The use of Mid-infrared (mid-IR) imagers has great potential for a number of applications in gas sensing and medical diagnostics, but so far for many of those non-defence fields it has been significantly limited by their high price tag. One of the reasons behind the great cost of mid-IR imagers is that most of them need to operate at cryogenic temperatures. Thanks to more than half a century of research, state-of-the-art mid-IR photodetectors have finally achieved premium detection performance without the need for cryogenic cooling. Some of them have even demonstrated very promising results, suggesting room temperature operation is on the horizon. As a result, the cost associated with cooling equipment has been significantly suppressed. However, most mid-IR imagers are still based on hybrid technologies needing a great number of die-level process steps and being prone to connection failure during thermal cycles. The high manufacturing cost this entails is also preventing a wider diffusion of mid-IR imagers. Currently, there is still a lack of an effective monolithic approach able to achieve low-cost mass production of mid-IR imagers in the same way as monolithic integration has been widely used for imagers working at visible wavelengths. This thesis presents a novel monolithic approach for making mid-IR imagers based on co-integration of mid-IR photodetectors with GaAs-based MESFETs on the same chip. The initial focus of the project was the development of the fabrication steps for delivery of prototype devices. In order to achieve monolithic fabrication of pixel devices made in either indium antimonide (InSb) or indium arsenide antimonide (InAsSb) on a gallium arsenide (GaAs) substrate, various highly controllable etch processes, both wet and dry etch based, were established for distinct material layers. Moreover, low temperature annealed Ohmic contacts to both antimonide-based materials and GaAs were used. The processing temperatures used never exceeded 180˚C, preventing degradation of photodetector performance after fabrication of transistors, thus avoiding well-known thermal issues of InSb fabrication. Furthermore, an intermediate step based on polyimide was developed to provide a smoothing section between the lower MESFET and upper photodetector regions of the pixel device. The polyimide planarisation enabled metal interconnects between the fabricated devices regardless of the considerable etch step ( > 6 µm) created after multiple mesa etches. Detailed electrical and optical measurements demonstrated that the devices were sensitive to mid-IR radiation in the 3 to 5 µm range at room temperature, and that each pixel could be isolated from its contacts by switching off the co-integrated MESFET. Following the newly developed fabrication flow, InSb-based mid-IR imaging arrays (in two sizes, 4×4 and 8×8) are presented here for the first time, with pixel addressing achieved by monolithically integrated GaAs MESFETs. By demonstrating real-time imaging results obtained from these array devices at room temperature, implementation of a new type of monolithic focal plane array for mid-IR imaging has been confirmed. The device is suitable for further scaling (up to 64×64 pixel and beyond) and potential commercialisation. More importantly, the monolithic approach developed in this work is very flexible, as a number of III-V materials with mid-IR detecting capabilities can be grown on GaAs substrates, meaning alternative semiconductor layer structures could also be investigated in the near future.

