I don't know about etching Nb, however what your asking is not
unreasonable for an RIE.  The only way you would be getting
undercutting is if the etch ions are being deflected and are not
running perpendicular.  This itself can happen for several reasons:

-Charging on the substrate in your etch pits (which means that your
power is not high enough or you are not removing your reactants)
-Too high a pressure (but 10mTorr is probably too low unless you are
using an ICP-RIE)
-Not high enough substrate power

If changing the gas does not work for you I would suggest using a
higher power (maybe up to 100W).  Also after a quick Google search I
found that it is fairly common to use a chlorine plasma to etch Nb,
but be sure that your system can handle the use of chlorine plasmas
since it can be very corrosive.

Nicolas Duarte
Ph.D. Candidate at Penn State University

On Nov 8, 2007, at 5:34 PM,   wrote:

> Hi,
>
> I'm trying to make 70nm wide (50nm deep) Nb trenches separated by
> about 100 nm (edge-to-edge).  With Al2O3 as an etch mask, I've been
> using SF6/O2 RIE to try to etch away the Nb.
> (20sccm SF6 and 0.6  sccm O2, 60 mtorr, 60W)
>
> I think my Nb is being completely undercut (all Nb under the 70nm
> wide Al2O3 etch mask is being etched away).
>
> How can I rid of the undercut?  Is it possible?  Is my pattern too
> small to use RIE? I know I should reduce the pressure to less than
> 10mtorr, but what else? Should I try CHF3/CF4?