As for the Holy Trinity of dendritic growth, be careful about where you think the contamination element is coming from. I did some papers on this back in 94-97. (Electrolytic Metal Migration In Flexible Printed Circuits) Essentially what we found was that with perfectly prepared base circuits (no measurable contamination) the application of coverfilm alone could lead to potential dendritic growth problems. It was entirely dependent on the resin formulation of the adhesive in the coverfilm, and you could grow a beautiful set of arching dendrites (following the field lines) through the adhesive between traces. The problem was that the adhesive had enough inherent ionics in it to become a solid state electrolyte when exposed to moisture. The bad news was that the control in our test was the industry standard at the time, and there were tons of circuits being made (including for avionics applications) with the susceptible coverfilm formulation. 

The interesting thing about this growth mode was that it had a clear incubation period, followed by steady growth until the circuit closed. It the circuit was going to fail you knew it fairly quickly. If it was a good one, no matter how long you left it, it would never initiate dendrites. Depending on the current you could get the dendrites to act like a fuse. Once they closed the circuit, high enough current would flow to blow the fuse, and then a new dendrite would start growing to replace it. The adhesive in question wasn't fire resistant, and we wondered if it would ever start a fire. We found numerous charred spots, but never got one to torch off.

I always wanted to invent something that would take advantage of the solid state electrolyte effect, but for now I'm still sleeping on it.

All the best,
Andy Magee - Flex Guru
Senior Consultant - Bourton Group