Machine Learning for Enhancing Parameter Extraction by the Transfer Matrix Method and Applications to X-ray and Neutron Scattering

• Applicant: Dr. Alexander Hinderhofer
• Subject Area: Experimental Condensed Matter Physics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 557510229

In the field of wave propagation and interface interactions, the transfer matrix method (TMM) is an important mathematical method used to analyze the propagation of waves through layered structures, particularly useful in modelling optics, especially ellipsometry, acoustics neutrons and X-rays. Its widespread application in various scientific disciplines has significantly advanced our understanding of wave behavior at interfaces and employed this for advanced characterization of thin films and materials. However, the current challenge lies in the intricate process of fitting experimental data using TMM and extracting thin film parameters, demanding a considerable amount of expertise and time. Recognizing the limitations of this conventional approach, together with recent developments in machine learning a compelling opportunity to revolutionize the field emerges. By applying the capabilities of a machine learning model specifically designed for parameter prediction, we aim to streamline and enhance the simulation process, offering a more efficient and accessible alternative to TMM. This proposal aims to develop and implement a machine learning-driven approach that not only simplifies parameter prediction but also resolves ambiguity in fitting results and opens new avenues for advancements in wave simulation methodologies.

• DFG Programme: Research Grants