Optical monitoring of surface processes relevant to thin film growth by chemical vapour deposition

This thesis reports on the investigation of the use of reflectance anisotropy spectroscopy (RAS) as an in-situ monitor for the preparation and oxidation of GaAs(100) c(4x4) surfaces using a CVD 2000 MOCVD reactor. These surfaces were oxidised using air. It was found that it was possible to follow surface degradation using RA transients at 2.6eV and 4eV. From this data it was possible to speculate on the nature of the surface oxidation process. A study was performed into the rate of surface degradation under different concentrations of air, it was found that the relation between the air concentration and the surface degradation was complicated but that the behaviour of the first third of the degradation approximated a first order behaviour. An estimation of the activation energy of the process was then made, and an assessment of the potential use of the glove-box for STM studies which is an integral part of the MOCVD equipment was also made. Following this, a description is given of the construction of an interferometer for monitoring thin film growth. An investigation is also described into two techniques designed to evaluate the changes in reflected intensity as measured by an interferometer. The first technique uses an iteration procedure to determine the film thickness from the reflection data. This is done using a Taylor series expansion of the thin film reflection function to iterate for the thickness. Problems were found with the iteration when applied to noisy data, these were solved by using a least squares fit to smooth the data. Problems were also found with the iteration at the turning points these were solved using the derivative of the function and by anticipating the position of the turning points. The second procedure uses the virtual interface method to determine the optical constants of the topmost deposited material, the virtual substrate, and the growth rate. This method is applied by using a Taylor series expansion of the thin film reflection function for each of the variables, this is then fitted to the data like a least squares fit. The fit parameters are then used to modify the constants, this process is the repeated to iterate for the actual values. These procedures are tested on simulated data with simulated noise based on the standard distribution curve. These routines were implemented using Visual Basic and C++. (author)SIGLEAvailable from British Library Document Supply Centre- DSC:DXN056386 / BLDSC - British Library Document Supply CentreGBUnited Kingdo