Electronics can fail due to overheating, overvoltage, and mechanical stress. Are there other causes? [closed]

Question

My experience with power electronics failures have generally fallen into three categories: overvoltage, overheating, or mechanical failures.

Each might have manifold sub-causes, of course. Overvoltage might be inductive kick, common mode noise, line transients, or static shock. Overheating might be caused by high frequency oscilliation of a switching device, or by high RMS currents, or by unexpected ambient conditions. Mechanical failure can be caused by vibration, or incorrect assembly, or just dropping the thing on the floor. But all failures I've seen seem to fit one of these three categories.

I can imagine a few others like radiation or aging, but I haven't personally seen them. Are there others? Is there a definitive list of ways to kill electronic components?

Answer 1

Yes, there are other mechanisms. In fact, it's impossible to give a complete list. This is softof the same problem as proving a negative. The best anyone can do is list the mechanisms they know about. Some additional ones are:

Chemical. Dropping most electronic things into a vat of boiling acid is going to kill them, but chemical processes can be more subtle too. Nothing is ever absolutely sealed. Corrosive agents will diffuse thru any barrier given enough time. Even just one assembly line worker with a little lunchtime salted peanuts residue on his hands can contaminate a whole batch of epoxy used to seal chips. Those chips then have a high failure rate months to years later in warm and humid environments. That's just one example, which actually happened.

Loss of essential substance. This is not really a mechanical failure since in some cases it is know this will happen. Electrolytic capacitors suffer from this, for example, even when everything works as designed. Given long enough, the electrolyte will diffuse out. Certain gas-filled components have the same problem, like neon bulbs.

Someting is consumed as part of normal operation. Batteries are a obvious example. Incandescent bulbs is another less obvious one. The temperature necessary to cause the intended black body radiation will also cause fillament molecules to evaporate off its surface, eventually causing it to fail. As another example, vacuum tube cathode coatings degrade over time with use.

Diffusion. Even without essential stuff leaving or bad stuff getting in, the stuff inside can move around. Short of freezing something to absolute 0, you can't keep molecules from moving around. Most of the time, these molecules will move around so slowly that something else will go wrong long long before this causes problems. For example, P and N type dopants in raw semiconductor are essential to how transistors work. These were diffused into the silicon crystal at high temperatures over hours. At ordinary temperatures, the diffusion rate is so small that other things will happen before most transistors die this way. However, the rate is not 0, and as transistor sizes decrease, this will eventually become a limit on lifetime.

Migration. Diffusion is one substance moving thru another, but migration is outright bulk movement of stuff. This is already a issue as the nm scale of modern chips. The interconnects don't stay where you put them due to applied voltage and current flowing thru them. This is something that has to be considered for very fine scale structures.

You didn't specify lifetime, so very long term there are other effects, like radio-active decay. Everything except iron will decay, eventually to iron. Most of the elements we build things with have such long half-lives that this is irrelevant on a human scale. Still, we have harnessed this effect deliberately. There have been spacecraft powered from the heat of radioactive decay (of strontium 90 if I remember right).

Answer 2

1. Temperature:
   Temperature can do a number of things to cause the circuit to fail, such as

   • If you have ceramic capacitors that have a certain dielectric (i.e. X7R) and the environmental temperatures exceed the dielectric limits, the capacitor value/behavior is undefined and will not operate per design.

   • Transistors! Notably bipolar junction transistors (BJTs). For a silicon BJT, the base emitter voltage will drop -2.5mV per every degree C rise in temperature if not compensated. If your transistor isn't biased properly over the full temperature range of your product specification, the circuit will not behave as designed.

2. Vibration
   Say you manufacture a widget. You have a contract manufacturer build it. How will it get to your customers? You ship it, right? If the quality of the solder connections is good, then the vibration as it applies to an IEC standard of the equivalent vibration of a transportation test should show no loose connections and the circuit will operate as designed from a manufacturability perspective. If there are things like cold solder joints (due to a poor manufacturing quality, or not adhering to the temperature profile of the components, OR if there is a LOT of copper on the board in the area of a component and the needed solder temperature for that component is higher due to needing to head not only the pads, but the copper plane), THAT could cause a cold solder joint too.

3. Compliance

   • ESD (Electrostatic Discharge)!
     This can happen a number of ways. The human body can generate charge (i.e. in a dry environment) and discharge to any exposed electrical connections to a circuit in the form of an arc (zap). In the case of CMOS, this can cause either non-catastrophic damage (micro-perforations in the insulating oxide layer), or catastrophic damage, which will add leakage to the gate (by a major perforation in the oxide layer) and preventing the junction from working at all.

   • RF Immunity
     In the presence of radio waves (i.e. keying a walkie talkie microphone), the energy from the antenna of the walkie talkie can couple to ANYTHING that resembles an antenna in your circuit. This may be enough to glitch a signal to cause undesired results.

   • Static
     Similar to ESD, a mere static electric field (for example in the presence of a MOSFET) that can cause it to enable in a glitchy way. The tricky thing is here, there need not be an actual discharge, but just the mere presence of the field in the vicinity of CMOS.

4. Mis-design (SCR Latchup)
   In electronics, due to the way in which transistors are manufactured there are certain parasitic unintended paths in which current can flow. For instance if some circuits are not biased properly or power sequenced properly, the current can flow instead of via the intended path in semiconductors, it can flow in an unintended parasitic path. This current flow in an unintended path typically generates a great deal of heat. This, like ESD, is a destructive process. While SCR latchup may appear to occur several times without consequence, the reality is that the component is weakened because of it. As a result, it will definitely fail prematurely at some point. Likely just when you least expect it.