Why do LEDs generate heat?

[Commander Kinsey]

Why do LEDs generate heat? I want a technical answer not "because they're inefficient". And will we ever make them more efficient?

[whisky-dave]

That is the technical answer just lioke why does a wire get hot when curremnt passes through it.

Probably.

[Commander Kinsey]

No, the technical answer would explain what part of it has resistance and if it can be overcome by using different materials. And are you sure it's even resistance? It could be some photons are reabsorbed before they escape the LED, generating heat.

[Commander Kinsey]

I got useful answers from Quora:

"LEDs are ever more and more efficient. In the last 40 years, tremendous strides have been made. They generate heat because they are conducting electricity through semiconductors. Unlike metals which have very little resistance to electric currents, semiconductors offer more resistance. Not as much as true nonmetals, but still more than metals. It is the resistance of the semiconductor layers, both N and P, and the resistance of the junction itself, that generate the heat."

"Every electronic device is less than 100 percent efficient. On a low level, it is due to the law of probability, or as the physicists call it, entropy. The odds of all those electrons conveying their energy into photons is very low. Some are always making random transitions, generating heat instead of light."

[whisky-dave]

Every conductor has resistance apart from perhaps a superconductor. yuo also have theb curent passing from one type of conductor to another. One of those conductors will convert the energy to light and heat. They won;t create much IR or UV because the material chosen was chosen because it emits the required wavelengh of light and not much eles.

No they have been trying for years and until someone can come up with a magic substance that has zero resistance that is.

No that would be IR 'heat' , and why would photons be re-absorbed anyway ?

[whisky-dave]

pretty much what I"ve just said. without Quora.

[Commander Kinsey]

The answers I pasted were far more detailed. I did say I wanted a technical answer.

[Martin Brown]

It is resistance in the sense that there is some frictional losses to the movement of current in the crystal lattice. Early LEDs you could bump up the quantum efficiency by stiffening it - immersing in LN2 worked a few times before thermal cycling killed it stone dead.

White LEDs rely on a yellow phosphor absorbing and re-emitting blue photons to make a perceived white light. Coloured LEDs typically have a forward voltage related to the energy of photon that they emit.

There is a hit for doing that that limits ultimate efficiency to something like 40% of power consumed out as useful light.

For comparison a tungsten light bulb is only about 2% efficient at making visible light.

Quantum efficiency and efficacy of LEDs has been improving with time. Cree have production models at 100Lm/W another factor of 2.5 improvement is theoretically possible and the odd sample has been made but the problems of making such a device in production quantities isn't cost effective at present. Cost per lumen and total flux graphs show how much improvement there has been since the first LED indicators in 1970.

[Commander Kinsey]

I assumed that the layers were so thin, resistance should be minimal. I guess transistors are the same and they get hot. Semiconductors probably have quite a high resistance?

Well they've certainly improved the efficiency over the years. The question is how far can they get without using them at absolute zero? Or perhaps even a completely different way of generating light?

Because they might not come out of the LED into the room, but hit another part of it internally.

[Commander Kinsey]

I didn't realise that, I thought white LEDs were designed to directly emit a handful of different visible light wavelengths to make white. Is there a reason this can't be done? An LED can make anything in the visible light spectrum, so surely a mixture of them would be more efficient than using phosphors? There could even be seperate LEDs within the housing, like growlamps which have visible, IR, and UV LEDs.

That graph looks very promising.

[whisky-dave]

Depending on yuor refernce point for thin and minimal.

Well yes compared to low or zero.

There's bio-luminescence and using a lattice structure like butterfly wings that could be used and have been used in high end modern displays. But bio-luminescence is diffiuclt to control at speed.

White LEDs do that and some other colours but I doubt much heat is generated.

[Brian Gaff]

One obviously has inefficiencies or the perpetual motion machine would be everywhere.

Brian

[Commander Kinsey]

Well they're measured in microns, not mm.

I was thinking "enough to make them increase in temperature by many C". My domestic LED lamps get just over body temperature. That's 20C of heating.

We wouldn't need speed for room lighting.

[Commander Kinsey]

[Martin Brown]

Tricolour can also be done but at the moment super efficient blue photon production and then down convert to yellow to get white is best.

You can get tricolour high power LEDs that can be mixed to generate any colour in consumer grade and theatre grade lighting systems too.

Magenta Growlights that only have blue and red LEDs are also now fairly common. No point in having any green since chlorophyll reflects it.

[Max Demian]

You can have red, green and blue LEDs to make up white, but it's less efficient and usually not worth doing unless you want to vary the hue.

[Commander Kinsey]

I see.

Yes, Phillips make them ("hue"?). I don't know how efficient they are.

Yes I think there's no green in mine, it appears purple to look at it. Probably R, B, UV, and IR. I wasn't aware plants could use IR, but it said that on the instructions. I guess there's IR in sunlight so plants maybe make use of it. I wonder why they don't use green? Maybe creating black chlorophyll would be impossible.

[Commander Kinsey]

I see. So for some reason blue LEDs are more efficient? (Even after wasting energy converting it with phosphors?) Do you know why?

[Martin Brown]

A guy got the Nobel prize for his contribution to making super efficient blue LEDs - mainly making the right substrate and getting it to work.

A lot of others had tried for decades and failed dismally.

There was a prehistoric dim blue LED based on silicon carbide used as indicators in a handful of expensive 1970's power amplifiers. They were very dim even by the dim standards of LEDs of that era, but blue.

The odd one pops up from time to time as a novelty.

[Commander Kinsey]

Perhaps one day there will be just as efficient LEDs of all colours and we can do away with the phosphor?