Understanding of plasma-material interactions – from now to future

Plasma processing technologies, which are widely used for semiconductor device manufacturing, are now facing a paradigm shift as Moore’s law is near its end and new three-dimensional structures and nonconventional materials are employed in advanced semiconductor devices. Highly innovative plasma processing techniques with atomic-scale accuracy must be developed expeditiously and cost-effectively. For this purpose, a large part of the conventional try-and-error approach must be replaced with more logical approaches. One of such approaches is to gain a better understanding of the fundamental mechanisms of processing, especially plasma-material interactions involved in the processes. The question is whether this academic approach is also sufficient. What if conventional wisdom may not answer our totally new questions? In this presentation, the author will first present recent analyses of plasma-material interaction, using atomic-layer etching (ALE), i.e., cyclic removal of atomic layers from a material surface, as an example of a modern etching process, based on atomic-scale simulations combined with highly controlled beam experiments. This is a conventional academic approach. As to a possible new approach, the author will briefly discuss a data-driven approach based on machine learning (ML), possibly coupled with prior knowledge of existing processes, for the development of novel plasma processing.