Sequential modeling of via geometry in photosensitive MCM dielectric materials using neural networks

Abstract

Via formation using photosensitive polymer technology reduces process cost and is hence of great interest in electronics packaging substrate fabrication. However, to overcome process complexity issues and to facilitate low-cost manufacturing, process optimization and control are required. In this paper, a modeling approach for via formation in MCM dielectric layers composed of photosensitive benzocyclobutene (BCB) is presented. A series of designed experiments are used to characterize the via formation work-cell (which consists of the spin coat, soft bake, expose, develop, cure, and plasma de-scum unit process steps). Sequential neural network process models are then constructed to characterize entire via formation process. In the sequential scheme, each work-cell sub-process is modeled individually, and each sub-process model is linked to previous sub-process outputs and subsequent sub-process inputs. This modeling scheme is compared with two other modeling approaches to evaluate model prediction capability. The sequential method shows superior prediction capability. This modeling structure will be useful for feedback and feed-forward process control, and it will eventually be used for development of supervisory process control scheme.