Fusion UV bake system

Paul Rissman rissman at stanford.edu
Thu Jan 11 11:53:50 PST 2007


Dear Labmembers:

UV bake is a standard industry process for hardening of resist used 
in harsh conditions, e.g. aggressive plasma etch processes and high 
dose implants.  This process uses a short UV exposure to create a 
skin on the resist to fix the profile, followed by an extended bake 
of up to 200 degrees with UV exposure.  The efficacy of this process 
can be found in the literature in papers such as Allen, ECS, 1982:

http://scitation.aip.org/getpdf/servlet/GetPDFServlet?filetype=pdf&id=JESOAN000129000006001379000001&idtype=cvips&prog=normal

SNF has a Fusion UV bake system and Cesar Baxter has gotten the 
system operational.  I have done some preliminary testing and have 
found the following:

1. I verified that resist could be baked to 200 degrees with no 
pattern flow using the standard process.
2. With Nancy Latta's help, we were able to demonstrate increase in 
selectivity for two kinds of etches.  On stsetch, Nancy ran a 10 
minute "deep" program.  The consumption of resist on the UV baked 
resist was about 70 nm (1 um starting film) and the consumption of 
resist on an unbaked wafer was 620 nm, or about a 9 times 
decrease.  On the amtetcher, Nancy ran a 40 minute oxide etch.  This 
consumed about 400 nm of resist on a UV baked sample and totally 
consumed the resist on the control, unbaked wafer.
3. Since the Fusion is a broad spectrum UV source, we can use it to 
expose PMMA.  A 30 second exposure with the bulb on low was 
sufficient to expose a standard PMMA film, which developed away in 
1:3::MIBK:IPA.

I have to this point had the following negative results.

1. Ed Myers ran the "shallow" program on stsetch2 for 10 minutes (120 
cycles), and the resist consumption for the control and UV baked 
sample were comparable.
2. I have been unable to crosslink PMMA with UV exposure.  The PMMA 
is consumed (either sublimation or oxidation) at a dose which should 
be sufficient from crosslinking.

I am presently looking for a few users who would like to try the 
Fusion with their processes.  As we get more data on the system, we 
can turn it over to general use.  Please contact me if you have 
processes which you think will benefit from using the Fusion system 
so we can arrange for a time to work together.

Paul Rissman 




More information about the labmembers mailing list