Nichia reaches 169 lm/W!

See the abstract here. A few of their numbers are very interesting. The first sample was 161 lm/W and 9.89 lm at 20 mA. This implies a forward voltage of 3.071V and an input power of 61.4 mW. They mentioned the output of the blue chip used in this LED was 42.2 mW, hence a WPE of 68.7%. This is the highest WPE I've ever heard for a blue LED.

The second sample is the one which achieved 169 lm/W. I found it interesting that they consider ~2.8V as nearly equal to the theoretical limit for the blue emitter used. Assuming similar phosphor conversion efficiencies as the 161 lm/W sample, this blue LED has achieved efficiencies of roughly 72%. In my opinion, most of the 28% which is not converted to light is probably simply not being extracted. In any case these lab results show that 75% to 80% emitter efficiency should be achievable in the near future, at least with blue. If this can be done with green and reds, then 300+ lm/W RGB-based white is possible.

The power LED results were also interesting. At 134 lm/W they break Cree's record. The blue emitter has an output of 651 mW, for an overall efficiency of just over 60% at 350 mA. That's not much worse than their 20 mA results. In any case it looks like the 150 lm/W barrier will soon be broken for power LEDs, and we're closing in on 200 lm/W for indicator-type LEDs.

Thanks for the heads up.
Hopefully they bring some more competition into the current power LED market. Nichia doesn't seem to have much of a presence in power LEDs as far as us flashaholics are concerned.
They also need to get to work on those blasted green LEDs. A high efficiency RGB in the market would really stir things up around here.

very interesting. thanks :thumbsup:

The second sample is the one which achieved 169 lm/W. I found it interesting that they consider ~2.8V as nearly equal to the theoretical limit for the blue emitter used.

Click to expand...

Assuming they are using 450nm blue:

1240 eV*nm / 450nm (blue) = 2.756V theoretical

would suggest that in the case of every successful photon conversion, only 3% of the voltage is dropped due to resistance and 97% goes into the conversion process.

A 530nm green LED at this sort of efficiency would be emitting 400 lumens/watt. I agree that a high efficiency RGB power LED would be fairly nice, or even better a RAGB (including amber) arranged into a quad-die package -- the addition of amber would imrpove color rendition if the LED were to be used as a light-source.

Or even if that doesn't happen, I still want to see more work done with multi-phosphor LEDs that are still based on blue LEDs, but produce light with lower color temeprature and higher CRI. Hopefully once theoretical efficiencies start to plateau with the single-phosphor approach more effort will be put into making white LEDs with a higher quality of light.


A flashlight with between 3500K and 4200K color temp using dual phosphor -- essentially neutral white with high color rendition -- would be awesome flashlight use.

Can someone explain to me how it can be that, within the same product line, the blue LED which is the basis of the white LED can be far less efficient than the white? For examle the Cree XR series cool white puts out 46lm at 350mA, whereas the blue only does 15lm. The phosphor layer traps and reflects some of the light coming from the blue dye below it, so if anything, it should be the other way around.

Another questions: Proponents of white RGB lighting complain that the main reason holding back progress of RGB lighting is the poor efficiency of green LEDs. Well, every data sheet I see puts the greens ahead of the reds and well ahead of the blues. So what am I missing here?

frenzee said:

Can someone explain to me how it can be that, within the same product line, the blue LED which is the basis of the white LED can be far less efficient than the white? For examle the Cree XR series cool white puts out 46lm at 350mA, whereas the blue only does 15lm. The phosphor layer traps and reflects some of the light coming from the blue dye below it, so if anything, it should be the other way around.

Click to expand...

The lumen unit is weighted based on the eye's sensitivity to different wavelengths (luminosity function). Our eyes are not very sensitive to blue light -- they are much more sensitive to the yellow/green light produced by the phosphor, so even though less total energy is being emitter from a phosphor LED, it appears brighter. The concept is even easier to understand with a UV based phosphor -- with no phosphor the UV emitter would be invisible regardless of efficiency.

Another questions: Proponents of white RGB lighting complain that the main reason holding back progress of RGB lighting is the poor efficiency of green LEDs. Well, every data sheet I see puts the greens ahead of the reds and well ahead of the blues. So what am I missing here?

Click to expand...

Green LEDs have poor power efficiency, but still produce more lumens as the eye has it's sharpest sensitivity to green light. There is more room for improvement with green than any other emitter.

wow , that is so strange, i wondered the exact same thing as frenzee.
the RGB items like say this
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=160176407230

red 25
Green 50
Blue 8

it looks in the specs like they dont have enough blue, or enough red for that matter, and when you light them with same current, they seem just a bit deficient in blue, but completly sufficent. send the same Wattage to each emitter, and they are high in red.

so they are Weighting how WE percieve the colors, that makes perfect sence. I though they were using a Color blind Meter. because when they rate some Aufully colored "white" spectrums , they rate them high.

i dont want to put down a lot of money for RGB led, because the NUMBERS in the specs look wackey, and all this talk about fixing stuff, and you say they can fix the Green most how confusing. although the "prolight rgb" did have the same offset, and the spectrum from it was awesome.

the atlas http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=160176407230 which is double dies
R 59
G 98
B 19

so that of course brings up one more question. the rgb doesnt have phosphors, so its not as efficent, but how efficent ARE these 2 RGB items shown? How many more Watts of these, making white from colors would i need, vrses a White with phosphors?

Maybe I've been spoiled by Cree, but I just don't get excited by paper launches anymore. :shrug:

It will be encouraging to see Nichia enter the power LED arena, though. The more competition the better!

Martini said:

Maybe I've been spoiled by Cree, but I just don't get excited by paper launches anymore. :shrug:

Click to expand...

Its at least encouraging to see what is going to be possible in the near future. Though it will probably be cree that does it.

saabluster said:

Its at least encouraging to see what is going to be possible in the near future. Though it will probably be cree that does it.

Click to expand...

That's really what I was taking out of it as well. We have yet to see Nichia commercially release even the 100 lm/W LEDs which they said we would have in April. At least with this latest lab test we know we can go beyond 150 lm/W.

jtr1962;

A couple of quotes from another reference to the same work were also very interesting:

(1) Low Power White - "The first is the white LED, which has a high general color rendering index (Ra) of 97 and CRI-No. 9 of 96. The CRI-No. 9 denotes the color reproduction in the red region. These values are higher than those of a tri-phosphor fluorescent lamp (Ra = 85 and CRI-No. 9 = 8)."

(2) High Power White Reference- "247 lm and 402 lm at 1 A and 2 A are obtained, respectively. at 2 A is equivalent to the total flux of a 30 W incandescent lamp."

Better color rendition is important aspect for the adaption of LED's to indoor illumination.

402 lumens is also a good achievement for a single LED.

Larry Cobb

Wow. A CRI of 97 is a major improvement. The incan guys might even like those.

Martini said:

Wow. A CRI of 97 is a major improvement. The incan guys might even like those.

Click to expand...

Yeah, right. :shakehead

Martini said:

Wow. A CRI of 97 is a major improvement. The incan guys might even like those.

Click to expand...

Seeing is believing! :candle: