Abstract
In this thesis we assess the use of a-Si photodiodes for medical X-ray imaging. The design of an a-Si X-ray imager based on an active matrix, suitable for medical imaging is outlined. The key points in the design are highlighted, and the role of the a-Si photodiode is discussed. We outline the underlying physics of photodiodes, with particular attention to a-Si, and review the current literature. Based on this a model suitable for simulation has been developed. This model is able to replicate the thickness independence of a-Si diodes without resorting to an arbitrary charge at the p/i interface. Because individual pixels must be isolated, a certain fraction of any pixel will have to be contactless. This causes a considerable loss of signal. The possibility of retrieving this lost signal by using a lateral field was examined using the above model. Five structures were examined. (1) A basic photodiode with a longer top contact than the bottom contact. (2) A junction focussed photodiode where an addition junction contact is placed between the contacts on the bottom to induce a lateral field. (3) An insulated focussed photodiode where an insulated contact is placed between the contacts on the bottom of the diode to induce a lateral field. (4) A skewed photodiode, where the top and bottom contacts are deliberately misaligned to generate a lateral potential. (5) A skewed photodiode with additional insulating contacts on the top and bottom to improve the lateral potential. The focussing elements in these devices were show to have varying degrees of efficiency. It was noted that the current models for recombination would not be suitable for transient simulations. Also, as they neglect the presence of amphoteric defects it seems likely that they would underestimate the recombination. An examination of the current dangling bond models was undertaken with a view to using these in simulations. However, application of these models to the problem at hand revealed limitations in their use. A new approach was tried, using a transient analysis, but this was also found to be wanting. Finally, possible paths for future work are discussed, along with the apparatus constructed for measurements to corroborate the simulation results.