etching through wafer

Kim, Han-jun at
Thu Apr 8 11:08:52 PDT 2004


I'd agree with Chris that holes are more difficult 
than trenches. I personally didn't have any problem 
in etching 40:1 AR features (0.5 um wide & 20 um deep),
with CD loss around 5%, using the current "old" STS. 

But they were "trenches" with 500 um of length, not 
"holes".  When you etch "deep" holes, you may want to 
take microloading and sidewall profile into account. 
I think you will better achieve your process goal by 
modifying the recipes of your current use. Let me tell
you a guideline about it briefly:  

Horizontal etching may be minimized by increasing 
C4F8 flow rate and decreasing power.  When you add 
some oxygen, the surface is also passivated with 
Si(x)O(y)F(z) along with the CF(X) on it. But the 
passivation is compromised with the enhanced etch rate. 

If you're concerned that adding oxygen etches resist 
way faster, you may introduce a hard mask. Finally,
lowering pressure helps regulate the gas density that 
the resulting etch becomes more uniform. This feature 
helps especially when the exposed areas are small. 
But, this is at the cost of reduced etch rate.

Good luck on your process!  


 Han-Jun Kim                  
 Hewlett-Packard Lab.
 1501 Page Mill Rd, MS 1198
 Palo Alto, CA 94304
 (650) 857-8525 / 8948 FAX        
 hjkim at          

-----Original Message-----
From: Chris Kenney [mailto:kenney at]
Sent: Thursday, April 08, 2004 10:12 AM
To: Andrew Arthur Davenport
Cc: stsetch at
Subject: Re: etching through wafer

Hi Andrew,

Our group has been etching small, high-aspect-ratio features
since the machine was installed.

When you say a 8 micrn hole, do you mean a circular cross
section or a trench?

For circular features we obtain aspect ratios of 15 to 1
after some tuning and some luck.

During the etch the diameter tends to widen.

Several years ago we etched small holes through a 300 micron thick
wafer. The holes were drawn as 15 micron diameter circles on the
mask. As I recall it took 4-5 hours to etch through and the
holes ended up around 22-23 microns in diameter after the etch.

Most papers use trenches when discussing process development
for a DRIE machine. Trenches are easier to measure and
one can obtain higher aspect ratios with trenches.

Unless a paper is very explicit in specifying that circular
holes were used I'd assume they used trenches.



On Thu, 8 Apr 2004, Andrew Arthur Davenport wrote:

> stsetch users,
>    I'm trying to etch a 8um hole through a 325um wafer.  I've tried the
> standard recipies (deep and fastdeep), and also 7um and 10um resist.  None
> of these seem to be able to etch through within 4 hours (by that time the
> resist is burnt).  Have tried bonding to another wafer with photoresist - it
> improves the resist lifetime but will still burn eventually.  Just wondering
> if anyone had experience with small deep features (or aspect ratios >30).
> I'm currently trying one of the recipies in the Ayon paper listed on the SNF
> website, but am uncertain of whether it will work.
> Thanks,
> Andrew Davenport

More information about the stsetch mailing list