Unlocking the Potential of Data Science: Optimizing Chemical Vapour Deposition and Beyond with Integrated Modeling and Machine Learning
Koronaki E.1,
Loachamin Suntaxi G.2,
Giovanis D.3,
Kathrein M.4,
Czettl C.5,
Boudouvis A.6,
Bordas S.P.A.1
1University of Luxembourg, Luxembourg
2University of Luxembourg and National Technical University of Athens, Luxembourg
3Johns Hopkins University, USA
4Ceratizit S.á r.l., Luxembourg
5Ceratizit GmbH, Austria
6National Technical University of Athens, Greece
Abstract
Data Science (DS) is able to enhance product- and technology-driven developments by improving quality and cost efficiency. Focusing on challenges in Chemical Vapor Deposition (CVD), such as temperature control, gas flux, and chemical reactions, the study demonstrates how DS can optimize complex manufacturing processes. Applied to CVD the methods show significant improvements in forecasting, sensitivity analysis and optimization. These techniques are also extendable to other deposition methods, e.g. PECVD, MOCVD, ALD, and other fields, including sintering and furnace construction. However, challenges remain in integrating DS into technical problem-solving. Technological gaps with main among them the lack of sensors, is a critical obstacle, which needs to be addressed as a problem in its own merrit. The pathway for unlocking the full potential of Data Science in technical domains, is combining traditional, equation-based, modeling, machine learning and natural language processing. This approach opens new prospects, beyond the technical applications, enhancing fields such as neuroscience or business management.