Is it possible that electromigration cause a short circuit in a PCB? Theoretically due to chemical potential difference between gold and nickel (1.68 V) it's known that metal ions can penetrate into the isolation and form a bridge. Is that really a practical threat and such a process is really causing short circuit in PCB devices? Is that process occurring only when the device is turned off? Could that make a short circuit between two close pads of a component for example a smd capacitor 0603?

  • \$\begingroup\$ From the wiki article it looks like electromigration is a) mainly happening when high currents are present, so not when the device is shut off; and, b) more an issue in on-die interconnect than pcb level interconnect. \$\endgroup\$ – The Photon Jun 2 '16 at 16:29
  • 1
    \$\begingroup\$ For PCB level failure mechanisms, you might want to look into dendrites and tin whiskers. \$\endgroup\$ – The Photon Jun 2 '16 at 16:30
  • 1
    \$\begingroup\$ look into CAF.. \$\endgroup\$ – JonRB Jun 2 '16 at 16:37
  • 1
    \$\begingroup\$ You may find this document interesting. \$\endgroup\$ – somers Jun 2 '16 at 16:37

If you have a mechanism for electromigration, than yes I suppose its possible. Possible mechanisms involve battery-like processes and\or an electric current. There are certain materials that develop "tin wiskers". If you are exposing your PCB to chemicals that could aid in the breakdown of metals and cause corrosion\reformation of a metal. If you have a really high current then this is also possible because you turn the metal into a plasma and it deposits elsewhere, like a sputtering machine. Metals don't move around on there own, they are stable, they need energy to move.

Keep in mind that just because its possible doesn't mean its probable. If you are having issues with shorting in a PCB I would first suspect the manufacturer, get an unpopulated PCB and inspect it. If its on the top layer then it should be easy to spot and measure on an unpopulated PCB. If its an internal plane sand off the problem area very carefully to expose the problem area (you can see through the PCB in a lot of cases by sanding off the solder mask, you can even sand through planes.) If it is a problem then contact the manufacturer.

If your having an issue with a component, then it could be the assembly process. When you get your boards back from the assembly, do a visual inspection. Even better, pull the capacitor off the board and measure the impedance and make sure its open. Then put it back on the board and make sure there are no solder bridges.

If your having a cap short out, if it's a polarized cap, you probably have a polarity problem. Tantalum caps should be rated to ~70% of their working voltage. Ceramic caps voltage rating should not be exceeded. Keep in mind there is also esr in a cap, it is possible to have some self heating depending on the conditions.

  • \$\begingroup\$ Would a conformation coating remove 99% the electromigration? \$\endgroup\$ – judoka_acl Jun 3 '16 at 9:22

It is very unlikely that you would see real electromigration in something as "gigantic" (relatively speaking) as a PC board. We typically speak of electromigration at nearly microscopic dimensions (such as inside an integrated circuit).

OTOH, tin-whiskers are becoming a serious problem with the mandating of lead-free solder. ROHS is creating a whole new class of "planned obsolescence". In fact critical products (medical, aerospace, military, etc.) are exempt from the ROHS lead-free rules specifically because of the tin-whisker problem.

  • \$\begingroup\$ Most whiskering so far has been observed in pure Sn lead finishes, not in SnAg/SnCu/SnCuAg solder joints -- this means that we may see RoHS hi-rel if the industry can get their arses behind NiPdAu/Ni-barrier terminations instead of those blasted matte Sn ones. \$\endgroup\$ – ThreePhaseEel Jun 3 '16 at 0:57
  • \$\begingroup\$ "The Galaxy IV was a telecommunications satellite that was disabled and lost due to short circuits caused by tin whiskers in 1998. ...The manufacturer, Hughes, has moved to nickel plating, rather than tin, to reduce the risk of whisker growth. The tradeoff has been an increase in weight, adding 100 to 200 pounds per payload." - Wireless Week, May 1999 \$\endgroup\$ – Richard Crowley Jun 3 '16 at 18:08
  • \$\begingroup\$ Then why, over 15 years after Hughes' switch to Ni-plating on component terminations, can I not get Ni-barrier or NiPdAu terminations across the board on components? (ICs are hit-or-miss at best, discrete semiconductors are fugetboutit, and passives are fairly rare -- I've only heard of it on higher end chip resistors and caps) \$\endgroup\$ – ThreePhaseEel Jun 3 '16 at 22:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.