Since the first contact and fusion of an egg and sperm and throughout development, a cell lives a life of constant communication with its environment. Cells interact with the external environment via a layer of proteins and respond to not only biochemical cues but also physical properties including stiffness and topography of adjacent surfaces. However, even though polymeric biomaterials have been described as one of the cornerstones of tissue engineering, the effect of an intrinsic polymer property known as mobility on cell behaviour is poorly characterised. Mobility is a physical property of polymers inversely proportional to the glass transition temperature (Tg); the temperature at which polymers undergo a transition between a rubbery viscous state to a glassy brittle solid. Therefore, films of four poly(alkyl acrylates) with similar surface chemistry but different glass transition temperatures achieved by varying branch chain lengths (1, 2, 4 or 6 methyl groups) were used in this work to investigate the role of polymer mobility on cell behaviour. I verified using atomic force microscopy the similarity in topography and stiffness between the four substrate surfaces and ascertained that fibronectin molecules adsorb in a globular conformation on the polymer with the shortest side chain (1 methyl group) compared with a more extended conformation on the rest of the polymers. My study of the fibronectin coatings using fluorescence recovery after photobleaching (FRAP) on the different polymer surfaces suggested that the mobility of the polymer substrate is translated to the interfacial protein layer. This interesting finding highlighted a possible pathway for cells cultured on fibronectin coated polymer surfaces to detect the underlying polymer mobility via the fibronectin coating. The interaction of cells with surfaces occurs via membrane proteins which interact with specific structural sites within extracellular matrix proteins; these include the cell binding site (RGD: Arginine Glycine Aspartic acid amino acid motif) and the Synergy site (PHSRN: Proline Histidine Serine Asparagine amino acid motif). My ELISA analyses indicated a higher exposure of these important cell-binding sites on the more extended fibronectin compared with the globular one however, this did not correlate to the mobility of polymers or the mobility of the fibronectin layer. This was also the case for myogenic cell differentiation, which was indiscriminately higher on polymers with extended fibronectin, however, cytoskeletal contractility was found to play an essential role in the myogenic differentiation of cells on these polymers in a mobility dependent manner. We then sought to understand the role of 21 mobility in modulating osteogenic differentiation of human MSCs in the presence and absence of stimulation with BMP-2. The Fibronectin network-forming polymer with the lowest mobility (side chain of 2 methyl groups) induced the highest expression of osteogenic markers in the absence of BMP-2 stimulation. My mechanistic studies using specific inhibitors also revealed that the Erk1/2 pathway was required for this increase in osteogenic markers, while contractility, unlike in myogenesis produced only minimal effects on osteogenic differentiation. In this set of polymers, mobility increases with side chain length, while all the polymers with more than one methyl group in their side chain induced the independent formation fibronectin networks upon adsorption. The polymer with two methyl groups in its side chain is characterised with the lowest mobility among the three fibrillogenesis - inducing polymers, and the highest expression of osteogenic markers in the absence of BMP-2. In the presence of BMP-2, smad phosphorylation was also higher on this polymer suggesting a combined synergistic effect towards osteogenic differentiation provided by the simultaneous activation of the Erk1/2 pathway and high phosphorylation of smad1/5/8. My observations suggest that fibronectin fibrils coating a polymer with low mobility may be most suited for osteogenic differentiation of hMSCs by simultaneously exposing cell-binding sites to a higher degree. Thus, inducing Erk1/2 signalling and presenting BMP-2 in a manner that stimulates the highest phosphorylation of smad1/5/8 hence achieving a stronger synergistic effect on the overall expression of osteogenic markers. The findings from this work strongly support previous studies suggesting that polymer mobility is a subtle change in the substrate with significant downstream biological significance and is crucial to understand to improve the application of polymeric biomaterials.

Silicon compatible light sources have been referred to as the \holy grail" for Si photonics. Such devices would give the potential for a range of applications; from optical interconnects on integrated circuits, to cheap optical gas sensing and spectroscopic devices on a Si platform. Whilst numerous heterogeneous integration schemes for integrating III-V lasers with Si wafers are being pursued, it would be far easier and cheaper to use the epitaxial tools already in complementary-metal-oxide-semiconductor (CMOS) lines, where Ge and SiGe chemical vapour deposition is used in a number of advanced technology nodes. Germanium is an effcient absorber, but a poor emitter due to a band-structure which is narrowly indirect, but by only 140 meV. Through the application of strain, or by alloying with Sn, the Ge bandstructure can be engineered to become direct bandgap, making it an effcient light emitter. In this work, silicon nitride stressor technologies, and CMOS compatible processes are used to produce levels of tensile strain in Ge optical micro-cavities where a transition to direct bandgap is predicted. The strain distribution, and the optical emission of a range of Ge optical cavities are analyzed, with an emphasis on the effect of strain distribution on the material band-structure. Peak levels of strain are reported which are higher than that reported in the literature using comparable techniques. Furthermore, these techniques are applied to GeSn epi-layers and demonstrate that highly compressive GeSn alloys grown pseudomorphically on Ge virtual substrates, can be transformed to direct bandgap materials, with emission >3 m wavelength { the longest wavelength emission demonstrated from GeSn alloys. Such emission is modeled to have a good overlap with methane absorption lines, indicating that there is huge potential for the such technologies to be used for low cost, Si compatible gas sensing in the mid-infrared.

