Dave,

I don't think any liquid solution will give you an optically smooth
finish after an etch back. Etchants are used to highlight silicon
defects in silicon wafers in a method of destructive testing, such as
Wright etch and Sirtle etch. (Is that name right?) The basic nature of
wet etch is to be selective and it varys due to minor differences and
defects and crystal faces.

I know in epitaxy, the first step is an HCl etch back of the silicon. It
was mostly a clean, and I don't know if it was that uniform. Mostly it
was to remove the surface layer and its contaminates. However, being
both a gas and at high temperature, 1050 it was not terribly selective
and didn't roughen the surface and wafers looked good coming out.  I ran
an epitaxy reactor for a few years.

I doubt 20 angstroms of silicon dioxide could be grown by even the
hottest sulfuric-peroxide solution. I think it would also behave like
any other liquid solution, pit your wafers if through 50 or 100 cycles
you actually get loss. I think the HF solution to strip the oxide would
pit the wafer also.

I suggest the following two things.

1. Oxidize the wafer in a furnace and strip the oxide. You would have
precise control over the thickness lost and it would be uniform. If you
desire to etch back is to get rid of contamination in the silicon, the
oxide will absorb many metals and some dopants preferentially.

2. Etch the surface at an elevated temperature . I think that a hot
etchant gas would do best, since it would have very poor selectivity. I
would say 600 deg. C or more and HCl in a tube. However, it is just a
guess. The HCl would also getter metals.

How do you plan to measure the depth of etch back in this process and
get the uniformity?

CMP might be another way to go with this wafer.

I am curious as to what would be the need to remove the surface of the
wafer in a process.

Ed

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Dave Goldstein
Sent: Tuesday, December 18, 2007 8:10 PM
To: [email protected]
Subject: [mems-talk] How to uniformly etch ~0.05 - 0.3um off a 3" Si
wafer?

I need to uniformly remove a thin layer of Si off a 3" Si wafer. No
structures, just the same material. No RIE. How to do it?

I've tried the HNA solution but it doesn't look very uniform and some
white-ish looking cloud was left on the wafer.

I have also tried using NaOH but it looked very bad after dipping the
wafer in there for couple of minutes. And doesn't look very uniform.

Right now I am trying using Piranha (H2SO4:H2O2=3:1) to chemically grow
a thin oxide and then etch off in Si. I am thinking that each cycle
should consume 2nm of Si then if I am doing this 30 times then I am
removing 60nm of material. But I have no way to verify this.

What do you suggest to use to remove a thin layer off a Si wafer in a
uniform fashion?

Dave