Subject: | |
From: | |
Date: | Tue, 29 Jul 1997 08:48:28 +0100 |
Content-Type: | multipart/mixed |
Parts/Attachments: |
|
|
Your lifted wedge bonds could have a number of causes. You don't
say if it is the gold, the aluminium wedges, or both that lift.
What is the gold thickness and what metal is underneath it. Very
thin immersion deposited gold (0.1 to 0.3 microns) are not gold
ball/wedge bondable with reasonable yields. I've not tried
wedge/wedge bonding gold, but this may well be the same.
Wire bonding requires cleanliness from the substrates. The
residues form no clean paste are difficult to remove. The alcohol
wash probably isn't making much of an impact. If these residues
are getting on the wire bond sites they can interfere with the
bond. Can you reverse the process flow and do the COB die attach
wire bond and glob top before attaching the SMD components?
If you do this and the COB components are on the same side of the
board as the SMT then you can't stencil print solder paste. You
will have to dispense solder paste instead. Reversing the flow
could also interfere with your test strategy as well, so it needs
some thinking about. Given the choice I would COB first, then
attach the surface mount parts.
Plasma cleaning can be used to improve COB yields by removing
organics from the metal surfaces. Epoxy die attach materials can
out gas during cure, the molecules depositing themselves on the
wire bond sites. I don't know if plasma will remove no clean flux
residues.
Look at the surface roughness of the wire bond sites. If the
surface is too rough then you will get some bonds lifting.
A final set of causes could be the wire bond process itself.
Process parameters, wire and capillary selection, or a machine
fault could all contribute to the problem.
Hope this helps.
Jeremy Drake
Design to Distribution Ltd
Stoke on Trent
England.
|
|
|