Fourth FAIRmat Users Meeting

Invited User Talk - Knowledge management and online AI training in Ga2O3 epitaxial growth development by FAIR data in NOMAD (hybrid)

Jun 13, 2024, 1:00 PM

30m

Lecture Hall F (Friedrich-Alexander-Universität Erlangen-Nürnberg)

User Talks (hybrid)

Speaker

Ta-Shun Chou (Leibniz-Institut für Kristallzüchtung (IKZ), Berlin, Germany)

Description

In the rapidly evolving field of materials science, efficient knowledge management and advanced AI training are pivotal for the development of innovative materials. This contribution explores the integration of knowledge management and online AI training in the context of Ga₂O₃ epitaxial growth development by metal organic vapor phase epitaxy (MOVPE), leveraging FAIR data principles within the NOMAD repository. We will discuss how the systematic organization and sharing of data through FAIR principles enhance the reliability and reproducibility of research outcomes.

Ga₂O₃, known for its exceptional electronic and optoelectronic properties, has significant potential in various applications such as power electronics and UV photodetectors. However, optimizing its epitaxial growth process requires a deep understanding of complex material behaviors and growth parameters. By implementing a robust knowledge management system, researchers can efficiently manage vast amounts (and forms) of experimental data and metadata along the years. This ensures that critical insights are preserved and made accessible to the scientific community, facilitating collaborative efforts and accelerating the R&D cycle.

Moreover, the application of AI methods to analyze FAIR data within the NOMAD repository enables the development of predictive models that can guide experimental efforts. Online AI training, powered by these rich datasets, can uncover hidden patterns and correlations, optimizing growth conditions and improving material quality. This integration of data-driven approaches and AI not only enhances the precision of epitaxial growth processes but also reduces the time and cost associated with traditional trial-and-error methods. Recently, we have successfully applied AI modeling to optimize the growth rate and predict the doping level in MOVPE-grown Si-doped β-Ga₂O₃ [1,2].

[1] Chou, T.-S. et al. Toward Precise n-Type Doping Control in MOVPE-Grown β-Ga2O3 Thin Films by Deep-Learning Approach. Crystals. 2022; 12(1), 8
[2] Chou, T.-S. et al. Machine learning supported analysis of MOVPE grown β-Ga2O3 thin films on sapphire. Journal of Crystal Growth. 2022; 126737