Atomically precise materials engineering through deep learning

Objetivo

Modern technology demands ever-increasing precision in materials engineering, especially in the creation of novel advanced materials where atomic structure dictates function. Traditional on-surface chemical synthesis allows the controlled formation of molecular structures, but lacks the precision to build molecules atom-by-atom. Scanning Probe Microscopy (SPM) has been a game- changer in nanoscale materials engineering by allowing high-resolution imaging and the ability to study and manipulate individual molecules. However, the manual manipulation of molecules using SPM remains painstakingly slow in large-scale engineering - automation is the essential breakthrough required for progress. ARCADE will solve this by creating a multi-component deep learning infrastructure to achieve the required SPM control. This revolutionary approach will enable three breakthrough functionalities. First, Autonomous High-Resolution SPM Scanning and Recognition where SPM can automatically scan and identify the molecular structures present. Second, Modelling Workflow Integration where ARCADE will seamlessly connect the material designer's intent with measurements and simulations. Third, Deep Reinforcement Learning (DRL) for Molecular Manipulation where ARCADE will utilise DRL to develop an intelligent SPM tip that can selectively control and manipulate atomic bonds. Together, these new functionalities will enable the precise, atom-by-atom construction of complex molecular structures. This level of control will foster breakthroughs in, for example, the development of energy harvesting, sensing, catalysis, optoelectronics and more generally in next generation computing devices. As a model system ARCADE focuses on spintronics technologies, where the manipulation of electron spin plays a crucial role in device functionality. ARCADE will pave the way for a future where materials engineering operates at the atomic level, ushering in a new era of material innovation.