Combining Numerical Simulations, Machine Learning and Genetic Algorithms for Optimizing a POCl3 Diffusion Process

Abstract

Advanced mathematical methods, like machine learning or genetic algorithms, have the potential to further accelerate the computer-aided optimization of processes. In this paper we combine the power of sophisticated numerical simulations with these modern concepts. The goal is to combine the strength of both approaches, high predictive quality from numerical models and fast prediction power of machine learning and genetic algorithms. We demonstrate this on a POCl3 diffusion process and optimize an industry relevant PERC solar cell up to 23.4%. This approach is not limited to POCl3 or PECR cells and can be applied to other cell architectures or processes.