Nobel Prize for Blue LEDs

The 2014 Nobel Prize in Physics has been awarded to Isamu Akasaki, Hiroshi Amano and Shuji Nakamura for the development of blue LED's. As always, this is kind of fascinating to watch evolve in the social media sphere, because as a genuinely unexpected big science story, journalists don't have pre-written articles based on an early copy of a embargoed paper. Which means absolutely everybody starts out using almost the exact words of the official Nobel press release, because that fills space while they frantically research the subject. Later in the day, you'll get some different framing, once writers get their head around it, and organizations like AIP run experts out for comment. A few organizations have the advantage of old stories, like this Physics World article by new laureate Nakamura or this old Scientific American piece about Nakamura's work.

Anyway, since I'm an optics guy by training, this is on the border of stuff I know about, so I'll offer a little off-the-cuff physics explanation, trying not to lean heavily on the Nobel Foundation's materials.

OK, so, this is a completely revolutionary new kind of light, or something? Well, no. The devices actually produce perfectly ordinary blue light, in the 400-500 nanometer range-- there's nothing exotic about the light. Or even the process by which it's produced, which is pretty much the same as for a red or green LED.

Let's pretend I don't know how those work. Because, you know, I don't. OK, well, it all goes back to guys like Planck and Einstein and Bohr. Quantum physics tells us that light is a stream of particles, each having a discrete energy that depends on the frequency. Short-wavelength photons like blue light have a fairly high energy, by the standards of atomic and condensed matter physics.

The emission of light generally involves an electron jumping between two allowed quantum states, with the energy difference producing a photon of the appropriate color.

Oh, so these guys just figured out how to collect a lot of atoms that emit blue light? Well, no, we already knew how to do that. It's part of what's going on in a fluorescent light bulb-- a vapor of atoms emits light at a few different wavelengths that combine to look mostly white to our eyes. That's not a very efficient process though, and requires a lot of vapor to make a useful amount of light. These guys used a solid-state system to produce lots of light from a tiny package. Very loosely speaking, if you're working with a solid rather than a gas, you have thousands of times the number of electrons packed into a given volume that can put out light.

So, they just made a solid lump of atoms that emit blue light? If only it were that easy. See, when you bring lots of atoms together, the behavior of the electrons changes pretty dramatically. A single electron doesn't have to be bound to a single atom any more, but can spread through the entire solid lump. Instead of the nice narrow states you get with atoms, you get broad "bands" of energy, with so many closely spaced energy levels that physicists stop trying to count them, and just treat it as a continuous blob. An electron could have any energy within that range.

But then, how do you get light out? Well, the bands are continuous, to a point. Within a band, an electron can easily move to a slightly higher or lower energy without emitting light, but there are "gaps" between the bands of allowed energies, a range of energies that are forbidden.

Why is that? Because of the wave nature of matter. Very loosely speaking, you can understand the concept by thinking about electrons passing through a regular crystal as a set of waves passing over a regular array of bumps. A little bit of the waves get reflected from each bump, which usually isn't a big deal, but for some particular wavelengths, the reflected waves and the incoming electron waves interfere with each other in a way that cancels out the waves. You can't have an electron moving through the material with that wavelength, which corresponds to a particular energy. Thus, a gap develops between otherwise continuous bands.

(This is very much a cartoon picture of what's going on, not a rigorous definition. I think it gets the right idea across, though, without going too much into the hairy math.)

So, an electron jumping across the gap between bands makes light? Exactly. If you have an electron in a high-energy band, and an open space in a low-energy band, the electron can drop down across the gap, and in the process emit some light. The wavelength of the emitted light is determined by the band gap.

And doesn't the gap have to do with the wavelengths the atoms would emit? Not as cleanly as you would like, no. The band gap depends on how the atoms in the material are organized into a crystal, which is a very complicated subject that I didn't do well on in graduate school. But it's not as simple as finding some atoms with a transition at a convenient wavelength and making them into a lump.

In order to get something that produces light of a given wavelength efficiently, you basically want to find a semiconductor material with the appropriate band gap, and make it into a diode. Which you do by butting two slightly different types of material up against each other (generally the same semiconducting material, with fraction-of-a-percent admixtures of other atoms "doping" the semiconductor). At the interface region, electron from one side drop across the band gap filling holes on the other side, and this process can be controlled very precisely.

So, that's how a red LED works? And these blue ones use some different method? No, the blue ones work the same way, just with different materials. Long-wavelength LEDs tend to use stuff like gallium arsenide doped with aluminum, while short-wavelength ones use gallium nitride doped with indium. The basic idea is the same, though: electrons come in one side, combine with holes at the boundary between materials, and emit light with a wavelength determined by the band gap.

That seems like a pretty small change, dude. How is that worth a Nobel? Well, because it's really frickin' hard to do. Gallium nitride is a more difficult material to work with than gallium arsenide (despite not being quite as toxic). Making LED's from gallium arsenide requires relatively small tweaks from the techniques used to make silicon computer chips, but making gallium nitride on the required scale and purity required a whole host of new techniques. These guys beat on the problem for a long time, developing entirely new methods of depositing thin layers of gallium nitride, controlling the doping to get the necessary properties to make a diode, and producing samples with few enough defects to have a reasonably long working life.

It's not a paradigm shift in terms of the basic physics, but it's a ton of hard work and new technological development, and richly deserves the Nobel.

OK, I guess. But why is this so much better than just getting a gas of blue-light-emitting atoms? Well, as I said above, the vastly higher density gets you an increase in the efficiency of the light emission. And they're just more compact-- to make a light bright enough to be useful out of an atomic vapor, you need a lot of it, which is why traditional fluorescent bulbs tend to be long tubes, and CFL bulbs are those funny spiraling coils. The light-emitting region of an LED is something like a hundred microns across, about the width of a hair. You can easily build those onto tiny chips that go into laser pointers, laptop screens, and cell phones.

Which is why blue LEDs went from impossible-to-make in the mid-1990s to being absolutely everywhere by the early 2000s-- the first cell phone I got in 2002 was full of the things, and you could just feel the joy of the engineers who got to stick those in.

You know, dude, you're falling into the same framing as the journalists you were disparaging at the start of this post. Don't you have anything other than light bulbs and display screens to offer? Well, as an AMO physics guy, the biggest benefit of this is the creation of blue/violet diode lasers (which are just a small step up from blue LED's). Blue light used to be a gigantic pain in the ass to generate in the lab, because you pretty much had to start with infrared light and then use a non-linear material to double the frequency up into the blue range. That's hard to do efficiently enough to get a lot of laser power.

These days, if I want blue light, say, to make a parametric down-conversion light source for producing photon pairs to use in quantum optics experiments (as my current thesis student is doing), I call up ThorLabs, and order a blue laser diode. I pop it into a diode mount, and boom! tens of milliwatts of blue light, a perfect source for downconversion experiments. It's a game-changer, bringing what used to be really difficult experiments well within the reach of undergraduate teaching labs.

Not a planet-saving innovation, I'll grant, but it's a major development in the technology AMO/ quantum optics people have available to investigate the universe. So I'd be all for this prize even without the whole energy-saving, portable-display-enabling thing.

Again, this isn't a revolutionary breakthrough in terms of the fundamental physics involved-- it's just a huge amount of hard work on the basic materials science and chemistry of semiconductor fabrication. But it's a dramatic breakthrough in terms of practical technology, and entirely deserving of a Nobel.