LED failure modes

[lumengeek]

Hi all,

I was speaking to a work colleague the other day about white LEDs and their rated life of so many thousand hours, but the question is what is the failure mode of the LED beyond it's rated lifetime? (assuming it is cooled properly and not overdriven). I can only assume that the phosphor degrades and the light output becomes more blue, but what else? and what is the reason for failure (other than runtime of course )

Would be interesting to see if anyone here has any experience of LEDs failing 'properly'

 

[JohnR66]

The epoxy encapsulated LEDs, like the common 5mm type dim due to the epoxy becoming degraded by the light and heat. Their life as rated by some product manufacturers is grossly overstated. Some claim 100,000 hours. When the actual life is just a few thousand hours, sometimes barely over 1000 hours. This is fine for flashlights, but life it is too short in lighting products where the LED is non replaceable, such as yard accent lighting.

Shame on manufacturers for selling over driven LED products that fade in just a few hundred hours.

The power LEDs, like the Luxeon and Cree XR-E and such have better thermal management and lenses so the life is claimed 50,000 hours to 75% output. I can't dispute this claim. I recall one CPF member running a Luxeon for several years (?) and claims very little light loss.

I'd like to know myself what die level degradation occurs in LEDs after normal use.

 

[JWP_EE]

Check out this site.

http://www.molalla.net/~leeper/5mmdeg.htm

 

[csshih]

too much heat in a 5mm will actually fog up the plastic quite a bit, make it turn yellow, etc.

 

[Gunner12]

I remember someone here running a Cree XR-E and someone else running some Luxeon LEDs(low bin) at high current without heatsinking just to see what happens.

At first they are fine.

Then the LED heats up and the tint might start to shift.

It gets to a point when blowing on the LED(and cooling it a bit) will make the LED brighter.

The the LED would start to smoke and dim a lot as the die dies.

The the LED is dead.

I think he saw burn marks on the LED due to the heat.

 

[lumengeek]

thanks for the replies so far, it's interesting to find that the primary reason for failure seems to be the epoxy degrading, and not the chip itself!

Could it be that the yellowing encapsulation also compensates for a degrading phosphor? (in other words, the light output from the chip becomes more blue, but you don't notice)

I would have thought that the phosphor would degrade quite a bit due to self heating from the downconversion it provides, but this seems not to be the case? TV picture tubes are rated for something like 8000 hours, and I'd guess that is due to the phosphors

So far the only 'die level' degradation I can assume applies to LEDs is thermal cycling, just as in power devices. It looks like the die itself is the least of our worries afterall?!

 

[SemiMan]

Failure modes.....

- Phosphor degredation in a white LED causing color shift.
- Minor degredation of the encapsulation (even a bit for silicone, but not nearly as severe as epoxy). However since generally we are talking power LEDs now, the epoxy discussion is not really needed as much.
- Actual degredation of the die. Thought it is a semiconductor, they do actually wear out. It is a function of temperature and current density. There are a host of ways/reasons this happens. Part of the problem is that it is impossible to make absolutely perfect ICs with absolutely pure material. What starts as small defects over time grow worse. There are practically books on the subject alone. In the simplest fashion, thinks of wherever there is a defect\impurity that in that very small area, the device is working beyond its rated design. What starts as a "small" failure point(s) grows and the light gets less and less. Does that help? Without getting into a lot more detail, I think is as simple as it can be made.

Semiman

 

[lumengeek]

yes that has just about summed it up, thanks. I was starting to think the die was the most robust part of the LED.

I guess due to the rated lifetime of thse things, no-one has really observed any 'true' failures other than epoxy darkening on 5mm leds, and premature die/phosphor failure on power leds due to overcurrent/overheating? oo:

 

[nbp]

I remember someone here running a Cree XR-E and someone else running some Luxeon LEDs(low bin) at high current without heatsinking just to see what happens.

At first they are fine.

Then the LED heats up and the tint might start to shift.

It gets to a point when blowing on the LED(and cooling it a bit) will make the LED brighter.

The the LED would start to smoke and dim a lot as the die dies.

The the LED is dead.

I think he saw burn marks on the LED due to the heat.

I am curious if this is what happened to my L2T. When I turn it on, it is quite dim and puts out verrrrrry yellow light. Looking directly at the die it is yellow/brown in color, almost burnt in appearance. Do you think that it just cooked itself?

Other people have used it, so I don't know if perhaps someone left it on in turbo for an extended period of time and it just overheated and burned.

Thoughts?

 

[Gunner12]

I am curious if this is what happened to my L2T. When I turn it on, it is quite dim and puts out verrrrrry yellow light. Looking directly at the die it is yellow/brown in color, almost burnt in appearance. Do you think that it just cooked itself?

Other people have used it, so I don't know if perhaps someone left it on in turbo for an extended period of time and it just overheated and burned.

Thoughts?

I don't see why it could have happened. There is more then enough metal in the L2T to prevent that from happening.

IIRC the LED that was tested was not mounted to anything and was overdriven at 2 amp current(I could be wrong about the current).

Maybe someone accidentally placed the batteries in backwards and fried the driver. Then 2 AA batteries might direct drive the LED at a low level.