Oak Ridge National Laboratory develops AI for nanoscale microscopy

Researchers at the Department of Energy’s Oak Ridge National Laboratory (ORNL) have developed SimuScan, an artificial intelligence framework designed to automate nanoscale feature detection in atomic force microscopy (AFM).

Traditionally, operating an AFM requires significant expert judgment to select scan paths, adjust settings, and identify which features warrant closer study. This reliance on specialized expertise can slow large-scale research and lead to inconsistent results. SimuScan addresses these bottlenecks by autonomously targeting the most informative areas of a sample for detailed analysis.

A major hurdle in applying AI to AFM has been the scarcity of high-quality, labeled training data. To overcome this, the researchers utilized a synthetic training data approach. By incorporating real-world scanning imperfections—such as electronic noise, tip geometry, and drift—into synthetic datasets, the framework can achieve high-accuracy generalization in physical laboratory environments.