Spectroscopic investigations of SiGe nano-structures

Abstract

This work investigates the growth of several novel structures using a variety of spectroscopic techniques. Strained Si caps deposited on a relaxed SiGe layer are grown with resulting dislocations found to be perfectly confined to the underlying graded layer. Surface morphology is investigated and Moiré Fringes are used in conjunction with Raman spectroscopic techniques to provide an overall picture of the stress present in the Si cap. The Si cap contains stress values greatly exceeding previously reported stress levels in the literature for similar structures. Ge p-n photodiode structures are grown using both novel and conventional techniques with the difference in resulting structure analysed in terms of dislocation density and stress variation of the top layers. The influence of foil thickness on elemental mapping is proven, while the confinement of dislocations in the graded layer sample is in excellent agreement with expected results. Strained Si quantum wells are grown using two distinct novel techniques, with the removal of the conventional SiGe graded layer shown to have negligible impact on the density of threading dislocations, with excellent layer growth reported. The use of a low temperature virtual substrate identified the importance of optimal temperature choice, with Raman techniques displaying a strong correlation between line width and relaxation levels. A generic interface is developed that allows for the integration of various data from a wide variety of disparate techniques. Statistical analysis is performed on Raman phonon positions within thin SiGe layers, with a new equation proposed that allows for more accurate Raman measurements in line with benchmarked X-Ray Diffraction (XRD) values. Subsequent validation of the proposed model revealed excellent agreement with XRD measurements. Finally Micro-Electro-Mechanical-Systems (MEMS) are characterised using both Raman and white light interferometry techniques. Suitable shape variation is identified in line with expected stress variations across the device. Detailed stress mapping, suitable to the development of stress templates at manufacturing level, is proposed using micro-Raman techniques.