Integration of reactive process modeling into semiconductor technology development

Abstract

Understanding the physico-chemical phenomena at play in semiconductor processing is critical in making important product development decisions. Product performance and yield for even the best-designed chips can be traced back through transistor and interconnect performance to material and structure quality resulting from complex chemical processes. Despite the importance of factoring process character into design and production decisions, no complete framework exists either to connect design ideals to process realities or to quantify the impact of process uncertainty on product performance. Hence, process, device, circuit and design decisions are all made with incomplete or imprecise information leading to sub-optimal products and inefficient use of limited resources.A software-based approach to filling the significant gap at the chemical process end of the spectrum will be described. First, a hybrid approach must be created that integrates modeling with empirical methods to overcome the insufficiency of either by itself. Second, this capability must be directly accessible by technology managers, developers and practitioners to limit the decay of its decision-making value with increasing distance from its source. Third, the impact of the uncertainty associated with assumptions and model inputs must be quantified to optimize resource allocation and minimize risk. Finally, the chemical process models must be integrated with materials models and linked to solid-state process and device models while quantifying total uncertainty.