Heteroepitaxy and Properties of III-V Materials Grown on Nanoscopically Roughened (001) Si.

Abstract

Polar-on-nonpolar heteroepitaxy of mainly III-V materials on Si by organometallic chemical vapor deposition (OMCVD) has attracted much interest due to its potential technological importance. The objectives of this work are (1) to study OMCVD growth mechanisms, focusing on the lattice-matched but chemically mismatched heteroepitaxy of GaP on Si, and (2) to explore the possibility of improving the quality of the deposited GaP. We examine particularly the possibility of improving GaP quality by using Si substrates that were deliberately roughened on the scale of nm. We also study comparatively on the chemically compatible but lattice-mismatched system of GaP on GaAs, and the polar-on-polar system of GaP on thermally generated amorphous SiO2. Inadvertent but nonetheless important data in establishing growth details were also obtained from the surface of the Mo susceptor surrounding the sample. We use spectroscopic polarimetry (SP) to follow growth chemistry in real time, and then analyze the deposited material with various techniques including atomic force microscopy (AFM) and spectroscopic ellipsometry (SE). To understand the growth process quantitatively, we investigated in detail the gas-phase kinetics inside our vertical-flow, low-pressure OMCVD reactor chamber. We calculated various kinetics parameters and characteristic dimensionless numbers based on simple hard-sphere approximations. According to the results, we now have a significantly better understanding of the processes as taking place inside the reactor chamber, and can design growth conditions to optimize any specific task that we would like to achieve. We have proposed a model for the growth of GaP in OMCVD using the precursors trimethylgallium (TMG) and PH3. This model is primarily based on our observations that different regions of the growth surface are in communication through the gas phase. The model considers the effects of the substrate and the formation of reactants that are directly responsible for GaP growth. Our model, when used together with the thermodynamics and kinetic theories of nucleation and epitaxy, provides a consistent explanation of the various growth behaviors that we have observed. We have found that GaP films grown in our reactor usually have nonuniform thicknesses, which may be thinner or thicker at the edges, but in all cases shows an exponential dependence on radius. The causes all trace back to the differences in chemical properties of the various surfaces investigated here, more specifically, to the different catalytic effects that these surfaces exert on PH3 decomposition and their tendencies to achieve good-quality GaP heteroepitaxy. These data provide unambiguous evidence that deposition occurs via a precursor that involves both Ga and P, and is formed by heterogeneous catalysis. It addition, this precursor is likely to be largely desorbed and to decompose in the gas phase. Starting with the diffusion equation, we derive analytic expressions that describe the variations in terms of the diffusion parameters and evaluate the diffusion length quantitatively. These results coincide with our observations, and show that different parts of the surface – including that of the susceptor – are in constant contact with each other during growth through gas-phase diffusion.