Ted,
We have examined a failure mode known as CAF (conductive anodic filament 
formation) and this situation sounds very much like it.  The failure's we 
have observed are on FR-4 boards and polyimide should be less  susceptible. 
  However, the voltage gradient you report in your note makes this a 
possibility, at least worth
looking at.

The failure mechanism was first reported on in the mid to late 1970's by 
Bell Labs.  We have
recently observed this failure mechanism enhanced by chemicals in the fusing 
fluid or soldering flux.
Please e-mail me directly or phone me at (404)894-9073 to discuss.  I would 
be interested in having a
sample of the failured circuitry to analysis.

Laura Turbini
Georgia Institute of Technology
Professor, Materials Science and Engineering
 ----------
>From: TED JONES
>To: TechNet
>Subject: Military Polyimide Board Failures
>Date: Friday, March 29, 1996 8:57AM
>
>The Navy has recently experienced field failures (shorts) on three
>7-layer polyimide PWBs.  The boards had operated for 300 hours in a
>sealed chamber.  The chamber is maintained at 50C, with constant
>humidity.
>
>The shorts (about 8 ohms of resistance) developed between a PTH and
>ground plane on layer 5.  There is a voltage potential of 30 volts
>(+15/-15) between the PTH and plane.  The clearance between the PTH's
>pad and the plane is 10 mils.
>
>Horizontal microsections of the failed areas do not show any obvious
>migration path.  The only anomaly noted on the failed pad is that it
>demonstrates much more pink ring than pads on adjacent PTHs.
>
>We're looking for ideas on the most likely cause of the shorts.
>
>Have there been any studies done that indicate pink ring will or will
>not contribute to this type of failure?
>
>Has anyone seen similar failures?
>
>