LED damage/failure mode Internal Parallel Resistance.

[Brlux]

So I have experienced this Phenomenon with LEDs over the past Decade but did not understand it until recently. When an LED gets damaged it can create a internal parallel resistance with with the die. Resulting in some amount of power being wasted in heat that has no light production property. It also makes a minimum current that must pass through the device before it will start to emit light. I first noticed this nearly a decade ago with some of the first power LEDs, 1W batwing Luxeon. I have them installed 3 in series on a heat sink and used them for home lighting. One one particular panel when dimmed down really low one LED was completely off, one was very dim, and the 3rd was about twice as bright as the dim one. Keep in mind these were all wired in series. A decade later of 24/7 operation it is still working the same as it was the day I built it.

I next observed the phenomena tiring to hand solder the Luxeon Rebel flip chip version (No top mounted bond wires) when they first came out. The hand soldering thermal cycle would damage the chip and produce a severe effect where hundreds of mA would be burnt in parallel resistance and it got worse rapidly as the parts were burn in. Under a microscope when driver to barely illuminate you could see an uneven lighting of the die almost like there were cracks or faults in it. Soldering them with even temperature in an oven had no such effect.

Recently I saw a video of the 100W Chinese LED panel (100 x 1W dies in a 10s 10P configuration) that illustrates the effect and helped me better understand what was happening. So thanks to the Guy who put the video together.

Video demonstrating the problem. Demonstration starts about 3:20.


Video explaining the problem.


Does anyone else have experience with this failure mode, or information on what causes it and how to avoid it. I have been searching the web and have not seem much on the subject. For as much as we love and use our LEDs it seems like it should be a more understood failure mode.

O and other commonly observed failure modes in 5mm LEDs is the bond wire separation form the die that causes the LED to oscillate (Flash) due to the temperature cycling of the part as it heats up and cools. It is commonly observed in LED traffic controlled devices.

Thanks

 

[Steve K]

We had a fellow at work who came from a semiconductor manufacturer and had a pretty good knowledge of how semiconductors fail. He would analyze the failed semiconductors in our returned parts. I asked him once why zener diodes nearly always failed shorted. His reply was that when the silicon gets hot enough, it becomes liquid (or nearly so?), and the impurities that created the N and P areas got to flow around and now produced a simple resistive material.

I suspect that a similar sort of thing is going on in the LEDs.

Just another reason why heat is not the friend of your LEDs or any other semiconductor.

For what it's worth, my recollection is that impurities migrate in the silicon lattice at more typical temperatures too, which is one way that silicon can wear out. The heat may just increase the rate of migration in addition to the problem of silicon changing to a liquid state.