How to detremine the lifetime of a LED

[shen3]

Theoratically a LED can last for 100,000 hours, but how can i prove that this is true? I've found out that the malfunction of a LED is caused by the degradation of the active region in it, but how can i determine when it is going to malfunction?

 

[mgb_phys]

Life time testing is ussually done at higher temperatures.
You measure the life of a sample of parts at a number of higher than normal temperatures, you determine the function relating the lifetime to temperature and use it to calculate what the lifetime would be at room temperaure.

 

[CraigD]

Just turn it on and wait for it to fail.

CraigD, AMInstP
www.cymek.com

P.S. you may want to bring a six pack.

 

[mgb_phys]

Just turn it on and wait for it to fail.

That's the problem with the perpetual motion machine invented by my great-great-great grandfather, we are still waiting for it to stop!

 

[ranger]

That's the problem with the perpetual motion machine invented by my great-great-great grandfather, we are still waiting for it to stop!

This article may give you some hints:
http://www.netl.doe.gov/ssl/PDFs/lifetimeWhiteLEDs_aug16_r1.pdf

 

[shen3]

Life time testing is ussually done at higher temperatures.
You measure the life of a sample of parts at a number of higher than normal temperatures, you determine the function relating the lifetime to temperature and use it to calculate what the lifetime would be at room temperaure.

Then how do i calculate what the lifetime would be at room temperaure?

 

[mgb_phys]

It depends on the function relating increased temperature and lifetime.
Eg. if you heat it to 100C and it last 50K hours, at 200C it lasts 20K you can plot a graph and see how long it would last at 20C.
Since you don't know that the relationship is linear, it is probably exponential, you would have to try a number of temperatures and see what shape the curve is.
You also have to do this for a number of devices to get an average behaviour and since you are running the device until it dies you only get one data point/device.

 

[Dario78]

It depends on the function relating increased temperature and lifetime.
Eg. if you heat it to 100C and it last 50K hours, at 200C it lasts 20K you can plot a graph and see how long it would last at 20C.
Since you don't know that the relationship is linear, it is probably exponential, you would have to try a number of temperatures and see what shape the curve is.
You also have to do this for a number of devices to get an average behaviour and since you are running the device until it dies you only get one data point/device.

..or, you can check for a producer site, example: osram, lumileds. They surely have all this info.

I have a question, what is the physically reason for the intensity degradation of a LED?

 

[mgb_phys]

..or, you can check for a producer site, example: osram, lumileds. They surely have all this info.

The OP wanted to know if the specs were true

I have a question, what is the physically reason for the intensity degradation of a LED?

In theory - dopeants migrating out of the active region of the diode.
In practice - corrosion of the bond wires because of contaminants when it was made, chemical changes in the encapsualting plastic from the atmopshere, mechancially breaking the leads where the enter the plastic

 

[Andy Resnick]

Theoratically a LED can last for 100,000 hours, but how can i prove that this is true? I've found out that the malfunction of a LED is caused by the degradation of the active region in it, but how can i determine when it is going to malfunction?

Lifetime testing is an interesting excersise. Typically, tests are performed at high temperatures to accelerate the process- to simulate the rapid passage of time. Also, products typically fail with a distribution known as the "bathtub curve" or Weibull chart:

http://en.wikipedia.org/wiki/Bathtub_curve

I forget what "lifetime of 100,000 hours" means in terms of MTBF or the Weibull chart, but it's a statistical measure, in the end. Your device may last longer or shorter.

 

[Lambduh]

I think for some LED's they rate the lifetime as the point when they only put out either 75% or 50% of the light they did when they started.

http://www.philipslumileds.com/pdfs/WP12.pdf

/shrug