Hi, I want to etch Si3N4 away from Si substrate. Does anybody know proper
solution which only attacks Si3N4 but Si ( or very weakly). Thanks in advance.
Cheers.

Try hot phosphoric acid (carefully) ! It will etch Si3N4, but not Si or
SiO2.

A few days ago I mentioned hot phosphoric acid as a Si3N4 etch. I have
had a number of enquiries about this, so I'll reply here.

We use a Phosphoric acid water mix to give a boiling point of 180 deg.
C. We find this operating point by starting with an excess of water, and
slowly increasing the temperature, boiling off the water until we reach
180 deg.

This etches nitride at 80 ang./min, SiO2 (thermal) at about 4 ang./min
and we can't measure the Si etch rate.

The equipment is set into an extracted bench, and the etch tank is a
chemical resistent polymer.

This mixture is prone to superheating as the water boils off, so beware.

If you are not confident about seting this sort of thing up, _please_
seek further advice.

This information is given in good faith, but I don't guarantee results !

The etch rate of oxide seems pretty steep. However the nitride etch rate is
pretty impressive. Also I am not sure why water is beingadded to phosphoric
acid, unless you want to heat it very fast (unsafe).

I use phosphoric acid at 140 C to etch off the nitride before LOCOS. The oxide
is grown on top of nitride and is used as mask. The oxide thickness is about
200-250 A. However it does not etch off even after 4 hours (nitride thickness
2500A). Condesor Lid and a temperature controlled bath is recommended for
saftey reasons. Also wafers should be very dry when they go in to the acid.

Depending on the conditions the selectivity Nitride to Oxide can range
anywhere from 15 to 250 (or even better?). If I remember correctly,
"old" etch solution has better selectivity _and_ higher etch rates for
Nitride. Don't ask me why. Only the water that boils off should be
replaced (condensing lids or water dripper), failure to do so will
overheat the bath. Of course, you don't pour water right into the
acid (we all had that lesson in chemistry class, didn't we?).

Phosphoric Acid

You can also use HF or BOE. It etches nitride, albeit slowly, but not Si. However, HF will roughen the Si surface.
Good luck

How about the reverse trick?

Can I etch silicon in chlorine plasma using nitride hardmask?

You can try adding N2 or He-O2 to Cl2/HBr chemistry.

We had some trouble about using HF to clean silicon wafer backside, but sometimes, the HF maybe contaminate wafer surface, and it will attack passivation(Si3N4), does anybody provide reaction formula to me, thanks very much