@misc{oai:repo.qst.go.jp:00083932,
 author = {Hideaki, Iwasawa and Tetsuro, Ueno and Masui, Takahiko and Tajima, Setsuko and Hideaki, Iwasawa and Tetsuro, Ueno},
 month = {Nov},
 note = {Angle-resolved photoemission spectroscopy (ARPES) is a powerful experimental technique in modern materials science because it can directly probe electronic states, which are deeply related to the physical properties of materials. Among the advanced ARPES techniques, spatially-resolved ARPES has recently attracted growing interest because of its capability to obtain local electronic information at the micro- or nano-metric length scales by utilizing a well-focused light source [1]. On the other hand, it is not trivial to analyze and understand the spatial variation of electronic states against massive datasets, typically in 4-dimensional space (energy, momentum, and two spatial axes). 
   In this work, we will present unsupervised learning using K-means and fuzzy-c-means clustering methods on spatial mapping dataset taken from Y-based high-Tc cuprate superconductor (YBa2Cu3O7-) by micro-ARPES. The spatial mapping dataset clearly showed spatial inhomogeneity on electronic structures due to multiple surface terminations due to BaO or CuO layers on a cleavage (001) plane [2]. We will present how the clustering analysis enables the visualization and identification of such spatial inhomogeneity on the local electronic structures. The advantages and disadvantages of these clustering methods will be detailed, with a comparison of the conventional analysis method.

[1] Hideaki Iwasawa, Electronic Structure 2, 043001 (2020).
[2] H. Iwasawa et al., Phys. Rev. B 98, 081112(R) (2018)., The 9th International Symposium on Surface Science (ISSS-9)},
 title = {Unsupervised Learning for Identifying Surface Inhomogeneity on Electronic Structures of High-Tc Cuprate},
 year = {2021}
}