of this work but other mixtures could be used. The membrane synthesis route is derived from the sol-gel technique. It relies upon micropores being produced within the membrane and this is accomplished by the thermal removal of organic ligands (the "templates"). The thermal stability and structural evolution with temperature of these materials has been characterised with TGA, DTA, FTIR, 13C CP MAS NMR 11B MAS NMR and 29Si MAS NMR investigations. The research was demarcated into the comparisons between two systems: non-borosilicate and borosilicate. The borosilicate systems were thought to merit special investigation due to the known property of the boron atom in borosiloxane bonds to act as a network enhancer. Three different organic ligands; methyl, ethyl and phenyl have been investigated. The higher thermal stability (~770K) and the known CO2 affinity of the phenyl ligand, led to the production of materials containing both the methyl ligand (to generate porosity) and the phenyl ligand (to hopefully provide CO2 affinity). Other structures with methyl as a backbone but containing boron were found to have superior performance in terms of separation factors, robustness and durability. The permeability of CO2, N2 Ar and He was measured through all the membranes systems, as a function of pressure, temperature and time. In both the borosilicate and non-borosilicate systems, CO2 was found to permeate preferentially over He in the best specimens. This was despite its much larger molecular diameter and for both classes of system, permeance was observed to decrease with elevated temperatures. The general conclusion that for both classes of system the mechanism of preferential CO2 transport is activated surface diffusion. Evidence of gradual adsorption of CO2 by the non-borosilicate systems was indicated by their steady decrease in performance with time when exposed to this species. (Such degradation in permeance performance was not observed for those non-borosilicate systems that had not been exposed to CO2 but just N2, He or Ar. The borosilicate systems however, were far more robust. Any decrease in permeance with time, after exposure to CO2 under pressure, was orders of magnitude slower than with the non-borosilicate systems. For the non-borosilicate systems the decrease in permeability is deemed to be due to CO2 chemisorption and must be related to the surface diffusion. For the non-borosilicate systems however, chemisorption appears to play a far less important role. Structural studies (NMR and FTIR) of all the systems indicated that the pyrolysis of the organic templates produces both siloxane and in the case of the borosilicate systems, borosiloxane linkages as well. These are assumed to be the generators of the sites through which surface diffusion occurs. For the non-borosilicate systems, surface diffusion seems to be improved by the incorporation of phenyl ligands within the siloxane network. However, this is associated with accelerated adsorption and decrease in overall performance. For the borosilicate systems, the most successful system had a methyl backbone and decreased in performance very gradually and after that remained constant except for long-term modulations which were mirrored by the inert species as well. Thermally rejuvenating the degraded non-borosilicate membranes did not meet with success. However, the borosilicate systems did partially respond to this treatment and regained a significant fraction of their original performance. The conclusion is that in the non-borosilicate system chemisorption dominates over physisorption as a CO2 selectivity mechanism, whilst for the borosilicate systems the reverse appears to be true.

This thesis describes the development of silicon planar micro-calorimetric gas sensors employing a nanostructured palladium (Pd) catalyst. Present commercial, bead-type calorimetric sensors have been manufactured for nearly forty years and are used in many applications, such as mining, water treatment and emergency services, with an estimated European market value of €221M by 2004. However, recent advances in both silicon micro-machining and nano materials have created the technologies necessary to transform the present labour-intensive fabrication process in to a new low-cost batch production. In addition, a reduction in power consumption, improved sensitivity and increased poisoning resistance of the sensor can also be achieved. Here, two generations of micro-calorimeter have been designed and fabricated comprising a silicon membrane structured micro-hotplate that can reach up to a temperature of 870'C without failure and an ultra-high surface area nanoporous Pd catalyst (about 20 m2/g), typically 25 run thick, deposited electrochemically on top of a gold electrode above the micro-heater. The exothermic reaction caused by the target gas (e.g. methane) interacting with the Pd catalyst results in an increase in the temperature and so resistance of the micro-heater. A Wheatstone bridge interface circuit is normally used to detect and measure the fractional resistance change. Full 3-D thermo-mechanical simulations have been performed employing experimental data in order to establish a simulation database for future developments. The differences between simulated and experimental results were found to be as low as 4.6%. The response of the sensors has been characterised in both continuous powering mode and pulse modulation powering mode. Device power consumption is only 50mW at 500'C in continuous mode, which is up to 100mW lower than that for commercial sensors. Typical response times of 2ms have been measured and so further power saving can be achieved when the sensors are operated in a pulse mode, e.g. 50% duty-cycle at 10Hz. Hence, an overall power saving of 75% could be achieved compared to commercial product. Infrared thermography revealed that a centre hot spot, commonly found with meander style micro-heaters, has been eliminated by the new drive-wheel micro-heater design. The sensitivity of the sensors has also been improved, up to a factor of 4 at 500'C ((60 mV/mm2)/%CH4), by the nanoporous catalyst and by heating it more isothermally. Furthermore, improvements have also been found on the poisoning resistance. Therefore, the potential commercialisation of the micro-calorimeter is very promising.

