Artificial Neural Network Discrimination for Parameter Estimation and Optimal Product Design of Thin Films Manufactured by Chemical Vapor Deposition

Abstract

Industrial production of valuable chemical products often involves the manipulation of phenomena evolving at different temporal and spatial scales. Product properties can be captured accurately using computationally expensive stochastic multiscale models that explicitly consider the feedbacks between different scales. However, product design quality is often tampered by uncertainties affecting process operation. In this work, we used artificial neural networks (ANNs) to estimate an uncertain parameter, accurately predict product properties under uncertainty, and achieve orders-of-magnitude computational savings of a multiscale model of thin film formation by chemical vapor deposition. ANNs were trained using multiple realizations of the uncertain parameter to capture the behavior of the thin film’s two key microscale properties: roughness and growth rate. Next, mean square error and maximum likelihood estimation were used for parameter estimation and to find the ANN that could generate the closest predictions to the real-time measurements collected from the process in the presence of uncertainty. The chosen ANNs were employed to seek for the optimal operating conditions to enable the fabrication process to meet product quality specifications. ANNs are a promising technique for product property prediction and efficient decision making in the design of optimal operating conditions for chemical processes under uncertainty.

Monte Carlo and Kinetic Monte Carlo Models for Deposition Processes: A Review of Recent Works

“…Besides 2D materials, in the context of nanoparticle and nanorod growth, MC simulations [12,40], multiscale modeling [41,42] and enhanced multiscale modeling with Artificial Neural Network (ANN) [43][44][45][46] have been performed. "Classical" kMC models have also used for the deposition of diamond [47], AlN [48] and GaAs [49] film, plasma enhanced a-Si:H CVD [50], hybrid MD/kMC [51] and for growth in extereme pressure conditions [52].…”

Physics-Informed Online Machine Learning and Predictive Control of Nonlinear Processes with Parameter Uncertainty

“…In real production systems, the process dynamics is constantly evolving and changing with respect to time, and these uncertainties in the system thus pose a significant hindrance and deter effective control by MPC. To address this issue, some recent works have developed MPC under uncertainty using ML methods. , For example, in ref the authors proposed a robust economic model predictive controller using the deep deterministic policy gradient framework under disturbances and parametric uncertainty. In addition to the design of robust controllers the quantification of model uncertainty and process disturbances resembles an inverse problem of PI learning, where the time-varying process parameters can be estimated based on the latest process data retrieved.…”

Machine Learning-Based Operational Modeling of an Electrochemical Reactor: Handling Data Variability and Improving Empirical Models

“…In addition, an operational model was constructed for this process using a hybrid approach of a deep neural network and a first-principles model . ANNs were further used to determine optimum operating conditions for chemical and industrial processes, − which contributed to maximizing the feasibility of novel processes from economic and safety perspectives.…”