The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Semiconductor Devices >> LEVEL=49 model
https://designers-guide.org/forum/YaBB.pl?num=1157437946

Message started by ywguo on Sep 4^th, 2006, 11:32pm

Title: LEVEL=49 model
Post by ywguo on Sep 4^th, 2006, 11:32pm

Hi Guys,

I find some parameters named STIMOD, sa0, sb0, wlod .... they are named STRESS model. It is a BSIM3V3.2 model.

Obviously, my simulator ignores those parameters. I wonder if this will affect the accuracy of the simulation.

Any comments are appreciated.

Yawei

Title: Re: LEVEL=49 model
Post by Geoffrey_Coram on Sep 5^th, 2006, 6:25am

In 90nm, I think someone found discrepancies of up to 10% in id_sat. The LOD or STI stress model captures the hydrodynamic stress that the oxide exerts on the channel because of the difference in lattice constants. The threshold voltage and mobility are subject to change. You should compare the sa0, sb0 values (which are for the device that was measured to extract the model parameters) with the sa,sb values in your layout. SA,SB = distance from gate edge to STI edge on the two sides of the device (there is no association of SA or SB with the drain or source in the Berkeley stress model).

Title: Re: LEVEL=49 model
Post by loose-electron on Sep 5^th, 2006, 9:30pm

One addendum as well -

A handful of the BSIM model paramaeters are frequently ignored in model fitting. There is a mix of trying to make the model fit device physics vs. having a set of realistic parameters that can be measured and fitted in the lab.

Some parameters can be dropped (not all) and get a reasonable set of bias curve fitting and transient responses when you have extensive limited geometry binning.

Not an ideal thing, but that is the reality of the matter. If you argue the validity of this, take a close look at some of the numbers in some models. Many parameters have been left unspecified and compensated for elsewhere in the model, elsewhere, some of the basic parameters that should not change (mobility is a favorite that gets pointed out) changes from model to model.

Welcome to reality...