AutoLab: A Vision-Guided Robotic Platform for Reproducible 2D Material Stacking

Two-dimensional (2D) materials such as graphene, hexagonal boron nitride, and transition-metal dichalcogenides can be assembled into artificial heterostructures with atomically sharp interfaces and tunable electronic properties. Yet the fabrication of these structures remains manual, relying on visual inspection to identify, align, and stack individual flakes. This skill-dependent workflow limits reproducibility and throughput, constraining systematic exploration of layer combinations and twist angles.

We present AutoLab, a compact robotic platform that automates both flake detection and van der Waals stacking through vision-guided and temperature-coordinated control. A top-view microscope scans exfoliated wafers, while a computer vision algorithm identifies and catalogs candidate flakes. During stacking, a side-view camera assists coarse approach of the polymer stamp onto the wafer, and coordinated control of seven motorized axes with a PID-regulated heater modulates the contact area between polymer and wafer, mimicking the dexterity of an experienced operator. Built entirely from standard optomechanical components, he core hardware components of AutoLab fits on an 18-inch optical breadboard, offering a practical route toward reproducible, wafer-scale 2D material based device fabrication.