Experimental characterization of flow dynamics of pulsed-chemical vapour deposition.
Degree GrantorUniversity of Canterbury
Degree NameMaster of Engineering
This research is a study of the precursor mass transport, the first variable that affects the film deposition rate, uniformity, coverage, and microstructure of resulted films on substrates inside Chemical Vapour Deposition (CVD) reactors. The Pulsed-CVD reactant flow field uniformities in pulse flow were compared to equivalent steady flow regimes. For mass limited transport CVD processes this represents an important matter, as precursor flux increase leads directly to increased deposition rates. The objective of the research was to develop design relations and define operational parameter ranges to achieve flow field uniformity through experimental investigations. Metered gaseous N2 reactant quantities were injected at equal time intervals into the continuously evacuated reactor. The resulting reactor pressure cycle crosses all the three pressure flow regimes, from viscous, to transition and finally to molecular flow. Nondimensional flow parameters for this unique pulse pressure flow regime were developed from first principles and were studied for relation to design and operation of Pulsed-CVD equipment and processes. Because of the reactor low pressures and non-steady conditions, temperature induced buoyancy driven flows have low effect on the flow field dynamics of the gaseous N2 flow (low Grashof number). Thus this research into pulsed pressure flow field uniformity was conducted for isothermal reactor conditions, without the heater powered. For the reactor flow field uniformity determination, the naphthalene sublimation technique has been employed. This method is usually employed in viscous flow for the determination of the convective heat transfer coefficient through the heat and mass transfer analogy. In this research a method was developed to use the sublimation rate of several samples placed at different locations in the reactor volume to measure the relative convective and pressure conditions, and thus the uniformity of the reactor flow field. xvii Experiments have been run by subsequently varying the pulsing cycle length, the reactor pressure (implicitly the injected reactant mass), and the deposition substrate geometry. The rest of the deposition variables have been kept constant. The experimental results show that cycle time greater than or equal to four times the reactor molecular time constant lead to best pulse flow uniformities, and that for these cycle times the 3D flow field uniformities in pulse flow regimes are always better than in equivalent steady flow ones. Comparable uniformities in both flows between stacked wafer substrates have been determined, with slightly better uniformities in pulse flow than in equivalent steady flow experiments. In order to determine the steady flow field uniformities inside the experimental reactor, as well as when varying its geometrical characteristics, the gas flow was simulated using the finite volume Computational Fluid Dynamics (CFD) method and the commercial software Fluent 6.1. Design and process parameters are proposed, and the reactor pressure is analytically modelled for the pulse flow regime.