LED Depreciation

D

**DONOTDELETE**

Guest
Does anybody know if it is the white LED's that degrade over time , or do all LED's have this problem?
 
Mike,

They all degrade over time. The Nichia white have an expected half life of 50,000 to 100,000 hours if they aren not over driven or subjected to high operating temperatures. UV and heat will age and yellow the epoxy housing. In many of the flashlights, the LED's life might will be in the hundreds to thousands of hours due to the high current passing through the LED's.
 
Actually, they are MUCH worse than that. Especially white (and blue also) ones. Popular 5mm white ones, if driven normally, meaning no overdrive and proper thermo design, they will loose half the brightness in about 5-6 thousand hours. If you overdrive them, they go MUCH quicker, depending on how badly they are overdriven. If you use them in high temperature, or poor heat sinking, they also go quickly.

I am aware that a lot of places and people say 100K life, but that is simply not correct. Most of them are not an expert, and never tested them (btw, I am and I did).

If you are interested in technical details, one handy information is available at www.lumileds.com/pdfs/AB07.PDF. This is an application brief for lumen maintenance. One testing I was involved, white LEDs were driven at 20mA at 60degree C ambient, they lost about 30% of brightness in 750 hours.

Luxeon LEDs seem to be an exception here. They were designed to overcome those issues. Read that application brief.
 
OOops, for some reason, the link does not work. You can get there by www.luxeon.com > products > documentation > and find AB07.
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by CCW:
Actually, they are MUCH worse than that. Especially white (and blue also) ones.<HR></BLOCKQUOTE>

I've seen white LED flashlights that dimmed noticeably by the FIRST battery change! They would go even faster if driven to the point of where the LEDs gets that pissed off sky blue color. I'm running an informal test now on a light that has an LED that does that. Two of its LEDs have a normal color, one has a blue-aqua tint, and the other has that really angry blue color during the first couple of hours of battery life. I want to see how fast that LED craps out.
 
Ok, I'm not an expert and I haven't tested any of these LED's for thousands of hours because my attension span can be measured in nano-seconds.

However, Nichia LED's are the primary LED in question here and I believe the specs as stated by Nichia. These specs will back up what CCW has stated above. I believe I have posted a link elsewhere in some thread but again, refer to:

presumption of life PDF file
 
Well, that seals it then! I'm not turning mine on anymore!

I keep expecting my purple nighlight LEDs to dim, but then haven't yet. They're on 12 hr/day.
 
Ok Darell...you made me realize something with that comment. I realize that Red LEDs are the most...stable but I've got an optical mouse and my computer is on for more than 12 hours a day usually and, wait lemme check...nope still red.
 
We wouldn't even dare going there...
shocked.gif
 
I'm a risk-taker. What can I say?
shocked.gif


I remember a while back, Craig was saying that he has seen most of his purple dim rapidly... I haven't yet seen it in mine.
 
I think that refers to a Pink LED they had for a short while. But, after Craig`s brief testing that made the por quality organic pink dye in the phosohor degrade badly, they discontinued them.


Purple LEDs sufer a similar failure mode to UV LEDs. The high-energy photons emitted by the die are enough to f*** up the junction and wire bond, until after a few hundred hours at most, they`ll quit. Essentially they self-destruct. Longer wavelength Purple will live longer than the shortest UV but almost certainly won`t give a hundred thousand hours. Few thousand maybe? I don`t know.


shocked.gif
confused.gif
 
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by Chris M.:
I think that refers to a Pink LED they had for a short while. But, after Craig`s brief testing that made the por quality organic pink dye in the phosohor degrade badly, they discontinued them<HR></BLOCKQUOTE>

ETG is offering a pink and a lavender phosphor LED that uses a stable inorganic phosphor that should keep the LED reasonably close to its starting color through the 10,000 hour mark. Brightness will probably fall off, but the color shouldn't fade like it did with those Roithner pink lEDs.

Purple LEDs won't simply stop working; it's a slow burn all the way to the bottom that starts the monent you turn one on. The high-energy photons initiate a cascade reaction that involves dissociation (breaking apart) of atomic bonds in the quantum GaN lattice. This process occurs at the subatomic level, and at present, nothing can be done to stop it.

In simpler terms, the rude little photons of UV light smack the GaN molecules in the puss so hard, it knocks their little blocks off and effectively kills them. Once the atoms have been dissociated (ripped apart) in this manner, they are unable to recombine into GaN within the existing bandgap. It's a cumulative process which is irreversible.

