BSIM3v3 model... Lint with negative values?

[spectrallypure]

Hi all! I am trying to do some hand calculations for design, and need
to estimate the effective channel width and length of MOS transistors.
I was suggested to use:

Leff = Ldrawn - 2Lint; Weff = Wdrawn - 2Wint

where Lint and Wint are taken from the design kit's BSIM3v3 model
files. However, for the technology in which I am working (0.35um
CMOS), these models for NMOS and PMOS have both NEGATIVE values for
Lint!!!:

lint=-5.005e-08 (NMOS), lint=-7.130e-08 (PMOS)

Being Lint somewhat related to the length source and drain overlaps
with the gate, how come can it be negative? Is this just the result of
model optimization? Is this meaningful?

Thanks in advance for any suggestions.

Jorge.

 

[Ozgur.Ates]

On 17 Haziran, 11:10, spectrallypure <jorgela...@gmail.com> wrote:

Jorge, BSIM3v3 model extraction programs try to match measured device
characteristics to model behaviour within lots of MOS devices with
different lenghts and widths. Physically LINT is of course positive,
however for your case BSIM3v3 equations better fit to measured
characteristics with negative LINT value.

Ozgur

Hi all! I am trying to do some hand calculations for design, and need
to estimate the effective channel width and length of MOS transistors.
I was suggested to use:

Leff = Ldrawn - 2Lint; Weff = Wdrawn - 2Wint

where Lint and Wint are taken from the design kit's BSIM3v3 model
files. However, for the technology in which I am working (0.35um
CMOS), these models for NMOS and PMOS have both NEGATIVE values for
Lint!!!:

lint=-5.005e-08 (NMOS), lint=-7.130e-08 (PMOS)

Being Lint somewhat related to the length source and drain overlaps
with the gate, how come can it be negative? Is this just the result of
model optimization? Is this meaningful?

Thanks in advance for any suggestions.

Jorge.