Heat transfer and mixing in initiated Chemical Vapor Deposition analyzed by in-situ gas composition sensing

Abstract

Studies of iCVD kinetics often make assumptions about the iCVD vapor phase that have been challenging to validate experimentally. We address this gap by investigating heat transfer and mixing in the iCVD reactor using in-situ gas composition sensing. Our work allows practitioners of iCVD to estimate the degree of mixing and temperature profile in the vapor phase based on reactor dimensions and process variables such as chamber pressure. We begin by using dimensional analysis to simplify the parameter space governing heat transfer and mixing, identifying key dimensionless groups that capture the dominant physics. We find that the degree of mixing in the reactor is primarily determined by the Peclet number (Pe), and heat transfer is primarily determined by the Knudsen number (Kn). Multiphysics simulations of the iCVD reactor provide visualizations of non-ideal mixing and allow us to identify the critical Pe above which the well-mixed assumption may begin to break down. To verify that the well-mixed assumption applies to iCVD at low Pe, we measured the residence time distribution (RTD) for argon and isopropanol flowing at 1 sccm, finding a close match to the well-mixed model. We used a Pirani pressure gauge for composition sensing to measure the RTD. We demonstrate the Pirani gauge to be a precise gas composition sensor, which is more modular and less expensive than methods like in-situ infrared spectroscopy and mass spectrometry. We then address heat transfer by measuring thermal diffusion from the filament at a range of pressures in argon and isopropanol. We quantified the effects of non-ideal heat transfer at low pressures and identified a critical Kn, above which heat transfer is independent of pressure. Measurement of thermally driven pressure changes in batch mode enables us to define and model an "effective temperature" for the reactor, matching the thermal diffusion model.