The end result is the LED very slowly dims, losing the shortwave (violet) output more quickly; so after x number of hours, you're left with a dim, reddish, brownish, or purplish white LED. The whitish color stems from the existing broadband emission produced by lower energy bandgaps that exist as a result of crystalline defects already in the P-N junction, and shows up even in brand new violet LEDs.

In this final "white dwarf" state, the LED might still be bright enough to use as a pilot light on some instrument kept in a dark room, but not usable for much of anything else. Send it to the big dumpster in the sky and replace it.
 
This one must be an interesting subject. Maybe I'll add couple more Cs.

There's a junction, then phosphor for some (white for sure), and then encapsulation. Basically junction area will loose its efficiency to generate light gradually by damages from high energy photons. Shorter wave length light has a higher energy level, hence UV is more destructive than green, and blue is obviously ranks high too. Any severe mechanical or thermal abuse will destroy junction rapidly, if not instantly. Overdriving LED is essentially a thermal abuse. Phosphors have a variety of stability issues, some of them really bad. Encapsulation - typically a clear epoxy resin - tends to loose its transparency when exposed to high energy photons. Just like what UV does to your everyday plastics, turn them yellow and brittle (and causes cancer). Heat will accelerate the damage by UV. According to one insider chemical expert at big LED firm, current generation of epoxy resin used for LEDs are affected by blue lights and up. Green, he said, seems to be O.K., which I don't exactly understand what O.K. is.

In case of Nichia's white (and blue), main reason for lumen degradation is the encapsulation. They use basically the same epoxy as everybody else. They claim that their phosphor is very stable, even though they did not supply any data. However, I have a lots of reasons to believe that claim.

Some new LEDs are using non-epoxy encapsulation. Luxeon uses silicon, and it's not supposed to degrade like epoxy.

I can easily imagine some poor quality phosphor does some weird things, but how will you know who's using what and how it is?

Sounds like a LED horror story, but after all, it's not that bad. For me, I still love LEDs, and more so as each month goes by.

I have to admit that the whole story is somewhat oversimplified. But I am not writing a book here. Hope I didn't bore you guys too much.
 
I'm far from bored here! Never bored here, as a matter of fact - its like attending a fantastic LED seminar plus luminary board discussion all the time.

But here's another factor that might affect LED degradation to discuss - heat (from here on I refer mainly to Luxeons).

The Lumileds datasheet mentions (and we all probably know) that heat will dim an LED as the temp rises by decreasing its efficiency. Lumileds also goes on to say that this loss of luminous output is reversible. Some papers written on the use of LEDs in traffic lights have suggested, however, that repeated/constant high operating temperatures (such as being in the heat of the sun daily) can irreversibly degrade performance of LEDs.

Are they talking about the same thing?

In this case, will Luxeons be affected (those traffic lights used the normal 5mm packages I presume)?
 
you know, guys, I was thinking about this and wondering if it would be better to construct a "longer lasting" led flashlight out of a combination of green, red, yellow, etc. leds....you know, separate the colors. (since the uv in the white yellows the epoxy, etc.)

What do you think?
 
CCW

This test you were involved in doesn't reflect the actual environment the led is used in (IMO). 60 degress C is very hot and the led is used intermittently in real world scenario. Would'nt the led have a much longer run time in actual use?
smile.gif
 
Mike,

Yes and no. The test I mentioned was acually related to an application I was working on (not a consumer product). Most products are not likely to be used in such environment, and not continuously either.

But, I am still concerned on overdriving issue. With some LEDs driven 4-5 times of their rating, the result will most likely be worse than 20mA-60C combination. At the end, LED does not know how hot outside world is, only thing matters is junction temperature. Overdriving LED will generate a lot more heat than the device is designed for at the junction unless some clever heat sink mechanism was attached.

If you use only short period at a time, that will help.

All LEDs are recommended to be driven at reduced rate at high temperature. Nichia white LEDs are rated to 85degrees C, but at much reduced current. 20mA at 60 was obviously an abuse.

I cannot comment on traffic signal situation much. All I know is that some of them use Luxeon LEDs and other forms, and 5mm is not the only choice. All of them should be designed per each LED spec for optimal life, that is to reduce driving current basing on expected LED temperature exposure (combination of environment and heat sink design). I have no idea what real life situations are.
 
Top