Realization of closed-loop optimization of epitaxial titanium nitride thin-film growth via machine learning

• Published in: Materials today physics 2021-01, Vol.16, p.100296, Article 100296
• Main Authors: Ohkubo, I., Hou, Z., Lee, J.N., Aizawa, T., Lippmaa, M., Chikyow, T., Tsuda, K., Mori, T.
• Format: Article
• Language: English
• Subjects: Machine learning; Molecular beam epitaxy; Superconductors; Transition metal nitrides
• ISSN: 2542-5293; 2542-5293
• DOI: 10.1016/j.mtphys.2020.100296

Closed-loop optimization of epitaxial titanium nitride (TiN) thin-film growth was accomplished using metal-organic molecular beam epitaxy (MO-MBE) combined with a Bayesian machine-learning technique and reduced the required number of thin-film growth experiments. Epitaxial TiN thin films grown under the process conditions optimized by the Bayesian approach exhibited abrupt metal–superconductor transitions above 5 K, demonstrating a new approach to the efficient development of less-studied materials, such as transition metal nitrides. The combination of the thin-film growth technique and Bayesian approach is expected to pave the way toward accelerating the development of the automated operation of thin-film growth apparatuses. [Display omitted] •Closed-loop optimization of epitaxial TiN thin-film growth was demonstrated using a Bayesian machine-learning technique.•The suitable growth conditions were obtained after eleven thin-film growth experiments.•Superconducting transitions appear above 5 K. High-quality epitaxial transition metal nitrides can be grown via MO-MBE.