Sensitivity-based optimization and statistical analysis of microwave semiconductor devices through multidimensional physical simulation (invited article)

A new, efficient approach to the sensitivity analysis of majority carrier and bipolar microwave semiconductor devices based on multidimensional physical models is discussed. The approach is exploited in the parametric optimization and statistical analysis of microwave semiconductor devices. The present technique is based on Branin's method for the sensitivity analysis of electrical networks. With respect to the adjoint method, formerly proposed for the sensitivity analysis of majority-carrier devices, the new approach enables the straightforward treatment of bipolar physical models. Some examples of application concerning GaAs microwave FET's are presented: the gradient-based parametric optimization of the doping profile so as to obtain maximum linearity and the statistical characterization of the electrical device response on the basis of the statistical spread of technological parameters, which is an intermediate step in yield-driven MMIC optimization based on physical models. © 1997 John Wiley & Sons, Inc. Int J Microwave Millimeter-Wave CAE 7: 129–143, 1997.