Hi all, a question about process standard in Sansen's book

[nnmate]

Why do people for CMOS process use n-welll process not p-well?
thanks

Walter

 

[erikl]

Re: Hi all, a question about process standard in Sansen's bo

Why do people for CMOS process use n-well process not p-well?

Because most CMOS processes use p- substrates (or p- epi on p+ substrate wafers). Only few CMOS technologies use n- substrates (and then use p-wells, of course). There are also CMOS technologies which use both n- & p-wells on (any type of) substrates (twin well or triple well processes).

 

[nnmate]

Re: Hi all, a question about process standard in Sansen's bo

hi erikl, thank you very much for ur replying
i know exactly the situation by foundry that u described. i think, it is better to modify my question like this, why do people use p-substrate usually, which advantages do we obtain, when we deposite Bor and Galium (+3 material) at first.

 

[erikl]

Re: Hi all, a question about process standard in Sansen's bo

... why do people use p-substrate usually, which advantages do we obtain, when we deposite Bor and Galium (+3 material) at first.

Highest purity silicon is always p-doped, because Boron has the largest segregation coefficient of all dopants in silicon, i.e. is the most difficult dopant to be removed from silicon during the zone refining process. If n-substrates are required, this has to be done by counter-doping with P or As, which means over-compensation of the p-dopant. Over-compensated semiconductor material usually shows shorter carrier life-time, which may - or may not - be desirable, depending on the specific application.
Over-compensation is an additional process step, which probably leads to higher silicon wafer cost (just a guess). May be this is the reason why most CMOS processes use p- substrates. On the other side, shorter carrier life-time in an (n-) substrate helps against latch-up problems in CMOS; this, however, can also be achieved by using p+ substrate wafers with a p- epi layer (which also raises wafer cost, of course).
(p+ substrates also show a lower carrier life-time than p- substrates.)

 

[nnmate]

Re: Hi all, a question about process standard in Sansen's bo

thanks and merry xmas!
could not hope for a better reply.