(Invited) Predicting Voxel Scale Geometry and Properties in Vat Photopolymerization Via Microscopic and Machine Learning Methods

Abstract

Additive manufacturing (AM) promises disruption of traditional manufacturing supply chains, democratizing and distributing manufacturing, enhancing flexibility and enabling previously impossible geometries and customization. Applications range from aerospace to regenerative medicine, leveraging the numerous benefits of the technology. Within polymer AM, vat photopolymerization, enabled by advances in high-resolution display technologies (e.g. liquid crystal display - LCD, digital light processing - DLP) to locally cure liquid resin into solid polymer, prints millions of voxels per layer simultaneously in a few seconds or less. Typically, these layers are combined step-wise until a 3-dimensional part is formed. Fundamental understanding and control of the printing process with geometrically and mechanically precise voxels requires characterization of parts at voxel and sub-voxel length scales. Despite the precise, high-resolution light engines, resultant parts exhibit defects such as over/under-polymerization, 3-dimensional anisotropy, weakened layer interfaces and more. In this presentation, we examine the formation of these defects starting from individual voxels and simple patterns, but extending to complex voxel-voxel interactions. We discuss novel, multiscale measurement tools such as atomic force microscopy and laser scanning confocal microscopy that elucidate the small-scale printing process in both space and time. The novel characterizations are used to develop new models of the printing process and enable precise geometric and mechanical control of the individual voxels. Finally, we introduce the use of neural network models and big-data to predict the output of arbitrary photomasks.