Autonomous robotic mechanical exfoliation of two-dimensional semiconductors combined with Bayesian optimization

TL;DR

This paper presents an autonomous robotic system combined with Bayesian optimization to efficiently produce high-quality monolayer 2D semiconductors, significantly reducing experimental trials and operator effort.

Contribution

It introduces a novel robotic-experimental framework that automates and optimizes the mechanical exfoliation process for 2D materials using Bayesian methods.

Findings

Optimized exfoliation parameters identified in just 30 trials

Achieved large-area monolayer WSe₂ with high efficiency

Reduced experimental parameter space exploration by 97.5%

Abstract

Simple mechanical exfoliation of layered materials is the most frequently employed method for producing high-quality monolayers of two-dimensional semiconducting materials. However, the mechanical exfoliation by human hands is a microscopically sophisticated process with a large number of microscopic parameters, which requires significant operator efforts and limits the reproducibility in achieving high-quality and large-area monolayer semiconducting materials. Herein, we have proposed a new strategy for the mechanical exfoliation by combining a developed robotic system and Bayesian optimization. We have demonstrated that it is possible to explore the optimized experimental conditions among a large number of parameter combinations for mechanical exfoliation in a relatively small number of experimental trials. Moreover, the entire mechanical exfoliation process from preparation to detection of monolayer semiconductors was performed by the developed autonomous robotic system. The optimized experimental condition was determined through only 30 trials of mechanical exfoliation experiments, representing 2.5% of all experimental parameter conditions. As a result, the critical parameters for the efficient fabrication of large-area monolayer WSe$_2$ were elucidated.