Understanding the growth of thin films produced by Atomic Layer Deposition (ALD) and Chemical Vapor Deposition (CVD) has been one of the most important challenge for surface chemists over the last two to three decades. There has been a lack of complete understanding of the surface chemistry behind these systems due to the dearth of experimental reaction kinetics data available. The data that do exist are generally derived through quantum computations. Thus, it becomes ever so important to develop a deposition model which not only predicts the bulk film chemistry but also explains its self-limiting nature and growth surface stability without the use of reaction rate data.

The reaction network analysis tools developed in this thesis are based on a reaction factorization approach that aims to decouple the reaction rates by accounting for the chemical species surface balance dynamic equations. This process eliminates the redundant dynamic modes and identifies conserved modes as reaction invariants. The analysis of these invariants is carried out using a Species-Reaction (S-R) graph approach which also serves to simplify the representation of the complex reaction network. The S-R graph is self explanatory and consistent for all systems. The invariants can be easily extracted from the S-R graph by following a set of straightforward rules and this is demonstrated for the CVD of gallium nitride and the ALD of gallium arsenide. We propose that understanding invariants through these S-R graphs not only provides us with the physical significance of conserved modes but

also give us a better insight into the deposition mechanism.