Multilayer Ceramic Capacitor (MLCC) failure root cause analysis

I have a consumer product that we have been manufacturing for some time. All of a sudden in one batch, we have about a 4% rate of a single individual capacitor on the board generating allowing a severe leakage current, in the range of 0.1 mA to 4 mA

The cap is a 10 uF MLCC, Y5V dialectric, manufactured by Samsung. There are three of the same part on the board, but only one specific cap seems to be failing. So it seems like something that is damaging that one cap during/after PCBA or FA.

We DID observe poor ESD control procedures (missing wrist straps) on the production line in this production run, but I doubt that is the cause since a Human Body Model 150 pF could barely generate a few volts on a 10 uF cap.

The only thing I've ready about that could cause failures like this are microcracks forming in the capacitor, either from thermal or mechanical stress. The problem is we haven't changed anything in our design, and this has never happened before in 4-5 batches of production. It is possible that the oven temps in the PCBA reflow were slightly different, who knows.

One interesting thing is that upon initial testing of the failed devices, we immediately tested quiescent current, because the failures presented themselves as dead batteries out of the box. Upon replacing the battery we noticed no problem (normal or nearly normal quiescent current, maybe 5 uA higher than usual). But then after leaving a battery plugged into the devices for 30 minutes or so, the quiescent current begins to creep up. After a day or more it can reach as high as 4 mA flowing through the cap (biased at 6.6 V).

Any ideas what could be causing this?

It sounds like a bad batch of caps. The caps should never pass DC so that's a bad part in any case. It could be that the cap is passing a little current initially and heats up, causing the problem to get worse. If that's the case then some freeze spray should drop the leakage for a bit. I assume the caps are bad. If you change the caps with ones from a different batch and the problem goes away that would confirm it.

Yeah, I would contact Samsung. Even if it isn't a Samsung problem, they might be able to help you diagnose the issue.

I would recommend measuring them on a frequency response analyzer. You can measure the ESR and various aspects of the capacitors and go back to them with some data. If you dont go back to the supplier with data then they might not be as responsive as if you go to them with some data.

I’m inclined to agree with that answer as well but also look at soldering temperatures too especially with this unleaded solder Its causes massive problem when taking off or assembling pcbs Hence why I don’t use this might also pay to look at working voltages of the capacitor I’ve noticed in many power supplies on the market that the working voltages are set to absolute minimum requirement bordering on stupidity some of this i.e we have a 10 volt rail thus we use a 10 volt working cap allowing for no tolerance when ideally the working voltage should be 16 vwg to 25 vwg ensuring far longer life on capacitors

Smaller internal IC interconnects and lower
energy consumption has reduced the power supply voltage and as a result, the degree of voltage
variation has been tightened and become more important but ...... they dont say and they dont tell you (Down to cost of course Manufacturing on the cheap)

They also dont allow for fault symptoms/ stabilisation at switch on where voltages produced can be somewhat larger damaging the capacitor on start up as in the case or a power supply or supply rail this becomes quite important , buck convertors included High intial overshoots can cause this damage as well

It may also be an idea to look at operating temperatures as well i.e high temperature ranges with a view to seeing where these are mounted and what next too or in multilayer pcbs on top of what with a view to components on the underside of the PCB

This I guess is the price you pay for miniaturization and multilayer surface mount densely packed PCB’s

Also to be a little bolder the price you pay for not paying for such technologies where you should be paying the same amount for that technology allwong far more improvements rather than cutting costs , cutting your own throats which allows the suppliers to put considerable more money into R & D Also training and decent manufacturing ( heard that before havent we hmmm !! Yes a 1000 billion times before ) but , buttter , butter , but, but, but,but hmmm Indeedie

Isnt that the truth of this as well ( age old story this you peanuts you get monkey's Want things cheap all the rime you must expect quality to fail ) Something somewhere has to go normally its quality

Of course the other answer is Its manufactured by Samsung

Samsung are having quite a few problems with capacitors in general on their power supplies particularly in LCD/ Plasma TV'S but they are not the only ones This is a fairly common problem with switch mode power supplies in general irregardless of MNFR

Still i dont suppose it will make a scrap of difference because after all YOU KNOW BETTER so there you go

The question is a year old, so I'd hope you've gotten some resolution on this already. But it's getting some discussion now, so I'd hope you could share what you've found.

Since it's just one cap out of three, and the problem asserts in non-operating conditions (if I'm reading correctly), that would lead be to also believe that it's a mechanical issue. If not reflow temps, I'd look at the depaneling process. Has it changed? Are the failures occurring on one particular board position out of the panel? Look at the board handling in other areas; test, final assembly, etc.

If it is the same 1 cap out of 3 on the board, consistently, then I would suspect damage from mechanical damage/stress. Damage could be from product assembly and stress sould be during PCA processing.

We had some passives which were damaged often enough to show up as top product test failures in a pareto chart. I narrowed it down to mechanical damage during hand screw installation of the pc board to the product chassis. There were about 20 screws, with all but 2 put in by robot. The damage was consistently around the 2 screws placed by hand. I changed the hand installed screw location to a less popolous area of the board and let the robots put in the screws in the crowded area. Problem solved.

Hi Patrick,

I'm just solving absolutely the same problem as you did. One of two capacitors increases after aprox 3 months its leakage up to 10mA. I cannot find the clue.
Have you found something? Can you advice me?

Thank you in advance,
Radek

Hi All,


Almost same problem, has anyone found the root cause or a solution for this mode of failure?

Matt

Hi Matt,

finally the problem was found on the PCB cutting machine. My PCB was 10x50mm and was bendet on separating machine. The caps were not cracked at all but afrer few weeks/moths absorbed some humidyty and problem came.

So our solution was to change the panel of pcbs - make bridges between single pcbs.

Hope it will help you.
Best
Radek