Modeling of via formation by photosensitive dielectric materials for MCM applications

The Packaging Research Center at the Georgia Institute of Technology focuses on thin-film multi-chip modules (MCMs) as the strategic solution in electronic packaging because this technology offers advantages in miniaturization, electrical and thermal performance, cost and reliability. Via formation using photosensitive polymer technology reduces process costs and hence is of great interest. 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 benzocyclobutene (BCB) is presented. A series of designed experiment is used to characterize the via formation workcell, and neural network process models are constructed for the various unit process (including spin coating, pre-bake, expose, develop, cure and de-scum). Film thickness, uniformity, via wall angle, via yield, and film retention are modeled. The neural network models will eventually be used for the development of a supervisory process control scheme.