Nanotechnology is currently emerging as the next industrial revolution. It enables the production of goods (i.e. nanoproducts, NPs) with enhanced functionalities, which have nonetheless caused mounting concerns about the potential implications they can have on the environment, economy and society. This thesis employs Multiple Criteria Decision Aiding (MCDA), one form of decision support, to aid the sustainable development of nanotechnology. The first original contribution of this doctoral research is the development of a framework of sustainability assessment criteria for NPs, through a three-phase procedure based on the MCDA process, including a literature review, a pilot and a main survey. It lead to a comprehensive framework of 68 criteria, ranked according to their relative importance, allocated to six main domain areas: (i) economic performance; (ii) environmental impacts; (iii) environmental risk assessment; (iv) human health risk assessment; (v) social implications; and (vi) technical performance. All the criteria are reliable and can be used in real case studies to increase the knowledge about the sustainability of NPs. The second original contribution presented in this thesis is a robust model (DRSA-based model) based on green chemistry principles implementation for the classification of synthesis processes of nanomaterials in preference-ordered classes. This tool was developed through knowledge elicitation techniques based on coconstructive MCDA with the collaboration of two experts (the decision makers) in synthesis of nanomaterials. The robustness of the ensuing model was assessed (and confirmed) by means of another model developed ad hoc (ELECTRE-based model), structured on an MCDA method implementing a stochastic multiple criteria classification strategy. The results confirm that MCDA is an effective decision support approach to foster sustainable development of nanotechnology, providing that the analysts who apply it take these considerations into account. They must ensure that (1) multidisciplinary teams are created to perform comprehensive and credible sustainability evaluations; (2) problem structuring and model construction are as important as (if not more important) than the results (i.e. decision recommendations) themselves; (3) identification of the appropriate MCDA method depends on the problem at hand and not vice-versa; and (4) the credibility of the decision recommendations is subject to the preferences of the decision-makers. If these considerations are accounted for, the possibility of advancing nanotechnology on a sustainable path is very concrete and realistic.

Surface water flooding due to the inadequacy of local drainage is a significant UK concern (DEFRA, 2005; POST, 2007). Urban flood modelling and mapping are typified by characteristic terrain difficulties (Charteris et al., 2001). These difficulties include topographic complexity (road crests and gutters; raised house pads); infrastructure (including houses, fences, garden beds, etc); and, complex underground pipe networks (Charteris et al., 2001; Mark et al., 2004a & b; Hunter et al., 2008; Neelz & Pender, 2007; Syme 2008). Designed to mitigate surface waters, drainage systems play a central role and include key features such as sewer pipes, kerbside gutter channels, gully pots and drains. These systems are often highly complex and spatially varied, simultaneously representing alternate flood pathways, hydraulic sinks and, potentially, additional flood sources (DEFRA, 2005). Accurate flood mapping (DEFRA, 2012) requires data rich modelling and, potentially, dynamic linking of models to form an integrated representation of urban surface and subsurface systems (DEFRA, 2005). Predicted data serves to estimate potential physical impacts, assessing likely damage to buildings and urban infrastructure (Kelman & Spence, 2004); or enhancing mapping further with peak velocities to provide detailed assessment of hazards to people (DEFRA, 2006; 2012). However such Integrated Urban Drainage (IUD; Gill, 2008) approaches are resource expensive (time and data requirements). Research into approaches offering more efficient representations is considered essential and timely. Therefore, the possible inclusion of proxy-model approaches offers an alternate tool for rapid hazard appraisal. Using a UK case-study approach, this Thesis addresses IUD modelling deficiencies through two specific aims: (i) examining IUD model development and impact on hazard prediction and, (ii) investigation of more resource-efficient proxy model approaches to the fully Integrated Urban Drainage model. Using TUFLOW hydrodynamic software (WBM, 2008), an IUD model of a dense UK urban area (2 sq km) is developed and examined. Firstly through small-scale IUD modelling showing improved IUD model performance with kerbside drainage and flow capture systems, particularly when based on depth-inflow criteria and, secondly, through enhanced infrastructure representation. Outputs are examined for both fluvial and pluvial source floods of the statutory 1% AEP event (HMSO, 2009). Data indicates significant IUD impact in terms of extent reductions of 56% (fluvial) and 30% (pluvial), and consequently mean peak depth reductions of 33% (fluvial) and 20% (pluvial) flood events. Velocity impacts are shown to be near negligible, recording less than 1% variation for each flood event. Examination of IUD proxy-model approaches identified inappropriate use for fluvial flood event modelling. Pluvial event surface water modelling identified approaches based on a uniformly applied adjustment of the 5% AEP Design Flood Frequency event (BS, 2008) showed most (95%) agreement to the full pluvial IUD model. This Thesis’ outcomes have supported current flood risk modelling and appraisal practice by Capita Symonds and WBM Pty Ltd (TUFLOW authors). Notable recent projects include Hereford, England (2010), Gold Coast City, Australia (2012) and Christchurch, New Zealand (2014).

