To me, the local spots of burnt (carbonized) epoxy sounds a lot like the
electromigration phenemona which others have previously suggested.

In the case of electromigration, the electromigration process will grow
metallic dendrites (tree-like) structure between conductors.  When the
dendrites grow enough to touch the opposite polarity conductor, they will
short the circuit momentarily and vaporizing themselves and possibly
leaving a local spot of burnt epoxy.

Comparing microscopic pictures of the initial board trace dimensions and
after aging in a high humidity, high temp enviornment with bias  (85 C,
85%RH with 5 volt bias is fairly standard) should show conductors growing
"fuzz" or dendrites if electromigration is the problem.

One of my co-workers also showed that monitoring the leakage current is an
effective predictor of electromigration. That is, place the sample in 85/85
and bias two adjacent but unconnected traces.  Monitor the current flow.  I
don't know what your specific design is, but if you have more than say a
microamp of current after the sample stabilizes at temp/humidty,
electromigration might be a problem.

As to solutions, electromigration may be occurring in the adhesive bulk, or
due to surface ionic contamination or possibly due to the substrate
material.  If for example you can find dendrites in the bulk of the
adhesive, the adhesive is the problem.  If the dendrites are at the
interface of the adhesive and one or both substrates, the surface
cleanliness and/or the adhesive may be at fault.

If the adhesive is at fault, you need to contact your adhesive vendor (and
you might want to actually start there) since they should know if their
adhesive has electromigration problems.  Unfortunately, if the adhesive is
the problem, the only solution is to change adhesives.  Surface coatings
will not significantly inhibit electromigration in my experience.

##############################################################
TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
##############################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
the body:
To subscribe:   SUBSCRIBE TECHNET <your full name>
To unsubscribe:   SIGNOFF TECHNET
##############################################################
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information.
If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
847-509-9700 ext.5365
##############################################################