Modeling via formation in photosensitive MCM dielectric materials using sequential neural networks

Abstract

Multichip module (MCM) technology is considered a strategic solution in electronics packaging because this approach offers significant advantages in terms of performance and reliability. However, manufacturing cost is a critical issue for mass production of high-performance MCM packages. Therefore, development of low-cost manufacturing technology is desirable. To realize this, process modeling, optimization, and control techniques 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 workcell (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 the entire process. In the sequential scheme, each workcell 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 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 will eventually be used for development of a supervisory process control scheme.