ABSTRACT
While both machine learning and numerical simulations have merits and demerits, machine learning models could be preferred over computational approach due to the following reasons. The adaptability of machine learning algorithm models to an unseen and novel dataset is a crucial benefit it can offer in ALD process by learning from new data and modifying their predictions to better comprehend the process and its final outcomes. It seeks to make machine learning models more transparent and interpretable to develop trust in them. It is crucial to know which physical and chemical variables significantly influence the deposition process and to decide on the best experimental setups to quantify those variables because finding the pertinent input features is one of the major hurdles in embedding domain knowledge into the ALD-based machine learning applications. The ability to optimize the deposition process by learning from sizable datasets produced by historical deposition experimental data is the key strength of utilizing machine learning algorithms in ALD.
