Geometric programming approach to doping profile design optimization of metal-oxide-semiconductor devices

Abstract

We study one-dimensional doping profile design optimization problem of metal-oxide-semiconductor (MOS) devices using a geometric programming (GP) technique. To model the explored optimal doping profile into a GP problem, the subthreshold swing is formulated as an objective function and the on- and off-state currents are considered as constraints for solving the corresponding optimal doping profile. The GP problem is a special type of convex optimization and is solved globally and efficiently using the existing numerical solvers in GGPLAB. The accuracy of optimized results is validated by comparing with numerical semiconductor device simulation. This approach provides a way to optimize doping problem which may benefit manufacturing of MOS devices. (C) 2012 Elsevier Ltd. All rights reserved.