Reactor Analysis of Chemical Vapor Deposition of Titanium Dioxide and Copper Thin Films.

Abstract

A set of reactor models for chemical vapor deposition (CVD) of titanium dioxide films from titanium tetraisopropoxide (TTIP) in a vertical cold-wall CVD reactor has been developed. The effects of the carrier gas on the deposition of TiO$\sb2$ films were examined. The reaction mechanism proposed by the earlier workers in our group was used for the reactor modeling (Siefering and Griffin, 1990a; 1990b). The kinetic parameters were estimated from the reactor models at three levels of approximation; i.e., a lumped parameter (LP) model, a one-dimensional stagnation point flow reactor (1-D SPFR) model, and a full 2-D SPFR model. The 2-D SPFR modeling equations were solved using a control volume based finite difference method (SIMPLE). Based on the 2-D SPFR model, a more accurate activation energy for gas phase reaction was estimated to be 55 kJ/mole. I showed that 1-D SPFR model provided a good agreement with the 2-D SPFR model with our reactor geometry. The relationship between the carrier gas and pure TTIP experiments was explained by the collision theory. Copper films with low resistivity have been deposited from copper(II) hexafluoroacetylacetonate (Cu(hfac)$\sb2$) diluted by H$\sb2$ in a horizontal warm-wall CVD reactor. The effects of substrate temperature and hydrogen pressure on the growth rate of Cu film were examined. A transport controlled engine was observed at substrate temperature of 350$\sp\circ$C and H$\sb2$ pressures above 300 Torr. On the other hand, a reaction controlled regime was observed at the substrate temperature of 250$\sp\circ$C and low H$\sb2$ pressure. A 2-D horizontal flow reactor model was developed to describe the temperature and concentration gradients. First, I calculated the surface concentration profiles by matching the measured growth rates with a power rate expression. Using the predicted surface concentration profiles, I proposed a non-competitive adsorption reaction mechanism to give a more physically realistic rate expression. Kinetic parameters were estimated based on both theory and measured results.