Pareto optimal driven automation framework for quantitative microstructure simulation towards spinodal decomposition

In this study, we developed a Pareto optimal driven automation framework for quantitative Cahn–Hilliard simulation of spinodal decomposition processes exploiting the scarce experimental information. Multiple characteristic microstructure data were adopted as the targets for reasoning the hyperparameters and uncertain parameters of the Cahn–Hilliard model. Advanced optimization algorithms and strategies for locating the Pareto frontiers were examined and discussed through applications in c-Ti 1 -x Al x N coatings. Validity of the proposed infrastructure was proved by successfully predicting the microstructure characteristics that were not used for parameter reasoning. It is anticipated that the developed auto-framework should be universal for various materials with spinodal decomposition. Graphical abstract