Image Generation of Ultra-Thin Polymer Films Using Diffusion Models from Tensile Testing for Mechanical Failure Prediction

Abstract

The mechanical reliability of ultra-thin polymer films, crucial in microelectronic applications, poses significant challenges. Polymethyl methacrylate (PMMA) is widely utilized in this context, yet conventional methods assessing mechanical reliability are limited in scope, often failing to predict failure accurately. Furthermore, these methods neglect the complex nanoscale physics of polymer films, which may lead to the formation of wrinkles during plastic deformation. In this paper, we present a novel approach that combines diffusion models with measured tensile properties to generate sequential PMMA images, enabling the prediction of mechanical failure in ultra-thin films. The proposed diffusion model not only accurately generates PMMA images from tensile testing but also simulates wrinkle formation, providing a more comprehensive and accurate assessment of PMMA thin film integrity. This research offers a promising avenue for enhancing the mechanical reliability of microelectronic devices.