AMAN-SPM: Autonomous Molecular and Atomic Nanofabrication

In the AMAN-SPM project, I will autonomously synthesize molecules on metal surfaces and arrange them into functional nanostructures inaccessible to conventional chemistry by developing an innovative machine learning-based approach. The new molecules will be created by synthesizing reactive precursor molecules, precisely positioning and orienting them in close proximity to each other, and inducing bond formation via voltage pulses applied with the tip of a scanning probe microscope (SPM). The main challenge of this task lies in the complexity of the interactions between the molecules, the surface and the probe tip. In order to accurately control on-surface synthesis, it is crucial to understand the outcome of each manipulation. However, predicting these outcomes is challenging because they are highly dependent on the manipulation parameters, and the underlying physics is often unknown. As a result, manually learning how to manipulate molecules is time-consuming and often beyond the reach of human experts. The AMAN-SPM project will allow to perform on-surface synthesis by using an SPM that operates autonomously, driven by decisions made by a machine learning algorithm. The ideal approach for learning on-surface synthesis is reinforcement learning (RL), as it requires no prior knowledge of physical models and learns the optimal manipulation parameters through experimentation. This method is easily adaptable to different systems and scalable from the synthesis of single molecules to the creation of complex nanostructures. Furthermore, the combination of experimental analysis and ab-initio calculations provides valuable insights into the interaction processes.