Why the capacitance of 180nm is the same as 32nm?

[john7796]

Hi all,

I find out the data following:

180nm process : the capacitance of M1 : 0.2fF/um.
32nm process : the capacitance of M1 : 0.2fF/um.

But the metal 1 line area of 32nm process should smaller much than 180nm process.

Why their capacitance is almost the same?

Please help me! Thanks.

John

 

[Dominik Przyborowski]

Probably both processes used the same BOX with the same electrical permittivity and the same distance between metal paths and substrate.
Notice that You mentioned about M1 capacitance per unit area.

 

[john7796]

Dominik Przyborowski,

My mentioned is the length of Metal line.

Could low-k process/high-k process at the same chip ?

 

[Dominik Przyborowski]

My mentioned is the length of Metal line.

With the same width?

Could low-k process/high-k process at the same chip ?

Everything is possible but nobody mixing two quite different processes - it's inpractical.

High-K processes was invited to overcame problems with very thin gate oxide, like tunnelling current, breakdown voltage, etc. But everything above transistors (burried oxide, metal layers distance) are quite similar.

 

[john7796]

Dominik Przyborowski,

Yes, in the same width, and should be metal 2.

Could High-K for transister, Low-K for metal-line?

 

[Dominik Przyborowski]

In each technology process You have two types of dielectric. Gate oxide (in old technologies it was only SiO_2 or SiON, while in high-K it's stack of two dielectrics like SiON and HfO2) and burried oxide for metal layers isolation. Of course we need high electrical permittivity materials for gate oxide for high current gain factor and low permittivity materials to avoid capacitive coupling between metal paths.

If You want to study about processes, I think the best knowledge mines are Solid state electronics, transactions on electron devices and electron devices letters journals.

 

[john7796]

Dominik Przyborowski,

Could you give me capacitance of every metal layers at 28nm ? Any FEB's data is ok.


John

 

[littlej_zju]

From 180nm to 32nm, although space is smaller, k is lower and metal thickness is thinner too. So, the capacitance would be not larger.
Ex., M1 thickness 5300(180nm) vs 900(28nm); k value 3.7(180nm) vs 2.63(28nm)

 

[john7796]

Littlej_zju,

Could you tell me the unit of thickness ? 5300 nm VS 900nm? but it seems too small?

 

[littlej_zju]

Not nm. It is A (10E-10).
Fab like A for thickness.

 

[john7796]

Littlej_zju,

Do you know the SRAM capacitance of 28nm process?

 

[littlej_zju]

I don't have such data on my hand. Maybe, you can get it from published papers.