AI-equipped Scanning Probe Microscopy for Self-driving Measurements at the Atomic Scale

Masayuki Abe1, Zhuo Diao1, Keiichi Ueda1,2, Linfeng Hou1, Fengxuan Li1, Hayato Yamashita1

1Osaka University
2Tokyo Metropolitan Industrial Technology, Research Institute

We present an AI-equipped scanning probe microscopy system (AI-SPM), which is designed to autonomously conduct measurements at the atomic scale. This advanced system can autonomously target specific atomic positions and perform tasks with atomic precision such as the acquisition of spectroscopic data and atom manipulations. One of the key features of our AI-SPM is its inherent ability to identify and circumvent (or target) areas with surface imperfections. In addition, the AI-SPM system has been designed to effectively compensate for common problems such as drifting from the target position and atomic fluctuations at the tip apex. These implementations are critical to maintaining the fidelity of site-specific surface measurements and analysis. We have tested our AI-SPM system in the harsh conditions imposed by room temperature experiments, in which thermal drift and atomic fluctuations at the tip apex are ubiquitous. In room-temperature experiments on the Si(111)-(7x7) surface, the system was able to autonomously distinguish areas free of adsorbates and defects and then automatically acquire thousands of current-voltage (I-V) spectroscopy measurements over the four different adatom sites, while compensating the thermal drift and monitoring the probe apex condition in an autonomous experiment. Such experiments yield big data with the statistical significance necessary for the reliable characterization of materials, which confirms the capabilities of our AI-SPM implementation for enhanced data collection. The incorporation of AI into the SPM field paves the way for more efficient, accurate, and reliable surface analyses at the atomic level, potentially transforming our approach to material characterization.