estimate Cox of a pmos

Hello all, I have a problem estimating the Cox = eox/tox of a pmos
(52/2) . I work on a umc 0.18 um. It may sound primar, but how do I
know these eox and tox parameters? it is important for me please!

thanx

 

Hi,

tox is the gate oxcide thickness, this is a process parameter that is
provided by your foundry. You can find this parameter in the Docs
provided by UMC, either the DRMs or the Electrical manual. Otherwise,
you can look at your Simulation Model Cards. This depends on your
simulator and model level. Assuming Spectre and BSIM3v3, you may look
at 'tox' parameter. Hope what I have said so far makes sens for you.
Another way of getting this tox information, is just to create a
simple simulation schematic for your Pmos, run a DC and then print the
Process Parameters for your MOS. Again, assuming Spectre and BSIM3v3,
go the the ADE -> Results -> Print -> Model Parameters and then select
your pmos. a list of model parameters should be printed, look for
'tox'.

As per eox, this is something you can calculate yourself.
eox=EpsilonOx*Epsilon0
EpsilonOx = Relative Permitivity of your gate oxide which is very
likely to be a Silicon Dioxide (SiO2). The value of this dielectric
constant at 300K is approximately 3.9 (no Units).
Epsilon0 = Vacuum Permitivity. This is a physical constant whose value
at 300K is 8.854e-12 F/M.
A quick calculation gives eox=3.45306e-11 F/M.
In BSIM3v3, this value is defined as EPSOX=3.453133e-11

Hope this answers your question.

Cheers,
Riad.

 

Hi Andrew,

I was looking at the 'Operating Point Parameters' section from spectre
-h bsim3v3. I was hoping to find something related to Cox. There is a
parameter called 'betaeff'.
Am I right in saying betaeff=Ueff*Cox ? A quick hand calculation on a
unity saturated transistor shows a kind of matching ...
I Couldn't find anything related to this betaeff in the spectre docs
I'm afraid. The literature is using betaeff as mu*cox off the top of
my head, not truly sure though.
Thanks for your enlightenment !

Cheers,
Riad.

 

Riad KACED wrote, on 08/24/09 13:26:

Riad,

I think it's

Ueff * Coxe * Weff / Leff

But I've not really checked very thoroughly... this was from a CCR from 4 years
ago which was to with fixing the betaeff output for bsim4 with m-factor (or nf
parameter, I can't remember).

Andrew.

 

Hi Andrew,

I would have given it a go actually, just try couple of Weff / Leff
and look at the betaff trend.
I'll be doing this whenever I got some spare time. I can try both both
BSIM3/BSIM4 and PSP.

Thanks for your updates, very much appreciated !

Regards,
Riad.

 

Riad KACED wrote, on 08/26/09 12:58:

Riad,

Likewise - I'd have given it a try. Just didn't have any spare time this week...

Andrew.