We study material properties in functional materials and devices using various imaging techniques combined with machine learning algorithms. In particular, we study material properties in the applications of information and energy technologies by advanced SPM techniques combined with machine learning algorithms. 

Machine learning algorithms for understanding material properties

(Left) Singular value decomposition and k-means clustering allow identifying each local microstructure of carbon steel. (Right) Further, the application of clustering to the 4D I-V data-set allows visualizing distribution of nanoscale components in fuel cells.

Deep learning algorithm for improving detection sensitivity and image quality of SPM data-set

The improvement of detection sensitivity of piezoresponse force microscopy was demonstrated by combination of band exciation and various machine learning algorithms including deep neural network.

Local evaluation of ferroelectricity and electrochemistry in oxides and metals

We explore local ferroelectricity in various materials systems (e.g. HfO2 and BaTiO3) for multiple applications (e.g. negative cap, FET, and MLCC). Furhter, we explore local electrochemistry in oxide and metal surfaces (e.g. steel, Al, hydrogen storage materials, TiO2, Li ion-based materials) for multiple applications (e.g. fuel cells, corroson, memories). To explore local ferroelectricity and electrochemistry, we apply machine learning algorithms to the SPM data-set combined with the advanced SPM techniques.

Synthesis of oxide materials 

We synthesize various oxide materials (e.g. HfO2, BaTiO3, and TiO2) using different kinds of synthesis methods (e.g. atomic layer deposition (ALD), spin coarter, sputter, and powder process).


We also study vaiours materials such as piezoelectrics and ferroelectrics for sensors, memories and capacitors by analyzing piezoelectric, ferroelectric and capacitive properites and analyze image and spectroscopy based on the machine learning algorithms.