An optimization approach to a finite dimensional parameter estimation problem in semiconductor device design

In this paper the parameter selection in semiconductor device design is pos ed as an optimization problem: given an ideal voltage-current (V-I) charact eristic, find one or more physical and geometrical parameters such that the V-I characteristic of the device matches the ideal one optimally with resp ect to a prescribed performance criterion. The voltage-current characterist ic of a semiconductor device is governed by a set of nonlinear partial diff erential equations (PDE), and thus a black-box approach is taken for the nu merical solution of the PDEs. Various existing numerical methods are propos ed for the solution of the nonlinear optimization problem. The Jacobian of the cost function is ill-conditioned and a scaling technique is thus propos ed to stabilize the resulting linear system. Numerical experiments, perform ed to show the usefulness of this approach, demonstrate that the approach a lways gives optimal or near-optimal solutions to the test problems in both two and three dimensions. (C) 1999 Academic Press.