White, Organic LEDs Achieving New Efficiency Levels

From the journal Nature, Reineke et al. (2009) report that they have successfully developed an organic, 'white'-light LED with superior efficiency to that of fluorescent tubes. They achieved efficiencies around 90 lumens/Watt, compared to fluroscent tubes at 70 lumens/Watt. In fact, if one is willing to accept less intense lighting, they were pushing 120 lumens/Watt.

Organic LEDs are more efficient when made in thinner layers, but this limits the total amount of light they can produce per unit area. So while you could technically paper the entire ceiling with them, as a manufacturer you wouldn't want to because the substrate costs money and so does shipping.

This report is really fascinating in terms of all the optical design elements they are incorporating to prevent wastage of electrons and photons. Or, at least, it is to me. Basically in order to get a material to emit light you have to have a bunch of energetic electrons. You have them decay/lose energy. One possible way to lose energy is in the form of a photon (i.e. light), but you could also shed energy as heat or just spread it out to other electrons (particularly if there are defects in the material). Or you could successfully emit the photon but it will just get trapped and absorbed by the LED before it gets into the air. For organic LEDs, photons being reabsorbed is the biggest problem.

The question-mark with organic LEDs remains lifetime, particularly for the blue wavelength versions. The ones discussed in this article only last a couple of hours. This was still the case when I first learned about them five years ago. Basically, they don't react well to oxygen.

Organic LEDs are not necessarily any better than conventional, semiconductor LEDs. They are being pursued because they are potentially very cheap and have the novelty of flexibility.

Night Lighting
The strategy behind efficiency in lighting is not simply in producing the most photons per Watt of applied power, but matching the emission spectrum to that of the human eye. The unit for this is the lumen, which is the perceived brightness.

Figure 1: Sensitivity of the human eye as a function of wavelength.

The objectives for day-vision, known as phototopic, and night-vision, known as scotopic, are not quite the same. Night-vision is actually more efficient, and it peaks at a wavelength of 507 nm, which is squarely in the green part of the colour spectrum. For reference, 450 nm is the centre of the blue spectrum and 630 nm would be red.

Of interest here is the use of yellow Sodium-vapour street lamps. Low-pressure sodium lamps are highly efficient in turning electricity into light, on the order of 50-80 %. However, the human eye is not very good at detecting the yellow light (589 nm) when using the rods in the eye for night-vision. As can be seen from Figure 1, night-vision is actually piss-poor at using yellow light so when driving or walking under street-lamps, you are actually using your day vision.

Table 1: Eye efficiency as a function of wavelength.

Wavelength (nm)	Photopic Efficiency (lum/W)	Scotopic Efficiency (lum/W)
470 (blue)	62	1150
507 (green)	303	1700
555 (green)	683	683
589 (yellow)	517	111

Compared to yellow sodium street lamps, a green LED could be potentially 3-times less efficient and still beat it in lumens per Watt. Of course, this isn't sufficient for driving. Depth perception requires phototopic vision, since the cones are concentrated at the centre of vision whereas night-vision is predominately peripheral. For walking paths and other applications, green LED lighting could potentially beat the pants off of sodium lamps. The ideal case would probably be a 507 nm LED with a phosphor that emits light at a longer, redder wavelength. Then both scotopic and photopic vision could be covered. Or you could just build an array that emits two wavelengths of light. In this case, the need for a blue wavelength is not quite so necessary.