Automating Plasma Focused Ion Beam Calibration with AI Powered Image Segmentation

The future of microscopy is automation, and the first step toward fully automated experimentation is calibration. Precise calibration of Plasma Focused Ion Beam (PFIB) instruments is essential for experimental reproducibility and data integrity. However, the current standard is a labor-intensive manual process requiring highly skilled operators and significant time, often resulting in suboptimal calibration frequency. To minimize instrument downtime and enhance experimental accuracy, we developed a closed feedback loop for automated calibration using the Thermo Fisher Scientific AutoScript API and cloud based, AI powered image segmentation. The PFIB system autonomously burns an array of spots, incrementally varying a chosen ion beam parameter. The array is imaged with the on-board scanning electron microscope (SEM) and uploaded to the cloud. Images are segmented using the Segment Anything Model (SAM) and analyzed via OpenCV and a closed-loop feedback system adjusts the instrument parameters to match the most circular spot burn. This fully autonomous procedure can execute overnight, ensuring consistent daily calibration. This approach significantly reduces experimental variability without requiring valuable researcher time.