NMOS CAP 0.18um layout tutorial

[diemilio]

MOSCAP TSMC 0.18

Hi everyone,

Can someone give me a general number for the the capacitance to area ratio for a MOSCAP for the 0.18 process in TSMC??... I know the capacitance is also a function of the voltage across it, so I will also appreciate if you could give me a rough idea of this dependence as well (linearity wise)

Thanks a lot,

diemilio

 

[erikl]

Re: MOSCAP TSMC 0.18

Gate capacitance is 8.4 fF/µm² (TOX=41Å). Find here via MOSIS freely accessible Wafer Electrical Test Data and SPICE Model Parameters for more cap parameters!

 

[diemilio]

Re: MOSCAP TSMC 0.18

Hi erikl,

Thanks for the info... What I'm trying to get is an idea of how big can I go in capacitance, meaning that how big can I make a MOSCAP without running into whatever issues I might run into (which I really don't know what they are cause I've never used MOSCAPs before).

Thanks again,

diemilio

 

[AdvaRes]

MOSCAP TSMC 0.18

Hi diemilio,
Hi Erikl,

Actually you can have a nice reply by simulation. Simulate a moscap and measure Cgg for each voltage Vg applied to the transistor gate. Plot Cgg versus Vg. So, a Capacitor C where C is n Times Cgg will be n time bigger in area than the simulated MOSCAP.

 

[erikl]

Re: MOSCAP TSMC 0.18

... an idea of how big can I go in capacitance, meaning that how big can I make a MOSCAP without running into whatever issues I might run into.

Hi diemilio,

this depends very much on the accuracy needed. If you need matching caps (e.g. for ADC, DAC, SC circuits), use MIM or MOM caps (just search for these keywords here in the forum). Cap values in the order of 0.1 .. 10pF.

But as you're asking for big cap values, I suppose you don't need very high accuracy, then you better use MOSCAPs aka gate caps, accumulation caps or enhancement caps. They use the thin gate oxide, which achieves the largest possible cap/area ratio, usually poly over n-well, essentially a pMOSFET with drain & source short-circuit. The disadvantage is its rather big voltage dependency (about a factor of 2 between 0 and VDD), and its unipolarity (like an ELKO). For the full cap value, its DC voltage > Vth. Perfect for VDD filtering; I use MOSCAPs at any left-over space (and the layout looks better ;-) ). If you need bipolar function, you have to use 2 antiparallel caps in series (thus halving the cap/area factor), sometimes needed for a floating compensation or filter cap.

I use MOSCAPs in the order of 1 .. 10pF for compensation applications, and hundreds of pFs of VDD filter caps on a chip (due to filling all the otherwise empty spaces - in case of a redesign you might be glad to use such space for additional circuitry ;-) .

For more info about MOSCAPs you'd perhaps might just search for this keyword, especially read

Ciao, erikl

Added after 18 minutes:

... So, a Capacitor C where C is n Times Cgg will be n time bigger in area than the simulated MOSCAP.

Interesting suggestion. Needs gain n; may be usable as a dynamic load capacitance, but how'd you use this Miller principle for a bipolar floating filter or compensation cap application?