The graphic on the left shows the anatomy of a state-of-the-art white solid-state lamp; and the graphic on the right shows in the black curve the spectral power density of the light that it emits. The lamp is basically a blue LED coated with green and red phosphors. Some of the blue light leaks through the phosphors, so you see this relatively narrower peak here in the blue part of the spectral power density. But some of the blue light is absorbed by the two phosphors and is re-emitted as green or red light, so you see these two relatively wider peaks here in the green and red parts of the spectral power density. The combination of blue, green, and red gives a white light that renders fairly faithfully the colors of objects in the environment around us.

The efficiency of this state-of-the-art SSL lamp is about 20%-25%, slightly better than that of a fluorescent lamp, but far from the 50%-100% efficiency that is, in principal, possible. Why is this? We can see why by looking at the lamp’s three associated component efficiencies.

Blue LED efficiency

The first component efficiency is that of the blue LED; it is ~43%. Late last year, Nichia reported a blue LED efficiency of 81%, so why isn’t this efficiency given at 81%? Well, Nichia’s 81% efficiency can only be achieved at a very low injection current of 25 mA into a mm2 chip. When driving the chip harder, a phenomenon called efficiency droop kicks in to decrease efficiency. To defray the chip’s cost, however, the manufacturer must drive the chip harder, at least at 700 mA, and ideally even harder. At these injection currents, efficiencies aren’t as high. So, a first SSL technology grand challenge is to eliminate blue LED efficiency droop to maintain 81% efficiency at high drive currents.

Phosphor and package efficiency

The second component efficiency is that of the phosphor and package; it is ~56%. Some of that is because the phosphors do not have perfect quantum efficiency. Quantum efficiencies are improving, eliminating the more fundamental loss- the so-called Stokes deficit (the quantum deficit associated with converting a blue photon into a red or green photon. Researchers are working to eliminate the Stokes deficit by eliminating phosphors entirely, and moving to semiconductor electroluminescence at all colors. The barrier is that, thus far, InGaN-based materials emit well in the blue, but not in the green and yellow, while InGaP-based materials emit well in the deep red but also not in the green and yellow. So, a second SSL technology grand challenge is to fill in this green-yellow gap in semiconductor electroluminescence.

Spectral efficiency

The third component efficiency is the match between the spectrum of white light and the human eye sensitivity. Because of the characteristics of the particular phosphors currently in use, this efficiency is ~85%. To see why, here on the right I’ve drawn in white the spectrum of a hypothetical 100%-efficient four-color RYGB source that gives the best combination of color rendering quality and match to the human eye response. You can see that the red phosphor emits farther into the deep red than the ideal red at a wavelength of 615nm or so. It’s such a deep red that the human eye isn’t very sensitive to it, as you can see from the human eye response illustrated just below. So, a third SSL technology grand challenge is to narrow the red phosphor emission but to keep it centered at this ideal 615 nm wavelength.

Functional efficiency

Now, one component efficiency is not listed here: intelligently matching light to its use (i.e., if the light isn’t being used, then it is turned off, and if it is being used, its intensity, chromaticity, perhaps even its color rendering quality and luminous efficacy, are all matched to how it is being used in time and space). So, a fourth SSL technology grand challenge is to create light with this kind of enhanced functionality.

Overall efficiency

So, we see why we are only at 20% overall efficiency. Individually these three component efficiencies do not seem too bad: 43%, 56% and 85%. However, once you multiply them all out, you end up with 20% overall efficiency. So, the goal is to improve each one of these component efficiencies to be as close to 100% as possible, and perhaps to even add functionality that enables us to achieve an “effective” efficiency even greater than 100%.