Optimal Testing in Semiconductor Manufacturing

We present an algorithm and discussion for minimizing processing and testing costs of a semiconductor manufacturing line. The core concept is that the testing at each process step has a nonzero cost. We derive the mathematical criteria to choose optimally, based on all future processing and testing costs and test failure probabilities, whether or not to perform testing at each process step. The formulation and implementation of the algorithm is necessarily an “engineering-side” activity. Yet, as the goal is the optimization of the business pertaining to processing step costs, the Chief Financial Officer has a critical role to play. Both Finance and Engineering must collaborate in the optimal testing platform we propose. Finance, rather than Engineering, bears the fiduciary and management responsibility to estimate and report processing and testing costs in the financial statements.