Automatic extraction of fine structural information in angle-resolved photoemission spectroscopy by multi-stage clustering algorithm

Unsupervised clustering method has shown strong capabilities in automatically categorizing the ARPES (ARPES: angle-resolved photoemission spectroscopy) spatial mapping dataset. However, there is still room for improvement in distinguishing subtle differences caused by different layers and substrates. Here, we propose a method called Multi-Stage Clustering Algorithm (MSCA). Using the K-means clustering results/metrics for real space in different energy-momentum windows as the input of the second round K-means clustering for momentum space, the energy-momentum windows that exhibit subtle inhomogeneity in real space will be highlighted. It recognizes different types of electronic structures both in real space and momentum space in spatially resolved ARPES dataset. This method can be used to capture the areas of interest, and is especially suitable for samples with complex band dispersions, and can be a practical tool to any high dimensional scientific data analysis. A bottleneck for the analysis of data produced by angle-resolved photoemission spectroscopy (ARPES) is the size of the data related to spatial resolution. Building on earlier work, the authors present a data processing method that adopts unsupervised machine learning-based tools to improve the accuracy and efficiency when analysing data produced by nano-ARPES measurements.