Slashdot Mirror
Researchers Conquer "LED Droop"
sciencehabit writes "Tiny and efficient, light-emitting diodes (LEDs) are supposed to be the bright future of illumination. But they perform best at only low power, enough for a flashlight or the screen of your cellphone. If you increase the current enough for them to light a room like an old-fashioned incandescent bulb, their vaunted efficiency nosedives. It's called LED droop, and it's a real drag on the industry. Now, researchers have found a way to build more efficient LEDs that get more kick from the same amount of current—especially in the hard-to-manufacture green and blue parts of the spectrum."
The solution is called "LED Viagra"?
I guess that's why their new LED burns-up 26 watts but only created the equivalent of a 100 watt bulb. They are losing efficiency because the LEDs are being driven to high powers. (Lower power 25W or 40W bulbs only use 3 and 6 watts.)
My AC stalker: " I personally agree with your posts most of the time, but that won't keep me from modding you troll"
Why must a single LED provide all the light? Couldn't an array of, say, four LEDs, each equivalent to a 25W incandescent and using mirrors and/or lenses to even out the light distribution, get the same efficiency and substitute for a 100W bulb? Am I missing something obvious?
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
But they perform best at only low power, enough for a flashlight or the screen of your cellphone. If you increase the current enough for them to light a room like an old-fashioned incandescent bulb, their vaunted efficiency nosedives.
For a second there, I had images of LEDs hanging droopily over the edges of tables and tree branches.
When our name is on the back of your car, we're behind you all the way!
Have gnu, will travel.
Ahite LEDs of various color spectrums are totally different. They usually have a phosphorescent coating that creates the white light. The blue and green LEDs mentioned are single spectrum blue and green lights.
That really sounds annoying.
If you increase the current enough for them to light a room like an old-fashioned incandescent bulb, their vaunted efficiency nosedives.
Apparently this droop issue is only a problem for non-blue wavelengths. At least if my subwoofer, PC and external HDD are anything to go by...
My eyes hurt.
"Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
Typically a "green" produced by GaN is fairly easy to manufacture and fairly efficient, but it is physically a very *hard* material. In contrast, the "blue-green" produced by InGaN (an alloy of a little bit of InN and base of GaN) isn't as efficient as it tends to have lots crystal defects and these defect cause brittle-ness and results in some electron-hole recombinations to be non-radiative (generating heat and not band-gap light emissions).
Regardless of this manufacturability issue, many white LEDs use an InGaN band-gap devices and create the "warmer" parts of the spectrum using phosphors. This makes most of the output light more blue-ish, but only the phosphor re-radiated (stoke's shifted) part in the warmer part of the spectrum where you pay the efficiency cost. For "cool" devices, less of the output is down-converted, so you have less efficiency loss. For "warmer" devices, more of the light is down converted and you pay for more conversion efficiency loss. Some warm devices actually have multiple LEDs (say a red, green, and blue), but color stability is generally hard to maintain over time and temperature, so these devices are generally less efficient and more expensive.
In any case, the effect that was described is that the currently "cheap" way of growing GaN base crystals for LEDs results in a polar orientation which is bad for high-current operation as it tends to generate a back field. This is described in more detail in this other site:
Most of the commercial GaN devices are grown along the [0001] direction, so-called “polar” or “c-plane” structures. However, there is an internal electric field perpendicular to the active regions in the c-plane devices as the c-axis is polar. This will result in band bending and a poor overlap of electron and hole wave-functions (the Quantum confined Stark effect, or QCSE), which reduces the radiative recombination efficiency and affects the device performance. In order to avoid (or reduce the effects of) the QCSE, GaN can be grown in “non-polar”, or “semi-polar”, orientations, in which there is no, or much less, internal polarization fields along the growth direction. In theory, this should increase the efficiency of light emitting structures. The high density of structural defects (such as basal plane stacking faults and partial dislocations) in heteroepitaxially grown non-polar and semi-polar GaN results in low internal quantum efficiency and output power of the devices, as reported in the literature.
Of course the answer is to just grow low-defect GaN in a non-polar or semi-polar orientation, but that's currently hard to do. These UCSB researchers aren't the only group working on this problem, but they apparently have done some cooperation with people doing actual manufacturing (Mitsubishi Chemical).
My experiment with LED each of the first 3 bulbs I bought lasted between 500 and 2000 hours est. I bought a different brand, one died the first week. I now use CCFL simply because while they use slightly more energy, they are 10% of the price and also burn out way too soon, but not any sooner than LED and the light is better too..
