Identification of a Dynamic Model for a Thin Film Deposition Process using a Self-Organizing Map

Detailed molecular simulations, ranging from the deposition of inorganic thin films to polymerization reactions, describe the position of each atom or molecule. These simulations have high computational requirements and they produce high dimensional data. On the other hand, simple models that describe the dependence of overall material properties on process conditions are needed to design and optimize processes and it is difficult to generate these types of models from the molecular simulation data. We present an algorithm to compute a simple dynamic model for the Kinetic Monte Carlo (KMC) simulation of a thin film deposition process by using principal component analysis (PCA), self organizing map (SOM) and cell-to-cell mapping.