The growing concerns over our exposure to hazardous substances have been addressed by stringent legislation to ensure air quality. A wide variety of applications have therefore arisen which require the reliable detection of hazardous gases. Hence, the motivation behind the research presented in this thesis was the aim of developing a portable gas monitor to detect nitrogen dioxide, carbon monoxide and volatile organic compounds (e.g. benzene, toluene). The need to improve gas sensor technology for suitability to this demanding application has been identified. Thus, the objectives were to develop a number of ultra-low power devices consisting of an array of chemoresistive gas sensors for incorporation into an intelligent sensor system. The operation of these sensors relies on the measurement of a change in resistance of a gas-sensitive material when exposed to specific gases. Silicon technology has been employed in order to obtain reproducible, miniaturised sensors with a low unit cost. Furthermore, chemoresistors employing metal oxide semiconductor (MOS), metal-substituted phthalocyanine (XPc) and conducting polymer (CP) materials have been used because of their sensitivity to the gases of interest. Common problems associated with these materials are poor specificity to a target gas and poor stability. However, the approach to minimising these problems was to design arrays of cross-sensitive chemoresistors for use in a microprocessor-based intelligent sensor system. The microprocessor applies a pattern recognition algorithm to the sensor outputs to extract the required information. This thesis describes the design, fabrication and characterisation of these sensor arrays. MOS and XPc materials have shown an optimum performance at elevated temperatures. Micromachining techniques have therefore been employed to integrate resistance heaters in a micro-hotplate structure, which can allow temperatures of 600°C to be attained in —15 ms with a typical power consumption of —150 mW/sensor. A pulsed mode of operation should provide average power consumptions of less than 1 mW. A low power consumption is critical for a portable batterypowered instrument. The design, modelling and characterisation of the micro-hotplate structures have also been described. The design and development of a novel automated gas sensor test system was also fundamental to this research, in order to accurately characterise sensor responses and to validate theoretical models. The research objectives have been fulfilled in that a number of sensor array devices have been produced, which are suitable for a portable intelligent instrument. The different designs and materials are compatible for integration into a hybrid sensor. The advancements achieved in sensor technology provide a foundation for future research into the production of a portable intelligent sensor system.

An investigation was conducted into the coating of metal substrates with a glass-ceramic enamel. Two metal types were coated, a 17% chrome-iron and a low carbon mild steel. The glass-ceramic was based on a complex lithium aluminosilicate glass. The enamel was applied using a vitreous enamelling coating technique, followed by conversion to a glass-ceramic. The coating process strongly influenced the microstructural form developed. For both metal substrate types it was possible to produce coatings which wet well and exhibit good adhesion. Crystalline substrate oxide is observed at the interfaces of these coatings. Abrasion prior to preoxidation is an essential requirement for good coating adhesion on a chrome-iron substrate. The interface region for a coating on chrome-iron exhibits little interaction or microstructural change. -However the coating on mild steel exhibits extensive interaction, with a reaction zone extending into the coating. The marked difference between the two coated substrate types can be explained by the different substrate oxide formed, solubility of the substrate oxide in the coating and nucleating ability of the substrate oxide surface. Addition of adherence oxides (NiO, CoO) to the coating on mild steel was examined. The adherence oxides participate in complex reactions which result in the formation of metallic alloys adjacent the interface. Both adherence oxides promote wetting under conditions where wetting is not possible if they are absent. Nickel oxide is detremental to coating adhesion. This may be related to its ability to cause a rapid dissolution of iron oxide present at the interface.