Charge trapping at the channel interface is a fundamental issue that adversely affects the reliability of metal-oxide semiconductor field effect transistor (MOSFET) devices. This effect represents a new source of statistical variability as these devices enter the nano-scale era. Recently, charge trapping has been identified as the dominant phenomenon leading to both random telegraph noise (RTN) and bias temperature instabilities (BTI). Thus, understanding the interplay between reliability and statistical variability in scaled transistors is essential to the implementation of a ‘reliability-aware’ complementary metal oxide semiconductor (CMOS) circuit design. In order to investigate statistical reliability issues, a methodology based on a simulation flow has been developed in this thesis that allows a comprehensive and multi-scale study of charge-trapping phenomena and their impact on transistor and circuit performance. The proposed methodology is accomplished by using the Gold Standard Simulations (GSS) technology computer-aided design (TCAD)-based design tool chain co-optimization (DTCO) tool chain. The 70 nm bulk IMEC MOSFET and the 22 nm Intel fin-shape field effect transistor (FinFET) have been selected as targeted devices. The simulation flow starts by calibrating the device TCAD simulation decks against experimental measurements. This initial phase allows the identification of the physical structure and the doping distributions in the vertical and lateral directions based on the modulation in the inversion layer’s depth as well as the modulation of short channel effects. The calibration is further refined by taking into account statistical variability to match the statistical distributions of the transistors’ figures of merit obtained by measurements. The TCAD simulation investigation of RTN and BTI phenomena is then carried out in the presence of several sources of statistical variability. The study extends further to circuit simulation level by extracting compact models from the statistical TCAD simulation results. These compact models are collected in libraries, which are then utilised to investigate the impact of the BTI phenomenon, and its interaction with statistical variability, in a six transistor-static random access memory (6T-SRAM) cell. At the circuit level figures of merit, such as the static noise margin (SNM), and their statistical distributions are evaluated. The focus of this thesis is to highlight the importance of accounting for the interaction between statistical variability and statistical reliability in the simulation of advanced CMOS devices and circuits, in order to maintain predictivity and obtain a quantitative agreement with a measured data. The main findings of this thesis can be summarised by the following points: Based on the analysis of the results, the dispersions of VT and ΔVT indicate that a change in device technology must be considered, from the planar MOSFET platform to a new device architecture such as FinFET or SOI. This result is due to the interplay between a single trap charge and statistical variability, which has a significant impact on device operation and intrinsic parameters as transistor dimensions shrink further. The ageing process of transistors can be captured by using the trapped charge density at the interface and observing the VT shift. Moreover, using statistical analysis one can highlight the extreme transistors and their probable effect on the circuit or system operation. The influence of the passgate (PG) transistor in a 6T-SRAM cell gives a different trend of the mean static noise margin.

This research thesis presents an analysis of the torsional loads on V-band clamps. In some applications, the relative rotational movement of the flanges connected by V-band clamps can result in catastrophic system failure. The ability to understand the factors impacting on torsional load capacity is therefore essential. In this research project, a theoretical model of a V-band joint subjected to torsional loads was developed. This model is used to identify those parameters that will impact on the joint’s reliability. An experimental investigation was conducted to validate a theoretical model using a newly developed test rig. The development and features of this test rig are presented in this report. This experimental investigation also allowed the impact of those parameters that are difficult to control, to be determined. A total of three V-bands were used with different diameters but nominally identical cross sections were studied. In the research results, the initial slip point between flanges and the V-band clamp was identified by experimentation within this research project. Different sizes of Vbands were used under boundary conditions and loads. From the simulation results it was determned that the friction effect on the V-band depends on the size of the V-band. For the largest size of V-band, there was moderate correlation of the experimental and theoretical results. For the smallest size, the results suggest that with band tightening, flange contact is localised, rather than being throughout the band’s entire circumference. The research demonstrated the significant relevance of the band and flanges’ contact points and the coefficient of friction, especially that between the flanges, on the V-band clamp’s theoretical torsional load capacity.

