Research
Evolution of Advanced Devices and Materials
Advanced semiconductor devices, indispensable for social infrastructure systems, must be more sophisticated and energy efficient. As a result, device structures and fabrication processes are becoming more complex, and new materials are essential to add new functions. Developing and introducing new analysis and measurement technologies is indispensable for such research and development.
Advances in materials analysis equipment/measurement instruments
Materials analysis and measurement instruments are evolving digitally, with each device becoming computer-controlled and storage capacity increasing. This has made it possible to acquire large amounts of data at high speeds. In addition, it is now possible to measure multiple parameters simultaneously. On the other hand, while it is now possible to make high-resolution measurements quickly, the influence of individual measurement noise is also becoming significant. Therefore, advanced estimation processes are essential to predict results from large amounts of but incomplete information, to integrate multiple measurement results, and to supplement interpretations with prior knowledge.
Purpose of Metrology Informatics
The goal is to achieve highly accurate and automated measurements, analyze and interpret the meaning of measurement data using AI and machine learning, and perform autonomous measurements. To accomplish this, data must be analyzed using data science and mathematical statistics, and methods must be developed that enable rapid and highly accurate analysis. Another critical issue is the construction of a system infrastructure that enables the discovery of new phenomena, the elucidation of principles, and the accumulation and utilization of data.
Various kinds of Informatics
In recent years, informatics technology has been increasingly used in the research, development, and manufacturing of materials and devices. As shown in the figure, we consider metrology informatics to be positioned as a supporting critical and essential technology for (narrowly defined) materials informatics, process informatics, and manufacturing informatics (virtual metrology, soft sensors).
Some Applications of Metrology Informatics
Image Segmentation
Automatic image processing of large amounts of data in a short time
Freeing from simple tasks and giving more time to be creative
Extraction of hidden infomation
Extract physically and chemically useful data from complex and large amounts of spectral data.
Lead to to derive new indicators that are consistent with physical and chemical knowledge.
Peak sepration/fitting
Automatic and accurate determination of the number of peaks in noisy spectra, etc.
Differences in judgment between people can be avoided. A large amount of data can be processed.
Multimodal measurement method
Organically combine multiple advanced measurement methods to obtain knowledge that was previously unavailable.
Research Themes
3D segmentation x 3D point cloud data multimodal analysis
We are developing a method to determine the precise three-dimensional microstructure and composition/impurity distribution of advanced semiconductor devices by combining transmission electron microscopy and three-dimensional atom probe techniques.
Multiscale and Multimodal Spectral Imaging Analysis
For advanced semiconductor materials such as compound semiconductors like gallium nitride and two-dimensional semiconductors, we are developing multi-scale and multi-modal spectral imaging methods that combine microscopic photoluminescence at different spatial scales and cathodoluminescence using scanning electron microscopy to We deepen our understanding of physical properties.
Spectral Analysis by Bayesian Inference
Photoluminescence/cathodoluminescence spectral data of two-dimensional electron gas in gallium nitride semiconductors allow us to explore carrier behavior under photo- and electron-beam excitation.
Metrology Informatics Infrastructure
To efficiently advance research and development of materials and devices, the experimental cycle of measurement →, design →, synthesis/process must be effectively executed. For this purpose, a database that automatically accumulates experimental data is indispensable. In addition, the development of technologies for automatic analysis of measurement data is also required. For this purpose, we will work on high-speed automatic analysis technology using such as Vision Transformer and other spectral data technologies.