Fibres were produced from glasses with the following compositions: A= 25% Li20 74% SiO2 1% P205 B= 29% LiO2 68% SiO2 1% P205 1% ZnO 1% K20 by drawing the fibres from a single tip platinum bushing. The fibres, in the form of tows, were then subjected to a series of heat treatments in which the nucleation and crystallization temperatures and times were varied. The tensile strength and elastic modulus was determined for each sample of fibres and use was made of microscopy and X-ray diffraction to determine the microstructure resulting from the subsequent heat treatments. The data obtained from the tests performed in this investigation required a statistical analysis because of the variance in any one measured parameter. This is often found to be the case when investigating small volumes of material such as fibres. A model was proposed, for each composition, relating the strength to crytallization time and an attempt was made to justify each model on the basis of the empirical data obtained and indirect evidence arising from the statistical interpretation of this data. The strength of crystallized fibres produced from composition A. decreased with crystallization time. This was explained by assuming that the strength controlling flaws were introduced by the appearance of a surface crystal layer and that the flaw size was proportional to the thickness of the layer. Evidence was produced to reinforce this argument. Two proposals regarding the elastic modulus, were tested and it was found that observed moduli could beat be described by considering the fibre to be a composite cylinder with an outer annulus of lithium disilicate surrounding a core of glass. Glass B. fibres did not behave in the same way as A. after crystallization except at high crystallization temperature. In general the strength would decrease thenincrease again slightly before finally decreasing. This behaviour was explained by assuming that the mean intercrystal spacing controlled flaw size until the depth of the surface crystal layer became large enough to dominate. Some research is described in which an attempt is made to inhibit surface nucleation/crystallization by using a vapour phase ion-exchange treatment. This work remains incomplete but there was some indication that it would be successful in allowing higher strengths to be achieved in glass-ceramic fibres after prolonged crystallization.

The study presented here is focused on the use of aqua-ammonia solution in a novel solar-powered refrigeration cycle intended to be suitable for use in the rural areas of developing countries. The cycle is referred to as a `intermittent regenerative' (IR) cycle, the term regenerative meaning in this context the use of heat recovery or recuperation. The first chapter describes the three better known cycles which may be considered for this application. The IR cycle is introduced as a hybrid development of these which offers the significant advantages of high efficiency while minimising complexity. Chapter 1 provides a methodology by which the novel aqua-ammonia system can be evaluated in comparison with existing systems. The second chapter surveys previous experimental work on solar driven aqua-ammonia cycles. Chapter 3 consists of a detailed design study of the new IR cycle based on computer modelling techniques. The study serves as an analysis of the cycle and allows the performance of the cycle, together with design features and component sizes, to be simulated in a variety of meteorological conditions. A number of original design proposals are evaluated through the modelling exercise. Chapter 4 summarises the results of a second separate modelling exercise which investigates the absorption phase of the cycle. Chapters 5 and 6 describe experimental work. The results of laboratory tests are compared with the predictions of the computer model and in the event serve to validate the theoretical characterisation made in chapter 3 of the performance of key components of the system. The energy efficiency of the system as measured by experiment is proved to correspond to theoretical prediction, so representing a significant advance on the performance of alternative systems. Chapter 7 addresses itself to the wider question of the social and economic validity of a device with the performance and cost of the IR device. A case study is undertaken which explores the potential role of the device in the fish trading economy of Zambia. The study provides data valuable in assessing the usefulness of the technology in helping to stimulate the under-capitalised rural economy of a developing country and in improving local food resource utilisation. Chapter 8 draws together the conclusions of the various chapters and provides an overall conclusion and comment on the value of the IR system. It is proved to have a high efficiency but not to have the robustness nor portability demanded for widespread application in remote locations. Nevertheless the likely life-time cost-effectiveness of the system is judged to be an improvement on existing alternatives and suggestions are made for further improvement.