The control of a moving structure in an unbounded flow has numerous applications in engineering, such as the aileron on an airplane. Here an approach is proposed where a CFD method is coupled with a controller to provide a qualitative flow model, and a tool for the development and the validation of the control scheme. A rotating rigid flat plate in transverse flow is considered. For the CFD, a discrete vortex method is used due to its relevance for separated flows, which implies approximating the flat plate by a thin ellipse. The simulation for a fixed plate has been completed with a plate approximated by a 20:1 ellipse and placed in an inviscid flow. A comparison with an image method is also undertaken. The results show encouraging features for modelling the vortex street, but also problems in the transient behaviour of the flow. The control method is based on fuzzy logic, and has shown a remarkable ability to adapt to the nonlinear nature of the force generated by the flow/structure system. Comparison is made with more classical schemes such as a controller based on optimal control theory using an intermediary flow/structure model, similar to a gain scheduling model, instead of the full simulation.

This thesis describes the investigation of the electrochemistry principles, technology, construction and composition of the electrode materials, electrolyte and additives used in redox flow batteries. The aim was to study a flow battery system with an appreciable working performance. The study explores and compares mainly three different redox flow battery technologies; all-vanadium, soluble lead-acid and a novel copper-lead dioxide flow batteries. The first system is based in sulfuric acid electrolyte environment whilst the other two are in methanesulfonic acid. Various cell parameters such as cell voltage, individual electrode potentials, flow rate and efficiencies (in particular voltage, charge and energy) have been utilized to compare. Further research in other redox couples and comparative study towards the design, construction and electrochemistry, along with the performance of these three batteries in relation to other electrochemical energy storage technologies available was also discussed. These technological studies are of particular interest for applications in the renewable energy storage (offshore and onshore) and sustainable energy research (grid integration and micro generation).

The extreme sport of Skeleton faces driver-training constraints. GB Skeleton athletes are limited to 120 minutes of track time per year. This PhD focuses on the learning effects from virtual environment (VE) training. It was hypothesised that learning and skill-acquisition within the sport of Skeleton could be accelerated through the use of physical simulation and VE training. This has been investigated by linking various perceptual learning paradigms to the design of VE training. A review of previous simulator development and learning intervention research found an apparent lack of task specificity in VE design. This led to initial perceptual learning theories that focused on task specific stimuli cueing. Two on-ice track-testing sessions were conducted. They primarily provided the acquisition of on-track sled dynamic measurements and athlete subjective data, which helped formulate hypotheses around which cueing stimuli was important within a VE training scenario. Test results showed: dominance of g-force proprioceptor stimuli, and an inhibited visual stimuli domain. Following these tests an experiment was conducted using a prototype simulator to investigate if proprioceptor stimuli had an effect on athlete sliding performance and in-simulator learning rates. The results from the 5-subject experiment concluded that mechanically-applied proprioceptor stimuli during virtual training showed improvement of in-simulator learning rates and task performance. The findings promoted the development of new VE feature theories. These theories promoted efficient VE design to maximise learning by customising task specific cueing systems. A novel virtual simulator was developed to investigate these theories. The simulator included; a real-time physics engine for athlete interaction, virtual graphics with 180deg field-of-view vision, high (5-80Hz) and low (up to 5Hz) frequency proprioceptor stimulation using vibration and forcefeedback controlled body loading, vestibular system stimulation from whole body roll motion, and virtual audio generation. A motion cueing experiment was conducted, aimed at identifying if the addition of whole-body roll motion led to an increased learning rate. Subject learning rate and performance was evaluated from steer timing error measurements. Subjective feedback was provided which supported the measured results. The experiment showed that motion cueing accelerated the rate of in-simulator learning for task specific events where vestibular stimulation led to faster perceptual information processing. Three verification methods for simulator evaluation were used to investigate the effectiveness of skill transfer from simulated to real-world environment. Subjective and objective measures assessed the effects each VE subsystem had on subjects’ ability to perform the task. The system’s applicability and validity as a perceptual priming tool was demonstrated and shown; from pre and post simulator intervention of athlete improvement comparisons, real-vs-simulated sled dynamics, and subject matter expert opinion. Compelling evidence was presented to suggest that positive transfer of training occurred. It is concluded that a low-cost VE for Olympic Skeleton training did accelerate the rate of learning and increase athlete performance.