Any suggestions on deep SiO2 etches (400um)?

[OhulahanBass]

I've been struggling and wasting mask materials trying to etch SiO2 wafers. The orientation of the etch is not a huge issue. I need to perform a detailed anisotropic etch on my feature side and etch the bulk depth from the other side.

For my bulk etch I have tried multiple masks in BOE with little success. My wafer is 500um thick and I want a bulk etch of 300-400um (future wafers may be thinner).

The wafer material is Fused Quartz/Silica (SiO2).

Masks attempted were Cr-Au-Cr-Au, NR9 & SU8.

Any suggestions or shared experience would be helpful. I don't think many have tried such etches and for good reason. I may be trying the impossible.

 

[Mapes]

When I last worked in microfabrication (2005), essentially the only approach was to mechanically powder blast the glass wafer. You can get ~300μm diameter through-holes in a 500μm wafer this way, etching halfway through each side with an elastomer mask. At the time, researchers were trying to develop a DRIE process for SiO₂, but the molecule's so stable that there's really no analogue to the use of SF₆ to etch Si.

On the wet side, 49% HF will etch oxide at about 1μm/min, but as you know, it's hard on masks. You might try <4nm Cr or Ti under Au; the adhesion layer here is so thin that kinetic limitations prevent fast undercutting. You'll get some delamination around the opening, but it may be acceptable.

 

[OhulahanBass]

When I last worked in microfabrication (2005), essentially the only approach was to mechanically powder blast the glass wafer. You can get ~300μm diameter through-holes in a 500μm wafer this way, etching halfway through each side with an elastomer mask. At the time, researchers were trying to develop a DRIE process for SiO₂, but the molecule's so stable that there's really no analogue to the use of SF₆ to etch Si.

On the wet side, 49% HF will etch oxide at about 1μm/min, but as you know, it's hard on masks. You might try <4nm Cr or Ti under Au; the adhesion layer here is so thin that kinetic limitations prevent fast undercutting. You'll get some delamination around the opening, but it may be acceptable.

The mask I am trying to perfect is Cr-Au-Cr-Au with Au covering Cr sidewalls for delamination prevention then SU8 on top to prevent pin holes. The cleaner I make the process the better the adhesion so far. I don't think I have powder blast available to me at CU but with some expensive training I could get my hands on a Focused Ion Beam which may be the only, much more expensive, option.

Thanks for insight. After much journal research the deepest published etch I've found in silica is 100um. Perhaps I was dreaming to attempt 300-400um.

 

[Mapes]

I don't think I have powder blast available to me at CU but with some expensive training I could get my hands on a Focused Ion Beam which may be the only, much more expensive, option.

You'd outsource the powder blasting. (I had the vendor Sensor Prep make arrays of through-holes in a standard 4" Pyrex wafer.) FIB is a serial process and would only be viable if you want just a few holes.

Have you read all of F.E.H. Tay's work on wet etching of glass wafers? Lot of good information on Au and a-Si masks there, if you can live with the undercut from 49% HF etching. (And it looks like I was low on the etch rate; it's reported to be >10μm/min for Pyrex.)

Finally: another option is laser drilling, but I'm not familiar with the difficulties/cost.

 

[OhulahanBass]

I have looked at many papers on Au and Si masks for SiO2 and Pyrex. Unfortunately Pyrex has a much weaker dielectric strength than pure silica, I need the dielectric properties similar to pure silica.

Thank you for the lead on F.E.H. Tay, I've found a few papers of his to review this weekend.

Perhaps this is a topic for a different thread, but I have found manufacture websites claim Pyrex has a much smaller dielectric strength than Fused Silica but the wiki sites say the opposite...?

 

[OhulahanBass]

Wow, read some Tay papers this weekend and I think your referral may be saving me hundreds of dollars by switching to a cheaper substrate and eliminating Au entirely. With a D263 substrate and Mo mask I should be able to reproduce similar success to Tay's. I'll try a similar method with the silica substrates this week and order D263 subs if that doesn't work.

Funny thing is he referenced many of the papers I had already read how ever I never came upon any of his work...I need to revise my journal searching methods.

Thanks again.

 

[Mapes]

Good luck!