5.4.1 Process Modeling

For the modeling of the reference device the tool makedevice [3] is used. This tool provides several doping elements placed anywhere in the device. For the modeling of the profiles only the following elements are used.

Constant Distribution used for modeling background doping

Gaussian Distribution with parameters n, x₀, y₀, $\sigma_x$, $\sigma_y$, l_x, l_y. For a purely vertical Gaussian distribution l_x is equal to the length of the device and l_y is zero.

Pearson/Error-function Distribution with parameters n, x₀, y₀, $\sigma_x$, l_y, $\sigma_y$, $\gamma$, and $\beta$. Vertical dependence is modeled using a Pearson Type IV^5.2 distribution and the lateral dependence is described by an Error-function^5.3.

For the doping characterization of the device only the types of doping are important. That is the reason why not every doping material is represented by a separate profile and only acceptor and donor dopings are generated.

This analytical device generator is sufficient for the inverse modeling. Another benefit is that makedevice is extremely fast compared to a complex process simulation. For the purpose of inverse modeling where the I/V-curves of a simulated device were compared to the measured datasets, this procedure is sufficient if the artificial device adequately represents the simulated device.

Footnotes

... IV^5.2

The Pearson Type IV distribution is explained in Appendix B.6

... Error-function^5.3

The Error-function is explained in Appendix B.7