The problem with photoresist as a mask in HF solutions for a long time is
that it tends to lift off because the interface is attacked.
Adhesion promoters will certainly help reduce this problem.
I think it's easier to use a metal stack like Cr/Au, or polysilicon.
See papers by Simpson, Woolley, and Mathies on fabricating devices in glass.

Whatever you choose, the glass will be etched both downward and laterally,
under the mask.
For large depths, you will be able to see the overhanging mask material.
If the mask is in tensile stress, it can tear.
If it's in compressive stress, it will wrinkle.
After a lot of undercut, the mask may fold over and partially block further
undercutting.
All of these can result in channels that are not of uniform width.

    --Kirt Williams

----- Original Message -----
From: "Gareth Jenkins" 
To: "General MEMS discussion" 
Sent: Wednesday, August 30, 2006 5:38 AM
Subject: Re: [mems-talk] Regarding Glass lithography


> Spin coating a resist (with an adhesion promoter) may be good enough for
> 50 microns depth in glass but I am not sure about 200 microns (at least in
> my experience). I think an additional metal masking layer is required for
> this kind of depth.
>
>
> Bill Moffat wrote:
>> Use vacuum vapor prime deposition of HMDS to make the resist stick
>> forever.