Bridging time and length scales in semiconductor process model development

The development of new semiconductor device materials is currently dominated by costly and time-consuming experimental investigations in which growth parameters are varied and correlated, ex situ, to device characteristics. Model-based process development has the potential to assist greatly in optimizing growth parameters and providing models that are suitable for use in in-situ process control. Researchers have used this approach sparingly, however, because robust, predictive models that describe the growth of semiconductor materials are neither well-developed nor validated. This situation is changing, however, due to the availability of high-resolution microscopy and improved modeling capabilities resulting from greatly increased computer power and more efficient numerical algorithms. The focus in this paper is on III-V semiconductor molecular beam epitaxy (MBE), which is the preferred technique for growing complex heterostructure device materials with many critical interfaces. The challenge is to understand the connection between macroscopic reactor conditions and microscopic film properties such as interface thickness and morphology.