Barrier Lowering and Backscattering Extraction in Short-Channel MOSFETs

TL;DR

This paper introduces a fully experimental method to extract barrier lowering and backscattering ratios in short-channel MOSFETs using the Lundstrom model, validated through simulations and measurements, and extends the model to 2D geometries.

Contribution

It presents a consistent, fully experimental extraction method for barrier lowering and backscattering in short-channel MOSFETs, including an extension to 2D device geometries.

Findings

Validated the extraction method with device simulations and measurements.

Found that 2D backscattering differs significantly from 1D estimates near the interface.

Extended the Lundstrom model to account for 2D geometries in MOSFETs.

Abstract

In this work we propose a fully experimental method to extract the barrier lowering in short-channel saturated MOSFETs using the Lundstrom backscattering transport model in a one sub-band approximation and carrier degeneracy. The knowledge of the barrier lowering at the operative bias point in the inversion regime is of fundamental importance in device scaling. At the same time we obtain also an estimate of the backscattering ratio and of the saturation inversion charge. Respect to previously reported works on extraction of transport parameters based on the Lundstrom model, our extraction method is fully consistent with it, whereas other methods make a number of approximations in the calculation of the saturation inversion charge which are inconsistent with the model. The proposed experimental extraction method has been validated and applied to results from device simulation and measurements on short-channel poly-Si/SiON gate nMOSFETs with gate length down to 70 nm. Moreover we propose an extension of the backscattering model to the case of 2D geometries (e.g. bulk MOSFETs). We found that, in this case, the backscattering is governed by the carrier transport in a few nanometers close to the silicon/oxide interface and that the value of the backscattering ratio obtained with a 1D approach can be significantly different from the real 2D value.