How hot phosphoric acid wet etch of Silicon Nitride can go wrong.

A hot phosphoric wet etch of Silicon nitride can go wrong in different
ways. Most of the following examples I have known to have occurred. The
following is a list of causes.

0. Your phosphoric acid is dried out. You need water in the phosphoric
acid to etch nitride. Semiconductor process phosphoric acid systems have
a control system where water is dripped into it and the solution is
heated and water is boiling off. It is a temperature/composition control
system. The phosphoric temperature is controlled by the water
composition and the water composition is controlled by feed back from
the temperature.

If you have a simple tank and it is heated, the phosphoric acid can dry
out. The nitride etch rate of phosphoric acid goes down if there is too
much water and goes down if there is too little. With out water, the
phosphoric acid molecule can't hydrolyze. You have H2PO4, and not H3O+
and PO4= ions. Without water there is no pH.

So if you have heated your acid and let it sit around, it might have
dried out.

1. Your CVD silicon nitride isn't. That is, your silicon nitride is
Silicon OxyNitride and is more oxide rather than nitride or at least
some oxide. This happens more often than people might think. A small air
leak and your nitride is more oxide. O2 takes precedence is reacting
with the silicon over ammonia.

The way to check for this is either the index of refraction or wet etch
with BOE 1235. BOE 1235 is a formulation that should etch thermal oxide
at 1235 ang./min. If your etch rate is 400 or 900 etc., you have a
silicon oxynitride film. If I remember correctly, it should be 50 or
less. There will be some etch rate, since your nitride is not likely

2. A common process to etch silicon nitride is to CVD deposit nitride,
thermal oxidize it, mask the wafer, etch the thermal oxide, then use the
patterned oxide for the wet etch of the nitride.

The juncture between the thermal oxide and the nitride isn't sharp. You
have an intermediate range. So you need to over etch the oxide until you
reach the mostly nitride level. Then use your patterned oxide as a mask
for nitride etch.

Regardless of the theory about the thermal oxide/nitride interface, you
need to over etch the oxide to get good nitride etch.

3. Your nitride may have a thermal oxide or oxynitride skin. The
solution is to over etch the oxide again. Depending on your process
sequence the surface of your nitride might be oxidized or the last
portion of the film might be deposited oxide. This might be some post
processing of the nitride, or recipe process problem or error. A check
for the index of refracton might not detect the skin, and wet etch with
BOE would tend to say your film is good.

4. Stoichiometric ratios: I am just guessing here, but if it is silicon
rich, the silicon is going to oxidize with the exposure to water. So the
surface of your silicon nitride at the interface for etching would be a
silicon oxynitride. Phosphoric acid will have trouble etching
oxynitride.

5. Plasma deposited silicon nitride. The stoichiometric composition of
this material can vary widely from having oxide, having hydrogen,
silicon rich. I would really check out what it is made of.

Best of luck.

Ed

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Nor Hafizah Ngajikin
Sent: Friday, October 30, 2009 1:36 AM
To: [email protected]
Subject: [mems-talk] wet etch of 2um Si3N4


Dear all,

I've tried to etch 2 micron thickness of Si3N4 using hot H3PO4. The
concentration of H3PO4 is 85w%. I heat the H3PO4 at 150 degree celcius
for 1 hour. Unfortunately, this experiment didn't work as expected. None
of the nitride is etch away.

Can anybody share with me your experience on nitride etch using H3PO4.

Thank you.

Hafizah