Joachim,
One way to make a gradual slope is to use a bi-layer material consisting of a
material with a low etch rate for the lower layer and a  layer that etches
very
fast laterally (undercuts) for the top material layer (both materials etch in
the same etchant). I have previously made structures of this type using
thermal
oxide for the lower layer and phosphous doped undensified CVD oxide for the
top
layer that was etched using a dilute HF etch.

F.J. Kub, "Methods for making a sloped insulator for solid state devices"
U.S. Patent No. 4,098,638, Date 7/4/78


1. Grow a thick thermal oxide (etch rate of      50nm/minute in dilute HF) or
deposit a thick CVD oxide and densify on a silicon substrate.
2. Deposit a CVD phosphorous doped undensified oxide (etch rate
2000nm/min in
dilute HF)
3. Optionally deposit a LPCVD nitride layer (the purpose of this layer is so
that the etchant will undercut this layer. It may be sufficient to use a
photoresist masking layer.)
4. Photostep to define to define an opening.
- etch the bi-layer in dilute HF

I believe that the slope is approximately related to ratio of the etch
rates of
the two materials.

The principle of the above bi-layer structure should work for other material
combinations if you can find a material combination that has both a high and
slow etch rate in the same etchant. It may be possible to use a resist
material
as you suggest if you can find two polymer materials that have a factor of 10
difference in the developing rate, but obviously this is a guess since it has
not been experimentally verified.

Laser micromachining with the appropriate wavelength laser (for ablation of
polymer materials) may be another option.

Regards,
Fritz Kub
Code 6813
Naval Research Laboratory
4555 Overlook Ave. SW
Washington, DC 20375
Phone: 202-767-3862
FAX:    202-404-7194
email:  kub@nrl.navy.mil

>
> Hej MEMS-researcher,
>
> I have to create a stepless sacrificial layer (of any suitable
> material like photoresist, ...), which means that the thickness of
> the (patterned) layer on a step has not an abrupt rise/fall, but is
> going smoothly from the thickness of the layer to zero and vice
> versa. The width of this slope should be possible to process between
> 50 and 100 um; the thickness of the layer as much as possible (up to
> 10 um or even more). I've heard about "gray-masks", multiexposure
> (with stepwise shifted mask). If someone has some
> information/experiences about that or any other idea, I'll be
> grateful to hear from you.
>
> Thanks and have a nice weekend (relaxing from work...),
>
> Joachim.
> --
>
-----------------------------------------------------------------------------
> -
> Joachim Oberhammer, Dipl.-Ing.
>
> Royal Institute of Technology (KTH)   Phone:    +46/(0)8 790 6250
> Dep. of Signals, Sensors and Systems  Fax:      +46/(0)8 10 0858
> Microsystem Technology (MST)          Mobile:   +46/(0)70 692 1858
>                                       e-mail:
joachim.oberhammer@s3.kth.se
> Osquldas vdg 10                       homepage: http://www.s3.kth.se/mst/
> SE-100 44 Stockholm, Sweden
> _______________________________________________
> mems-talk mailing list
> mems-talk@memsnet.org
> To unsubscribe or change your list options, use:
>
> http://fab.mems-e
> xchange.org/mailman/listinfo/mems-talk




-- 

Code 6813
Naval Research Laboratory
4555 Overlook Ave. SW
Washington, DC 20375
Phone: 202-767-3862
FAX: 202-404-7194
email: kub@nrl.navy.mil
Code 6813 Naval Research Laboratory 4555 Overlook Ave. SW Washington, DC 20375 Phone: 202-767-3862 FAX: 202-404-7194 email: kub@nrl.navy.mil