No description available.

No

Ce projet de recherche vise à caractériser l'influence de divers traitements de passivation de surface de l'arséniure de gallium (GaAs) sur les propriétés électriques et optiques de ce matériau. Les procédés de passivation étudiés sont les traitements au soufre (NH[indice inférieur 4])[indice inférieur 2]S et les dépôts de nitrure de silicium SiN[indice inférieur x] et trois types de substrat ont été utilisés à titre comparatif, un type N (10[indice supérieur 16]), un type N+ (10[indice supérieur 18]) et un non dopé. Dans ce dernier cas, un système de déposition chimique en phase vapeur assistée par plasma (PECVD) a été utilisé et l'influence de la fréquence de la source d'alimentation AC du plasma a été étudiée. Des techniques de caractérisation électrique, optique et électro-optique ont été utilisées pour l'étude. Des structures métal-isolant-semiconducteur (MIS) ont été réalisées pour les mesures AC et DC de capacité à plusieurs fréquences. L'analyse des mesures électriques a permis de démontrer un plus grand détachement du niveau de Fermi pour les échantillons passivés avec un dépôt de nitrure de silicium SiN[indice inférieur x] à basse fréquence plutôt qu'à haute fréquence. Des mesures optiques en continu et résolue en temps ont été effectuées sur une série d'échantillons de GaAs présentant différents niveaux de dopage et différents traitements de surface. Les mesures de photoluminescence en continu et les mesures résolues en temps montrent que les propriétés optiques des dispositifs dépendent grandement du type de substrat utilisé. Plus d'information sur le champ surfacique des dispositifs est nécessaire pour conclure sur l'efficacité de la passivation. Pour obtenir cette information, des mesures de photoluminescence, continues et résolues en temps, ont aussi été effectuées sur les structures MIS en présence d'un champ électrique. Ces mesures n'ont pas permis de mettre en évidence l'influence de la modification du champ de surface sur l'intensité du signal de luminescence, et ce peu importe le procédé de traitement de surface utilisé. Finalement, des antennes THz ont été fabriquées sur un substrat de SI-GaAs passivé par le traitement PECVD à basse fréquence. Ces antennes émettent un champ THz plus intense et avec un plus grand contenu fréquentiel que celle fabriquée sans traitement de passivation.

In an attempt to evaluate the role of gangliosides as receptors for fibronectin, a series of Balb/c 3T3 variant cell lines, with a reduced ability to bind the ganglioside-specific ligand cholera toxin (CT), were examined. Initial characterisation showed that the cell lines displayed a generalised reduction in the synthesis of gangliosides more complex than GM3, but not of cell surface glycoproteins. There was no reduction in the levels of fibronectin found at the surface of the variants as compared with the parental cell line and all were able to spread and form focal contacts on fibronectin-coated substrates. These results suggest that complex glycolipids of the 'ganglio' series are not essential for Balb/c 3T3 cells to interact with fibronectin. The role of gangliosides as receptors for heat-labile toxin of E.coli (H-LT) was investigated using CT as a ganglioside-specific control. Both toxins bound to ganglioside GM1 and to Balb/c 3T3 cell membranes. Binding of 125I-labelled toxins was inhibited by either unlabelled toxin. There was no evidence to suggest that CT or H-LT recognised receptor(s), in Balb/c 3T3 cells, in addition to GM1. In rabbit intestinal brush borders at 0°C, there were more binding sites for H-LT than CT and 125I-H-LT binding could not be inhibited by unlabelled CT. At higher temperatures there was some inhibition of 125I-H-LT binding by CT. In Western blots H-LT recognised proteins co-migrating with the major brush border galacto-proteins. Toxin binding to brush borders from the Wistar strain of rat was similar. One of the 125I-H-LT binding sites may be the sucrase-isomaltase complex, since the toxin bound to brush border fractions enriched for enzyme activity. The data suggest that the major binding sites for H-LT in brush borders are not ganglioside in nature but may be glycoproteins.

