Can a MOS capacitance be negative?

[tangqin55]

MOS capacitances

After simulating one MOS transistor, the Results will provide ids and 16 intrinsic caps like cgs and cgd.

I feel streng 'cause they are all negative.

please help me to figure it out!

 

[tangqin55]

MOS capacitances

?
Nobody can answer???

 

[sowmya005]

Re: MOS capacitances

Hi,
I am not sure if a MOS can have negative mos caps, but surely it can have negative Ids.
what is the type of MOS? what are u trying to simulate?
please provide more details.

thanks,
Sowmya

 

[tangqin55]

MOS capacitances

I just simulate a 45nm MOS device for Ids and caps.
I planed to use table-based data to model the intrinsic caps of a transistor, but after Spectre simulation, I found all the caps have negative values.

I read some papers about table-based MOS models, almost all of them use charge tables for modeling.
I just feel confused why not directly using cap table model?

Thanks so much for your reply!!!

 

[love_analog]

The caps got from spectre are based on mathematical abstractions. It is the differential of charge wrt Voltage.

For instance if Cdg is -ve, that doesn't mean the cap from drain to gate is negative.

Search on this board, I remember someone had an easier to understand explanation

 

[tangqin55]

then, if I want to get Cgs, Cgd look-up tables wrt Vgs and vds, can I use the value provided after Spectre simulation?

 

[narenderbalguri]

i think it will not possible to have -ve capacitance...
Cgs =(2/3 )W LCox +W Cgso=2/3 W LCox +W LD Cox
Cgd =W Cgdo=W LD Cox

 

[tangqin55]

For 45nm technology and below, the caps are derived from charge. Since the charge expressions are really complicated so I want to get the cap value directly.

I read some papers about table-based transistor model, but all of them use charge tables. I can't understand why not using cap tables 'cause Q=CV.

is there any way to simulation terminal charge by using spectre or spice?