Up to 400% increase in LED efficiency

Holes Placed Periodically in Substrate Boost LED Efficiency
LIGHTimes Staff
May 27, 2005...A research team from Kyoto University and Japan Science and Technology Agency has developed a design that significantly boosts the efficiency of light emitting diodes according to a NikkeiNet Interactive article. The article sites research appearing in the May 27 edition of the journal Science which discusses how tiny holes every 390-480nm in the crystalline structure of a thin film InGaAsP substrate of LEDs can boost light emitting 300 to 400%. The article points out that only about 20% of the light manages to escape a typical LED, but the rest remains trapped within the substrate and dissipates. The periodic holes open up escape routes for the trapped light. The researchers have theorized that such structure might help make LEDs that are close to 100% efficient.

Light emitting efficiency boosts with hole filled GaAs substrates were reported by researchers at UCLA in 1993.

I think about a possible 400% increase in the efficency of, say, a surefire L2 or U2 and I'm off on a fantasy of having a 100+ lumen light that runs for 2 months straight on one or two CR123 batts. <sits down with a dreamy look on face>

This is great news! It appears to be a relatively "free" increase in efficiency. I wonder how quickly manufacturers will be able to develop a process for this and ramp up production.

As an added bonus, flashlights may get 0.000001 grams lighter. /ubbthreads/images/graemlins/wink.gif

Those guys should work harder still ... I am growing older and need more light /ubbthreads/images/graemlins/grin.gif /ubbthreads/images/graemlins/grin.gif ... bring'em on guys, your 100% efficient LEDs ... /ubbthreads/images/graemlins/ooo.gif
bernie

That'd be sweet, then I might spend $100+ on an LED flashlight knowing that it wouldn't likely become obsolete in 12 months.

So much for my prediction that LED efficiency would top out in the 60% to 80% range maybe sometime within the next decade or two. Still, I love to be proven wrong like this. /ubbthreads/images/graemlins/smile.gif Near 100% efficiency-it just doesn't get any better than that.

Gimme a 500 lumen Surefire U2 LED cannon and then my wallet will breathe so much easier knowing that I no longer need anything as old fashioned as the Surefire M6 /ubbthreads/images/graemlins/nana.gif /ubbthreads/images/graemlins/grin.gif

Plus it's easier to pocket carry than an M6..

thanks NewBie, I always enjoy your postings. I think we are seeing the light at the end of the tunnel. Just waiting for a modest 40 lumen keychain light

a single CR123 light with 100 lm output... yummy

Hey,
We don't have to wait for the manufacturers! We can do simple mods ourself! Small 20 nm diameter bit and a tiny CNC mill with 5 nm repeatability and we're set! Pop the lens and silicone off the Luxeon, add holes and reassemble! Piece of cake! /ubbthreads/images/graemlins/icon15.gif

If this was reported in 1993, either some folks have been asleep at the wheel or the swiss cheese is not as trivial as we might hope.

Hey I've got a White Lux1 with the cap off! If someone has a 20nm bit, send it on over!
</sarcasm>

Anyway, I don't think we're going to be seeing LEDs with this soon. The numbers are pretty crazy.

Lets say you had a magic CNC mill, with a magic 20nm dill bit and you wanted to do this to a Luxeon (I, or III) die you would need to make 5E10 20nm holes in the 1000^2 micron luxeon die.

I bet someone (with alot of extra time) could hack up a dvd writer to do something like this. Just as long as you know how long to keep the laser on for to make the correct depth hole and assuming you have full control over the laser head AND add an extra axis of movement.
Neat idea, micro CNC laser mill /ubbthreads/images/graemlins/naughty.gif

you can Chemically "itch" the holes.. similar to the electrical lines that are laid in Intel's Processors. What they do is they simply "print" the itching thing on the Silicon Die. This Printing has a very high resolution... that makes the transistors smaller and smaller on the Processor Chip.

Same Techineque can be used to drill these holes on a Mass Produced GaAs or the GAPII Wafers.

For White LED's we will still require More dia Drills, because Blue Die + Yello Phosper kinda structure...

the Blue Die is Transperent (as far as i know) so posibility of "photons" getting trapped inside is less.. i guess....

May i get one of them.

This would need to be done with something like x-ray lithography. Likely to add a couple of steps in the manufacturing process, I would think.
While this has been known since '93, I am strongly inclined to think that the reason it hasn't been implemented yet, is the difficulty in manufacturing. Lithography and manufacturing technology and efficiency has greatly increased since '93. I think it's feasible now to make these.

We're doing 90nm feature etching for mass-production CPUs TODAY. 350nm (0.35u) CPUs were being made in the mid-late 90's. It'd be dead simple to etch 400-450nm features on an LED die. Especially something as simple as a smattering of holes.

The press release said "tiny holes every 390-480nm". If the holes are tiny in comparison to 390-480nm that would imply to me at least an order of magnitude smaller than the spacing, i.e. 39-48nm. If it means two orders of magnitude we are looking at 4-5nm holes. Can this be done with present technology?

As the crystal is being grown, just give it gas! The bubbles will become holes after it's sliced! I think a nano bean burrito would be the ticket! /ubbthreads/images/graemlins/thumbsup.gif

Yeah, I guessed that 20 nm might be too big of drill but what do I know?!? (I think the answer to that is obvious! /ubbthreads/images/graemlins/blush.gif )

Just a mathematical point about a 400 percent increase. A 100 percent increase would be 2x as efficient. Thus a 400 percent increase means 5x as efficient. IIRC, current Luxeons are like 10 percent efficient (???) is that right? Or maybe it's higher. But say it's 10 percent. This 400 percent increase would then mean an LED that was 50 percent efficient at converting electrical energy to visible light.

So this goes towards what someone said above about 60-80 percent within the next decade. That still could be right, although I suspect we'll see that sooner.

In any case, what wavelength does a "InGaAsP" LED produce? Because for a direct applicability to a Luxeon style white LED, it would need to be a blue-wavelength producing substrate. Right? Of course, it would seem that this same principle would apply equally to all LED's. But who knows?

I for one am looking forward to the nano beam burrito McGizmo LED (Trade-Mark protected). /ubbthreads/images/graemlins/nana.gif /ubbthreads/images/graemlins/thumbsup.gif

I can picture it now, and 500+ lumen KL4 head on E1e body driven by a R123. Hell, put one in the RAW body and drive with a RCR2. Damn, if this happens soon I'm gonna ditch my incadescent hobby soon...

i'm agree with JS... 100% efficient leds are near impossible... this 400% more at what is refered? 1 year ago a luxeon have 25lm/W... 5 times more is 125lm/W, ok this is a super-excellent result, but far from 100% efficiency.

and when they become available on the market? is a year that we talk about that luxeon IV with 60lm/W of efficiency but is still in our dreams...