Balanced truncation on empirical gramians for model-order-reduction of Non-Quasi-Static effects in MOSFETs

In this paper, an empirical truncated balanced realization (TBR) approach is introduced to reduce the model order of non-quasi-static (NQS) effects in MOSFETs. In the PSP model (an industrial standard in compact modeling of MOSFETs), a simple spline-collocation (SC) approach is most commonly used to compute NQS. The SC technique, however, suffers from relatively high computing effort. To the best of our knowledge, this work is the first application of the TBR technique to reduce model order for the NQS effects. We compared the accuracy and efficiency of this new technique against the SC results. Our results show that TBR requires significantly less computing efforts due to smaller number of discretization points along the MOS channel.