Expression set for epitaxial layer design in power MOSFET

Abstract

In this work, the reason for the difference between the simulated and experimental breakdown voltage is confirmed in power MOSFETs. Theoretical study predicts two factors. One is that the recombination phenomenon is ignored in the simulation. Another problem is that the accuracies of commonly used impact ionization models are not qualified. The widely used impact ionization models, which are the key factor to influence the breakdown voltage, are analyzed theoretically and numerically. Meanwhile, the validities of these models are verified experimentally. It is proved that the Lackner model is more precise. With the help of qualifying and with the recombination phenomenon considered, the conventional ionization integral is modified. The relations between the breakdown voltage, the specific on-resistance, the thickness and the doping concentration of the epitaxial layer are then simulated and fitted to the non-punch-through (NPT) structure. While the breakdown voltage of the NPT structure using the Lackner model varies from 182 V to 1471 V, the value using the Chynoweth model, which is in common use in simulation, is underestimated by 6.0%(171 V) to 9.5%(1332 V). Based on these improvements, an expression set is presented for the epitaxial layer parameters selection. The re-established formula set has more accuracy and obvious difference from the results obtained with the conventional Chynoweth models. For optimum designed punch-though (PT) structures, the analytical results indicate that the specific on-resistance can be reduced by about 12.0% compared with the NPT structure. The PT structure also allows the epitaxial layer to be decreased by about 25.3% in thickness and 8.3% in doping concentration.