Copper is added for electro migration, Silicon is added to keep the
Aluminum from attacking the silicon during heating. Gold bonding is used
for all sorts devices with aluminum pads of all sorts of compositions. I
was never in the back end, but I can tell you that wafers could be
stored for months before sorting, dicing and bonding.

Just keep your pads in dry air and you will not have a problem. Make
sure you problem isn't due to something you are doing during metal dep
or wet etching or storage. I have seen pads go bad with contamination
from wafer boats, boxes, and polyimide outgassing prior to cure.

Common pad problems, since our assembly people would complain about them
and I often worked on them are:

1. Your metal dep is too cold and at a high pressure and the aluminum
looks like a solid sheet but it is composed of columnar grains and
doesn't laterally conduct. (Rare).

2. Your metal target has worn through and you have junk in the matal.

3. Your wet etch for the pad mask has attacked the aluminum and has
roughened it. I don't think it is a problem but assembly people complain
about it. Simple to fix and then you don't have to listen to them
complain. Also, excessive time in some developers might attack the
metal. Just don't try to fix incomplete develop with overdevelop or over
etch. Don't over etch excessively.

4. Polyimide in a closed box prior to polyimide cure. Condenses on the
pads, cures, forms barrier to bonding.

5. Boat or box outgassing on grains and then the contamination being
burnt on during anneal or other themal treatments.

6. Pads have been etched out due to misprocessing. (It happens).

If your pads look fine under low magnification 5X objective, 10X eye
piece, they will likely bond just fine and assembly needs to fix their
tools and stop complaining. The Au bonding is a very robust process.

As stated earlier Au has been bonded to aluminum for decades and has not
been a difficult or challenging process.

Edward H. Sebesta
Independent Semiconductor Engineer

-----Original Message-----
From: mems-talk-bounces@memsnet.org
[mailto:mems-talk-bounces@memsnet.org] On Behalf Of Jordi Teva
Sent: Tuesday, January 27, 2009 11:20 AM
To: General MEMS discussion
Subject: Re: [mems-talk] Aluminum Bond pad reference


the Al bond pads provided by the semiconductor industry are not pure Al,
they are alloys with others metals, mostly Cu, just to avoid the problem
you have.. I do agree with Morten¨'s proposal with the Gold layer..,  I
think you will have a layer of alumina as soon as you place your chips
in the ambient air...

Jordi