The research presented is part of a larger study, dedicated to investigating flow boiling in small to microchannels. The test facility, originally designed by Huo (2005) and since used by Chen (2006) and Mahmoud (2011), has been used to investigate flow boiling of R134a across a range of channel diameters and both seamless cold drawn and welded channels. These previous studies concluded that one of the reasons for discrepancies in reported data is the result of surface characteristics. The objective of this current study is to further investigate the effect of channel characteristics and changing the refrigerant to R245fa. Surface characteristics are investigated with stainless steel, copper and brass channels, all seamless cold drawn and 1.1 mm internal diameter. Experiments using R245fa were initially conducted in the same stainless steel channel used with R134a by Mahmoud (2011). This allowed for the surface characteristics to be negated and the comparison to be based purely on the changes in the thermophysical properties between R134a and R245fa. Experiments were conducted at inlet pressures of 1.85 and 2.45 bar, mass fluxes of 100 – 400 kg/m2s, heat fluxes from 1 – 60 kW/m2 and vapour qualities from 0 – 0.95. The test section surfaces were evaluated based on scanning electron microscopy (SEM) and confocal laser microscopy (CFLSM). SEM allowed for a visual inspection of the channel surface, with clear differences in the surface stricter evident. The surfaces were then compared based on two CFLSM profilers. The values of the surface parameters differed between the two profilers but the same trend was seen, brass being the roughest surface and copper the smoothest. Changes in the surface parameter values were found to be a function of the scan area, scan resolution and cut-off value. A borosilicate glass tube, at the test section exit, allowed for flow visualisation. Mahmoud (2011) reported bubbly, slug, churn and annular flow for R134a, with no effect of hysteresis. Churn and annular flow were present for R245fa with an increasing heat flux. This was a result of a higher surface tension for R245fa which facilitates annular flow. Hysteresis was evident for R245fa, with bubbly, slug, churn and annular flow seen with a decreasing heat flux. The hysteresis effect is a result of nucleation sites activating during the increase in heat flux and remaining activated as the heat flux is decreased. The activation of nucleation sites depends on the size, which was constant due to the same channel being used, and the wall superheat. The wall superheat is lower for R245fa which does not allow for the nucleation sites to be initially activated with an increasing heat flux. The same effect of hysteresis was evident for copper and brass. Differences in the exit vapour quality and heat flux at which flow patterns occurred were seen between the three materials. The heat transfer coefficient varied in both magnitude and trend between R134a and R245fa. Mahmoud (2011) reported an almost constant heat transfer coefficient with vapour quality at a higher magnitude than seen for R245fa. R245fa showed an increasing trend with vapour quality. Peaks in the heat transfer coefficient were seen to be a result of surface flaw, evident when plotting as a function of the axial location. The test section was reversed in orientation, moving the location of the peak from near the entry of the test section to near the exit. A similar heat transfer coefficient peak was seen at the same axial location, near the exit of the test section, confirming that the peak was a result of a surface flaw and a result of the flow developing. The heat transfer coefficient changed in magnitude and trend for copper and brass. The magnitude of the recorded heat transfer coefficient did not follow the same trend as the surface parameters. The heat transfer correlations in literature did not predict the increase in the heat transfer with vapour quality, performing poorly compared with R134a. The best correlation for the prediction of both refrigerants was that of Mahmoud and Karayiannis I (2012). The pressure drop for R245fa was over 300 % higher than that of R134a, with a steeper increase with heat flux. This is attributed to a higher liquid viscosity and lower vapour density for R245fa. The pressure drop was highest for the roughest channel, brass, but lowest for stainless steel which had the intermediate roughness. The smoothest channel, copper, showed the largest difference in the effect of inlet pressure on the measured pressure drop and the roughest surface, brass, the smallest difference. The effect of surface characteristics on pressure drop is greater than the effect of changes in the fluid properties with inlet pressure. Pressure drop correlations performed poorly for R245fa in comparison with R134a, with the majority under predicting the pressure drop. Only one pressure drop correlation included a function of the surface parameters, Del Col et al. (2013), but this correlation under predicted the effect of the surface parameters on pressure drop. There was no one correlation which gave satisfactory results for all three materials.

Riemann surfaces frequently possess automorphisms which can be exploited to simplify calculations. However, existing computer software (Maple in particular) is designed for maximum generality and has not yet been extended to make use of available symmetries. In many calculations, the symmetries can be most easily used by choosing a specific basis for H₁(Σ,Z) under which the automorphism group acts neatly. This thesis describes a Maple library, designed to be used in conjunction with the existing algcurves, which allows such a choice to be made. In addition we create a visual tool to simplify the presentation of Riemann surfaces as sheeted covers of C and the creation of homology bases suitable for use in the Maple library. Two applications are considered for these techniques, first Klein's curve and then Bring's. Both of these possess maximal symmetry groups for their genus, and this fact is exploited to obtain neat algebraic homology bases. In the Klein case the basis is novel; Bring's is derived from work in the hyperbolic setting by Riera. In both cases previous hyperbolic work and calculations are related to the algebraic results. Vectors of Riemann constants are calculated for both curves, again exploiting the symmetry. Finally this thesis moves back to simpler cases, and presents a general algorithm to convert results from general genus 2 curves into results based on a symmetric basis if one exists. This is applied to algebraic and numeric examples where we discover an elliptic surface covered in each case.

Scuffing is a severe form of surface damage which limits the performance of lubricated sliding machine components. Empirical work has shown that failure either occurs under relatively mild elastohydrodynamic conditions with barely modified surfaces or under severe conditions with the surfaces well run-in. Two hypotheses exist which may explain these experimental differences. This thesis examines their relevance. The first hypothesis is that, under elastohydrodynamic lubrication, the surface asperities either remain rigid or become elastically deformed-micro-elastohydrodynamic lubrication. A non-dimensional plot, developed by Baglin, predicts the occurrence of the regimes. An experimental study of running-in and scuffing for tests initially operating in the different regimes is described. Tests were run on a two disc machine with incremental loading. Running-in occurred both when tests started in the micro-ehl regime and when they apparently entered it during operation. High sliding prevented entry into micro-ehl; scuffing occurred with barely modified surfaces. This hypothesis discriminates between failure types but cannot alone predict scuffing. The second hypothesis, by Crook and Shotter, is that scuffing represents an inbalance between the rate of film thinning with increasing load and the rate of running-in. Increasing load increases the temperature which, due to its effect on viscosity, controls film thinning. Knowledge of the machine's thermal behaviour is required. A model is developed to predict temperature in a finite length cylinder subject to a discrete rotating heat source and convective cooling. Steps to apply the theory to a two disc machine are detailed and the results compared to previous experimental temperatures. Methods of changing thermal response are considered and preliminary tests with the discs insulated to increase the temperature rise are described. A marked reduction in scuffing load emphasises the importance of thermal design. Further experimentation is necessary to determine whether the Crook and Shotter hypothesis can quantify scuffing failure.

No description available.

No description available.

No description available.