King's LED Hat
Big Gav perked up my interest in revisiting some old research topics by posting on the commercialization of white LEDs. Given that incandescent lights consume a huge fraction of the USA's electrical energy output, clearly any kind of efficiency improvements here would help us soften our landing. From the reports it certainly looks like good news, so I became intrigued in finding out what they use for the underlying technology, especially the blue part of the spectrum (the long sought Holy Grail of the necessary red/green/blue trilogy of attaining white light).
Having spent the better part of the 1980's firmly entrenched in the land of III-V semiconductor research, but then leaving it to seek out a more varied career, I could get a good sense of how the research turned into commercial technology with a bit of digging. For technical background, this white paper by the whirling dervish experimentalist Hadish Morkoc1 of the University of Illinois (now at VCU), though dated, provides the basics.
In the world of semiconductor LEDs and lasers, the key to finding a match to a specific emission color involves searching the periodic table for combinations of elements that (1) form a crystalline lattice, (2) is semiconducting, (3) the semiconducting lattice forms a direct bad gap, and (4) that energy band gap has to match to the wavelength of a specific color (Ebg=hc/lambda). As this rules out just about all possible combinations of elements, researchers typically hammer on a subset of material compounds over and over. For red and green wavelengths, such crystalline compounds have long been known (Nick Holonyak another Illinois prof, basically invented the LED some 45 years ago, and should have probably won a Nobel prize). In the 80's, many thought that the material Zinc Selenide (
ZnSe) held out the most promise for the missing jigsaw puzzle piece of blue.
Several of my grad school colleagues at around that time completed their thesis and went to work for a SBIR-sized company that had this goal in mind. I haven't really kept in touch with them at all, but they evidentally did hammer on that material at will, as well as working on the intriguing but frustrating Gallium Nitride family (
GaN). (Another company out east worked on the similarly frustrating Silicon Carbide
SiC). These guys could themselves have made the blue light breakthrough, but it took a Japanese researcher, S. Nakamura, to really hit on the recipe for a successful blue
GaNLED. Check out the interview in that link.
In retrospect, looking at the way the research progressed, Nakamura had it in the bag. One of the problems of an experimental research career path is that grad students become expert technicians within a highly refined specialization. Because the specific tools they use have such a high cost, their marketable skills cause them to peg to a very narrow job description. For example, if the tool they used did not have the right capability, they by definition would never have a chance to succeed. Nakamura happened to know more about chemical-based epitaxial reactors, while my former colleagues stayed with the molecular beam variant of epitaxy. Part of the reason I left that type of research early on (before I became typecast), was the possibility of having to follow expensive equipment around the country, a slave to the machine so-to-speak.
But back to the
GaNitself, early on people noticed that though it could form a crystalline lattice and met all the requirements for blue spectrum (with mixtures of aluminum and indium), it had lots of defects and was tough to work with (the closest compound on the periodic table to
GaNis Boron Nitride, the stuff that high temperature crucibles are typically made of). Defects basically create extra energy states within the band gap and reduce the efficiency of carrier recombination leading to lowered light output. For the laser variant, the poor prognosis would also include material breakdown as the lasing would locally heat up the defected and dislocated regions causing the dislocations to enlarge (and so on in a classical runaway scenario). Kudos to Nakamura for keeping on this research in the face of great odds. As Morkoc and Nakamura both note, the defects strangely, and fortunately, don't seem to matter as much for Gallium Nitride as it would for other materials, e.g. Gallium Arsenide.
The final bit of irony: the toughness of this material, once thought to hinder any kind of easy processing (high temperature growth, etc) makes it very useful for things like automobile headlights and streetlights. Yes, streetlights, the big energy sucking sound that the USA has to deal with.
Welcome to the real world of research. A big, freakin' crap-shoot.
1Having met Morkoc at a conference, I was under the naive impression that to be a prodigious researcher also required you to be a spead freak.