Reading time ( words)
I recently spoke with Foresite CEO Terry Munson during the IPC High-Reliability Forum and Microvia Summit in Baltimore. Terry discussed his presentation on the causes of conductive anodic filament (CAF), the dangers of resin starvation, and what advice he’d give to PCB designers to avoid those types of failures.
Andy Shaughnessy: Can you give us a rundown of your presentation?
Terry Munson: Thanks for the opportunity to share. I talked today about CAF and what’s causing those types of failures. We’re seeing an increase in those type of failures in automotive, medical, and even some of the server board construction. Class 2 and 3 hardware are both experiencing similar issues. And the difference between CAF and electrochemical migration boils down to this: CAF is inner layer, and the electrochemical migration is external.
Shaughnessy: I thought the dendrites and CAF were almost interchangeable.
Munson: Yes, dendrites. We’ve seen a lot of external failures; it’s harder to find those internal failures when things are shorted and when there are two vias that are shorted together or to the ground plane. Dendrites grow on the surface due to flux residues, board fabrication residues, external moisture, and micro-condensation—droplets of water sitting there. Internally, the only thing you have as a contaminant source is the etching materials for the prepreg, the desmear process, and then the plating process. In the barrel itself—if the resin flows well and there isn’t any resin starvation in the weave—you have no place for that plating chemistry to go, so it just gets flushed out and is relatively clean behind the barrel.
When you have resin starvation, and fiberglass is exposed, that fiberglass bundle becomes a collector. All that low surface tension plating chemistry works right in. Now, when you plate a big copper barrel right next to that, the chemistry is going to do what chemistry is designed to do: dissociate the copper and create a conductive opportunity. We’re still figuring out how we go from the anode to the cathode without the normal pathway. You have dendrite growth that transports through the metal salt, which collects on the surface. With a CAF, you always seem to have the power effect; the dendrite is thicker on the power side. So, it is plating to that crystal area and then putting to a ground plane and finding a pathway through that weave—that structure—or at micro-separation where you have two layers that can separate.
In my presentation, I showed both of the open weave exposures, which are very visible in cross-section. Then, I showed the micro-separation at the prepreg level where the wrong material was used for low-flow or no flow resin. Because it was the wrong material, there was a gap between the two layers. The plating chemistry fit into that space between the via and the ground plane. I had beautiful dendrite photos showing that. It worked out extremely well.
We also have inner layer shorts that occur at the power trace to ground due to poor rinsing of the prepreg itself or poor control. And when we split a board apart and use a localize C3 extraction where we can still extract the surface area of the inside of the board and compare the microvias to areas where there are no microvias, we see a difference between 3 μg/in2 of sulfate to 74 μg/in2 sulfate where we see dendrites. Areas that have low levels of sulfate resin from the methane sulfonic acid plating have very low levels of contamination. Areas where we see dendrites growing around the microvias come from very high doses of contaminants from the trapped residue from the plating.
Now that we’ve split the board open, we’re able to then take a look at the physical. Most people are surprised you can cleave it and look at the inner layers. It splits relatively easy, especially if there is a weak layer; then, it will split along that interface. We have created a tool to allow us to do that with a press and a special blade that we have. We can go in and cleave a board open and see any weave that has nearly no resin flow into it. We can see the impression made into the resin and confirm that it did not flow into it. We can compare that to other areas of a board where we have good resin flow and show there is no problem.
To read the full article, which appeared in the July 2019 issue of PCB007 Magazine, click here.