Parametric yield optimization with mesh adaptive direct search

Random manufacturing process variations can affect the performance of an integrated circuit to the extent that a significant number of the manufactured circuits must be discarded because they fail to satisfy the specifications. To increase the yield, random variations must be taken into account in the design phase. This can be achieved by choosing appropriate values for the parameters accessible to the circuit designer. This process (circuit sizing) can be automated by means of parametric optimization. As simulators do not compute sensitivities, derivative-free optimization algorithms, like mesh adaptive direct search (MADS), are well suited for optimizing circuits. We propose an MADS-based approach for finding a circuit that satisfies the minimum yield requirement. The approach is tested on two integrated circuit-sizing problems. The results demonstrate its effectiveness and speed.