5.3 Inverse Modeling of Analytical Doping Profiles

In this application inverse modeling is used to simulate the actual doping profile in a deep sub-micron device. This is an important task since for small gate length devices the actual doping profile differs from the simulated data because of various reasons. This can be caused by, for example, missing or uncalibrated models of physical effects in the process simulators. Also vertical SIMS measurements of the manufactured device do not have the necessary accuracy for performance analysis.

In our inverse modeling strategy we make the following assumption. Electrical data measured from manufactured devices -- particular output and transfer I/V characteristics -- are a very strong description of a transistor. This electrical behavior is caused by a specific doping profile.

The task is divided in two subproblems:

1. Fit analytical doping profiles to the output of the process simulation.

2. Fit the evaluated results to the electrical parameters by fine-tuning the doping profile.

These two tasks were performed using the optimization capabilities of the SIESTA simulation environment.