Mastering processing-microstructure complexity through the prediction of thin film structure zone diagrams by generative machine learning models

TL;DR

This paper introduces a generative machine learning approach combining variational autoencoders and GANs to predict and optimize thin film microstructure diagrams, significantly reducing experimental costs in materials design.

Contribution

It presents a novel deep learning framework that accurately predicts structure zone diagrams, enabling efficient exploration of synthesis and composition parameters for microstructure control.

Findings

Generative models produce high-quality structure zone diagrams.

The approach reduces experimental efforts in materials discovery.

Models effectively capture chemical and processing complexities.

Abstract

Thin films are ubiquitous in modern technology and highly useful in materials discovery and design. For achieving optimal extrinsic properties their microstructure needs to be controlled in a multi-parameter space, which usually requires a too-high number of experiments to map. We propose to master thin film processing microstructure complexity and to reduce the cost of microstructure design by joining combinatorial experimentation with generative deep learning models to extract synthesis-composition-microstructure relations. A generative machine learning approach comprising a variational autoencoder and a conditional generative adversarial network predicts structure zone diagrams. We demonstrate that generative models provide a so far unseen level of quality of generated structure zone diagrams comprising chemical and processing complexity for the optimization of chemical composition and processing parameters to achieve a desired microstructure.