Dear Leon,

I've never tried bonding aluminum to silicon... so I may be way off, but,
how are you cleaning your substrates? Do you know your surfaces are
really clean?

For me, the main problem was surface cleanliness. I've found that a
good RCA1 (and perhaps RCA2 if you're using the interface for carrier
transport) clean does wonders for bonding substrates. You may also be
able to get by using solvents (acetone and IPA in that order) and
scrubbing with cotton tipped q-tips. Another reported method is using
a CO2 snow gun but I haven't had much luck with that one. You will
probably also want to plasma clean your surfaces before bonding - I've
used a parallel plate RIE with ~12" diameter platens with ~100-200 W
power for about 30 secs in argon or nitrogen to activate the surface.

Good luck.

Felix


On Jul 6, 2009, at 12:25 PM, Joseph Grogan wrote:

> Hi Leon,
>
> I haven't done any bonding of materials with mismatched thermal
> expansion coefficients, but my understanding is that the trick is to
> heat and cool VERY slowly. Maybe 1 C/min.
>
> I haven't done Al-Si bonding either, but here's some general advice
> that might apply to you. A critical issue I encountered in oxide-
> oxide bonding was in regards to wafer specs. I was blindly using 100
> mm Prime grade wafers thinking Prime was Prime and that was as good
> as I could hope for. But not all Prime grade wafers are created
> equal. The standard specifies a range for specs like TTV & Bow/Warp.
> After several seemingly inconsistent bond attempts, I took wafers
> from two different batches (both Prime) and had them characterized.
> It turned out the wafer from the batch that consistently bonded had
> low Bow & Warp (~3 um) and the wafer from the batch that
> consistently gave me trouble had high Bow & Warp (~ 25 um). Both
> were within spec for Prime (and even within generally recommended
> specs for wafer bonding), but one worked well and the other gave me
> a lot of trouble. It can be hard to distinguish between a problem
> with your process and a problem with your samples. So my advice is
> to make sure the substrate you are starting with is of the highest
> quality so as to avoid difficult to diagnose problems down the line.
> Surface roughness, TTV, and Bow/Warp are critical. If those specs
> are good enough, you don't even need a bonding tool, simply place
> the wafers together and they'll stick. Many wafer manufacturers will
> work with you to produce wafers with tighter specs - that's the path
> I'm pursuing now.
>
> good luck,
> Joe Grogan

Felix Lu
felix_lu@yahoo.com