An illustration depicting the novel automated PLD platform developed at ORNL that enables AI-driven optimization of growth conditions for thin films, establishing a novel workflow applicable to any materials that can be synthesized using PLD.
Motivation: Synthesis of thin films (for basic scientific study of 2D, quantum, and electronic materials) has traditionally relied upon slow, sequential processes with substantial human intervention.
Approach: For the first time, a self-driving experiment grew ultrathin films of WSe2. LLMs augmented with literature knowledge are used to generate hypotheses, self-driving synthesis systems autonomously run the experiments, and other AI models correlate complex data to thin film characteristics.
Impacts:
- Results demonstrated 10-100× faster exploration and synthesis.
- The novel approach enables universal accelerated discovery and autonomous growth materials for electronics, energy, etc.
S&T Challenge: Aligns with “Recentering Microelectronics in America,” “Designing Materials with Predictable Functionality,” and “Achieving AI-Driven Autonomous Laboratories”