Presentation Information
[3F10]Proposal of an XPS data-driven surrogate model: a novel analysis framework utilizing background information
*Shunichi Yoneda1, Ryo Murakami2, Hiroshi Shinotsuka2, Hideki Yoshikawa2, Shigeo Tanuma2, Hiromi Tanaka3, Hayaru Shouno1, Kenji Nagata2 (1. University of Electro-Communications, 2. National Institute for Materials Science, 3. National Institute of Technology, Yonago College)
X-ray Photoelectron Spectroscopy (XPS) is crucial for non-destructive depth analysis. While angle-resolved XPS estimates depth, its accuracy diminishes for deeper layers. Analyzing wide spectra (including background) offers a solution, but SESSA simulations are computationally expensive, hindering practical use. This research proposes a surrogate model for wide XPS spectra to overcome SESSA's computational cost. The model uses pre-calculated response functions to efficiently describe how photoelectron intensity changes with film thickness. We found this dependence is well-approximated by an exponential function, simplifying complex scattering. This allows accurate wide spectra reconstruction, improving calculation speed by approximately six orders of magnitude compared to direct SESSA simulations. Combined with Bayesian estimation, the method enables fast and accurate film thickness estimation. We demonstrate its effectiveness using a simulated In2O3/Si data example.