On Thu, 03 Oct 2019 15:08:43 +0100, Martin Brown wrote:

On 03/10/2019 14:45, Commander Kinsey wrote:
On Thu, 03 Oct 2019 14:37:34 +0100, whisky-dave
wrote:

On Thursday, 3 October 2019 14:29:30 UTC+1, Commander Kinsey wrote:
Why do LEDs generate heat? I want a technical answer not "because
they're inefficient".

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

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

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.

sure it's even resistance? It could be some photons are reabsorbed
before they escape the LED, generating heat.

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.

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.

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

And will we ever make them more efficient?

Probably.

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.

https://en.wikipedia.org/wiki/Light-...bre akthrough

That graph looks very promising.