Rod
There are other droops in life I'd rather see solved...
No. Red LEDs were the first LEDs made, and still are the easiest and cheapest to make light brightly.
Since the have laws banning incandescent bulbs because they are inneficient, when are they going to do something about the large incandescent light source 92 million miles away? Not only is it inefficient, it is the major cause of global warming.
(PK so there might be some issues of jurisdiction, but the owner of said light source (Oracle) is in this country...
Slew is somewhat correct. The nonpolar/semipolar substrates are currently very expensive and small (1 sq. inch at best, compared to 12sq. inches for sapphire substrates that all commercial LED's, except Soraa's, are currently grown on). It has been prohibitively expensive for any other academic institution to do any meaningful research on nonpolar/semipolar GaN LED's and lasers. That is unlikely to change anytime soon. There are industrial companies working on it though. A few Japanese companies and the startup company owned by the UCSB professors, Soraa. This article does give a lot of hype, the 20-2-1 LED's aren't quite the magic bullet that it implies. But nonpolar/semipolar LED's probably are the future once Soraa, Ammono, or Mitsubishi Chemical figures out how to grow large bulk GaN nonpolar/semipolar crystals by the ammonothermal technique. Soraa is releasing (or already released?) an LED based on nonpolar/semipolar technology this year, and lasers probably later this year or next.
There are some good c-plane LED's for sure. Nichia's best c-plane LED's probably have 95% PEAK internal quantum efficiency, but they still have droop problems and are expensive.
Pushing the diode at 200mA only resulted in an actual drop of roughly 8 points from that 52 percent. That's better than current blues used in my panels, which are top-line and only roughly 35% efficient.
But that isn't solving droop. Droop is the speed at which an LED driven at higher currents loses light output, which is a secondary byproduct of this. This mitigates the hell out of it, but doesn't solve the overall issue of light output loss over operative time.
But the higher efficiency is very welcomed. Applying this to create white diodes will let us smack roughly 250 lux/w when all is said and done, and with that, HID lighting has finally met its match, for good.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
That severely limits the max power you can stick in a standard bulb size space. Keep in mind that each led doesn't only need space to be in, but also a driver circuit(wich is bigger than the led itself, if done correctly).
Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
Seems that white LEDs are dangerous...
Slashdot's rate-of-post filter: Preventing you from posting too many great ideas at once.
I'm curious... how would you manufacture efficient greens with GaN only? Unless you dope it with something to widen the wells, it's only going to give you 330nm. My experience with LEDs is that all commercial blue / green (anything from UV to say 540 green) are InGaN. Not the indium clusters you mention - they are unreliable as hell, as you mentioned, but as a dopant in the gaN structure.
I'm curious... how would you manufacture efficient greens with GaN only? Unless you dope it with something to widen the wells, it's only going to give you 330nm. My experience with LEDs is that all commercial blue / green (anything from UV to say 540 green) are InGaN. Not the indium clusters you mention - they are unreliable as hell, as you mentioned, but as a dopant in the gaN structure.
Unfortuantly, this was typo. I meant to say typically a "violet" produced by GaN (as used in blu-ray lasers). I don't think I mentioned Indium clusters, but InGaN alloys (perhaps that wasn't clear about InN and GaN being the alloy, but I was trying to say it in a way that is easier for most folks to understand and I though people would understand the word alloy rather than the complicated substrate/base/layer dopant/deposition growth technique that is used in commercial InGaN production to limit crystal defects, I still believe alloy of InN and GaN on base of GaN to buffer against crystal defects is a generally correct explaination to the first order approximation, but if I'm wrong, I'm wrong). I suppose I should proof read more before I hit submit, but this is of course slashdot and I usually just stream-of-unconcious type when I have something to say (including this post)...
Slew is somewhat correct. The nonpolar/semipolar substrates are currently very expensive and small (1 sq. inch at best, compared to 12sq. inches for sapphire substrates that all commercial LED's, except Soraa's, are currently grown on). .
I thought commercial LEDs had moved on to SiC substrates for the higher thermal conductivity, even though they're more expensive than sapphire. Most commercial GaN RF circuits and transistors (including Cree) moved on from sapphire a while ago are fabricated on SiC substrates, and a couple companies are doing GaN circuits on silicon substrates. I thought that Cree for sure would be growing their LED's on SiC given that they're the world's main source of SiC substrates, and based on their webpage that seems to be the case.