Fourier method modeling of semiconductor devices

A novel high-order approach to numerical modeling of semiconductor devices is presented. The new method is a combination of the classical Fourier-series Galerkin procedure, a special matrix calculus, and fast numerical pseudospectral techniques. The proposed algorithm renders the exact solution (machine precision) of the semiconductor equations in closed form of a trigonometric polynomial. The condition number of the diagonally dominant discrete equations is near unity. As a consequence, a highly accurate solution is achieved at moderate computer costs. The method has been implemented for oneand two-dimensional device models. Properties of the new procedure are demonstrated on examples.